forked from fastai/fastai
-
Notifications
You must be signed in to change notification settings - Fork 0
/
callback.html
1262 lines (852 loc) · 109 KB
/
callback.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
---
title: callback
keywords: fastai
sidebar: home_sidebar
summary: "Implementation of the callback system"
---
<!--
#################################################
### THIS FILE WAS AUTOGENERATED! DO NOT EDIT! ###
#################################################
# file to edit: docs_src/callback.ipynb
# instructions: https://docs.fast.ai/gen_doc_main.html
-->
<div class="container" id="notebook-container">
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<h2 id="Classes-for-callback-implementors">Classes for callback implementors<a class="anchor-link" href="#Classes-for-callback-implementors">¶</a></h2>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>fastai provides a powerful <em>callback</em> system, which is documented on the <a href="/callbacks.html#callbacks"><code>callbacks</code></a> page; look on that page if you're just looking for how to use existing callbacks. If you want to create your own, you'll need to use the classes discussed below.</p>
<p>A key motivation for the callback system is that additional functionality can be entirely implemented in a single callback, so that it's easily read. By using this trick, we will have different methods categorized in different callbacks where we will find clearly stated all the interventions the method makes in training. For instance in the <a href="/callbacks.lr_finder.html#LRFinder"><code>LRFinder</code></a> callback, on top of running the fit function with exponentially growing LRs, it needs to handle some preparation and clean-up, and all this code can be in the same callback so we know exactly what it is doing and where to look if we need to change something.</p>
<p>In addition, it allows our <a href="/basic_train.html#fit"><code>fit</code></a> function to be very clean and simple, yet still easily extended. So far in implementing a number of recent papers, we haven't yet come across any situation where we had to modify our training loop source code - we've been able to use callbacks every time.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h2 id="Callback" class="doc_header"><code>class</code> <code>Callback</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L159" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-pytest" style="float:right; padding-right:10px">[test]</a></h2><blockquote><p><code>Callback</code>()</p>
</blockquote>
<div class="collapse" id="Callback-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>Tests found for <code>Callback</code>:</p><ul><li><code>pytest -sv tests/test_callback.py::test_callbacks_fit</code> <a href="https://github.com/fastai/fastai/blob/master/tests/test_callback.py#L64" class="source_link" style="float:right">[source]</a></li><li><code>pytest -sv tests/test_callback.py::test_callbacks_learner</code> <a href="https://github.com/fastai/fastai/blob/master/tests/test_callback.py#L33" class="source_link" style="float:right">[source]</a></li></ul><p>To run tests please refer to this <a href="/dev/test.html#quick-guide">guide</a>.</p></div></div><p>Base class for callbacks that want to record values, dynamically change learner params, etc.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>To create a new type of callback, you'll need to inherit from this class, and implement one or more methods as required for your purposes. Perhaps the easiest way to get started is to look at the source code for some of the pre-defined fastai callbacks. You might be surprised at how simple they are! For instance, here is the <strong>entire</strong> source code for <a href="/train.html#GradientClipping"><code>GradientClipping</code></a>:</p>
<div class="highlight"><pre><span></span><span class="nd">@dataclass</span>
<span class="k">class</span> <span class="nc">GradientClipping</span><span class="p">(</span><span class="n">LearnerCallback</span><span class="p">):</span>
<span class="n">clip</span><span class="p">:</span><span class="nb">float</span>
<span class="k">def</span> <span class="nf">on_backward_end</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">clip</span><span class="p">:</span>
<span class="n">nn</span><span class="o">.</span><span class="n">utils</span><span class="o">.</span><span class="n">clip_grad_norm_</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">learn</span><span class="o">.</span><span class="n">model</span><span class="o">.</span><span class="n">parameters</span><span class="p">(),</span> <span class="bp">self</span><span class="o">.</span><span class="n">clip</span><span class="p">)</span>
</pre></div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>You generally want your custom callback constructor to take a <a href="/basic_train.html#Learner"><code>Learner</code></a> parameter, e.g.:</p>
<div class="highlight"><pre><span></span><span class="nd">@dataclass</span>
<span class="k">class</span> <span class="nc">MyCallback</span><span class="p">(</span><span class="n">Callback</span><span class="p">):</span>
<span class="n">learn</span><span class="p">:</span><span class="n">Learner</span>
</pre></div>
