-
Notifications
You must be signed in to change notification settings - Fork 362
/
Copy pathtorch_compile.html
883 lines (692 loc) · 51.1 KB
/
torch_compile.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
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="generator" content="Docutils 0.17.1: http://docutils.sourceforge.net/" />
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Torch-TensorRT torch.compile Backend — Torch-TensorRT v2.2.0.dev0+f617898 documentation</title>
<link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
<!-- <link rel="stylesheet" href="../_static/pygments.css" type="text/css" /> -->
<link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="../_static/sg_gallery.css" type="text/css" />
<link rel="stylesheet" href="../_static/sg_gallery-binder.css" type="text/css" />
<link rel="stylesheet" href="../_static/sg_gallery-dataframe.css" type="text/css" />
<link rel="stylesheet" href="../_static/sg_gallery-rendered-html.css" type="text/css" />
<link rel="stylesheet" href="../_static/collapsible-lists/css/tree_view.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.10.0-beta/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="../_static/css/custom.css" type="text/css" />
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="Torch-TensorRT Dynamo Backend" href="dynamo_export.html" />
<link rel="prev" title="Torch-TensorRT (FX Frontend) User Guide" href="getting_started_with_fx_path.html" />
<!-- Google Tag Manager -->
<script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src=
'https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);
})(window,document,'script','dataLayer','');</script>
<!-- End Google Tag Manager -->
<script src="../_static/js/modernizr.min.js"></script>
<!-- Preload the theme fonts -->
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-book.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-Medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium-italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-SemiBold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<!-- Preload the katex fonts -->
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.2/css/all.css" integrity="sha384-vSIIfh2YWi9wW0r9iZe7RJPrKwp6bG+s9QZMoITbCckVJqGCCRhc+ccxNcdpHuYu" crossorigin="anonymous">
</head>
<div class="container-fluid header-holder tutorials-header" id="header-holder">
<div class="container">
<div class="header-container">
<a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.org/get-started">Get Started</a>
</li>
<li>
<a href="https://pytorch.org/ecosystem">Ecosystem</a>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="resource-option with-down-arrow">
PyTorch Edge
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/edge">
<span class="dropdown-title">About PyTorch Edge</span>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/executorch">
<span class="dropdown-title">ExecuTorch</span>
</a>
</div>
</div>
</li>
<li>
<a href="https://pytorch.org/blog/">Blog</a>
</li>
<li>
<a href="https://pytorch.org/tutorials">Tutorials</a>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="resource-option with-down-orange-arrow">
Docs
</a>
<div class="resources-dropdown-menu">
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/docs/stable/index.html">
<span class="dropdown-title">PyTorch</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/audio/stable/index.html">
<span class="dropdown-title">torchaudio</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/text/stable/index.html">
<span class="dropdown-title">torchtext</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/vision/stable/index.html">
<span class="dropdown-title">torchvision</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torcharrow">
<span class="dropdown-title">torcharrow</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/data">
<span class="dropdown-title">TorchData</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torchrec">
<span class="dropdown-title">TorchRec</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/serve/">
<span class="dropdown-title">TorchServe</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torchx/">
<span class="dropdown-title">TorchX</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/xla">
<span class="dropdown-title">PyTorch on XLA Devices</span>
<p></p>
</a>
</div>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="resource-option with-down-arrow">
Resources
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/features">
<span class="dropdown-title">About</span>
<p>Learn about PyTorch’s features and capabilities</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/foundation">
<span class="dropdown-title">PyTorch Foundation</span>
<p>Learn about the PyTorch foundation</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/#community-module">
<span class="dropdown-title">Community</span>
<p>Join the PyTorch developer community to contribute, learn, and get your questions answered.</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/community-stories">
<span class="dropdown-title">Community Stories</span>
