forked from pytorch/pytorch.github.io
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtorch.nn.quantized.html
1772 lines (1507 loc) · 171 KB
/
torch.nn.quantized.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!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 name="robots" content="noindex">
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>torch.nn.quantized — PyTorch 1.7.1 documentation</title>
<link rel="canonical" href="https://pytorch.org/docs/stable/torch.nn.quantized.html"/>
<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="https://cdn.jsdelivr.net/npm/[email protected]/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="_static/css/jit.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/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" />
<link rel="next" title="torch.nn.quantized.dynamic" href="torch.nn.quantized.dynamic.html" />
<link rel="prev" title="torch.quantization" href="torch.quantization.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/[email protected]/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" 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>
<a href="https://pytorch.org/mobile">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="active">
<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/1.7.1/index.html">
<span class="dropdown-title">PyTorch</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/audio/0.7.0/index.html">
<span class="dropdown-title">torchaudio</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/text/0.8.1/index.html">
<span class="dropdown-title">torchtext</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/vision/0.8/">
<span class="dropdown-title">torchvision</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/elastic/">
<span class="dropdown-title">TorchElastic</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/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/#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/resources">
<span class="dropdown-title">Developer Resources</span>
<p>Find resources and get questions answered</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">
<a href='http://pytorch.org/docs/versions.html'>1.7.1 ▼</a>
</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"><span class="caption-text">Notes</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="notes/amp_examples.html">Automatic Mixed Precision examples</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/autograd.html">Autograd mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/broadcasting.html">Broadcasting semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cpu_threading_torchscript_inference.html">CPU threading and TorchScript inference</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cuda.html">CUDA semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/ddp.html">Distributed Data Parallel</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/extending.html">Extending PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/faq.html">Frequently Asked Questions</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/large_scale_deployments.html">Features for large-scale deployments</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/multiprocessing.html">Multiprocessing best practices</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/randomness.html">Reproducibility</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/serialization.html">Serialization semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/windows.html">Windows FAQ</a></li>
</ul>
<p class="caption"><span class="caption-text">Language Bindings</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="cpp_index.html">C++</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/javadoc/">Javadoc</a></li>
</ul>
<p class="caption"><span class="caption-text">Python API</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="torch.html">torch</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.html">torch.nn</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.functional.html">torch.nn.functional</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensors.html">torch.Tensor</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_attributes.html">Tensor Attributes</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_view.html">Tensor Views</a></li>
<li class="toctree-l1"><a class="reference internal" href="autograd.html">torch.autograd</a></li>
<li class="toctree-l1"><a class="reference internal" href="cuda.html">torch.cuda</a></li>
<li class="toctree-l1"><a class="reference internal" href="amp.html">torch.cuda.amp</a></li>
<li class="toctree-l1"><a class="reference internal" href="backends.html">torch.backends</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.html">torch.distributed</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributions.html">torch.distributions</a></li>
<li class="toctree-l1"><a class="reference internal" href="fft.html">torch.fft</a></li>
<li class="toctree-l1"><a class="reference internal" href="futures.html">torch.futures</a></li>
<li class="toctree-l1"><a class="reference internal" href="hub.html">torch.hub</a></li>
<li class="toctree-l1"><a class="reference internal" href="jit.html">torch.jit</a></li>
<li class="toctree-l1"><a class="reference internal" href="linalg.html">torch.linalg</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.init.html">torch.nn.init</a></li>
<li class="toctree-l1"><a class="reference internal" href="onnx.html">torch.onnx</a></li>
<li class="toctree-l1"><a class="reference internal" href="optim.html">torch.optim</a></li>
<li class="toctree-l1"><a class="reference internal" href="complex_numbers.html">Complex Numbers</a></li>
<li class="toctree-l1 current"><a class="reference internal" href="quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="rpc.html">Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="random.html">torch.random</a></li>
<li class="toctree-l1"><a class="reference internal" href="sparse.html">torch.sparse</a></li>
<li class="toctree-l1"><a class="reference internal" href="storage.html">torch.Storage</a></li>
<li class="toctree-l1"><a class="reference internal" href="bottleneck.html">torch.utils.bottleneck</a></li>
<li class="toctree-l1"><a class="reference internal" href="checkpoint.html">torch.utils.checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_extension.html">torch.utils.cpp_extension</a></li>
<li class="toctree-l1"><a class="reference internal" href="data.html">torch.utils.data</a></li>
<li class="toctree-l1"><a class="reference internal" href="dlpack.html">torch.utils.dlpack</a></li>
<li class="toctree-l1"><a class="reference internal" href="mobile_optimizer.html">torch.utils.mobile_optimizer</a></li>
<li class="toctree-l1"><a class="reference internal" href="model_zoo.html">torch.utils.model_zoo</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorboard.html">torch.utils.tensorboard</a></li>
<li class="toctree-l1"><a class="reference internal" href="type_info.html">Type Info</a></li>
<li class="toctree-l1"><a class="reference internal" href="named_tensor.html">Named Tensors</a></li>
<li class="toctree-l1"><a class="reference internal" href="name_inference.html">Named Tensors operator coverage</a></li>
<li class="toctree-l1"><a class="reference internal" href="__config__.html">torch.__config__</a></li>
</ul>
<p class="caption"><span class="caption-text">Libraries</span></p>
<ul>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/audio">torchaudio</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/text">torchtext</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/vision">torchvision</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/elastic/">TorchElastic</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/serve">TorchServe</a></li>
<li class="toctree-l1"><a class="reference external" href="http://pytorch.org/xla/">PyTorch on XLA Devices</a></li>
</ul>
<p class="caption"><span class="caption-text">Community</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="community/contribution_guide.html">PyTorch Contribution Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/governance.html">PyTorch Governance</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/persons_of_interest.html">PyTorch Governance | Persons of Interest</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><a href="quantization.html">Quantization</a> ></li>
<li>torch.nn.quantized</li>
<li class="pytorch-breadcrumbs-aside">
<a href="_sources/torch.nn.quantized.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="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="torch-nn-quantized">
<h1>torch.nn.quantized<a class="headerlink" href="#torch-nn-quantized" title="Permalink to this headline">¶</a></h1>
<p>This module implements the quantized versions of the nn layers such as
~`torch.nn.Conv2d` and <cite>torch.nn.ReLU</cite>.</p>
<div class="section" id="module-torch.nn.quantized.functional">
<span id="functional-interface"></span><h2>Functional interface<a class="headerlink" href="#module-torch.nn.quantized.functional" title="Permalink to this headline">¶</a></h2>
<p>Functional interface (quantized).</p>
<dl class="function">
<dt id="torch.nn.quantized.functional.relu">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">relu</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">inplace=False</em><span class="sig-paren">)</span> → Tensor<a class="reference internal" href="_modules/torch/nn/quantized/functional.html#relu"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.relu" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies the rectified linear unit function element-wise.