<p>Note that this allows the callback user to just pass your callback name to <code>callback_fns</code> when constructing their <a href="/basic_train.html#Learner"><code>Learner</code></a>, since that always passes <code>self</code> when constructing callbacks from <code>callback_fns</code>. In addition, by passing the learner, this callback will have access to everything: e.g all the inputs/outputs as they are calculated, the losses, and also the data loaders, the optimizer, etc. At any time:</p>
<ul>
<li>Changing self.learn.data.train_dl or self.data.valid_dl will change them inside the fit function (we just need to pass the <a href="/basic_data.html#DataBunch"><code>DataBunch</code></a> object to the fit function and not data.train_dl/data.valid_dl)</li>
<li>Changing self.learn.opt.opt (We have an <a href="/callback.html#OptimWrapper"><code>OptimWrapper</code></a> on top of the actual optimizer) will change it inside the fit function.</li>
<li>Changing self.learn.data or self.learn.opt directly WILL NOT change the data or the optimizer inside the fit function.</li>
</ul>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>In any of the callbacks you can unpack in the kwargs:</p>
<ul>
<li><code>n_epochs</code>, contains the number of epochs the training will take in total</li>
<li><code>epoch</code>, contains the number of the current</li>
<li><code>iteration</code>, contains the number of iterations done since the beginning of training</li>
<li><code>num_batch</code>, contains the number of the batch we're at in the dataloader</li>
<li><code>last_input</code>, contains the last input that got through the model (eventually updated by a callback)</li>
<li><code>last_target</code>, contains the last target that got through the model (eventually updated by a callback)</li>
<li><code>last_output</code>, contains the last output spitted by the model (eventually updated by a callback)</li>
<li><code>last_loss</code>, contains the last loss computed (eventually updated by a callback)</li>
<li><code>smooth_loss</code>, contains the smoothed version of the loss</li>
<li><code>last_metrics</code>, contains the last validation loss and metrics computed</li>
<li><code>pbar</code>, the progress bar</li>
<li><a href="/train.html#train"><code>train</code></a>, flag to know if we're in training mode or not</li>
<li><code>stop_training</code>, that will stop the training at the end of the current epoch if True</li>
<li><code>stop_epoch</code>, that will break the current epoch loop</li>
<li><code>skip_step</code>, that will skip the next optimizer step</li>
<li><code>skip_zero</code>, that will skip the next zero grad</li>
</ul>
<p>When returning a dictionary with those key names, the state of the <a href="/callback.html#CallbackHandler"><code>CallbackHandler</code></a> will be updated with any of those changes, so in any <a href="/callback.html#Callback"><code>Callback</code></a>, you can change those values.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<h3 id="Methods-your-subclass-can-implement">Methods your subclass can implement<a class="anchor-link" href="#Methods-your-subclass-can-implement">¶</a></h3>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>All of these methods are optional; your subclass can handle as many or as few as you require.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_train_begin" class="doc_header"><code>on_train_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L162" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_train_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_train_begin</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_train_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_train_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_train_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>To initialize constants in the callback.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here we can initiliaze anything we need.
The optimizer has now been initialized. We can change any hyper-parameters by typing, for instance:</p>
<pre><code>self.opt.lr = new_lr
self.opt.mom = new_mom
self.opt.wd = new_wd
self.opt.beta = new_beta</code></pre>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_epoch_begin" class="doc_header"><code>on_epoch_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L165" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_epoch_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_epoch_begin</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_epoch_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_epoch_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_epoch_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>At the beginning of each epoch.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>This is not technically required since we have <code>on_train_begin</code> for epoch 0 and <code>on_epoch_end</code> for all the other epochs,
yet it makes writing code that needs to be done at the beginning of every epoch easy and more readable.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_batch_begin" class="doc_header"><code>on_batch_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L168" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_batch_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_batch_begin</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_batch_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_batch_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_batch_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Set HP before the output and loss are computed.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the perfect place to prepare everything before the model is called.