<p>Learn how our community solves real, everyday machine learning problems with PyTorch.</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/resources">
<span class="dropdown-title">Developer Resources</span>
<p>Find resources and get questions answered</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/events">
<span class="dropdown-title">Events</span>
<p>Find events, webinars, and podcasts</p>
</a>
<a class="nav-dropdown-item" href="https://discuss.pytorch.org/" target="_blank">
<span class="dropdown-title">Forums</span>
<p>A place to discuss PyTorch code, issues, install, research</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/hub">
<span class="dropdown-title">Models (Beta)</span>
<p>Discover, publish, and reuse pre-trained models</p>
</a>
</div>
</div>
</li>
<li>
<a href="https://github.com/pytorch/pytorch">GitHub</a>
</li>
</ul>
</div>
<a class="main-menu-open-button" href="#" data-behavior="open-mobile-menu"></a>
</div>
</div>
</div>
<body class="pytorch-body">
<div class="table-of-contents-link-wrapper">
<span>Table of Contents</span>
<a href="#" class="toggle-table-of-contents" data-behavior="toggle-table-of-contents"></a>
</div>
<nav data-toggle="wy-nav-shift" class="pytorch-left-menu" id="pytorch-left-menu">
<div class="pytorch-side-scroll">
<div class="pytorch-menu pytorch-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<div class="pytorch-left-menu-search">
<div class="version">
v2.2.0.dev0+f617898
</div>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../search.html" method="get">
<input type="text" name="q" placeholder="Search Docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
<p class="caption" role="heading"><span class="caption-text">Getting Started</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../getting_started/installation.html">Installation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../getting_started/getting_started_with_python_api.html">Using Torch-TensorRT in Python</a></li>
<li class="toctree-l1"><a class="reference internal" href="../getting_started/getting_started_with_cpp_api.html">Using Torch-TensorRT in C++</a></li>
<li class="toctree-l1"><a class="reference internal" href="../getting_started/getting_started_with_windows.html">Building Torch-TensorRT on Windows</a></li>
<li class="toctree-l1"><a class="reference internal" href="../getting_started/getting_started_with_windows.html#building-with-visual-studio-code">Building With Visual Studio Code</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">User Guide</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="creating_torchscript_module_in_python.html">Creating a TorchScript Module</a></li>
<li class="toctree-l1"><a class="reference internal" href="creating_torchscript_module_in_python.html#working-with-torchscript-in-python">Working with TorchScript in Python</a></li>
<li class="toctree-l1"><a class="reference internal" href="creating_torchscript_module_in_python.html#saving-torchscript-module-to-disk">Saving TorchScript Module to Disk</a></li>
<li class="toctree-l1"><a class="reference internal" href="getting_started_with_fx_path.html">Torch-TensorRT (FX Frontend) User Guide</a></li>
<li class="toctree-l1 current"><a class="current reference internal" href="#">Torch-TensorRT <cite>torch.compile</cite> Backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo_export.html">Torch-TensorRT Dynamo Backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="ptq.html">Post Training Quantization (PTQ)</a></li>
<li class="toctree-l1"><a class="reference internal" href="runtime.html">Deploying Torch-TensorRT Programs</a></li>
<li class="toctree-l1"><a class="reference internal" href="saving_models.html">Saving models compiled with Torch-TensorRT</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamic_shapes.html">Dynamic shapes with Torch-TensorRT</a></li>
<li class="toctree-l1"><a class="reference internal" href="use_from_pytorch.html">Using Torch-TensorRT Directly From PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="using_dla.html">DLA</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Tutorials</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../tutorials/serving_torch_tensorrt_with_triton.html">Serving a Torch-TensorRT model with Triton</a></li>
<li class="toctree-l1"><a class="reference internal" href="../tutorials/notebooks.html">Example notebooks</a></li>
<li class="toctree-l1"><a class="reference internal" href="../tutorials/_rendered_examples/dynamo/torch_compile_resnet_example.html">Compiling ResNet using the Torch-TensorRT <cite>torch.compile</cite> Backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="../tutorials/_rendered_examples/dynamo/torch_compile_transformers_example.html">Compiling a Transformer using torch.compile and TensorRT</a></li>
<li class="toctree-l1"><a class="reference internal" href="../tutorials/_rendered_examples/dynamo/torch_compile_advanced_usage.html">Torch Compile Advanced Usage</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Python API Documenation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../py_api/torch_tensorrt.html">torch_tensorrt</a></li>
