forked from Xilinx/brevitas
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbrevitas.function.html
667 lines (522 loc) · 43.2 KB
/
brevitas.function.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
<!DOCTYPE html>
<html class="writer-html5" lang="en" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>brevitas.function package — Brevitas 0.2.0b1.dev0+gc32fb31.d20210127 documentation</title>
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<!--[if lt IE 9]>
<script src="_static/js/html5shiv.min.js"></script>
<![endif]-->
<script type="text/javascript" id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
<script src="_static/jquery.js"></script>
<script src="_static/underscore.js"></script>
<script src="_static/doctools.js"></script>
<script type="text/javascript" src="_static/js/theme.js"></script>
<link rel="index" title="Index" href="genindex.html" />
<link rel="search" title="Search" href="search.html" />
</head>
<body class="wy-body-for-nav">
<div class="wy-grid-for-nav">
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
<div class="wy-side-scroll">
<div class="wy-side-nav-search" >
<a href="index.html" class="icon icon-home"> Brevitas
</a>
<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>
<div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<!-- Local TOC -->
<div class="local-toc"><ul>
<li><a class="reference internal" href="#">brevitas.function package</a><ul>
<li><a class="reference internal" href="#submodules">Submodules</a></li>
<li><a class="reference internal" href="#module-brevitas.function.autograd_ste_ops">brevitas.function.autograd_ste_ops module</a></li>
<li><a class="reference internal" href="#module-brevitas.function.ops">brevitas.function.ops module</a></li>
<li><a class="reference internal" href="#module-brevitas.function.ops_ste">brevitas.function.ops_ste module</a></li>
<li><a class="reference internal" href="#module-brevitas.function.shape">brevitas.function.shape module</a></li>
<li><a class="reference internal" href="#module-brevitas.function">Module contents</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
</nav>
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
<nav class="wy-nav-top" aria-label="top navigation">
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
<a href="index.html">Brevitas</a>
</nav>
<div class="wy-nav-content">
<div class="rst-content">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="wy-breadcrumbs">
<li><a href="index.html" class="icon icon-home"></a> »</li>
<li>brevitas.function package</li>
<li class="wy-breadcrumbs-aside">
<a href="_sources/brevitas.function.rst.txt" rel="nofollow"> View page source</a>
</li>
</ul>
<hr/>
</div>
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<div class="section" id="brevitas-function-package">
<h1>brevitas.function package<a class="headerlink" href="#brevitas-function-package" title="Permalink to this headline">¶</a></h1>
<div class="section" id="submodules">
<h2>Submodules<a class="headerlink" href="#submodules" title="Permalink to this headline">¶</a></h2>
</div>
<div class="section" id="module-brevitas.function.autograd_ste_ops">
<span id="brevitas-function-autograd-ste-ops-module"></span><h2>brevitas.function.autograd_ste_ops module<a class="headerlink" href="#module-brevitas.function.autograd_ste_ops" title="Permalink to this headline">¶</a></h2>
<p>Implementation of various torch.autograd.Function with straight-through estimators.</p>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.abs_binary_sign_grad_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">abs_binary_sign_grad_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.abs_binary_sign_grad_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.binary_sign_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">binary_sign_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.binary_sign_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.ceil_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">ceil_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.ceil_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.floor_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">floor_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.floor_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.round_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">round_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.round_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.round_to_zero_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">round_to_zero_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.round_to_zero_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.scalar_clamp_min_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">scalar_clamp_min_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.scalar_clamp_min_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.scalar_clamp_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">scalar_clamp_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.scalar_clamp_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.tensor_clamp_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">tensor_clamp_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.tensor_clamp_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
<dl class="py function">
<dt id="brevitas.function.autograd_ste_ops.ternary_sign_ste_impl">
<code class="sig-prename descclassname">brevitas.function.autograd_ste_ops.</code><code class="sig-name descname">ternary_sign_ste_impl</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.autograd_ste_ops.ternary_sign_ste_impl" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>
</div>
<div class="section" id="module-brevitas.function.ops">
<span id="brevitas-function-ops-module"></span><h2>brevitas.function.ops module<a class="headerlink" href="#module-brevitas.function.ops" title="Permalink to this headline">¶</a></h2>
<p>Implementation of various core operations often performed as part of quantization.