Example: change the values of the hyperparameters (if we don't do it on_batch_end instead)</p>
<p>At the end of that event <code>xb</code>,<code>yb</code> will be set to <code>last_input</code>, <code>last_target</code> of the state of the <a href="/callback.html#CallbackHandler"><code>CallbackHandler</code></a>.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_loss_begin" class="doc_header"><code>on_loss_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L171" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_loss_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_loss_begin</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_loss_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_loss_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_loss_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called after forward pass but before loss has been computed.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed after the output has been computed but before the
loss computation.
Example: putting the output back in FP32 when training in mixed precision.</p>
<p>At the end of that event the output will be set to <code>last_output</code> of the state of the <a href="/callback.html#CallbackHandler"><code>CallbackHandler</code></a>.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_backward_begin" class="doc_header"><code>on_backward_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L174" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_backward_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_backward_begin</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_backward_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_backward_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_backward_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called after the forward pass and the loss has been computed, but before backprop.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed after the loss has been computed but before the gradient computation.
Example: <code>reg_fn</code> in RNNs.</p>
<p>At the end of that event the output will be set to <code>last_loss</code> of the state of the <a href="/callback.html#CallbackHandler"><code>CallbackHandler</code></a>.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_backward_end" class="doc_header"><code>on_backward_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L177" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_backward_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_backward_end</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_backward_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_backward_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_backward_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called after backprop but before optimizer step. Useful for true weight decay in AdamW.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed after the gradients have been computed but
before the optimizer is called.</p>
<p>If <code>skip_step</code> is <code>True</code> at the end of this event, the optimizer step is skipped.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_step_end" class="doc_header"><code>on_step_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L180" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_step_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_step_end</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_step_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_step_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_step_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called after the step of the optimizer but before the gradients are zeroed.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed after the optimizer step but before the gradients
are zeroed.</p>
<p>If <code>skip_zero</code> is <code>True</code> at the end of this event, the gradients are not zeroed.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_batch_end" class="doc_header"><code>on_batch_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L183" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_batch_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_batch_end</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_batch_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_batch_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_batch_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called at the end of the batch.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed after a batch is fully done.
Example: change the values of the hyperparameters (if we don't do it on_batch_begin instead)</p>
<p>If <code>end_epoch</code> is <code>True</code> at the end of this event, the current epoch is interrupted (example: lr_finder stops the training when the loss explodes).</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_epoch_end" class="doc_header"><code>on_epoch_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L186" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_epoch_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_epoch_end</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_epoch_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_epoch_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_epoch_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Called at the end of an epoch.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to run some code that needs to be executed at the end of an epoch.
Example: Save the model if we have a new best validation loss/metric.</p>
<p>If <code>end_training</code> is <code>True</code> at the end of this event, the training stops (example: early stopping).</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.on_train_end" class="doc_header"><code>on_train_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L189" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-on_train_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_train_end</code>(<strong>**<code>kwargs</code></strong>:<code>Any</code>)</p>
</blockquote>
<div class="collapse" id="Callback-on_train_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-on_train_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_train_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Useful for cleaning up things and saving files/models.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>Here is the place to tidy everything. It's always executed even if there was an error during the training loop,
and has an extra kwarg named exception to check if there was an exception or not.