<li class="toctree-l1"><a class="reference internal" href="../py_api/logging.html">torch_tensorrt.logging</a></li>
<li class="toctree-l1"><a class="reference internal" href="../py_api/ptq.html">torch_tensorrt.ptq</a></li>
<li class="toctree-l1"><a class="reference internal" href="../py_api/ts.html">torch_tensorrt.ts</a></li>
<li class="toctree-l1"><a class="reference internal" href="../py_api/fx.html">torch_tensorrt.fx</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">C++ API Documenation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_api/torch_tensort_cpp.html">Torch-TensorRT C++ API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_api/namespace_torch_tensorrt.html">Namespace torch_tensorrt</a></li>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_api/namespace_torch_tensorrt__logging.html">Namespace torch_tensorrt::logging</a></li>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_api/namespace_torch_tensorrt__torchscript.html">Namespace torch_tensorrt::torchscript</a></li>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_api/namespace_torch_tensorrt__ptq.html">Namespace torch_tensorrt::ptq</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">CLI Documenation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../cli/torchtrtc.html">torchtrtc</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Contributor Documentation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../contributors/system_overview.html">System Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../contributors/writing_converters.html">Writing Converters</a></li>
<li class="toctree-l1"><a class="reference internal" href="../contributors/writing_dynamo_aten_lowering_passes.html">Writing Dynamo ATen Lowering Passes</a></li>
<li class="toctree-l1"><a class="reference internal" href="../contributors/useful_links.html">Useful Links for Torch-TensorRT Development</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Indices</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../indices/supported_ops.html">Operators Supported</a></li>
</ul>
</div>
</div>
</nav>
<div class="pytorch-container">
<div class="pytorch-page-level-bar" id="pytorch-page-level-bar">
<div class="pytorch-breadcrumbs-wrapper">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="pytorch-breadcrumbs">
<li>
<a href="../index.html">
Docs
</a> >
</li>
<li>Torch-TensorRT <cite>torch.compile</cite> Backend</li>
<li class="pytorch-breadcrumbs-aside">
<a href="../_sources/user_guide/torch_compile.rst.txt" rel="nofollow"><img src="../_static/images/view-page-source-icon.svg"></a>
</li>
</ul>
</div>
</div>
<div class="pytorch-shortcuts-wrapper" id="pytorch-shortcuts-wrapper">
Shortcuts
</div>
</div>
<section data-toggle="wy-nav-shift" id="pytorch-content-wrap" class="pytorch-content-wrap">
<div class="pytorch-content-left">
<!-- Google Tag Manager (noscript) -->
<noscript><iframe src="https://www.googletagmanager.com/ns.html?id="
height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript>
<!-- End Google Tag Manager (noscript) -->
<div class="rst-content">
<div role="main" class="main-content" itemscope="itemscope" itemtype="http://schema.org/Article">
<article itemprop="articleBody" id="pytorch-article" class="pytorch-article">
<section id="torch-tensorrt-torch-compile-backend">
<span id="torch-compile"></span><h1>Torch-TensorRT <cite>torch.compile</cite> Backend<a class="headerlink" href="#torch-tensorrt-torch-compile-backend" title="Permalink to this headline">¶</a></h1>
<span class="target" id="module-torch_tensorrt.dynamo"></span><p>This guide presents the Torch-TensorRT <cite>torch.compile</cite> backend: a deep learning compiler which uses TensorRT to accelerate JIT-style workflows across a wide variety of models.</p>
<section id="key-features">
<h2>Key Features<a class="headerlink" href="#key-features" title="Permalink to this headline">¶</a></h2>
<p>The primary goal of the Torch-TensorRT <cite>torch.compile</cite> backend is to enable Just-In-Time compilation workflows by combining the simplicity of <cite>torch.compile</cite> API with the performance of TensorRT. Invoking the <cite>torch.compile</cite> backend is as simple as importing the <cite>torch_tensorrt</cite> package and specifying the backend:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch_tensorrt</span>
<span class="o">...</span>
<span class="n">optimized_model</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">compile</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">backend</span><span class="o">=</span><span class="s2">"torch_tensorrt"</span><span class="p">,</span> <span class="n">dynamic</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span>
</pre></div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Many additional customization options are available to the user. These will be discussed in further depth in this guide.</p>
</div>
<p>The backend can handle a variety of challenging model structures and offers a simple-to-use interface for effective acceleration of models. Additionally, it has many customization options to ensure the compilation process is fitting to the specific use case.</p>
