-
Notifications
You must be signed in to change notification settings - Fork 257
/
Copy pathmobile_perf.html
898 lines (723 loc) ยท 66.1 KB
/
mobile_perf.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
<!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="viewport" content="width=device-width, initial-scale=1.0">
<title>PyTorch ๋ชจ๋ฐ์ผ ์ฑ๋ฅ ๋ ์ํผ — PyTorch Tutorials 1.10.2+cu102 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/copybutton.css" type="text/css" />
<link rel="stylesheet" href="../_static/gallery.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="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="../_static/katex-math.css" type="text/css" />
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<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.kr/" aria-label="PyTorch"></a>
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.kr/get-started">์์ํ๊ธฐ</a>
</li>
<li class="active">
<a href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ</a>
</li>
<li>
<a href="https://pytorch.kr/hub">ํ๋ธ</a>
</li>
<li>
<a href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ</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">
1.10.2+cu102
</div>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../search.html" method="get">
<input type="text" name="q" placeholder="Search Tutorials" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ๋ ์ํผ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="recipes_index.html">๋ชจ๋ ๋ ์ํผ ๋ณด๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../prototype/prototype_index.html">๋ชจ๋ ํ๋กํ ํ์
๋ ์ํผ ๋ณด๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ์์ํ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/intro.html">ํ์ดํ ์น(PyTorch) ๊ธฐ๋ณธ ์ตํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/quickstart_tutorial.html">๋น ๋ฅธ ์์(Quickstart)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/tensorqs_tutorial.html">ํ
์(Tensor)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/data_tutorial.html">Dataset๊ณผ DataLoader</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/transforms_tutorial.html">๋ณํ(Transform)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/buildmodel_tutorial.html">์ ๊ฒฝ๋ง ๋ชจ๋ธ ๊ตฌ์ฑํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/autogradqs_tutorial.html"><code class="docutils literal notranslate"><span class="pre">torch.autograd</span></code>๋ฅผ ์ฌ์ฉํ ์๋ ๋ฏธ๋ถ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/optimization_tutorial.html">๋ชจ๋ธ ๋งค๊ฐ๋ณ์ ์ต์ ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/saveloadrun_tutorial.html">๋ชจ๋ธ ์ ์ฅํ๊ณ ๋ถ๋ฌ์ค๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">Introduction to PyTorch on YouTube</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt.html">Introduction to PyTorch - YouTube Series</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/introyt1_tutorial.html">Introduction to PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/tensors_deeper_tutorial.html">Introduction to PyTorch Tensors</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/autogradyt_tutorial.html">The Fundamentals of Autograd</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/modelsyt_tutorial.html">Building Models with PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/tensorboardyt_tutorial.html">PyTorch TensorBoard Support</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/trainingyt.html">Training with PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/captumyt.html">Model Understanding with Captum</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ๋ฐฐ์ฐ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deep_learning_60min_blitz.html">PyTorch๋ก ๋ฅ๋ฌ๋ํ๊ธฐ: 60๋ถ๋ง์ ๋์ฅ๋ด๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/pytorch_with_examples.html">์์ ๋ก ๋ฐฐ์ฐ๋ ํ์ดํ ์น(PyTorch)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/nn_tutorial.html"><cite>torch.nn</cite> ์ด <em>์ค์ ๋ก</em> ๋ฌด์์ธ๊ฐ์?</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/tensorboard_tutorial.html">TensorBoard๋ก ๋ชจ๋ธ, ๋ฐ์ดํฐ, ํ์ต ์๊ฐํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">์ด๋ฏธ์ง/๋น๋์ค</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/torchvision_tutorial.html">TorchVision ๊ฐ์ฒด ๊ฒ์ถ ๋ฏธ์ธ์กฐ์ (Finetuning) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/transfer_learning_tutorial.html">์ปดํจํฐ ๋น์ (Vision)์ ์ํ ์ ์ดํ์ต(Transfer Learning)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/fgsm_tutorial.html">์ ๋์ ์์ ์์ฑ(Adversarial Example Generation)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/dcgan_faces_tutorial.html">DCGAN ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/vt_tutorial.html">๋ฐฐํฌ๋ฅผ ์ํ ๋น์ ํธ๋์คํฌ๋จธ(Vision Transformer) ๋ชจ๋ธ ์ต์ ํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">์ค๋์ค</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_io_tutorial.html">Audio I/O</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_resampling_tutorial.html">Audio Resampling</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_data_augmentation_tutorial.html">Audio Data Augmentation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_feature_extractions_tutorial.html">Audio Feature Extractions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_feature_augmentation_tutorial.html">Audio Feature Augmentation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_datasets_tutorial.html">Audio Datasets</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/speech_recognition_pipeline_tutorial.html">Speech Recognition with Wav2Vec2</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/speech_command_classification_with_torchaudio_tutorial.html">Speech Command Classification with torchaudio</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/text_to_speech_with_torchaudio.html">Text-to-speech with torchaudio</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/forced_alignment_with_torchaudio_tutorial.html">Forced Alignment with Wav2Vec2</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ
์คํธ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/transformer_tutorial.html">nn.Transformer ์ TorchText ๋ก ์ํ์ค-ํฌ-์ํ์ค(Sequence-to-Sequence) ๋ชจ๋ธ๋งํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/char_rnn_classification_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: ๋ฌธ์-๋จ์ RNN์ผ๋ก ์ด๋ฆ ๋ถ๋ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/char_rnn_generation_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: ๋ฌธ์-๋จ์ RNN์ผ๋ก ์ด๋ฆ ์์ฑํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/seq2seq_translation_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: Sequence to Sequence ๋คํธ์ํฌ์ Attention์ ์ด์ฉํ ๋ฒ์ญ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/text_sentiment_ngrams_tutorial.html">torchtext ๋ผ์ด๋ธ๋ฌ๋ฆฌ๋ก ํ