Examples: save log_files, load best model found during training</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="Callback.get_state" class="doc_header"><code>get_state</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L196" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#Callback-get_state-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>get_state</code>(<strong><code>minimal</code></strong>:<code>bool</code>=<strong><em><code>True</code></em></strong>)</p>
</blockquote>
<div class="collapse" id="Callback-get_state-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#Callback-get_state-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>get_state</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Return the inner state of the <a href="/callback.html#Callback"><code>Callback</code></a>, <code>minimal</code> or not.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>This is used internally when trying to export a <a href="/basic_train.html#Learner"><code>Learner</code></a>. You won't need to subclass this function but you can add attribute names to the lists <code>exclude</code> or <code>not_min</code>of the <a href="/callback.html#Callback"><code>Callback</code></a> you are designing. Attributes in <code>exclude</code> are never saved, attributes in <code>not_min</code> only if <code>minimal=False</code>.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<h2 id="Annealing-functions">Annealing functions<a class="anchor-link" href="#Annealing-functions">¶</a></h2>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>The following functions provide different annealing schedules. You probably won't need to call them directly, but would instead use them as part of a callback. Here's what each one looks like:</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_png output_subarea ">
<img src="data:image/png;base64,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
"
>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="annealing_cos" class="doc_header"><code>annealing_cos</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L364" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#annealing_cos-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>annealing_cos</code>(<strong><code>start</code></strong>:<code>Number</code>, <strong><code>end</code></strong>:<code>Number</code>, <strong><code>pct</code></strong>:<code>float</code>) → <code>Number</code></p>
</blockquote>
<div class="collapse" id="annealing_cos-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#annealing_cos-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>annealing_cos</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Cosine anneal from <code>start</code> to <code>end</code> as pct goes from 0.0 to 1.0.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="annealing_exp" class="doc_header"><code>annealing_exp</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L361" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#annealing_exp-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>annealing_exp</code>(<strong><code>start</code></strong>:<code>Number</code>, <strong><code>end</code></strong>:<code>Number</code>, <strong><code>pct</code></strong>:<code>float</code>) → <code>Number</code></p>
</blockquote>
<div class="collapse" id="annealing_exp-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#annealing_exp-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>annealing_exp</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Exponentially anneal from <code>start</code> to <code>end</code> as pct goes from 0.0 to 1.0.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="annealing_linear" class="doc_header"><code>annealing_linear</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L358" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#annealing_linear-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>annealing_linear</code>(<strong><code>start</code></strong>:<code>Number</code>, <strong><code>end</code></strong>:<code>Number</code>, <strong><code>pct</code></strong>:<code>float</code>) → <code>Number</code></p>
</blockquote>
<div class="collapse" id="annealing_linear-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#annealing_linear-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>annealing_linear</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Linearly anneal from <code>start</code> to <code>end</code> as pct goes from 0.0 to 1.0.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="annealing_no" class="doc_header"><code>annealing_no</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L355" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#annealing_no-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>annealing_no</code>(<strong><code>start</code></strong>:<code>Number</code>, <strong><code>end</code></strong>:<code>Number</code>, <strong><code>pct</code></strong>:<code>float</code>) → <code>Number</code></p>
</blockquote>
<div class="collapse" id="annealing_no-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#annealing_no-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>annealing_no</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>No annealing, always return <code>start</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="annealing_poly" class="doc_header"><code>annealing_poly</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L372" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#annealing_poly-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>annealing_poly</code>(<strong><code>degree</code></strong>:<code>Number</code>) → <code>Number</code></p>
</blockquote>
<div class="collapse" id="annealing_poly-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#annealing_poly-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>annealing_poly</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Anneal polynomically from <code>start</code> to <code>end</code> as pct goes from 0.0 to 1.0.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h2 id="CallbackHandler" class="doc_header"><code>class</code> <code>CallbackHandler</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L224" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-pytest" style="float:right; padding-right:10px">[test]</a></h2><blockquote><p><code>CallbackHandler</code>(<strong><code>callbacks</code></strong>:<code>Collection</code>[<a href="/callback.html#Callback"><code>Callback</code></a>]=<strong><em><code>None</code></em></strong>, <strong><code>metrics</code></strong>:<code>Collection</code>[<a href="/callback.html#Callback"><code>Callback</code></a>]=<strong><em><code>None</code></em></strong>, <strong><code>beta</code></strong>:<code>float</code>=<strong><em><code>0.98</code></em></strong>)</p>
</blockquote>