See <a class="reference internal" href="#torch.nn.quantized.ReLU" title="torch.nn.quantized.ReLU"><code class="xref py py-class docutils literal notranslate"><span class="pre">ReLU</span></code></a> for more details.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input</p></li>
<li><p><strong>inplace</strong> – perform the computation inplace</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.linear">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">linear</code><span class="sig-paren">(</span><em class="sig-param">input: torch.Tensor</em>, <em class="sig-param">weight: torch.Tensor</em>, <em class="sig-param">bias: Optional[torch.Tensor] = None</em>, <em class="sig-param">scale: Optional[float] = None</em>, <em class="sig-param">zero_point: Optional[int] = None</em><span class="sig-paren">)</span> → torch.Tensor<a class="reference internal" href="_modules/torch/nn/quantized/functional.html#linear"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.linear" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a linear transformation to the incoming quantized data:
<span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>y</mi><mo>=</mo><mi>x</mi><msup><mi>A</mi><mi>T</mi></msup><mo>+</mo><mi>b</mi></mrow><annotation encoding="application/x-tex">y = xA^T + b</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.625em;vertical-align:-0.19444em;"></span><span class="mord mathnormal" style="margin-right:0.03588em;">y</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span></span><span class="base"><span class="strut" style="height:0.924661em;vertical-align:-0.08333em;"></span><span class="mord mathnormal">x</span><span class="mord"><span class="mord mathnormal">A</span><span class="msupsub"><span class="vlist-t"><span class="vlist-r"><span class="vlist" style="height:0.8413309999999999em;"><span style="top:-3.063em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mathnormal mtight" style="margin-right:0.13889em;">T</span></span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2222222222222222em;"></span><span class="mbin">+</span><span class="mspace" style="margin-right:0.2222222222222222em;"></span></span><span class="base"><span class="strut" style="height:0.69444em;vertical-align:0em;"></span><span class="mord mathnormal">b</span></span></span></span>
</span>.
See <a class="reference internal" href="#torch.nn.quantized.Linear" title="torch.nn.quantized.Linear"><code class="xref py py-class docutils literal notranslate"><span class="pre">Linear</span></code></a></p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Current implementation packs weights on every call, which has penalty on performance.
If you want to avoid the overhead, use <a class="reference internal" href="#torch.nn.quantized.Linear" title="torch.nn.quantized.Linear"><code class="xref py py-class docutils literal notranslate"><span class="pre">Linear</span></code></a>.</p>
</div>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – Quantized input of type <cite>torch.quint8</cite></p></li>
<li><p><strong>weight</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – Quantized weight of type <cite>torch.qint8</cite></p></li>
<li><p><strong>bias</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – None or fp32 bias of type <cite>torch.float</cite></p></li>
<li><p><strong>scale</strong> (<em>double</em>) – output scale. If None, derived from the input scale</p></li>
<li><p><strong>zero_point</strong> (<em>long</em>) – output zero point. If None, derived from the input zero_point</p></li>
</ul>
</dd>
</dl>
<dl class="simple">
<dt>Shape:</dt><dd><ul class="simple">
<li><p>Input: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo separator="true">,</mo><mi>i</mi><mi>n</mi><mi mathvariant="normal">_</mi><mi>f</mi><mi>e</mi><mi>a</mi><mi>t</mi><mi>u</mi><mi>r</mi><mi>e</mi><mi>s</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *, in\_features)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal">n</span><span class="mord" style="margin-right:0.02778em;">_</span><span class="mord mathnormal" style="margin-right:0.10764em;">f</span><span class="mord mathnormal">e</span><span class="mord mathnormal">a</span><span class="mord mathnormal">t</span><span class="mord mathnormal">u</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal">e</span><span class="mord mathnormal">s</span><span class="mclose">)</span></span></span></span>
</span> where <cite>*</cite> means any number of
additional dimensions</p></li>
<li><p>Weight: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>o</mi><mi>u</mi><mi>t</mi><mi mathvariant="normal">_</mi><mi>f</mi><mi>e</mi><mi>a</mi><mi>t</mi><mi>u</mi><mi>r</mi><mi>e</mi><mi>s</mi><mo separator="true">,</mo><mi>i</mi><mi>n</mi><mi mathvariant="normal">_</mi><mi>f</mi><mi>e</mi><mi>a</mi><mi>t</mi><mi>u</mi><mi>r</mi><mi>e</mi><mi>s</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(out\_features, in\_features)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord mathnormal">o</span><span class="mord mathnormal">u</span><span class="mord mathnormal">t</span><span class="mord" style="margin-right:0.02778em;">_</span><span class="mord mathnormal" style="margin-right:0.10764em;">f</span><span class="mord mathnormal">e</span><span class="mord mathnormal">a</span><span class="mord mathnormal">t</span><span class="mord mathnormal">u</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal">e</span><span class="mord mathnormal">s</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal">n</span><span class="mord" style="margin-right:0.02778em;">_</span><span class="mord mathnormal" style="margin-right:0.10764em;">f</span><span class="mord mathnormal">e</span><span class="mord mathnormal">a</span><span class="mord mathnormal">t</span><span class="mord mathnormal">u</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal">e</span><span class="mord mathnormal">s</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p>Bias: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>o</mi><mi>u</mi><mi>t</mi><mi mathvariant="normal">_</mi><mi>f</mi><mi>e</mi><mi>a</mi><mi>t</mi><mi>u</mi><mi>r</mi><mi>e</mi><mi>s</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(out\_features)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord mathnormal">o</span><span class="mord mathnormal">u</span><span class="mord mathnormal">t</span><span class="mord" style="margin-right:0.02778em;">_</span><span class="mord mathnormal" style="margin-right:0.10764em;">f</span><span class="mord mathnormal">e</span><span class="mord mathnormal">a</span><span class="mord mathnormal">t</span><span class="mord mathnormal">u</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal">e</span><span class="mord mathnormal">s</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p>Output: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo separator="true">,</mo><mi>o</mi><mi>u</mi><mi>t</mi><mi mathvariant="normal">_</mi><mi>f</mi><mi>e</mi><mi>a</mi><mi>t</mi><mi>u</mi><mi>r</mi><mi>e</mi><mi>s</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *, out\_features)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">o</span><span class="mord mathnormal">u</span><span class="mord mathnormal">t</span><span class="mord" style="margin-right:0.02778em;">_</span><span class="mord mathnormal" style="margin-right:0.10764em;">f</span><span class="mord mathnormal">e</span><span class="mord mathnormal">a</span><span class="mord mathnormal">t</span><span class="mord mathnormal">u</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal">e</span><span class="mord mathnormal">s</span><span class="mclose">)</span></span></span></span>
</span></p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.conv1d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">conv1d</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">weight</em>, <em class="sig-param">bias</em>, <em class="sig-param">stride=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">padding_mode='zeros'</em>, <em class="sig-param">scale=1.0</em>, <em class="sig-param">zero_point=0</em>, <em class="sig-param">dtype=torch.quint8</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#conv1d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.conv1d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 1D convolution over a quantized 1D input composed of several input
planes.</p>