์คํธ ๋ถ๋ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/translation_transformer.html">nn.Transformer์ torchtext๋ก ์ธ์ด ๋ฒ์ญํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">๊ฐํํ์ต</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/reinforcement_q_learning.html">๊ฐํ ํ์ต (DQN) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/mario_rl_tutorial.html">Train a Mario-playing RL Agent</a></li>
</ul>
<p class="caption"><span class="caption-text">PyTorch ๋ชจ๋ธ์ ํ๋ก๋์
ํ๊ฒฝ์ ๋ฐฐํฌํ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/flask_rest_api_tutorial.html">Flask๋ฅผ ์ฌ์ฉํ์ฌ Python์์ PyTorch๋ฅผ REST API๋ก ๋ฐฐํฌํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/Intro_to_TorchScript_tutorial.html">TorchScript ์๊ฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/cpp_export.html">C++์์ TorchScript ๋ชจ๋ธ ๋ก๋ฉํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/super_resolution_with_onnxruntime.html">(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">Code Transforms with FX</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/fx_conv_bn_fuser.html">(๋ฒ ํ) FX์์ ํฉ์ฑ๊ณฑ/๋ฐฐ์น ์ ๊ทํ(Convolution/Batch Norm) ๊ฒฐํฉ๊ธฐ(Fuser) ๋ง๋ค๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/fx_profiling_tutorial.html">(beta) Building a Simple CPU Performance Profiler with FX</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ๋ก ํธ์๋ API</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/memory_format_tutorial.html">(๋ฒ ํ) PyTorch๋ฅผ ์ฌ์ฉํ Channels Last ๋ฉ๋ชจ๋ฆฌ ํ์</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/forward_ad_usage.html">Forward-mode Automatic Differentiation (Beta)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/cpp_frontend.html">PyTorch C++ ํ๋ก ํธ์๋ ์ฌ์ฉํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/torch-script-parallelism.html">TorchScript์ ๋์ ๋ณ๋ ฌ ์ฒ๋ฆฌ(Dynamic Parallelism)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/cpp_autograd.html">C++ ํ๋ก ํธ์๋์ ์๋ ๋ฏธ๋ถ (autograd)</a></li>
</ul>
<p class="caption"><span class="caption-text">PyTorch ํ์ฅํ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/custom_function_double_backward_tutorial.html">Double Backward with Custom Functions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/custom_function_conv_bn_tutorial.html">Fusing Convolution and Batch Norm using Custom Function</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/cpp_extension.html">Custom C++ and CUDA Extensions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/torch_script_custom_ops.html">Extending TorchScript with Custom C++ Operators</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/torch_script_custom_classes.html">์ปค์คํ
C++ ํด๋์ค๋ก TorchScript ํ์ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/dispatcher.html">Registering a Dispatched Operator in C++</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/extend_dispatcher.html">Extending dispatcher for a new backend in C++</a></li>
</ul>
<p class="caption"><span class="caption-text">๋ชจ๋ธ ์ต์ ํ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/profiler.html">PyTorch ๋ชจ๋ ํ๋กํ์ผ๋ง ํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/tensorboard_profiler_tutorial.html">PyTorch Profiler With TensorBoard</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/hyperparameter_tuning_tutorial.html">Hyperparameter tuning with Ray Tune</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/vt_tutorial.html">๋ฐฐํฌ๋ฅผ ์ํ ๋น์ ํธ๋์คํฌ๋จธ(Vision Transformer) ๋ชจ๋ธ ์ต์ ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/parametrizations.html">Parametrizations Tutorial</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/pruning_tutorial.html">๊ฐ์ง์น๊ธฐ ๊ธฐ๋ฒ(Pruning) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/dynamic_quantization_tutorial.html">(๋ฒ ํ) LSTM ๊ธฐ๋ฐ ๋จ์ด ๋จ์ ์ธ์ด ๋ชจ๋ธ์ ๋์ ์์ํ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dynamic_quantization_bert_tutorial.html">(๋ฒ ํ) BERT ๋ชจ๋ธ ๋์ ์์ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/quantized_transfer_learning_tutorial.html">(๋ฒ ํ) ์ปดํจํฐ ๋น์ ํํ ๋ฆฌ์ผ์ ์ํ ์์ํ๋ ์ ์ดํ์ต(Quantized Transfer Learning)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/static_quantization_tutorial.html">(beta) Static Quantization with Eager Mode in PyTorch</a></li>
</ul>
<p class="caption"><span class="caption-text">๋ณ๋ ฌ ๋ฐ ๋ถ์ฐ ํ์ต</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/dist_overview.html">PyTorch Distributed Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/model_parallel_tutorial.html">๋จ์ผ ๋จธ์ ์ ์ฌ์ฉํ ๋ชจ๋ธ ๋ณ๋ ฌํ ๋ชจ๋ฒ ์ฌ๋ก</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/ddp_tutorial.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ ์ฒ๋ฆฌ ์์ํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dist_tuto.html">PyTorch๋ก ๋ถ์ฐ ์ดํ๋ฆฌ์ผ์ด์
๊ฐ๋ฐํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_tutorial.html">Getting Started with Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_param_server_tutorial.html">Implementing a Parameter Server Using Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dist_pipeline_parallel_tutorial.html">Distributed Pipeline Parallelism Using RPC</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_async_execution.html">Implementing Batch RPC Processing Using Asynchronous Executions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/rpc_ddp_tutorial.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ(DDP)๊ณผ ๋ถ์ฐ RPC ํ๋ ์์ํฌ ๊ฒฐํฉ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/pipeline_tutorial.html">ํ์ดํ๋ผ์ธ ๋ณ๋ ฌํ๋ก ํธ๋์คํฌ๋จธ ๋ชจ๋ธ ํ์ต์ํค๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/ddp_pipeline.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ ์ฒ๋ฆฌ์ ๋ณ๋ ฌ ์ฒ๋ฆฌ ํ์ดํ๋ผ์ธ์ ์ฌ์ฉํ ํธ๋์คํฌ๋จธ ๋ชจ๋ธ ํ์ต</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/generic_join.html">Distributed Training with Uneven Inputs Using the Join Context Manager</a></li>
</ul>
<p class="caption"><span class="caption-text">Mobile</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deeplabv3_on_ios.html">iOS์์์ ์ด๋ฏธ์ง ๋ถํ DeepLapV3</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deeplabv3_on_android.html">์๋๋ก์ด๋์์์ ์ด๋ฏธ์ง ๋ถํ DeepLapV3</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">
Tutorials
</a> >
</li>
<li>PyTorch ๋ชจ๋ฐ์ผ ์ฑ๋ฅ ๋ ์ํผ</li>
<li class="pytorch-breadcrumbs-aside">
<a href="../_sources/recipes/mobile_perf.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">
<div class="pytorch-call-to-action-links">
<div id="tutorial-type">recipes/mobile_perf</div>
<div id="google-colab-link">
<img class="call-to-action-img" src="../_static/images/pytorch-colab.svg"/>
<div class="call-to-action-desktop-view">Run in Google Colab</div>