<div class="collapse" id="CallbackHandler-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>CallbackHandler</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Manage all of the registered <code>callbacks</code> and <code>metrics</code>, smoothing loss by momentum <code>beta</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>You probably won't need to use this class yourself. It's used by fastai to combine all the callbacks together and call any relevant callback functions for each training stage. The methods below simply call the equivalent method in each callback function in <a href="/callbacks.html#callbacks"><code>self.callbacks</code></a>.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_backward_begin" class="doc_header"><code>on_backward_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L288" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_backward_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_backward_begin</code>(<strong><code>loss</code></strong>:<code>Tensor</code>) → <code>Tuple</code>[<code>Any</code>, <code>Any</code>]</p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_backward_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_backward_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_backward_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle gradient calculation on <code>loss</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_backward_end" class="doc_header"><code>on_backward_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L295" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_backward_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_backward_end</code>() → <code>Any</code></p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_backward_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_backward_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_backward_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle end of gradient calculation.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_batch_begin" class="doc_header"><code>on_batch_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L275" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_batch_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_batch_begin</code>(<strong><code>xb</code></strong>:<code>Tensor</code>, <strong><code>yb</code></strong>:<code>Tensor</code>, <strong><code>train</code></strong>:<code>bool</code>=<strong><em><code>True</code></em></strong>) → <code>Tuple</code>[<code>Any</code>, <code>Any</code>]</p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_batch_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_batch_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_batch_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle new batch <code>xb</code>,<code>yb</code> in <code>train</code> or validation.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_batch_end" class="doc_header"><code>on_batch_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L305" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_batch_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_batch_end</code>(<strong><code>loss</code></strong>:<code>Tensor</code>) → <code>Any</code></p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_batch_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_batch_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_batch_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle end of processing one batch with <code>loss</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_epoch_begin" class="doc_header"><code>on_epoch_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L270" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_epoch_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_epoch_begin</code>()</p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_epoch_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_epoch_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_epoch_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle new epoch.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_epoch_end" class="doc_header"><code>on_epoch_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L314" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_epoch_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_epoch_end</code>(<strong><code>val_loss</code></strong>:<code>Tensor</code>) → <code>bool</code></p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_epoch_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_epoch_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_epoch_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Epoch is done, process <code>val_loss</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_loss_begin" class="doc_header"><code>on_loss_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L282" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_loss_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_loss_begin</code>(<strong><code>out</code></strong>:<code>Tensor</code>) → <code>Any</code></p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_loss_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_loss_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_loss_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle start of loss calculation with model output <code>out</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_step_end" class="doc_header"><code>on_step_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L300" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_step_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_step_end</code>() → <code>Any</code></p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_step_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_step_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_step_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle end of optimization step.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_train_begin" class="doc_header"><code>on_train_begin</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L260" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_train_begin-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_train_begin</code>(<strong><code>epochs</code></strong>:<code>int</code>, <strong><code>pbar</code></strong>:<code>PBar</code>, <strong><code>metrics</code></strong>:<code>MetricFuncList</code>)</p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_train_begin-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_train_begin-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_train_begin</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>About to start learning.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.on_train_end" class="doc_header"><code>on_train_end</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L321" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-on_train_end-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>on_train_end</code>(<strong><code>exception</code></strong>:<code>Union</code>[<code>bool</code>, <code>Exception</code>])</p>
</blockquote>
<div class="collapse" id="CallbackHandler-on_train_end-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-on_train_end-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>on_train_end</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Handle end of training, <code>exception</code> is an <code>Exception</code> or False if no exceptions during training.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="CallbackHandler.set_dl" class="doc_header"><code>set_dl</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L253" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#CallbackHandler-set_dl-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>set_dl</code>(<strong><code>dl</code></strong>:<a href="https://pytorch.org/docs/stable/data.html#torch.utils.data.DataLoader"><code>DataLoader</code></a>)</p>