<p>See <a class="reference internal" href="#torch.nn.quantized.Conv1d" title="torch.nn.quantized.Conv1d"><code class="xref py py-class docutils literal notranslate"><span class="pre">Conv1d</span></code></a> for details and output shape.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input tensor of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>minibatch</mtext><mo separator="true">,</mo><mtext>in_channels</mtext><mo separator="true">,</mo><mi>i</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{minibatch} , \text{in\_channels} , iW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">minibatch</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord text"><span class="mord">in_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>weight</strong> – quantized filters of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo separator="true">,</mo><mfrac><mtext>in_channels</mtext><mtext>groups</mtext></mfrac><mo separator="true">,</mo><mi>i</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels} , \frac{\text{in\_channels}}{\text{groups}} , iW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.4942159999999998em;vertical-align:-0.481108em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord"><span class="mopen nulldelimiter"></span><span class="mfrac"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:1.013108em;"><span style="top:-2.6550000000000002em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">groups</span></span></span></span></span><span style="top:-3.23em;"><span class="pstrut" style="height:3em;"></span><span class="frac-line" style="border-bottom-width:0.04em;"></span></span><span style="top:-3.527em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">in_channels</span></span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.481108em;"><span></span></span></span></span></span><span class="mclose nulldelimiter"></span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>bias</strong> – <strong>non-quantized</strong> bias tensor of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mclose">)</span></span></span></span>
</span>. The tensor type must be <cite>torch.float</cite>.</p></li>
<li><p><strong>stride</strong> – the stride of the convolving kernel. Can be a single number or a
tuple <cite>(sW,)</cite>. Default: 1</p></li>
<li><p><strong>padding</strong> – implicit paddings on both sides of the input. Can be a
single number or a tuple <cite>(padW,)</cite>. Default: 0</p></li>
<li><p><strong>dilation</strong> – the spacing between kernel elements. Can be a single number or
a tuple <cite>(dW,)</cite>. Default: 1</p></li>
<li><p><strong>groups</strong> – split input into groups, <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mtext>in_channels</mtext></mrow><annotation encoding="application/x-tex">\text{in\_channels}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.00444em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">in_channels</span></span></span></span></span>
</span> should be divisible by the
number of groups. Default: 1</p></li>
<li><p><strong>padding_mode</strong> – the padding mode to use. Only “zeros” is supported for quantized convolution at the moment. Default: “zeros”</p></li>
<li><p><strong>scale</strong> – quantization scale for the output. Default: 1.0</p></li>
<li><p><strong>zero_point</strong> – quantization zero_point for the output. Default: 0</p></li>
<li><p><strong>dtype</strong> – quantization data type to use. Default: <code class="docutils literal notranslate"><span class="pre">torch.quint8</span></code></p></li>
</ul>
</dd>
</dl>
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.nn.quantized</span> <span class="kn">import</span> <span class="n">functional</span> <span class="k">as</span> <span class="n">qF</span>
<span class="gp">>>> </span><span class="n">filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">33</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">3</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="gp">>>> </span><span class="n">inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">20</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">50</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="gp">>>> </span><span class="n">bias</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">33</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="go">>>></span>
<span class="gp">>>> </span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span> <span class="o">=</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span>
<span class="gp">>>> </span><span class="n">dtype_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quint8</span>
<span class="gp">>>> </span><span class="n">dtype_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">qint8</span>
<span class="go">>>></span>
<span class="gp">>>> </span><span class="n">q_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">filters</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_filters</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">q_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">inputs</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_inputs</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">qF</span><span class="o">.</span><span class="n">conv1d</span><span class="p">(</span><span class="n">q_inputs</span><span class="p">,</span> <span class="n">q_filters</span><span class="p">,</span> <span class="n">bias</span><span class="p">,</span> <span class="n">padding</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="o">=</span><span class="n">zero_point</span><span class="p">)</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.conv2d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">conv2d</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">weight</em>, <em class="sig-param">bias</em>, <em class="sig-param">stride=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">padding_mode='zeros'</em>, <em class="sig-param">scale=1.0</em>, <em class="sig-param">zero_point=0</em>, <em class="sig-param">dtype=torch.quint8</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#conv2d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.conv2d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 2D convolution over a quantized 2D input composed of several input
planes.</p>
<p>See <a class="reference internal" href="#torch.nn.quantized.Conv2d" title="torch.nn.quantized.Conv2d"><code class="xref py py-class docutils literal notranslate"><span class="pre">Conv2d</span></code></a> for details and output shape.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input tensor of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>minibatch</mtext><mo separator="true">,</mo><mtext>in_channels</mtext><mo separator="true">,</mo><mi>i</mi><mi>H</mi><mo separator="true">,</mo><mi>i</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{minibatch} , \text{in\_channels} , iH , iW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">minibatch</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord text"><span class="mord">in_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>weight</strong> – quantized filters of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo separator="true">,</mo><mfrac><mtext>in_channels</mtext><mtext>groups</mtext></mfrac><mo separator="true">,</mo><mi>k</mi><mi>H</mi><mo separator="true">,</mo><mi>k</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels} , \frac{\text{in\_channels}}{\text{groups}} , kH , kW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.4942159999999998em;vertical-align:-0.481108em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord"><span class="mopen nulldelimiter"></span><span class="mfrac"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:1.013108em;"><span style="top:-2.6550000000000002em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">groups</span></span></span></span></span><span style="top:-3.23em;"><span class="pstrut" style="height:3em;"></span><span class="frac-line" style="border-bottom-width:0.04em;"></span></span><span style="top:-3.527em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">in_channels</span></span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.481108em;"><span></span></span></span></span></span><span class="mclose nulldelimiter"></span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>bias</strong> – <strong>non-quantized</strong> bias tensor of shape <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mclose">)</span></span></span></span>
</span>. The tensor type must be <cite>torch.float</cite>.</p></li>
<li><p><strong>stride</strong> – the stride of the convolving kernel. Can be a single number or a
tuple <cite>(sH, sW)</cite>. Default: 1</p></li>
<li><p><strong>padding</strong> – implicit paddings on both sides of the input. Can be a
single number or a tuple <cite>(padH, padW)</cite>. Default: 0</p></li>
<li><p><strong>dilation</strong> – the spacing between kernel elements. Can be a single number or
a tuple <cite>(dH, dW)</cite>. Default: 1</p></li>
<li><p><strong>groups</strong> – split input into groups, <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mtext>in_channels</mtext></mrow><annotation encoding="application/x-tex">\text{in\_channels}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.00444em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">in_channels</span></span></span></span></span>
</span> should be divisible by the
number of groups. Default: 1</p></li>
<li><p><strong>padding_mode</strong> – the padding mode to use. Only “zeros” is supported for quantized convolution at the moment. Default: “zeros”</p></li>
<li><p><strong>scale</strong> – quantization scale for the output. Default: 1.0</p></li>