<div class="call-to-action-mobile-view">Colab</div>
</div>
<div id="download-notebook-link">
<img class="call-to-action-notebook-img" src="../_static/images/pytorch-download.svg"/>
<div class="call-to-action-desktop-view">Download Notebook</div>
<div class="call-to-action-mobile-view">Notebook</div>
</div>
<div id="github-view-link">
<img class="call-to-action-img" src="../_static/images/pytorch-github.svg"/>
<div class="call-to-action-desktop-view">View on GitHub</div>
<div class="call-to-action-mobile-view">GitHub</div>
</div>
</div>
<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">
<div class="section" id="pytorch">
<h1>PyTorch ๋ชจ๋ฐ์ผ ์ฑ๋ฅ ๋ ์ํผ<a class="headerlink" href="#pytorch" title="Permalink to this headline">ยถ</a></h1>
<div class="section" id="id1">
<h2>์๊ฐ<a class="headerlink" href="#id1" title="Permalink to this headline">ยถ</a></h2>
<p>์ ๋ถ๋ ์๋์ง๋ง, ๋ชจ๋ฐ์ผ ๊ธฐ๊ธฐ์์์ ์ ํ๋ฆฌ์ผ์ด์
๊ณผ ML ๋ชจ๋ธ ์ถ๋ก ์ฌ์ฉ ์ฌ๋ก์
์ฑ๋ฅ(์ง์ฐ์๊ฐ)์ ๋งค์ฐ ์ค๋ํ ์ฌํญ์
๋๋ค.</p>
<p>์ค๋๋ PyTorch๋ GPU, DSP, NPU์ ๊ฐ์ ํ๋์จ์ด ๋ฐฑ์๋๊ฐ ์ฌ์ฉ ๊ฐ๋ฅํ ๋๊น์ง
CPU ๋ฐฑ์๋์์ ๋ชจ๋ธ์ ์คํํฉ๋๋ค.</p>
<p>์ด ๋ ์ํผ์์ ๋ฐฐ์ธ ๋ด์ฉ์:</p>
<ul class="simple">
<li>๋ชจ๋ฐ์ผ ๊ธฐ๊ธฐ์์ ์คํ ์๊ฐ์ ์ค์ด๋๋ฐ ๋์์ด ๋ (์ฑ๋ฅ์ ๋์ด๊ณ , ์ง์ฐ์๊ฐ์ ์ค์ด๋) ๋ชจ๋ธ ์ต์ ํ ๋ฐฉ๋ฒ</li>
<li>๋ฒค์น๋งํน(์ต์ ํ๊ฐ ์ฌ์ฉ ์ฌ๋ก์ ๋์์ด ๋์๋์ง ํ์ธ) ํ๋ ๋ฐฉ๋ฒ</li>
</ul>
</div>
<div class="section" id="id2">
<h2>๋ชจ๋ธ ์ค๋น<a class="headerlink" href="#id2" title="Permalink to this headline">ยถ</a></h2>
<p>๋ชจ๋ฐ์ผ ๊ธฐ๊ธฐ์์ ์คํ ์๊ฐ์ ์ค์ด๋๋ฐ ๋์์ด ๋ (์ฑ๋ฅ์ ๋์ด๊ณ , ์ง์ฐ์๊ฐ์ ์ค์ด๋)
๋ชจ๋ธ์ ์ต์ ํ๋ฅผ ์ํ ์ค๋น๋ถํฐ ์์ํฉ๋๋ค.</p>
<div class="section" id="id3">
<h3>์ค์ <a class="headerlink" href="#id3" title="Permalink to this headline">ยถ</a></h3>
<p>์ฒซ๋ฒ์งธ๋ก ์ ์ด๋ ๋ฒ์ ์ด 1.5.0 ์ด์์ธ PyTorch๋ฅผ conda๋ pip์ผ๋ก ์ค์นํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">conda</span> <span class="n">install</span> <span class="n">pytorch</span> <span class="n">torchvision</span> <span class="o">-</span><span class="n">c</span> <span class="n">pytorch</span>
</pre></div>
</div>
<p>๋๋</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">pip</span> <span class="n">install</span> <span class="n">torch</span> <span class="n">torchvision</span>
</pre></div>
</div>
<p>๋ชจ๋ธ ์ฝ๋:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">torch</span>
<span class="kn">from</span> <span class="nn">torch.utils.mobile_optimizer</span> <span class="kn">import</span> <span class="n">optimize_for_mobile</span>
<span class="k">class</span> <span class="nc">AnnotatedConvBnReLUModel</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">Module</span><span class="p">):</span>
<span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="nb">super</span><span class="p">(</span><span class="n">AnnotatedConvBnReLUModel</span><span class="p">,</span> <span class="bp">self</span><span class="p">)</span><span class="o">.</span><span class="fm">__init__</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">conv</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">Conv2d</span><span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="n">bias</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">bn</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">BatchNorm2d</span><span class="p">(</span><span class="mi">5</span><span class="p">)</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">relu</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">ReLU</span><span class="p">(</span><span class="n">inplace</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">quant</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">QuantStub</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">dequant</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">DeQuantStub</span><span class="p">()</span>
<span class="k">def</span> <span class="nf">forward</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">x</span><span class="p">):</span>
<span class="n">x</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">contiguous</span><span class="p">(</span><span class="n">memory_format</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">channels_last</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">quant</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">conv</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">bn</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">dequant</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="k">return</span> <span class="n">x</span>
<span class="n">model</span> <span class="o">=</span> <span class="n">AnnotatedConvBnReLUModel</span><span class="p">()</span>
</pre></div>
</div>
<p><code class="docutils literal notranslate"><span class="pre">torch.quantization.QuantStub</span></code> ์ <code class="docutils literal notranslate"><span class="pre">torch.quantization.DeQuantStub()</span></code> ์ ๋ฏธ์ฌ์ฉ ์คํ
(stub)์ด๋ฉฐ, ์์ํ(quantization) ๋จ๊ณ์ ์ฌ์ฉํฉ๋๋ค.</p>
</div>
<div class="section" id="torch-quantization-fuse-modules-fuse">
<h3>1. <code class="docutils literal notranslate"><span class="pre">torch.quantization.fuse_modules</span></code> ์ด์ฉํ์ฌ ์ฐ์ฐ์ ๊ฒฐํฉ(fuse)ํ๊ธฐ<a class="headerlink" href="#torch-quantization-fuse-modules-fuse" title="Permalink to this headline">ยถ</a></h3>
<p>fuse_modules์ ์์ํ ํจํค์ง ๋ด๋ถ์ ์๋ค๋ ๊ฒ์ ํผ๋ํ์ง ๋ง์ญ์์ค.
fuse_modules์ ๋ชจ๋ <code class="docutils literal notranslate"><span class="pre">torch.nn.Module</span></code> ์์ ๋์ํฉ๋๋ค.</p>
<p><code class="docutils literal notranslate"><span class="pre">torch.quantization.fuse_modules</span></code> ์ ๋ชจ๋๋ค์ ๋ฆฌ์คํธ๋ฅผ ํ๋์ ๋ชจ๋๋ก ๊ฒฐํฉํฉ๋๋ค.
์ด๊ฒ์ ์๋ ์์์ ๋ชจ๋๋ค๋ง ๊ฒฐํฉ์ํต๋๋ค:</p>
<ul class="simple">
<li>Convolution, Batch normalization</li>
<li>Convolution, Batch normalization, Relu</li>
<li>Convolution, Relu</li>
<li>Linear, Relu</li>
</ul>
<p>์ด ์คํฌ๋ฆฝํธ๋ ์ด์ ์ ์ ์ธ๋ ๋ชจ๋ธ์์ Convolution, Batch Normalization, Relu๋ฅผ ๊ฒฐํฉํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">fuse_modules</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="p">[[</span><span class="s1">'conv'</span><span class="p">,</span> <span class="s1">'bn'</span><span class="p">,</span> <span class="s1">'relu'</span><span class="p">]],</span> <span class="n">inplace</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
</pre></div>
</div>
</div>
<div class="section" id="id4">
<h3>2. ๋ชจ๋ธ ์์ํํ๊ธฐ<a class="headerlink" href="#id4" title="Permalink to this headline">ยถ</a></h3>
<p>PyTorch ์์ํ์ ๋ํ ๋ด์ฉ์
<a class="reference external" href="https://pytorch.org/blog/introduction-to-quantization-on-pytorch/">the dedicated tutorial</a> ์์ ์ฐพ์ ์ ์์ต๋๋ค.</p>
<p>๋ชจ๋ธ์ ์์ํ๋ ์ฐ์ฐ์ int8๋ก ์ฎ๊ธฐ๋ฉด์
๋์คํฌ์์ ๋ชจ๋ธ ํฌ๊ธฐ๋ฅผ ์ค์ด๊ธฐ๋ ํฉ๋๋ค.
์ด๋ฐ ํฌ๊ธฐ ๊ฐ์๋ ๋ชจ๋ธ์ ์ฒ์ ์ฝ์ด ๋ค์ผ ๋ ๋์คํฌ ์ฝ๊ธฐ ์ฐ์ฐ์ ์ค์ด๋๋ฐ ๋์์ ์ฃผ๊ณ ๋จ(RAM)์ ์ด๋๋ ์ค์
๋๋ค.