</blockquote>
<div class="collapse" id="CallbackHandler-set_dl-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#CallbackHandler-set_dl-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>set_dl</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Set the current <code>dl</code> used.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h2 id="OptimWrapper" class="doc_header"><code>class</code> <code>OptimWrapper</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L9" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-pytest" style="float:right; padding-right:10px">[test]</a></h2><blockquote><p><code>OptimWrapper</code>(<strong><code>opt</code></strong>:<a href="https://pytorch.org/docs/stable/optim.html#torch.optim.Optimizer"><code>Optimizer</code></a>, <strong><code>wd</code></strong>:<code>Floats</code>=<strong><em><code>0.0</code></em></strong>, <strong><code>true_wd</code></strong>:<code>bool</code>=<strong><em><code>False</code></em></strong>, <strong><code>bn_wd</code></strong>:<code>bool</code>=<strong><em><code>True</code></em></strong>)</p>
</blockquote>
<div class="collapse" id="OptimWrapper-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>OptimWrapper</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Basic wrapper around <code>opt</code> to simplify hyper-parameters changes.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing text_cell rendered"><div class="inner_cell">
<div class="text_cell_render border-box-sizing rendered_html">
<p>This is a convenience class that provides a consistent API for getting and setting optimizer hyperparameters. For instance, for <a href="https://pytorch.org/docs/stable/optim.html#torch.optim.Adam"><code>optim.Adam</code></a> the momentum parameter is actually <code>betas[0]</code>, whereas for <a href="https://pytorch.org/docs/stable/optim.html#torch.optim.SGD"><code>optim.SGD</code></a> it's simply <code>momentum</code>. As another example, the details of handling weight decay depend on whether you are using <code>true_wd</code> or the traditional L2 regularization approach.</p>
<p>This class also handles setting different WD and LR for each layer group, for discriminative layer training.</p>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="OptimWrapper.clear" class="doc_header"><code>clear</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L67" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-clear-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>clear</code>()</p>
</blockquote>
<div class="collapse" id="OptimWrapper-clear-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-clear-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>clear</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Reset the state of the inner optimizer.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="OptimWrapper.create" class="doc_header"><code>create</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L19" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-create-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>create</code>(<strong><code>opt_func</code></strong>:<code>Union</code>[<code>type</code>, <code>Callable</code>], <strong><code>lr</code></strong>:<code>Union</code>[<code>float</code>, <code>Tuple</code>, <code>List</code>[<code>T</code>]], <strong><code>layer_groups</code></strong>:<code>ModuleList</code>, <strong><code>wd</code></strong>:<code>Floats</code>=<strong><em><code>0.0</code></em></strong>, <strong><code>true_wd</code></strong>:<code>bool</code>=<strong><em><code>False</code></em></strong>, <strong><code>bn_wd</code></strong>:<code>bool</code>=<strong><em><code>True</code></em></strong>) → <a href="https://pytorch.org/docs/stable/optim.html#torch.optim.Optimizer"><code>Optimizer</code></a></p>
</blockquote>
<div class="collapse" id="OptimWrapper-create-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-create-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>create</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Create an <a href="https://pytorch.org/docs/stable/optim.html#torch.optim.Optimizer"><code>optim.Optimizer</code></a> from <code>opt_func</code> with <code>lr</code>. Set lr on <code>layer_groups</code>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="OptimWrapper.new" class="doc_header"><code>new</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L29" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-new-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>new</code>(<strong><code>layer_groups</code></strong>:<code>ModuleList</code>, <strong><code>split_no_wd</code></strong>:<code>bool</code>=<strong><em><code>True</code></em></strong>)</p>
</blockquote>
<div class="collapse" id="OptimWrapper-new-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-new-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>new</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Create a new <a href="/callback.html#OptimWrapper"><code>OptimWrapper</code></a> from <code>self</code> with another <code>layer_groups</code> but the same hyper-parameters.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="OptimWrapper.read_defaults" class="doc_header"><code>read_defaults</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L110" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-read_defaults-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>read_defaults</code>()</p>
</blockquote>
<div class="collapse" id="OptimWrapper-read_defaults-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-read_defaults-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>read_defaults</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Read the values inside the optimizer for the hyper-parameters.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">
<div class="output_markdown rendered_html output_subarea ">
<h4 id="OptimWrapper.read_val" class="doc_header"><code>read_val</code><a href="https://github.com/fastai/fastai/blob/master/fastai/callback.py#L140" class="source_link" style="float:right">[source]</a><a class="source_link" data-toggle="collapse" data-target="#OptimWrapper-read_val-pytest" style="float:right; padding-right:10px">[test]</a></h4><blockquote><p><code>read_val</code>(<strong><code>key</code></strong>:<code>str</code>) → <code>Union</code>[<code>List</code>[<code>float</code>], <code>Tuple</code>[<code>List</code>[<code>float</code>], <code>List</code>[<code>float</code>]]]</p>
</blockquote>
<div class="collapse" id="OptimWrapper-read_val-pytest"><div class="card card-body pytest_card"><a type="button" data-toggle="collapse" data-target="#OptimWrapper-read_val-pytest" class="close" aria-label="Close"><span aria-hidden="true">×</span></a><p>No tests found for <code>read_val</code>. To contribute a test please refer to <a href="/dev/test.html">this guide</a> and <a href="https://forums.fast.ai/t/improving-expanding-functional-tests/32929">this discussion</a>.</p></div></div><p>Read a hyperparameter <code>key</code> in the optimizer dictionary.</p>
</div>
</div>
</div>
</div>
</div>
<div class="cell border-box-sizing code_cell rendered">
<div class="output_wrapper">
<div class="output">
<div class="output_area">