<li><p><strong>zero_point</strong> – quantization zero_point for the output. Default: 0</p></li>
<li><p><strong>dtype</strong> – quantization data type to use. Default: <code class="docutils literal notranslate"><span class="pre">torch.quint8</span></code></p></li>
</ul>
</dd>
</dl>
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.nn.quantized</span> <span class="kn">import</span> <span class="n">functional</span> <span class="k">as</span> <span class="n">qF</span>
<span class="gp">>>> </span><span class="n">filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">8</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</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="gp">>>> </span><span class="n">inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">5</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="gp">>>> </span><span class="n">bias</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">8</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="go">>>></span>
<span class="gp">>>> </span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span> <span class="o">=</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span>
<span class="gp">>>> </span><span class="n">dtype_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quint8</span>
<span class="gp">>>> </span><span class="n">dtype_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">qint8</span>
<span class="go">>>></span>
<span class="gp">>>> </span><span class="n">q_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">filters</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_filters</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">q_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">inputs</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_inputs</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">qF</span><span class="o">.</span><span class="n">conv2d</span><span class="p">(</span><span class="n">q_inputs</span><span class="p">,</span> <span class="n">q_filters</span><span class="p">,</span> <span class="n">bias</span><span class="p">,</span> <span class="n">padding</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="o">=</span><span class="n">zero_point</span><span class="p">)</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.conv3d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">conv3d</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">weight</em>, <em class="sig-param">bias</em>, <em class="sig-param">stride=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">padding_mode='zeros'</em>, <em class="sig-param">scale=1.0</em>, <em class="sig-param">zero_point=0</em>, <em class="sig-param">dtype=torch.quint8</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#conv3d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.conv3d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 3D convolution over a quantized 3D input composed of several input
planes.</p>
<p>See <a class="reference internal" href="#torch.nn.quantized.Conv3d" title="torch.nn.quantized.Conv3d"><code class="xref py py-class docutils literal notranslate"><span class="pre">Conv3d</span></code></a> for details and output shape.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input tensor of shape
<span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>minibatch</mtext><mo separator="true">,</mo><mtext>in_channels</mtext><mo separator="true">,</mo><mi>i</mi><mi>D</mi><mo separator="true">,</mo><mi>i</mi><mi>H</mi><mo separator="true">,</mo><mi>i</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{minibatch} , \text{in\_channels} , iD , iH , iW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">minibatch</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord text"><span class="mord">in_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.02778em;">D</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>weight</strong> – quantized filters of shape
<span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo separator="true">,</mo><mfrac><mtext>in_channels</mtext><mtext>groups</mtext></mfrac><mo separator="true">,</mo><mi>k</mi><mi>D</mi><mo separator="true">,</mo><mi>k</mi><mi>H</mi><mo separator="true">,</mo><mi>k</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels} , \frac{\text{in\_channels}}{\text{groups}} , kD , kH , kW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.4942159999999998em;vertical-align:-0.481108em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord"><span class="mopen nulldelimiter"></span><span class="mfrac"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:1.013108em;"><span style="top:-2.6550000000000002em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">groups</span></span></span></span></span><span style="top:-3.23em;"><span class="pstrut" style="height:3em;"></span><span class="frac-line" style="border-bottom-width:0.04em;"></span></span><span style="top:-3.527em;"><span class="pstrut" style="height:3em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord text mtight"><span class="mord mtight">in_channels</span></span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.481108em;"><span></span></span></span></span></span><span class="mclose nulldelimiter"></span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.02778em;">D</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>bias</strong> – <strong>non-quantized</strong> bias tensor of shape
<span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>out_channels</mtext><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{out\_channels})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">out_channels</span></span><span class="mclose">)</span></span></span></span>
</span>. The tensor type must be <cite>torch.float</cite>.</p></li>
<li><p><strong>stride</strong> – the stride of the convolving kernel. Can be a single number or a
tuple <cite>(sD, sH, sW)</cite>. Default: 1</p></li>
<li><p><strong>padding</strong> – implicit paddings on both sides of the input. Can be a
single number or a tuple <cite>(padD, padH, padW)</cite>. Default: 0</p></li>
<li><p><strong>dilation</strong> – the spacing between kernel elements. Can be a single number or
a tuple <cite>(dD, dH, dW)</cite>. Default: 1</p></li>
<li><p><strong>groups</strong> – split input into groups, <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mtext>in_channels</mtext></mrow><annotation encoding="application/x-tex">\text{in\_channels}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.00444em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">in_channels</span></span></span></span></span>
</span> should be
divisible by the number of groups. Default: 1</p></li>
<li><p><strong>padding_mode</strong> – the padding mode to use. Only “zeros” is supported for
quantized convolution at the moment. Default: “zeros”</p></li>
<li><p><strong>scale</strong> – quantization scale for the output. Default: 1.0</p></li>
<li><p><strong>zero_point</strong> – quantization zero_point for the output. Default: 0</p></li>
<li><p><strong>dtype</strong> – quantization data type to use. Default: <code class="docutils literal notranslate"><span class="pre">torch.quint8</span></code></p></li>
</ul>
</dd>
</dl>
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.nn.quantized</span> <span class="kn">import</span> <span class="n">functional</span> <span class="k">as</span> <span class="n">qF</span>
<span class="gp">>>> </span><span class="n">filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">8</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</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="gp">>>> </span><span class="n">inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">5</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="gp">>>> </span><span class="n">bias</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">8</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="go">>>></span>
<span class="gp">>>> </span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span> <span class="o">=</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span>
<span class="gp">>>> </span><span class="n">dtype_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quint8</span>
<span class="gp">>>> </span><span class="n">dtype_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">qint8</span>
<span class="go">>>></span>