์ด๋ฌํ ๋ ์์์ ๋ชจ๋ฐ์ผ ์ ํ๋ฆฌ์ผ์ด์
์ ์ฑ๋ฅ์ ๋งค์ฐ ์ค์ํ ์ ์์ต๋๋ค.
์ด ์ฝ๋๋ ๋ชจ๋ธ ๋ณด์ (calibration) ํจ์๋ฅผ ์ํด ์คํ
์ ์ฌ์ฉํด์ ์์ํ๋ฅผ ํฉ๋๋ค. <a class="reference external" href="https://tutorials.pytorch.kr/advanced/static_quantization_tutorial.html#post-training-static-quantization">์ฌ๊ธฐ</a> ์์ ๊ด๋ จ๋ ์ฌํญ์ ์ฐพ์ ์ ์์ต๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">qconfig</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">get_default_qconfig</span><span class="p">(</span><span class="s1">'qnnpack'</span><span class="p">)</span>
<span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">prepare</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">inplace</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="c1"># ๋ชจ๋ธ ๋ณด์ </span>
<span class="k">def</span> <span class="nf">calibrate</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">calibration_data</span><span class="p">):</span>
<span class="c1"># ๋ชจ๋ธ ๋ณด์ ์ฝ๋</span>
<span class="k">return</span>
<span class="n">calibrate</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="p">[])</span>
<span class="n">torch</span><span class="o">.</span><span class="n">quantization</span><span class="o">.</span><span class="n">convert</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">inplace</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
</pre></div>
</div>
</div>
<div class="section" id="torch-utils-mobile-optimizer">
<h3>3. torch.utils.mobile_optimizer ์ฌ์ฉํ๊ธฐ<a class="headerlink" href="#torch-utils-mobile-optimizer" title="Permalink to this headline">ยถ</a></h3>
<p>Torch mobile_optimizer ํจํค์ง๋ ์คํฌ๋ฆฝํธ๋ ๋ชจ๋ธ์ ์ด์ฉํด์ ๋ช ๊ฐ์ง ์ต์ ํ๋ฅผ ์ํํ๊ณ ,
์ด๋ฌํ ์ต์ ํ๋ conv2d์ ์ ํ ์ฐ์ฐ์ ๋์์ด ๋ฉ๋๋ค.
์ด ํจํค์ง๋ ์ต์ ํ๋ ํ์์ผ๋ก ๋ชจ๋ธ ๊ฐ์ค์น๋ฅผ ์ฐ์ ํจํค์งํ๋ฉฐ(pre-packs)
๋ค์ ์ฐ์ฐ์ด relu์ด๋ฉด ์์ ์ฐ์ฐ๋ค๊ณผ relu ์ฐ์ฐ์ ๊ฒฐํฉ ์ํต๋๋ค.</p>
<p>๋จผ์ ์ด์ ๋จ๊ณ์์๋ถํฐ ๊ฒฐ๊ณผ ๋ชจ๋ธ์ ์์ฑํฉ๋๋ค:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">torchscript_model</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">jit</span><span class="o">.</span><span class="n">script</span><span class="p">(</span><span class="n">model</span><span class="p">)</span>
</pre></div>
</div>
<p>๋ค์์ <code class="docutils literal notranslate"><span class="pre">optimize_for_mobile</span></code> ์ ํธ์ถํ๊ณ ๋์คํฌ์ ๋ชจ๋ธ์ ์ ์ฅํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">torchscript_model_optimized</span> <span class="o">=</span> <span class="n">optimize_for_mobile</span><span class="p">(</span><span class="n">torchscript_model</span><span class="p">)</span>
<span class="n">torch</span><span class="o">.</span><span class="n">jit</span><span class="o">.</span><span class="n">save</span><span class="p">(</span><span class="n">torchscript_model_optimized</span><span class="p">,</span> <span class="s2">"model.pt"</span><span class="p">)</span>
</pre></div>
</div>
</div>
<div class="section" id="channels-last-tensor">
<h3>4. Channels Last Tensor ๋ฉ๋ชจ๋ฆฌ ํ์ ์ ํํ๊ธฐ<a class="headerlink" href="#channels-last-tensor" title="Permalink to this headline">ยถ</a></h3>
<p>Channels Last(NHWC) ๋ฉ๋ชจ๋ฆฌ ํ์์ PyTorch 1.4.0์์ ๋์
๋์์ต๋๋ค. ์ด ํ์์ ์ค์ง 4์ฐจ์ ํ
์๋ง์ ์ง์ํฉ๋๋ค. ์ด ๋ฉ๋ชจ๋ฆฌ ํ์์ ๋๋ถ๋ถ์ ์ฐ์ฐ์, ํนํ ํฉ์ฑ๊ณฑ ์ฐ์ฐ์ ๋ ๋์ ๋ฉ๋ชจ๋ฆฌ ์ง์ญ์ฑ์ ์ ๊ณตํฉ๋๋ค. ์ธก์ ๊ฒฐ๊ณผ๋ MobileNetV2 ๋ชจ๋ธ์์ ๊ธฐ๋ณธ Channels First(NCHW) ํ์์ ๋นํด 3๋ฐฐ์ ์๋ ํฅ์์ ๋ณด์ฌ ์ค๋๋ค.</p>
<p>์ด ๋ ์ํผ๋ฅผ ์์ฑํ๋ ์์ ์์๋, PyTorch Android ์๋ฐ API๋ Channels Last ๋ฉ๋ชจ๋ฆฌ ํ์์ผ๋ก ๋ ์
๋ ฅ์ ์ง์ํ์ง ์์ต๋๋ค. ํ์ง๋ง ๋ชจ๋ธ ์
๋ ฅ์ ์ํด ์ด ๋ฉ๋ชจ๋ฆฌ ํ์์ผ๋ก ๋ณํํ๋ฉด TorchScript ๋ชจ๋ธ ์์ค์์ ์ฌ์ฉ์ด ๊ฐ๋ฅํฉ๋๋ค.</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="k">def</span> <span class="nf">forward</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">x</span><span class="p">):</span>
<span class="n">x</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">contiguous</span><span class="p">(</span><span class="n">memory_format</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">channels_last</span><span class="p">)</span>
<span class="o">...</span>
</pre></div>
</div>
<p>์ด ๋ณํ์ ์
๋ ฅ์ด Channels Last ๋ฉ๋ชจ๋ฆฌ ํ์์ด๋ฉด ๋น์ฉ์ด ๋ค์ง ์์ต๋๋ค. ๊ฒฐ๊ตญ์๋ ๋ชจ๋ ์ฐ์ฐ์๊ฐ Channels Last ๋ฉ๋ชจ๋ฆฌ ํ์์ ์ ์งํ๋ฉด์ ์์
์ ํฉ๋๋ค.</p>
</div>
<div class="section" id="android">
<h3>5. Android - ์๋ฐฉํฅ ์ ๋ฌ์ ์ํ ํ
์ ์ฌ์ฌ์ฉํ๊ธฐ<a class="headerlink" href="#android" title="Permalink to this headline">ยถ</a></h3>
<p>๋ ์ํผ์์ ์ด ๋ถ๋ถ์ Android์๋ง ํด๋นํฉ๋๋ค.</p>
<p>๋ฉ๋ชจ๋ฆฌ๋ Android ์ฑ๋ฅ์ ๋งค์ฐ ์ค์ํ ์์์
๋๋ค. ์ค๋๋ ๋๋ฐ์ด์ค์์ ํนํ๋ ๋ ์ค์ํฉ๋๋ค.