</section>
<section id="customizeable-settings">
<h2>Customizeable Settings<a class="headerlink" href="#customizeable-settings" title="Permalink to this headline">¶</a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="torch_tensorrt.dynamo.CompilationSettings">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch_tensorrt.dynamo.</span></span><span class="sig-name descname"><span class="pre">CompilationSettings</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">precision:</span> <span class="pre">torch.dtype</span> <span class="pre">=</span> <span class="pre">torch.float32</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">debug:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">workspace_size:</span> <span class="pre">int</span> <span class="pre">=</span> <span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">min_block_size:</span> <span class="pre">int</span> <span class="pre">=</span> <span class="pre">5</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">torch_executed_ops:</span> <span class="pre">typing.Set[str]</span> <span class="pre">=</span> <span class="pre"><factory></span></span></em>, <em class="sig-param"><span class="n"><span class="pre">pass_through_build_failures:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_aux_streams:</span> <span class="pre">typing.Optional[int]</span> <span class="pre">=</span> <span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">version_compatible:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">optimization_level:</span> <span class="pre">typing.Optional[int]</span> <span class="pre">=</span> <span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_python_runtime:</span> <span class="pre">typing.Optional[bool]</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">truncate_long_and_double:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_fast_partitioner:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">enable_experimental_decompositions:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">device:</span> <span class="pre">torch_tensorrt._Device.Device</span> <span class="pre">=</span> <span class="pre"><factory></span></span></em>, <em class="sig-param"><span class="n"><span class="pre">require_full_compilation:</span> <span class="pre">bool</span> <span class="pre">=</span> <span class="pre">False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/torch_tensorrt/dynamo/_settings.html#CompilationSettings"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#torch_tensorrt.dynamo.CompilationSettings" title="Permalink to this definition">¶</a></dt>
<dd><p>Compilation settings for Torch-TensorRT Dynamo Paths</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>precision</strong> (<em>torch.dpython:type</em>) – Model Layer precision</p></li>
<li><p><strong>debug</strong> (<em>bool</em>) – Whether to print out verbose debugging information</p></li>
<li><p><strong>workspace_size</strong> (<em>python:int</em>) – Workspace TRT is allowed to use for the module (0 is default)</p></li>
<li><p><strong>min_block_size</strong> (<em>python:int</em>) – Minimum number of operators per TRT-Engine Block</p></li>
<li><p><strong>torch_executed_ops</strong> (<em>Sequence</em><em>[</em><em>str</em><em>]</em>) – Sequence of operations to run in Torch, regardless of converter coverage</p></li>
<li><p><strong>pass_through_build_failures</strong> (<em>bool</em>) – Whether to fail on TRT engine build errors (True) or not (False)</p></li>
<li><p><strong>max_aux_streams</strong> (<em>Optional</em><em>[</em><em>python:int</em><em>]</em>) – Maximum number of allowed auxiliary TRT streams for each engine</p></li>
<li><p><strong>version_compatible</strong> (<em>bool</em>) – Provide version forward-compatibility for engine plan files</p></li>
<li><p><strong>optimization_level</strong> (<em>Optional</em><em>[</em><em>python:int</em><em>]</em>) – Builder optimization 0-5, higher levels imply longer build time,
searching for more optimization options. TRT defaults to 3</p></li>
<li><p><strong>use_python_runtime</strong> (<em>Optional</em><em>[</em><em>bool</em><em>]</em>) – Whether to strictly use Python runtime or C++ runtime. To auto-select a runtime
based on C++ dependency presence (preferentially choosing C++ runtime if available), leave the
argument as None</p></li>
<li><p><strong>truncate_long_and_double</strong> (<em>bool</em>) – Whether to truncate int64/float64 TRT engine inputs or weights to int32/float32</p></li>
<li><p><strong>use_fast_partitioner</strong> (<em>bool</em>) – Whether to use the fast or global graph partitioning system</p></li>
<li><p><strong>enable_experimental_decompositions</strong> (<em>bool</em>) – Whether to enable all core aten decompositions
or only a selected subset of them</p></li>
<li><p><strong>device</strong> (<a class="reference internal" href="../py_api/torch_tensorrt.html#torch_tensorrt.Device" title="torch_tensorrt.Device"><em>Device</em></a>) – GPU to compile the model on</p></li>
<li><p><strong>require_full_compilation</strong> (<em>bool</em>) – Whether to require the graph is fully compiled in TensorRT.