<span class="gp">>>> </span><span class="n">q_filters</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">filters</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_filters</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">q_inputs</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="n">inputs</span><span class="p">,</span> <span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="p">,</span> <span class="n">dtype_inputs</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">qF</span><span class="o">.</span><span class="n">conv3d</span><span class="p">(</span><span class="n">q_inputs</span><span class="p">,</span> <span class="n">q_filters</span><span class="p">,</span> <span class="n">bias</span><span class="p">,</span> <span class="n">padding</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="n">scale</span><span class="p">,</span> <span class="n">zero_point</span><span class="o">=</span><span class="n">zero_point</span><span class="p">)</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.max_pool2d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">max_pool2d</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">stride=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">return_indices=False</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#max_pool2d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.max_pool2d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 2D max pooling over a quantized input signal composed of
several quantized input planes.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters are propagated to the output.</p>
</div>
<p>See <code class="xref py py-class docutils literal notranslate"><span class="pre">MaxPool2d</span></code> for details.</p>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.adaptive_avg_pool2d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">adaptive_avg_pool2d</code><span class="sig-paren">(</span><em class="sig-param">input: torch.Tensor</em>, <em class="sig-param">output_size: None</em><span class="sig-paren">)</span> → torch.Tensor<a class="reference internal" href="_modules/torch/nn/quantized/functional.html#adaptive_avg_pool2d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.adaptive_avg_pool2d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 2D adaptive average pooling over a quantized input signal composed
of several quantized input planes.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<p>See <code class="xref py py-class docutils literal notranslate"><span class="pre">AdaptiveAvgPool2d</span></code> for details and output shape.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>output_size</strong> – the target output size (single integer or
double-integer tuple)</p>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.avg_pool2d">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">avg_pool2d</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">stride=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">count_include_pad=True</em>, <em class="sig-param">divisor_override=None</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#avg_pool2d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.avg_pool2d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies 2D average-pooling operation in <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>k</mi><mi>H</mi><mo>×</mo><mi>k</mi><mi>W</mi></mrow><annotation encoding="application/x-tex">kH \times kW</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.77777em;vertical-align:-0.08333em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mspace" style="margin-right:0.2222222222222222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222222222222222em;"></span></span><span class="base"><span class="strut" style="height:0.69444em;vertical-align:0em;"></span><span class="mord mathnormal" style="margin-right:0.03148em;">k</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span></span></span></span>
</span> regions by step size
<span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>s</mi><mi>H</mi><mo>×</mo><mi>s</mi><mi>W</mi></mrow><annotation encoding="application/x-tex">sH \times sW</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.76666em;vertical-align:-0.08333em;"></span><span class="mord mathnormal">s</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mspace" style="margin-right:0.2222222222222222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222222222222222em;"></span></span><span class="base"><span class="strut" style="height:0.68333em;vertical-align:0em;"></span><span class="mord mathnormal">s</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span></span></span></span>
</span> steps. The number of output features is equal to the number of
input planes.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<p>See <code class="xref py py-class docutils literal notranslate"><span class="pre">AvgPool2d</span></code> for details and output shape.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input tensor <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mtext>minibatch</mtext><mo separator="true">,</mo><mtext>in_channels</mtext><mo separator="true">,</mo><mi>i</mi><mi>H</mi><mo separator="true">,</mo><mi>i</mi><mi>W</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(\text{minibatch} , \text{in\_channels} , iH , iW)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.06em;vertical-align:-0.31em;"></span><span class="mopen">(</span><span class="mord text"><span class="mord">minibatch</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord text"><span class="mord">in_channels</span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.08125em;">H</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">i</span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="mclose">)</span></span></span></span>
</span></p></li>
<li><p><strong>kernel_size</strong> – size of the pooling region. Can be a single number or a
tuple <cite>(kH, kW)</cite></p></li>
<li><p><strong>stride</strong> – stride of the pooling operation. Can be a single number or a
tuple <cite>(sH, sW)</cite>. Default: <code class="xref py py-attr docutils literal notranslate"><span class="pre">kernel_size</span></code></p></li>
<li><p><strong>padding</strong> – implicit zero paddings on both sides of the input. Can be a
single number or a tuple <cite>(padH, padW)</cite>. Default: 0</p></li>
<li><p><strong>ceil_mode</strong> – when True, will use <cite>ceil</cite> instead of <cite>floor</cite> in the formula
to compute the output shape. Default: <code class="docutils literal notranslate"><span class="pre">False</span></code></p></li>
<li><p><strong>count_include_pad</strong> – when True, will include the zero-padding in the
averaging calculation. Default: <code class="docutils literal notranslate"><span class="pre">True</span></code></p></li>
<li><p><strong>divisor_override</strong> – if specified, it will be used as divisor, otherwise
size of the pooling region will be used. Default: None</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.interpolate">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">interpolate</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">size=None</em>, <em class="sig-param">scale_factor=None</em>, <em class="sig-param">mode='nearest'</em>, <em class="sig-param">align_corners=None</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#interpolate"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.interpolate" title="Permalink to this definition">¶</a></dt>
<dd><p>Down/up samples the input to either the given <code class="xref py py-attr docutils literal notranslate"><span class="pre">size</span></code> or the given
<code class="xref py py-attr docutils literal notranslate"><span class="pre">scale_factor</span></code></p>
<p>See <a class="reference internal" href="nn.functional.html#torch.nn.functional.interpolate" title="torch.nn.functional.interpolate"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.nn.functional.interpolate()</span></code></a> for implementation details.</p>
<p>The input dimensions are interpreted in the form:
<cite>mini-batch x channels x [optional depth] x [optional height] x width</cite>.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only 2D/3D input is supported for quantized inputs</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only the following modes are supported for the quantized inputs:</p>
<ul class="simple">
<li><p><cite>bilinear</cite></p></li>
<li><p><cite>nearest</cite></p></li>
</ul>
</div>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – the input tensor</p></li>
<li><p><strong>size</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>] or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>] or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>]</em>) – output spatial size.</p></li>
<li><p><strong>scale_factor</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#float" title="(in Python v3.9)"><em>float</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#float" title="(in Python v3.9)"><em>float</em></a><em>]</em>) – multiplier for spatial size. Has to match input size if it is a tuple.</p></li>
<li><p><strong>mode</strong> (<a class="reference external" href="https://docs.python.org/3/library/stdtypes.html#str" title="(in Python v3.9)"><em>str</em></a>) – algorithm used for upsampling:
<code class="docutils literal notranslate"><span class="pre">'nearest'</span></code> | <code class="docutils literal notranslate"><span class="pre">'bilinear'</span></code></p></li>
<li><p><strong>align_corners</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#bool" title="(in Python v3.9)"><em>bool</em></a><em>, </em><em>optional</em>) – Geometrically, we consider the pixels of the
input and output as squares rather than points.