ํ
์๋ ์๋นํ ์์ ๋ฉ๋ชจ๋ฆฌ๋ฅผ ํ์๋ก ํ ์ ์์ต๋๋ค.
์๋ฅผ ๋ค์ด ํ์ค ์ปดํจํฐ ๋น์ ํ
์๋ 1*3*224*224๊ฐ์ ์์๋ฅผ ํฌํจํฉ๋๋ค.
๋ฐ์ดํฐ ํ์
์ด float์ด๊ณ 588kb ๋ฉ๋ชจ๋ฆฌ๊ฐ ํ์ํ๋ค๊ณ ๊ฐ์ ํ ๊ฒฝ์ฐ์
๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">FloatBuffer</span> <span class="n">buffer</span> <span class="o">=</span> <span class="n">Tensor</span><span class="o">.</span><span class="n">allocateFloatBuffer</span><span class="p">(</span><span class="mi">1</span><span class="o">*</span><span class="mi">3</span><span class="o">*</span><span class="mi">224</span><span class="o">*</span><span class="mi">224</span><span class="p">);</span>
<span class="n">Tensor</span> <span class="n">tensor</span> <span class="o">=</span> <span class="n">Tensor</span><span class="o">.</span><span class="n">fromBlob</span><span class="p">(</span><span class="n">buffer</span><span class="p">,</span> <span class="n">new</span> <span class="n">long</span><span class="p">[]{</span><span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">});</span>
</pre></div>
</div>
<p>์ฌ๊ธฐ์์ ๋ค์ดํฐ๋ธ ๋ฉ๋ชจ๋ฆฌ๋ฅผ <code class="docutils literal notranslate"><span class="pre">java.nio.FloatBuffer</span></code> ๋ก ํ ๋นํ๊ณ ์ ์ฅ์๊ฐ ํ ๋น๋ ๋ฒํผ์ ๋ฉ๋ชจ๋ฆฌ๋ฅผ ๊ฐ๋ฆฌํฌ <code class="docutils literal notranslate"><span class="pre">org.pytorch.Tensor</span></code> ๋ฅผ ๋ง๋ญ๋๋ค.</p>
<p>๋๋ถ๋ถ์ ์ฌ์ฉ ์ฌ๋ก์์ ๋ชจ๋ธ ์๋ฐฉํฅ ์ ๋ฌ์ ๋จ ํ ๋ฒ๋ง ํ์ง ์๊ณ , ์ผ์ ํ ๋น๋๋ก ํน์ ๊ฐ๋ฅํ ํ ๋นจ๋ฆฌ ์งํํฉ๋๋ค.</p>
<p>๋ง์ฝ ๋ชจ๋ ๋ชจ๋ ์๋ฐฉํฅ ์ ๋ฌ์ ์ํด ๋ฉ๋ชจ๋ฆฌ ํ ๋น์ ์๋ก ํ๋ค๋ฉด - ๊ทธ๊ฑด ์ต์ ํ๊ฐ ์๋๋๋ค.
๋์ ์, ์ด์ ๋จ๊ณ์์ ํ ๋นํ ๋์ผํ ๋ฉ๋ชจ๋ฆฌ์ ์ ๋ฐ์ดํฐ๋ฅผ ์ฑ์ฐ๊ณ ๋ชจ๋ ์๋ฐฉํฅ ์ ๋ฌ์ ๋์ผํ ํ
์ ๊ฐ์ฒด์์ ๋ค์ ์คํํจ์ผ๋ก์จ ๋์ผํ ๋ฉ๋ชจ๋ฆฌ๋ฅผ ์ฌ์ฌ์ฉ ํ ์ ์์ต๋๋ค.</p>
<p>์ฝ๋๊ฐ ์ด๋ค ์์ผ๋ก ๊ตฌ์ฑ์ด ๋์ด ์๋์ง๋ <a class="reference external" href="https://github.com/pytorch/android-demo-app/blob/master/PyTorchDemoApp/app/src/main/java/org/pytorch/demo/vision/ImageClassificationActivity.java#L174">pytorch android application example</a> ์์ ํ์ธํ ์ ์์ต๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">protected</span> <span class="n">AnalysisResult</span> <span class="n">analyzeImage</span><span class="p">(</span><span class="n">ImageProxy</span> <span class="n">image</span><span class="p">,</span> <span class="nb">int</span> <span class="n">rotationDegrees</span><span class="p">)</span> <span class="p">{</span>
<span class="k">if</span> <span class="p">(</span><span class="n">mModule</span> <span class="o">==</span> <span class="n">null</span><span class="p">)</span> <span class="p">{</span>
<span class="n">mModule</span> <span class="o">=</span> <span class="n">Module</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="n">moduleFileAbsoluteFilePath</span><span class="p">);</span>
<span class="n">mInputTensorBuffer</span> <span class="o">=</span>
<span class="n">Tensor</span><span class="o">.</span><span class="n">allocateFloatBuffer</span><span class="p">(</span><span class="mi">3</span> <span class="o">*</span> <span class="mi">224</span> <span class="o">*</span> <span class="mi">224</span><span class="p">);</span>
<span class="n">mInputTensor</span> <span class="o">=</span> <span class="n">Tensor</span><span class="o">.</span><span class="n">fromBlob</span><span class="p">(</span><span class="n">mInputTensorBuffer</span><span class="p">,</span> <span class="n">new</span> <span class="n">long</span><span class="p">[]{</span><span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">});</span>
<span class="p">}</span>
<span class="n">TensorImageUtils</span><span class="o">.</span><span class="n">imageYUV420CenterCropToFloatBuffer</span><span class="p">(</span>
<span class="n">image</span><span class="o">.</span><span class="n">getImage</span><span class="p">(),</span> <span class="n">rotationDegrees</span><span class="p">,</span>
<span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span>
<span class="n">TensorImageUtils</span><span class="o">.</span><span class="n">TORCHVISION_NORM_MEAN_RGB</span><span class="p">,</span>
<span class="n">TensorImageUtils</span><span class="o">.</span><span class="n">TORCHVISION_NORM_STD_RGB</span><span class="p">,</span>
<span class="n">mInputTensorBuffer</span><span class="p">,</span> <span class="mi">0</span><span class="p">);</span>
<span class="n">Tensor</span> <span class="n">outputTensor</span> <span class="o">=</span> <span class="n">mModule</span><span class="o">.</span><span class="n">forward</span><span class="p">(</span><span class="n">IValue</span><span class="o">.</span><span class="n">from</span><span class="p">(</span><span class="n">mInputTensor</span><span class="p">))</span><span class="o">.</span><span class="n">toTensor</span><span class="p">();</span>
<span class="p">}</span>
</pre></div>
</div>
<p>๋ฉค๋ฒ ๋ณ์ <code class="docutils literal notranslate"><span class="pre">mModule</span></code> , <code class="docutils literal notranslate"><span class="pre">mInputTensorBuffer</span></code> , <code class="docutils literal notranslate"><span class="pre">mInputTensor</span></code> ๋ ๋จ ํ ๋ฒ ์ด๊ธฐํ๋ฅผ ํ๊ณ
๋ฒํผ๋ <code class="docutils literal notranslate"><span class="pre">org.pytorch.torchvision.TensorImageUtils.imageYUV420CenterCropToFloatBuffer</span></code> ๋ฅผ ์ด์ฉํด์ ๋ค์ ์ฑ์์ง๋๋ค.</p>
</div>
</div>
<div class="section" id="id5">
<h2>๋ฒค์น๋งํน<a class="headerlink" href="#id5" title="Permalink to this headline">ยถ</a></h2>
<p>๋ฒค์น๋งํน(์ต์ ํ๊ฐ ์ฌ์ฉ ์ฌ๋ก์ ๋์์ด ๋์๋์ง ํ์ธ)ํ๋ ์ต๊ณ ์ ๋ฐฉ๋ฒ์ ์ต์ ํ๋ฅผ ํ๊ณ ์ถ์ ํน์ ํ ์ฌ์ฉ ์ฌ๋ก๋ฅผ ์ธก์ ํ๋ ๊ฒ์
๋๋ค. ์ฑ๋ฅ ์ธก์ ํ์๊ฐ ํ๊ฒฝ์ ๋ฐ๋ผ ๋ฌ๋ผ์ง ์ ์๊ธฐ ๋๋ฌธ์
๋๋ค.</p>
<p>PyTorch ๋ฐฐํฌํ์ ๋ชจ๋ธ ์๋ฐฉํฅ ์ ๋ฌ์ ์คํํ๋ ๋ฐฉ์์ ์ฌ์ฉํด์ ์ํ ๊ทธ๋๋ก์(naked) ๋ฐ์ด๋๋ฆฌ๋ฅผ ๋ฒค์น๋งํนํ๋ ์๋จ์ ์ ๊ณตํฉ๋๋ค.