Only applicable for <cite>ir=”dynamo”</cite>; has no effect for <cite>torch.compile</cite> path</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<section id="custom-setting-usage">
<h3>Custom Setting Usage<a class="headerlink" href="#custom-setting-usage" title="Permalink to this headline">¶</a></h3>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch_tensorrt</span>
<span class="o">...</span>
<span class="n">optimized_model</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">compile</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">backend</span><span class="o">=</span><span class="s2">"torch_tensorrt"</span><span class="p">,</span> <span class="n">dynamic</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span>
<span class="n">options</span><span class="o">=</span><span class="p">{</span><span class="s2">"truncate_long_and_double"</span><span class="p">:</span> <span class="kc">True</span><span class="p">,</span>
<span class="s2">"precision"</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">half</span><span class="p">,</span>
<span class="s2">"debug"</span><span class="p">:</span> <span class="kc">True</span><span class="p">,</span>
<span class="s2">"min_block_size"</span><span class="p">:</span> <span class="mi">2</span><span class="p">,</span>
<span class="s2">"torch_executed_ops"</span><span class="p">:</span> <span class="p">{</span><span class="s2">"torch.ops.aten.sub.Tensor"</span><span class="p">},</span>
<span class="s2">"optimization_level"</span><span class="p">:</span> <span class="mi">4</span><span class="p">,</span>
<span class="s2">"use_python_runtime"</span><span class="p">:</span> <span class="kc">False</span><span class="p">,})</span>
</pre></div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Quantization/INT8 support is slated for a future release; currently, we support FP16 and FP32 precision layers.</p>
</div>
</section>
</section>
<section id="compilation">
<h2>Compilation<a class="headerlink" href="#compilation" title="Permalink to this headline">¶</a></h2>
<p>Compilation is triggered by passing inputs to the model, as so:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch_tensorrt</span>
<span class="o">...</span>
<span class="c1"># Causes model compilation to occur</span>
<span class="n">first_outputs</span> <span class="o">=</span> <span class="n">optimized_model</span><span class="p">(</span><span class="o">*</span><span class="n">inputs</span><span class="p">)</span>
<span class="c1"># Subsequent inference runs with the same, or similar inputs will not cause recompilation</span>
<span class="c1"># For a full discussion of this, see "Recompilation Conditions" below</span>
<span class="n">second_outputs</span> <span class="o">=</span> <span class="n">optimized_model</span><span class="p">(</span><span class="o">*</span><span class="n">inputs</span><span class="p">)</span>
</pre></div>
</div>
</section>
<section id="after-compilation">
<h2>After Compilation<a class="headerlink" href="#after-compilation" title="Permalink to this headline">¶</a></h2>
<p>The compilation object can be used for inference within the Python session, and will recompile according to the recompilation conditions detailed below. In addition to general inference, the compilation process can be a helpful tool in determining model performance, current operator coverage, and feasibility of serialization. Each of these points will be covered in detail below.</p>
<section id="model-performance">
<h3>Model Performance<a class="headerlink" href="#model-performance" title="Permalink to this headline">¶</a></h3>
<p>The optimized model returned from <cite>torch.compile</cite> is useful for model benchmarking since it can automatically handle changes in the compilation context, or differing inputs that could require recompilation. When benchmarking inputs of varying distributions, batch sizes, or other criteria, this can save time.</p>
</section>
<section id="operator-coverage">
<h3>Operator Coverage<a class="headerlink" href="#operator-coverage" title="Permalink to this headline">¶</a></h3>
<p>Compilation is also a useful tool in determining operator coverage for a particular model. For instance, the following compilation command will display the operator coverage for each graph, but will not compile the model - effectively providing a “dryrun” mechanism:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch_tensorrt</span>
<span class="o">...</span>
<span class="n">optimized_model</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">compile</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">backend</span><span class="o">=</span><span class="s2">"torch_tensorrt"</span><span class="p">,</span> <span class="n">dynamic</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span>
<span class="n">options</span><span class="o">=</span><span class="p">{</span><span class="s2">"debug"</span><span class="p">:</span> <span class="kc">True</span><span class="p">,</span>
<span class="s2">"min_block_size"</span><span class="p">:</span> <span class="nb">float</span><span class="p">(</span><span class="s2">"inf"</span><span class="p">),})</span>
</pre></div>
</div>