If set to <code class="docutils literal notranslate"><span class="pre">True</span></code>, the input and output tensors are aligned by the
center points of their corner pixels, preserving the values at the corner pixels.
If set to <code class="docutils literal notranslate"><span class="pre">False</span></code>, the input and output tensors are aligned by the corner
points of their corner pixels, and the interpolation uses edge value padding
for out-of-boundary values, making this operation <em>independent</em> of input size
when <code class="xref py py-attr docutils literal notranslate"><span class="pre">scale_factor</span></code> is kept the same. This only has an effect when <code class="xref py py-attr docutils literal notranslate"><span class="pre">mode</span></code>
is <code class="docutils literal notranslate"><span class="pre">'bilinear'</span></code>.
Default: <code class="docutils literal notranslate"><span class="pre">False</span></code></p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.hardswish">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">hardswish</code><span class="sig-paren">(</span><em class="sig-param">input: torch.Tensor</em>, <em class="sig-param">scale: float</em>, <em class="sig-param">zero_point: int</em><span class="sig-paren">)</span> → torch.Tensor<a class="reference internal" href="_modules/torch/nn/quantized/functional.html#hardswish"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.hardswish" title="Permalink to this definition">¶</a></dt>
<dd><p>This is the quantized version of <a class="reference internal" href="nn.functional.html#torch.nn.functional.hardswish" title="torch.nn.functional.hardswish"><code class="xref py py-func docutils literal notranslate"><span class="pre">hardswish()</span></code></a>.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> – quantized input</p></li>
<li><p><strong>scale</strong> – quantization scale of the output tensor</p></li>
<li><p><strong>zero_point</strong> – quantization zero point of the output tensor</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.upsample">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">upsample</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">size=None</em>, <em class="sig-param">scale_factor=None</em>, <em class="sig-param">mode='nearest'</em>, <em class="sig-param">align_corners=None</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#upsample"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.upsample" title="Permalink to this definition">¶</a></dt>
<dd><p>Upsamples the input to either the given <code class="xref py py-attr docutils literal notranslate"><span class="pre">size</span></code> or the given
<code class="xref py py-attr docutils literal notranslate"><span class="pre">scale_factor</span></code></p>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>This function is deprecated in favor of
<a class="reference internal" href="#torch.nn.quantized.functional.interpolate" title="torch.nn.quantized.functional.interpolate"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.nn.quantized.functional.interpolate()</span></code></a>.
This is equivalent with <code class="docutils literal notranslate"><span class="pre">nn.quantized.functional.interpolate(...)</span></code>.</p>
</div>
<p>See <a class="reference internal" href="nn.functional.html#torch.nn.functional.interpolate" title="torch.nn.functional.interpolate"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.nn.functional.interpolate()</span></code></a> for implementation details.</p>
<p>The input dimensions are interpreted in the form:
<cite>mini-batch x channels x [optional depth] x [optional height] x width</cite>.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only 2D input is supported for quantized inputs</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only the following modes are supported for the quantized inputs:</p>
<ul class="simple">
<li><p><cite>bilinear</cite></p></li>
<li><p><cite>nearest</cite></p></li>
</ul>
</div>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – quantized input tensor</p></li>
<li><p><strong>size</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>] or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>] or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>]</em>) – output spatial size.</p></li>
<li><p><strong>scale_factor</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#float" title="(in Python v3.9)"><em>float</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#float" title="(in Python v3.9)"><em>float</em></a><em>]</em>) – multiplier for spatial size. Has to be an integer.</p></li>
<li><p><strong>mode</strong> (<em>string</em>) – algorithm used for upsampling:
<code class="docutils literal notranslate"><span class="pre">'nearest'</span></code> | <code class="docutils literal notranslate"><span class="pre">'bilinear'</span></code></p></li>
<li><p><strong>align_corners</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#bool" title="(in Python v3.9)"><em>bool</em></a><em>, </em><em>optional</em>) – Geometrically, we consider the pixels of the
input and output as squares rather than points.
If set to <code class="docutils literal notranslate"><span class="pre">True</span></code>, the input and output tensors are aligned by the
center points of their corner pixels, preserving the values at the corner pixels.
If set to <code class="docutils literal notranslate"><span class="pre">False</span></code>, the input and output tensors are aligned by the corner
points of their corner pixels, and the interpolation uses edge value padding
for out-of-boundary values, making this operation <em>independent</em> of input size
when <code class="xref py py-attr docutils literal notranslate"><span class="pre">scale_factor</span></code> is kept the same. This only has an effect when <code class="xref py py-attr docutils literal notranslate"><span class="pre">mode</span></code>
is <code class="docutils literal notranslate"><span class="pre">'bilinear'</span></code>.
Default: <code class="docutils literal notranslate"><span class="pre">False</span></code></p></li>
</ul>
</dd>
</dl>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>With <code class="docutils literal notranslate"><span class="pre">align_corners</span> <span class="pre">=</span> <span class="pre">True</span></code>, the linearly interpolating modes
(<cite>bilinear</cite>) don’t proportionally align the
output and input pixels, and thus the output values can depend on the
input size. This was the default behavior for these modes up to version
0.3.1. Since then, the default behavior is <code class="docutils literal notranslate"><span class="pre">align_corners</span> <span class="pre">=</span> <span class="pre">False</span></code>.