์ด ์ ๊ทผ๋ฒ์ ์ ํ๋ฆฌ์ผ์ด์
๋ด๋ถ์์ ์ํํ๋ ๋ฐฉ๋ฒ๋ณด๋ค ๋ ์์ ์ ์ธ ์ธก์ ์น๋ฅผ ์ ๊ณตํฉ๋๋ค.</p>
<div class="section" id="id6">
<h3>Android - ๋ฒค์น๋งํน ์ค์ <a class="headerlink" href="#id6" title="Permalink to this headline">ยถ</a></h3>
<p>๋ ์ํผ์์ ์ด ๋ถ๋ถ์ Android์๋ง ํด๋นํฉ๋๋ค.</p>
<p>๋ฒค์น๋งํน์ ์ํด ๋จผ์ ๋ฒค์น๋งํฌ ๋ฐ์ด๋๋ฆฌ๋ฅผ ๋น๋ํด์ผ ํฉ๋๋ค:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o"><</span><span class="n">from</span><span class="o">-</span><span class="n">your</span><span class="o">-</span><span class="n">root</span><span class="o">-</span><span class="n">pytorch</span><span class="o">-</span><span class="nb">dir</span><span class="o">></span>
<span class="n">rm</span> <span class="o">-</span><span class="n">rf</span> <span class="n">build_android</span>
<span class="n">BUILD_PYTORCH_MOBILE</span><span class="o">=</span><span class="mi">1</span> <span class="n">ANDROID_ABI</span><span class="o">=</span><span class="n">arm64</span><span class="o">-</span><span class="n">v8a</span> <span class="o">./</span><span class="n">scripts</span><span class="o">/</span><span class="n">build_android</span><span class="o">.</span><span class="n">sh</span> <span class="o">-</span><span class="n">DBUILD_BINARY</span><span class="o">=</span><span class="n">ON</span>
</pre></div>
</div>
<p>์ด ๊ณณ์ arm64 ๋ฐ์ด๋๋ฆฌ๊ฐ ์์ด์ผ ํฉ๋๋ค: <code class="docutils literal notranslate"><span class="pre">build_android/bin/speed_benchmark_torch</span></code> .
์ด ๋ฐ์ด๋๋ฆฌ๋ <code class="docutils literal notranslate"><span class="pre">--model=<path-to-model></span></code>, <code class="docutils literal notranslate"><span class="pre">--input_dim="1,3,224,224"</span></code> ์ ์
๋ ฅ์ ์ํ ์ฐจ์ ์ ๋ณด๋ก ๋ฐ๊ณ <code class="docutils literal notranslate"><span class="pre">--input_type="float"</span></code> ์ผ๋ก ์
๋ ฅ ํ์
์ ์ธ์๋ก ๋ฐ์ต๋๋ค.</p>
<p>Android ๋๋ฐ์ด์ค๋ฅผ ์ฐ๊ฒฐํ ์ ์ด ์์ผ๋ฉด,
speedbenchark_torch ๋ฐ์ด๋๋ฆฌ์ ๋ชจ๋ธ์ ํฐ์ผ๋ก ํธ์ํฉ๋๋ค:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">adb</span> <span class="n">push</span> <span class="o"><</span><span class="n">speedbenchmark</span><span class="o">-</span><span class="n">torch</span><span class="o">></span> <span class="o">/</span><span class="n">data</span><span class="o">/</span><span class="n">local</span><span class="o">/</span><span class="n">tmp</span>
<span class="n">adb</span> <span class="n">push</span> <span class="o"><</span><span class="n">path</span><span class="o">-</span><span class="n">to</span><span class="o">-</span><span class="n">scripted</span><span class="o">-</span><span class="n">model</span><span class="o">></span> <span class="o">/</span><span class="n">data</span><span class="o">/</span><span class="n">local</span><span class="o">/</span><span class="n">tmp</span>
</pre></div>
</div>
<p>์ด์ ๋ชจ๋ธ์ ๋ฒค์น๋งํนํ ์ค๋น๊ฐ ๋์์ต๋๋ค:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">adb</span> <span class="n">shell</span> <span class="s2">"/data/local/tmp/speed_benchmark_torch --model=/data/local/tmp/model.pt"</span> <span class="o">--</span><span class="n">input_dims</span><span class="o">=</span><span class="s2">"1,3,224,224"</span> <span class="o">--</span><span class="n">input_type</span><span class="o">=</span><span class="s2">"float"</span>
<span class="o">-----</span> <span class="n">output</span> <span class="o">-----</span>
<span class="n">Starting</span> <span class="n">benchmark</span><span class="o">.</span>
<span class="n">Running</span> <span class="n">warmup</span> <span class="n">runs</span><span class="o">.</span>
<span class="n">Main</span> <span class="n">runs</span><span class="o">.</span>
<span class="n">Main</span> <span class="n">run</span> <span class="n">finished</span><span class="o">.</span> <span class="n">Microseconds</span> <span class="n">per</span> <span class="nb">iter</span><span class="p">:</span> <span class="mf">121318.</span> <span class="n">Iters</span> <span class="n">per</span> <span class="n">second</span><span class="p">:</span> <span class="mf">8.24281</span>
</pre></div>
</div>
</div>
<div class="section" id="ios">
<h3>iOS - ๋ฒค์น๋งํน ์ค์ <a class="headerlink" href="#ios" title="Permalink to this headline">ยถ</a></h3>
<p>iOS์ ๊ฒฝ์ฐ , ๋ฒค์น๋งํน์ ๋๊ตฌ๋ก <a class="reference external" href="https://github.com/pytorch/pytorch/tree/master/ios/TestApp">TestApp</a> ์ ์ฌ์ฉํฉ๋๋ค.</p>
<p>๋จผ์ <code class="docutils literal notranslate"><span class="pre">optimize_for_mobile</span></code> ๋ฉ์๋๋ฅผ <a class="reference external" href="https://github.com/pytorch/pytorch/blob/master/ios/TestApp/benchmark/trace_model.py">TestApp/benchmark/trace_model.py</a> ์ ์๋ ํ์ด์ฌ ์คํฌ๋ฆฝํธ์ ์ ์ฉํฉ๋๋ค. ๊ฐ๋จํ ์๋์ ๊ฐ์ด ์ฝ๋๋ฅผ ์์ ํฉ๋๋ค.</p>