<p>If key operators for your model are unsupported, see <a class="reference internal" href="../contributors/fx_converters.html#dynamo-conversion"><span class="std std-ref">Dynamo Converters</span></a> to contribute your own converters, or file an issue here: <a class="reference external" href="https://github.com/pytorch/TensorRT/issues">https://github.com/pytorch/TensorRT/issues</a>.</p>
</section>
<section id="feasibility-of-serialization">
<h3>Feasibility of Serialization<a class="headerlink" href="#feasibility-of-serialization" title="Permalink to this headline">¶</a></h3>
<p>Compilation can also be helpful in demonstrating graph breaks and the feasibility of serialization of a particular model. For instance, if a model has no graph breaks and compiles successfully with the Torch-TensorRT backend, then that model should be compileable and serializeable via the <cite>torch_tensorrt</cite> Dynamo IR, as discussed in <a class="reference internal" href="dynamic_shapes.html#dynamic-shapes"><span class="std std-ref">Dynamic shapes with Torch-TensorRT</span></a>. To determine the number of graph breaks in a model, the <cite>torch._dynamo.explain</cite> function is very useful:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch</span>
<span class="kn">import</span> <span class="nn">torch_tensorrt</span>
<span class="o">...</span>
<span class="n">explanation</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">_dynamo</span><span class="o">.</span><span class="n">explain</span><span class="p">(</span><span class="n">model</span><span class="p">)(</span><span class="o">*</span><span class="n">inputs</span><span class="p">)</span>
<span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"Graph breaks: </span><span class="si">{</span><span class="n">explanation</span><span class="o">.</span><span class="n">graph_break_count</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
<span class="n">optimized_model</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">compile</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">backend</span><span class="o">=</span><span class="s2">"torch_tensorrt"</span><span class="p">,</span> <span class="n">dynamic</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">options</span><span class="o">=</span><span class="p">{</span><span class="s2">"truncate_long_and_double"</span><span class="p">:</span> <span class="kc">True</span><span class="p">})</span>
</pre></div>
</div>
</section>
</section>
<section id="dynamic-shape-support">
<h2>Dynamic Shape Support<a class="headerlink" href="#dynamic-shape-support" title="Permalink to this headline">¶</a></h2>
<p>The Torch-TensorRT <cite>torch.compile</cite> backend will currently require recompilation for each new batch size encountered, and it is preferred to use the <cite>dynamic=False</cite> argument when compiling with this backend. Full dynamic shape support is planned for a future release.</p>
</section>
<section id="recompilation-conditions">
<h2>Recompilation Conditions<a class="headerlink" href="#recompilation-conditions" title="Permalink to this headline">¶</a></h2>
<p>Once the model has been compiled, subsequent inference inputs with the same shape and data type, which traverse the graph in the same way will not require recompilation. Furthermore, each new recompilation will be cached for the duration of the Python session. For instance, if inputs of batch size 4 and 8 are provided to the model, causing two recompilations, no further recompilation would be necessary for future inputs with those batch sizes during inference within the same session. Support for engine cache serialization is planned for a future release.</p>
<p>Recompilation is generally triggered by one of two events: encountering inputs of different sizes or inputs which traverse the model code differently. The latter scenario can occur when the model code includes conditional logic, complex loops, or data-dependent-shapes. <cite>torch.compile</cite> handles guarding in both of these scenario and determines when recompilation is necessary.</p>
</section>
</section>
</article>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="dynamo_export.html" class="btn btn-neutral float-right" title="Torch-TensorRT Dynamo Backend" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
<a href="getting_started_with_fx_path.html" class="btn btn-neutral" title="Torch-TensorRT (FX Frontend) User Guide" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
</div>
<hr>
<div role="contentinfo">
<p>
© Copyright 2022, NVIDIA Corporation.
</p>
</div>
<div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</div>
</footer>
</div>
</div>
<div class="pytorch-content-right" id="pytorch-content-right">
<div class="pytorch-right-menu" id="pytorch-right-menu">
<div class="pytorch-side-scroll" id="pytorch-side-scroll-right">
<ul>
<li><a class="reference internal" href="#">Torch-TensorRT <cite>torch.compile</cite> Backend</a><ul>
<li><a class="reference internal" href="#key-features">Key Features</a></li>
<li><a class="reference internal" href="#customizeable-settings">Customizeable Settings</a><ul>
<li><a class="reference internal" href="#custom-setting-usage">Custom Setting Usage</a></li>