See <a class="reference internal" href="generated/torch.nn.Upsample.html#torch.nn.Upsample" title="torch.nn.Upsample"><code class="xref py py-class docutils literal notranslate"><span class="pre">Upsample</span></code></a> for concrete examples on how this
affects the outputs.</p>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.upsample_bilinear">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">upsample_bilinear</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">size=None</em>, <em class="sig-param">scale_factor=None</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#upsample_bilinear"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.upsample_bilinear" title="Permalink to this definition">¶</a></dt>
<dd><p>Upsamples the input, using bilinear upsampling.</p>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>This function is deprecated in favor of
<a class="reference internal" href="#torch.nn.quantized.functional.interpolate" title="torch.nn.quantized.functional.interpolate"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.nn.quantized.functional.interpolate()</span></code></a>.
This is equivalent with
<code class="docutils literal notranslate"><span class="pre">nn.quantized.functional.interpolate(...,</span> <span class="pre">mode='bilinear',</span> <span class="pre">align_corners=True)</span></code>.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only 2D inputs are supported</p>
</div>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – quantized input</p></li>
<li><p><strong>size</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>]</em>) – output spatial size.</p></li>
<li><p><strong>scale_factor</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>]</em>) – multiplier for spatial size</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.nn.quantized.functional.upsample_nearest">
<code class="sig-prename descclassname">torch.nn.quantized.functional.</code><code class="sig-name descname">upsample_nearest</code><span class="sig-paren">(</span><em class="sig-param">input</em>, <em class="sig-param">size=None</em>, <em class="sig-param">scale_factor=None</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/functional.html#upsample_nearest"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.functional.upsample_nearest" title="Permalink to this definition">¶</a></dt>
<dd><p>Upsamples the input, using nearest neighbours’ pixel values.</p>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>This function is deprecated in favor of
<a class="reference internal" href="#torch.nn.quantized.functional.interpolate" title="torch.nn.quantized.functional.interpolate"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.nn.quantized.functional.interpolate()</span></code></a>.
This is equivalent with <code class="docutils literal notranslate"><span class="pre">nn.quantized.functional.interpolate(...,</span> <span class="pre">mode='nearest')</span></code>.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The input quantization parameters propagate to the output.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only 2D inputs are supported</p>
</div>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>input</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – quantized input</p></li>
<li><p><strong>size</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em> or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>] or </em><em>Tuple</em><em>[</em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>, </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a><em>]</em>) – output spatial
size.</p></li>
<li><p><strong>scale_factor</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.9)"><em>int</em></a>) – multiplier for spatial size. Has to be an integer.</p></li>
</ul>
</dd>
</dl>
</dd></dl>
<span class="target" id="module-torch.nn.quantized"></span></div>
<div class="section" id="relu">
<h2>ReLU<a class="headerlink" href="#relu" title="Permalink to this headline">¶</a></h2>
<dl class="class">
<dt id="torch.nn.quantized.ReLU">
<em class="property">class </em><code class="sig-prename descclassname">torch.nn.quantized.</code><code class="sig-name descname">ReLU</code><span class="sig-paren">(</span><em class="sig-param">inplace=False</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/modules/activation.html#ReLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.ReLU" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies quantized rectified linear unit function element-wise:</p>
<p><span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mtext>ReLU</mtext><mo stretchy="false">(</mo><mi>x</mi><mo stretchy="false">)</mo><mo>=</mo><mi>max</mi><mo></mo><mo stretchy="false">(</mo><msub><mi>x</mi><mn>0</mn></msub><mo separator="true">,</mo><mi>x</mi><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">\text{ReLU}(x)= \max(x_0, x)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mord text"><span class="mord">ReLU</span></span><span class="mopen">(</span><span class="mord mathnormal">x</span><span class="mclose">)</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span></span><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mop">max</span><span class="mopen">(</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.30110799999999993em;"><span style="top:-2.5500000000000003em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight">0</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">x</span><span class="mclose">)</span></span></span></span>
</span>, where <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>x</mi><mn>0</mn></msub></mrow><annotation encoding="application/x-tex">x_0</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.58056em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.30110799999999993em;"><span style="top:-2.5500000000000003em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight">0</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span>
</span> is the zero point.</p>
<p>Please see <a class="reference external" href="https://pytorch.org/docs/1.7.1/nn.html#torch.nn.ReLU">https://pytorch.org/docs/stable/nn.html#torch.nn.ReLU</a>
for more documentation on ReLU.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>inplace</strong> – (Currently not supported) can optionally do the operation in-place.</p>
</dd>
</dl>
<dl class="simple">
<dt>Shape:</dt><dd><ul class="simple">
<li><p>Input: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mclose">)</span></span></span></span>
</span> where <cite>*</cite> means, any number of additional
dimensions</p></li>
<li><p>Output: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mclose">)</span></span></span></span>
</span>, same shape as the input</p></li>
</ul>
</dd>
</dl>
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">m</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">quantized</span><span class="o">.</span><span class="n">ReLU</span><span class="p">()</span>
<span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span>
<span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="nb">input</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</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">qint32</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">output</span> <span class="o">=</span> <span class="n">m</span><span class="p">(</span><span class="nb">input</span><span class="p">)</span>
</pre></div>
</div>
</dd></dl>
</div>
<div class="section" id="relu6">
<h2>ReLU6<a class="headerlink" href="#relu6" title="Permalink to this headline">¶</a></h2>
<dl class="class">
<dt id="torch.nn.quantized.ReLU6">
<em class="property">class </em><code class="sig-prename descclassname">torch.nn.quantized.</code><code class="sig-name descname">ReLU6</code><span class="sig-paren">(</span><em class="sig-param">inplace=False</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/modules/activation.html#ReLU6"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.ReLU6" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies the element-wise function:</p>