<div class="highlight-default 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">torchvision</span>
<span class="kn">from</span> <span class="nn">torch.utils.mobile_optimizer</span> <span class="kn">import</span> <span class="n">optimize_for_mobile</span>
<span class="n">model</span> <span class="o">=</span> <span class="n">torchvision</span><span class="o">.</span><span class="n">models</span><span class="o">.</span><span class="n">mobilenet_v2</span><span class="p">(</span><span class="n">pretrained</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="n">model</span><span class="o">.</span><span class="n">eval</span><span class="p">()</span>
<span class="n">example</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">rand</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">)</span>
<span class="n">traced_script_module</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">jit</span><span class="o">.</span><span class="n">trace</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">example</span><span class="p">)</span>
<span class="n">torchscript_model_optimized</span> <span class="o">=</span> <span class="n">optimize_for_mobile</span><span class="p">(</span><span class="n">traced_script_module</span><span class="p">)</span>
<span class="n">torch</span><span class="o">.</span><span class="n">jit</span><span class="o">.</span><span class="n">save</span><span class="p">(</span><span class="n">torchscript_model_optimized</span><span class="p">,</span> <span class="s2">"model.pt"</span><span class="p">)</span>
</pre></div>
</div>
<p>์ด์ <code class="docutils literal notranslate"><span class="pre">python</span> <span class="pre">trace_model.py</span></code> ๋ฅผ ์คํํฉ์๋ค. ๋ชจ๋ ๊ฒ์ด ์ ์๋ํ๋ค๋ฉด ๋ฒค์น๋งํน ๋๋ ํ ๋ฆฌ ๋ด๋ถ์ ์ต์ ํ๋ ๋ชจ๋ธ์ ์์ฑํ ์ ์์ด์ผ ํฉ๋๋ค.</p>
<p>๋ค์์ ์์ค์์๋ถํฐ PyTorch ๋ผ์ด๋ธ๋ฌ๋ฆฌ๋ฅผ ๋น๋ํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">BUILD_PYTORCH_MOBILE</span><span class="o">=</span><span class="mi">1</span> <span class="n">IOS_ARCH</span><span class="o">=</span><span class="n">arm64</span> <span class="o">./</span><span class="n">scripts</span><span class="o">/</span><span class="n">build_ios</span><span class="o">.</span><span class="n">sh</span>
</pre></div>
</div>
<p>์ด์ ์ต์ ํ๋ ๋ชจ๋ธ๊ณผ PyTorch๊ฐ ์ค๋น๋์๊ธฐ์ XCode ํ๋ก์ ํธ๋ฅผ ๋ง๋ค๊ณ ๋ฒค์น๋งํนํ ์๊ฐ์
๋๋ค. ์ด๋ฅผ ์ํด XCode ํ๋ก์ ํธ๋ฅผ ์ค์ ํ๋ ๋ฌด๊ฑฐ์ด ์์
์ ์ํํ๋ ๋ฃจ๋น ์คํฌ๋ฆฝํธ <cite>setup.rb</cite> ๋ฅผ ์ฌ์ฉํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">ruby</span> <span class="n">setup</span><span class="o">.</span><span class="n">rb</span>
</pre></div>
</div>
<p>์ด์ <cite>TestApp.xcodeproj</cite> ๋ฅผ ์ด๊ณ iPhone์ ์ฐ๊ฒฐํ๋ฉด ์ค๋น๊ฐ ๋๋ฌ์ต๋๋ค. ์๋๋ iPhoneX์์์ ์์ ๊ฒฐ๊ณผ์
๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">TestApp</span><span class="p">[</span><span class="mi">2121</span><span class="p">:</span><span class="mi">722447</span><span class="p">]</span> <span class="n">Main</span> <span class="n">runs</span>
<span class="n">TestApp</span><span class="p">[</span><span class="mi">2121</span><span class="p">:</span><span class="mi">722447</span><span class="p">]</span> <span class="n">Main</span> <span class="n">run</span> <span class="n">finished</span><span class="o">.</span> <span class="n">Milliseconds</span> <span class="n">per</span> <span class="nb">iter</span><span class="p">:</span> <span class="mf">28.767</span>
<span class="n">TestApp</span><span class="p">[</span><span class="mi">2121</span><span class="p">:</span><span class="mi">722447</span><span class="p">]</span> <span class="n">Iters</span> <span class="n">per</span> <span class="n">second</span><span class="p">:</span> <span class="p">:</span> <span class="mf">34.762</span>
<span class="n">TestApp</span><span class="p">[</span><span class="mi">2121</span><span class="p">:</span><span class="mi">722447</span><span class="p">]</span> <span class="n">Done</span><span class="o">.</span>
</pre></div>
</div>
</div>
</div>
</div>
</article>
</div>
<footer>
<hr class="rating-hr hr-top">
<div class="rating-container">
<div class="rating-prompt">Rate this Tutorial</div>
<div class="stars-outer">
<i class="far fa-star" title="1 Star" data-behavior="tutorial-rating" data-count="1"></i>
<i class="far fa-star" title="2 Stars" data-behavior="tutorial-rating" data-count="2"></i>
<i class="far fa-star" title="3 Stars" data-behavior="tutorial-rating" data-count="3"></i>
<i class="far fa-star" title="4 Stars" data-behavior="tutorial-rating" data-count="4"></i>
<i class="far fa-star" title="5 Stars" data-behavior="tutorial-rating" data-count="5"></i>
</div>
</div>
<hr class="rating-hr hr-bottom"/>
<div role="contentinfo">
<p>
© Copyright 2021, PyTorch & PyTorch Korea Community.