</ul>
</li>
<li><a class="reference internal" href="#compilation">Compilation</a></li>
<li><a class="reference internal" href="#after-compilation">After Compilation</a><ul>
<li><a class="reference internal" href="#model-performance">Model Performance</a></li>
<li><a class="reference internal" href="#operator-coverage">Operator Coverage</a></li>
<li><a class="reference internal" href="#feasibility-of-serialization">Feasibility of Serialization</a></li>
</ul>
</li>
<li><a class="reference internal" href="#dynamic-shape-support">Dynamic Shape Support</a></li>
<li><a class="reference internal" href="#recompilation-conditions">Recompilation Conditions</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
</section>
</div>
<script type="text/javascript" id="documentation_options" data-url_root="../" src="../_static/documentation_options.js"></script>
<script data-url_root="../" id="documentation_options" src="../_static/documentation_options.js"></script>
<script src="../_static/jquery.js"></script>
<script src="../_static/underscore.js"></script>
<script src="../_static/doctools.js"></script>
<script src="../_static/collapsible-lists/js/CollapsibleLists.compressed.js"></script>
<script src="../_static/collapsible-lists/js/apply-collapsible-lists.js"></script>
<script crossorigin="anonymous" integrity="sha256-Ae2Vz/4ePdIu6ZyI/5ZGsYnb+m0JlOmKPjt6XZ9JJkA=" src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.4/require.min.js"></script>
<script type="text/javascript" src="../_static/js/vendor/popper.min.js"></script>
<script type="text/javascript" src="../_static/js/vendor/bootstrap.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/1.5.0/list.min.js"></script>
<script type="text/javascript" src="../_static/js/theme.js"></script>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
<!-- Begin Footer -->
<div class="container-fluid docs-tutorials-resources" id="docs-tutorials-resources">
<div class="container">
<div class="row">
<div class="col-md-4 text-center">
<h2>Docs</h2>
<p>Access comprehensive developer documentation for PyTorch</p>
<a class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html">View Docs</a>
</div>
<div class="col-md-4 text-center">
<h2>Tutorials</h2>
<p>Get in-depth tutorials for beginners and advanced developers</p>
<a class="with-right-arrow" href="https://pytorch.org/tutorials">View Tutorials</a>
</div>
<div class="col-md-4 text-center">
<h2>Resources</h2>
<p>Find development resources and get your questions answered</p>
<a class="with-right-arrow" href="https://pytorch.org/resources">View Resources</a>
</div>
</div>
</div>
</div>
<footer class="site-footer">
<div class="container footer-container">
<div class="footer-logo-wrapper">
<a href="https://pytorch.org/" class="footer-logo"></a>
</div>
<div class="footer-links-wrapper">
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/">PyTorch</a></li>
<li><a href="https://pytorch.org/get-started">Get Started</a></li>
<li><a href="https://pytorch.org/features">Features</a></li>
<li><a href="https://pytorch.org/ecosystem">Ecosystem</a></li>
<li><a href="https://pytorch.org/blog/">Blog</a></li>
<li><a href="https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md">Contributing</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/resources">Resources</a></li>
<li><a href="https://pytorch.org/tutorials">Tutorials</a></li>
<li><a href="https://pytorch.org/docs/stable/index.html">Docs</a></li>
<li><a href="https://discuss.pytorch.org" target="_blank">Discuss</a></li>
<li><a href="https://github.com/pytorch/pytorch/issues" target="_blank">Github Issues</a></li>
<li><a href="https://pytorch.org/assets/brand-guidelines/PyTorch-Brand-Guidelines.pdf" target="_blank">Brand Guidelines</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title">Stay up to date</li>
<li><a href="https://www.facebook.com/pytorch" target="_blank">Facebook</a></li>
<li><a href="https://twitter.com/pytorch" target="_blank">Twitter</a></li>
<li><a href="https://www.youtube.com/pytorch" target="_blank">YouTube</a></li>
<li><a href="https://www.linkedin.com/company/pytorch" target="_blank">LinkedIn</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title">PyTorch Podcasts</li>
<li><a href="https://open.spotify.com/show/6UzHKeiy368jKfQMKKvJY5" target="_blank">Spotify</a></li>
<li><a href="https://podcasts.apple.com/us/podcast/pytorch-developer-podcast/id1566080008" target="_blank">Apple</a></li>
<li><a href="https://www.google.com/podcasts?feed=aHR0cHM6Ly9mZWVkcy5zaW1wbGVjYXN0LmNvbS9PQjVGa0lsOA%3D%3D" target="_blank">Google</a></li>
<li><a href="https://music.amazon.com/podcasts/7a4e6f0e-26c2-49e9-a478-41bd244197d0/PyTorch-Developer-Podcast?" target="_blank">Amazon</a></li>
</ul>
</div>
</div>
<div class="privacy-policy">
<ul>
<li class="privacy-policy-links"><a href="https://www.linuxfoundation.org/terms/" target="_blank">Terms</a></li>
<li class="privacy-policy-links">|</li>
<li class="privacy-policy-links"><a href="https://www.linuxfoundation.org/privacy-policy/" target="_blank">Privacy</a></li>
</ul>
</div>
<div class="copyright">
<p>© Copyright The Linux Foundation. The PyTorch Foundation is a project of The Linux Foundation.