<p><span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mtext>ReLU6</mtext><mo stretchy="false">(</mo><mi>x</mi><mo stretchy="false">)</mo><mo>=</mo><mi>min</mi><mo></mo><mo stretchy="false">(</mo><mi>max</mi><mo></mo><mo stretchy="false">(</mo><msub><mi>x</mi><mn>0</mn></msub><mo separator="true">,</mo><mi>x</mi><mo stretchy="false">)</mo><mo separator="true">,</mo><mi>q</mi><mo stretchy="false">(</mo><mn>6</mn><mo stretchy="false">)</mo><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">\text{ReLU6}(x) = \min(\max(x_0, x), q(6))</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mord text"><span class="mord">ReLU6</span></span><span class="mopen">(</span><span class="mord mathnormal">x</span><span class="mclose">)</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2777777777777778em;"></span></span><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mop">min</span><span class="mopen">(</span><span class="mop">max</span><span class="mopen">(</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.30110799999999993em;"><span style="top:-2.5500000000000003em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight">0</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal">x</span><span class="mclose">)</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord mathnormal" style="margin-right:0.03588em;">q</span><span class="mopen">(</span><span class="mord">6</span><span class="mclose">)</span><span class="mclose">)</span></span></span></span>
</span>, where <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>x</mi><mn>0</mn></msub></mrow><annotation encoding="application/x-tex">x_0</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.58056em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.30110799999999993em;"><span style="top:-2.5500000000000003em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight">0</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span>
</span> is the
zero_point, and <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>q</mi><mo stretchy="false">(</mo><mn>6</mn><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">q(6)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mord mathnormal" style="margin-right:0.03588em;">q</span><span class="mopen">(</span><span class="mord">6</span><span class="mclose">)</span></span></span></span>
</span> is the quantized representation of number 6.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>inplace</strong> – can optionally do the operation in-place. Default: <code class="docutils literal notranslate"><span class="pre">False</span></code></p>
</dd>
</dl>
<dl class="simple">
<dt>Shape:</dt><dd><ul class="simple">
<li><p>Input: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mclose">)</span></span></span></span>
</span> where <cite>*</cite> means, any number of additional
dimensions</p></li>
<li><p>Output: <span class="math"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mo stretchy="false">(</mo><mi>N</mi><mo separator="true">,</mo><mo>∗</mo><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">(N, *)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.10903em;">N</span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.16666666666666666em;"></span><span class="mord">∗</span><span class="mclose">)</span></span></span></span>
</span>, same shape as the input</p></li>
</ul>
</dd>
</dl>
<img alt="_images/ReLU61.png" src="_images/ReLU61.png" />
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">m</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">quantized</span><span class="o">.</span><span class="n">ReLU6</span><span class="p">()</span>
<span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span>
<span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">quantize_per_tensor</span><span class="p">(</span><span class="nb">input</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</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">qint32</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">output</span> <span class="o">=</span> <span class="n">m</span><span class="p">(</span><span class="nb">input</span><span class="p">)</span>
</pre></div>
</div>
</dd></dl>
</div>
<div class="section" id="elu">
<h2>ELU<a class="headerlink" href="#elu" title="Permalink to this headline">¶</a></h2>
<dl class="class">
<dt id="torch.nn.quantized.ELU">
<em class="property">class </em><code class="sig-prename descclassname">torch.nn.quantized.</code><code class="sig-name descname">ELU</code><span class="sig-paren">(</span><em class="sig-param">scale</em>, <em class="sig-param">zero_point</em>, <em class="sig-param">alpha=1.0</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/modules/activation.html#ELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.ELU" title="Permalink to this definition">¶</a></dt>
<dd><p>This is the quantized equivalent of <a class="reference internal" href="generated/torch.nn.ELU.html#torch.nn.ELU" title="torch.nn.ELU"><code class="xref py py-class docutils literal notranslate"><span class="pre">ELU</span></code></a>.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>scale</strong> – quantization scale of the output tensor</p></li>
<li><p><strong>zero_point</strong> – quantization zero point of the output tensor</p></li>
<li><p><strong>alpha</strong> – the alpha constant</p></li>
</ul>
</dd>
</dl>
</dd></dl>
</div>
<div class="section" id="hardswish">
<h2>Hardswish<a class="headerlink" href="#hardswish" title="Permalink to this headline">¶</a></h2>
<dl class="class">
<dt id="torch.nn.quantized.Hardswish">
<em class="property">class </em><code class="sig-prename descclassname">torch.nn.quantized.</code><code class="sig-name descname">Hardswish</code><span class="sig-paren">(</span><em class="sig-param">scale</em>, <em class="sig-param">zero_point</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/modules/activation.html#Hardswish"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.Hardswish" title="Permalink to this definition">¶</a></dt>
<dd><p>This is the quantized version of <a class="reference internal" href="generated/torch.nn.Hardswish.html#torch.nn.Hardswish" title="torch.nn.Hardswish"><code class="xref py py-class docutils literal notranslate"><span class="pre">Hardswish</span></code></a>.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>scale</strong> – quantization scale of the output tensor</p></li>
<li><p><strong>zero_point</strong> – quantization zero point of the output tensor</p></li>
</ul>
</dd>
</dl>
</dd></dl>
</div>
<div class="section" id="conv1d">
<h2>Conv1d<a class="headerlink" href="#conv1d" title="Permalink to this headline">¶</a></h2>
<dl class="class">
<dt id="torch.nn.quantized.Conv1d">
<em class="property">class </em><code class="sig-prename descclassname">torch.nn.quantized.</code><code class="sig-name descname">Conv1d</code><span class="sig-paren">(</span><em class="sig-param">in_channels</em>, <em class="sig-param">out_channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">stride=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">bias=True</em>, <em class="sig-param">padding_mode='zeros'</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/nn/quantized/modules/conv.html#Conv1d"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.nn.quantized.Conv1d" title="Permalink to this definition">¶</a></dt>
<dd><p>Applies a 1D convolution over a quantized input signal composed of
several quantized input planes.</p>
<p>For details on input arguments, parameters, and implementation see
<a class="reference internal" href="generated/torch.nn.Conv1d.html#torch.nn.Conv1d" title="torch.nn.Conv1d"><code class="xref py py-class docutils literal notranslate"><span class="pre">Conv1d</span></code></a>.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only <cite>zeros</cite> is supported for the <code class="xref py py-attr docutils literal notranslate"><span class="pre">padding_mode</span></code> argument.</p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Only <cite>torch.quint8</cite> is supported for the input data type.</p>
</div>
<dl class="field-list simple">
<dt class="field-odd">Variables</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>~Conv1d.weight</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – packed tensor derived from the learnable weight
parameter.</p></li>
<li><p><strong>~Conv1d.scale</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – scalar for the output scale</p></li>
<li><p><strong>~Conv1d.zero_point</strong> (<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><em>Tensor</em></a>) – scalar for the output zero point</p></li>
</ul>
</dd>
</dl>
<p>See <a class="reference internal" href="generated/torch.nn.Conv1d.html#torch.nn.Conv1d" title="torch.nn.Conv1d"><code class="xref py py-class docutils literal notranslate"><span class="pre">Conv1d</span></code></a> for other attributes.</p>
<p>Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">m</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">quantized</span><span class="o">.</span><span class="n">Conv1d</span><span class="p">(</span><span class="mi">16</span><span class="p">,</span> <span class="mi">33</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="n">stride</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span>
<span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">20</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">100</span><span class="p">)</span>