</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="#">PyTorch ๋ชจ๋ฐ์ผ ์ฑ๋ฅ ๋ ์ํผ</a><ul>
<li><a class="reference internal" href="#id1">์๊ฐ</a></li>
<li><a class="reference internal" href="#id2">๋ชจ๋ธ ์ค๋น</a><ul>
<li><a class="reference internal" href="#id3">์ค์ </a></li>
<li><a class="reference internal" href="#torch-quantization-fuse-modules-fuse">1. <code class="docutils literal notranslate"><span class="pre">torch.quantization.fuse_modules</span></code> ์ด์ฉํ์ฌ ์ฐ์ฐ์ ๊ฒฐํฉ(fuse)ํ๊ธฐ</a></li>
<li><a class="reference internal" href="#id4">2. ๋ชจ๋ธ ์์ํํ๊ธฐ</a></li>
<li><a class="reference internal" href="#torch-utils-mobile-optimizer">3. torch.utils.mobile_optimizer ์ฌ์ฉํ๊ธฐ</a></li>
<li><a class="reference internal" href="#channels-last-tensor">4. Channels Last Tensor ๋ฉ๋ชจ๋ฆฌ ํ์ ์ ํํ๊ธฐ</a></li>
<li><a class="reference internal" href="#android">5. Android - ์๋ฐฉํฅ ์ ๋ฌ์ ์ํ ํ
์ ์ฌ์ฌ์ฉํ๊ธฐ</a></li>
</ul>
</li>
<li><a class="reference internal" href="#id5">๋ฒค์น๋งํน</a><ul>
<li><a class="reference internal" href="#id6">Android - ๋ฒค์น๋งํน ์ค์ </a></li>
<li><a class="reference internal" href="#ios">iOS - ๋ฒค์น๋งํน ์ค์ </a></li>
</ul>
</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 type="text/javascript" src="../_static/jquery.js"></script>
<script type="text/javascript" src="../_static/underscore.js"></script>
<script type="text/javascript" src="../_static/doctools.js"></script>
<script type="text/javascript" src="../_static/clipboard.min.js"></script>
<script type="text/javascript" src="../_static/copybutton.js"></script>
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/katex.min.js"></script>
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/contrib/auto-render.min.js"></script>
<script type="text/javascript" src="../_static/katex_autorenderer.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>
<script>
//add microsoft link
if(window.location.href.indexOf("/beginner/basics/")!= -1)
{
var url="https://docs.microsoft.com/learn/paths/pytorch-fundamentals/?wt.mc_id=aiml-7486-cxa";
switch(window.location.pathname.split("/").pop().replace('.html',''))
{
case"quickstart_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/9-quickstart?WT.mc_id=aiml-7486-cxa";
break;
case"tensorqs_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/2-tensors?WT.mc_id=aiml-7486-cxa";
break;
case"data_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/3-data?WT.mc_id=aiml-7486-cxa";
break;
case"transforms_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/4-transforms?WT.mc_id=aiml-7486-cxa";
break;
case"buildmodel_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/5-model?WT.mc_id=aiml-7486-cxa";
break;
case"autogradqs_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/6-autograd?WT.mc_id=aiml-7486-cxa";
break;
case"optimization_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/7-optimization?WT.mc_id=aiml-7486-cxa";
break;
case"saveloadrun_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/8-inference?WT.mc_id=aiml-7486-cxa";
}
$(".pytorch-call-to-action-links").children().first().before("<a href="+url+' data-behavior="call-to-action-event" data-response="Run in Microsoft Learn" target="_blank"><div id="microsoft-learn-link" style="padding-bottom: 0.625rem;border-bottom: 1px solid #f3f4f7;padding-right: 2.5rem;display: -webkit-box; display: -ms-flexbox; isplay: flex; -webkit-box-align: center;-ms-flex-align: center;align-items: center;"><img class="call-to-action-img" src="../../_static/images/microsoft-logo.svg"/><div class="call-to-action-desktop-view">Run in Microsoft Learn</div><div class="call-to-action-mobile-view">Learn</div></div></a>')
}
</script>
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-71919972-3"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'UA-71919972-3');
</script>
<script>
$("[data-behavior='call-to-action-event']").on('click', function(){
ga('send', {
hitType: 'event',
eventCategory: $(this).attr("data-response"),
eventAction: 'click',
eventLabel: window.location.href
});
gtag('event', 'click', {
'event_category': $(this).attr("data-response"),
'event_label': $("h1").first().text(),
'tutorial_link': window.location.href
});
});
$("[data-behavior='tutorial-rating']").on('click', function(){
gtag('event', 'click', {
'event_category': 'Tutorial Rating',
'event_label': $("h1").first().text(),
'value': $(this).attr("data-count")
});
});
if (location.pathname == "/") {
$(".rating-container").hide();
$(".hr-bottom").hide();
}
</script>
<script type="text/javascript">
var collapsedSections = ['ํ์ดํ ์น(PyTorch) ๋ ์ํผ', 'ํ์ดํ ์น(PyTorch) ๋ฐฐ์ฐ๊ธฐ', '์ด๋ฏธ์ง/๋น๋์ค', '์ค๋์ค', 'ํ
์คํธ', '๊ฐํํ์ต', 'PyTorch ๋ชจ๋ธ์ ํ๋ก๋์
ํ๊ฒฝ์ ๋ฐฐํฌํ๊ธฐ', 'Code Transforms with FX', 'ํ๋ก ํธ์๋ API', 'PyTorch ํ์ฅํ๊ธฐ', '๋ชจ๋ธ ์ต์ ํ', '๋ณ๋ ฌ ๋ฐ ๋ถ์ฐ ํ์ต', 'Mobile'];
</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>๊ณต์ ๋ฌธ์ (์์ด)</h2>
<p>PyTorch ๊ณต์ ๋ฌธ์์
๋๋ค.</p>
<a id="orgTutorialLink" class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html" target="_blank">๊ณต์ ๋ฌธ์๋ก ์ด๋</a>
</div>
<div class="col-md-4 text-center">
<h2>ํ๊ตญ์ด ํํ ๋ฆฌ์ผ</h2>
<p>ํ๊ตญ์ด๋ก ๋ฒ์ญ ์ค์ธ PyTorch ํํ ๋ฆฌ์ผ์
๋๋ค.</p>
<a class="with-right-arrow" href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ๋ก ์ด๋</a>
</div>
<div class="col-md-4 text-center">
<h2>์ปค๋ฎค๋ํฐ</h2>
<p>๋ค๋ฅธ ์ฌ์ฉ์๋ค๊ณผ ์๊ฒฌ์ ๋๋ ๋ณด์ธ์!</p>
<a class="with-right-arrow" href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ๋ก ์ด๋</a>
</div>
</div>
</div>
</div>
<footer class="site-footer">
<div class="container footer-container">
<div class="footer-logo-wrapper">
<a href="https://pytorch.kr/" 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 ํํ์ด์ง (๊ณต์)</a></li>
<li><a href="https://pytorch.org" target="_blank">๊ณต์ ํํ์ด์ง</a></li>
<li><a href="https://pytorch.org/tutorials" target="_blank">๊ณต์ ํํ ๋ฆฌ์ผ</a></li>
<li><a href="https://pytorch.org/docs" target="_blank">๊ณต์ ๋ฌธ์</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="">ํ๊ตญ ์ฌ์ฉ์ ๋ชจ์</a></li>
<li><a href="https://pytorch.kr/about">์ฌ์ดํธ ์๊ฐ</a></li>
<li><a href="https://tutorials.pytorch.kr/">ํ๊ตญ์ด ํํ ๋ฆฌ์ผ</a></li>
<li><a href="https://github.com/9bow/PyTorch-tutorials-kr" target="_blank">ํ๊ตญ์ด ํํ ๋ฆฌ์ผ ์ ์ฅ์</a></li>
</ul>
</div>
</div>
<div class="trademark-disclaimer">
<ul>
<li>์ด ์ฌ์ดํธ๋ PyTorch ํ๊ตญ ์ฌ์ฉ์ ์ปค๋ฎค๋ํฐ๋ก Facebook, Inc์์ ์ด์ํ๋ ์ฌ์ดํธ๊ฐ ์๋๋๋ค. PyTorch, PyTorch ๋ก๊ณ ๋ฐ ๋ชจ๋ ๊ด๋ จ ํ๊ธฐ๋ Facebook, Inc์ ์ํ์
๋๋ค.</li>
<li>This site is a user community and is not operated by Facebook, Inc. PyTorch, the PyTorch logo and any related marks are trademarks of Facebook, Inc.</li>
</ul>
</div>
</div>
</footer>
<!-- 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.kr/" 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.kr/get-started">์์ํ๊ธฐ</a>
</li>
<li class="active">
<a href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ</a>
</li>
<li>
<a href="https://pytorch.kr/hub">ํ๋ธ</a>
</li>
<li>
<a href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ</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>