For web site terms of use, trademark policy and other policies applicable to The PyTorch Foundation please see
<a href="https://www.linuxfoundation.org/policies/">www.linuxfoundation.org/policies/</a>. The PyTorch Foundation supports the PyTorch open source
project, which has been established as PyTorch Project a Series of LF Projects, LLC. For policies applicable to the PyTorch Project a Series of LF Projects, LLC,
please see <a href="https://www.lfprojects.org/policies/">www.lfprojects.org/policies/</a>.</p>
</div>
</div>
</footer>
<div class="cookie-banner-wrapper">
<div class="container">
<p class="gdpr-notice">To analyze traffic and optimize your experience, we serve cookies on this site. By clicking or navigating, you agree to allow our usage of cookies. As the current maintainers of this site, Facebook’s Cookies Policy applies. Learn more, including about available controls: <a href="https://www.facebook.com/policies/cookies/">Cookies Policy</a>.</p>
<img class="close-button" src="../_static/images/pytorch-x.svg">
</div>
</div>
<!-- End Footer -->
<!-- Begin Mobile Menu -->
<div class="mobile-main-menu">
<div class="container-fluid">
<div class="container">
<div class="mobile-main-menu-header-container">
<a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
<a class="main-menu-close-button" href="#" data-behavior="close-mobile-menu"></a>
</div>
</div>
</div>
<div class="mobile-main-menu-links-container">
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.org/get-started">Get Started</a>
</li>
<li>
<a href="https://pytorch.org/ecosystem">Ecosystem</a>
</li>
<li>
<a href="">Mobile</a>
</li>
<li>
<a href="https://pytorch.org/blog/">Blog</a>
</li>
<li>
<a href="https://pytorch.org/tutorials">Tutorials</a>
</li>
<li class="resources-mobile-menu-title">
Docs
</li>
<ul class="resources-mobile-menu-items">
<li>
<a href="https://pytorch.org/docs/stable/index.html">PyTorch</a>
</li>
<li>
<a href="https://pytorch.org/audio/stable/index.html">torchaudio</a>
</li>
<li>
<a href="https://pytorch.org/text/stable/index.html">torchtext</a>
</li>
<li>
<a href="https://pytorch.org/vision/stable/index.html">torchvision</a>
</li>
<li>
<a href="https://pytorch.org/torcharrow">torcharrow</a>
</li>
<li>
<a href="https://pytorch.org/data">TorchData</a>
</li>
<li>
<a href="https://pytorch.org/torchrec">TorchRec</a>
</li>
<li>
<a href="https://pytorch.org/serve/">TorchServe</a>
</li>
<li>
<a href="https://pytorch.org/torchx/">TorchX</a>
</li>
<li>
<a href="https://pytorch.org/xla">PyTorch on XLA Devices</a>
</li>
</ul>
<li class="resources-mobile-menu-title">
Resources
</li>
<ul class="resources-mobile-menu-items">
<li>
<a href="https://pytorch.org/features">About</a>
</li>
<li>
<a href="https://pytorch.org/foundation">PyTorch Foundation</a>
</li>
<li>
<a href="https://pytorch.org/#community-module">Community</a>
</li>
<li>
<a href="https://pytorch.org/community-stories">Community Stories</a>
</li>
<li>
<a href="https://pytorch.org/resources">Developer Resources</a>
</li>
<li>
<a href="https://pytorch.org/events">Events</a>
</li>
<li>
<a href="https://discuss.pytorch.org/">Forums</a>
</li>
<li>
<a href="https://pytorch.org/hub">Models (Beta)</a>
</li>
</ul>
<li>
<a href="https://github.com/pytorch/pytorch">Github</a>
</li>
</ul>
</div>
</div>
</div>
<!-- End Mobile Menu -->
<script type="text/javascript" src="../_static/js/vendor/anchor.min.js"></script>
<script type="text/javascript">
$(document).ready(function() {
mobileMenu.bind();
mobileTOC.bind();
pytorchAnchors.bind();
sideMenus.bind();
scrollToAnchor.bind();
highlightNavigation.bind();
mainMenuDropdown.bind();
filterTags.bind();
// Add class to links that have code blocks, since we cannot create links in code blocks
$("article.pytorch-article a span.pre").each(function(e) {
$(this).closest("a").addClass("has-code");
});
})
</script>
</body>
</html>