The implemented functions adheres to the restriction imposed by Pytorch 1.1.0’s TorchScript compiler.</p>
<dl class="py function">
<dt id="brevitas.function.ops.binary_sign">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">binary_sign</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.binary_sign" title="Permalink to this definition">¶</a></dt>
<dd><p>Computes the 2-valued sign of an input tensor.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – input tensor.</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>the 2-valued sign tensor of the input tensor.</p>
</dd>
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">binary_sign</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([</span><span class="mf">2.1</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]))</span>
<span class="go">tensor([ 1., -1., 1.])</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.identity">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">identity</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.identity" title="Permalink to this definition">¶</a></dt>
<dd><p>Identity function.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – Input Tensor</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>THe input tensor x</p>
</dd>
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">identity</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mf">1.7</span><span class="p">))</span>
<span class="go">tensor(1.7)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.max_int">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">max_int</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">signed</span></em>, <em class="sig-param"><span class="n">narrow_range</span></em>, <em class="sig-param"><span class="n">bit_width</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.max_int" title="Permalink to this definition">¶</a></dt>
<dd><p>Compute the maximum integer representable by a given number of bits.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>signed</strong> (<em>bool</em>) – Indicates whether the represented integer is signed or not.</p></li>
<li><p><strong>narrow_range</strong> (<em>bool</em>) – Indicates whether to narrow the maximum unsigned value represented by 1.</p></li>
<li><p><strong>bit_width</strong> (<em>Tensor</em>) – Number of bits available for the representation.</p></li>
</ul>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>Maximum integer that can be represented according to the input arguments.</p>
</dd>
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">max_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(127)</span>
<span class="gp">>>> </span><span class="n">max_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(254)</span>
<span class="gp">>>> </span><span class="n">max_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(127)</span>
<span class="gp">>>> </span><span class="n">max_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(255)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.min_int">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">min_int</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">signed</span></em>, <em class="sig-param"><span class="n">narrow_range</span></em>, <em class="sig-param"><span class="n">bit_width</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.min_int" title="Permalink to this definition">¶</a></dt>
<dd><p>Compute the minimum integer representable by a given number of bits.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>signed</strong> (<em>bool</em>) – Indicates whether the represented integer is signed or not.</p></li>
<li><p><strong>narrow_range</strong> (<em>bool</em>) – Indicates whether to narrow the minimum value represented by 1.</p></li>
<li><p><strong>bit_width</strong> (<em>Tensor</em>) – Number of bits available for the representation.</p></li>
</ul>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>Maximum unsigned integer that can be represented according to the input arguments.</p>
</dd>
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">min_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(-127)</span>
<span class="gp">>>> </span><span class="n">min_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(0)</span>
<span class="gp">>>> </span><span class="n">min_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(-128)</span>
<span class="gp">>>> </span><span class="n">min_int</span><span class="p">(</span><span class="n">signed</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">narrow_range</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="n">bit_width</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">8</span><span class="p">))</span>
<span class="go">tensor(0)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.round_to_zero">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">round_to_zero</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.round_to_zero" title="Permalink to this definition">¶</a></dt>
<dd><p>Compute rounding towards zero.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – input tensor.</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>rounded input tensor.</p>
</dd>
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">round_to_zero</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([</span><span class="o">-</span><span class="mf">1.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">]))</span>
<span class="go">tensor([-1., -0., 0., 1.])</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.tensor_clamp">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">tensor_clamp</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em>, <em class="sig-param"><span class="n">min_val</span></em>, <em class="sig-param"><span class="n">max_val</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.tensor_clamp" title="Permalink to this definition">¶</a></dt>
<dd><p>Generalized clamp function with support for tensors as clamping values.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>x</strong> (<em>Tensor</em>) – Input on which to apply the clamp operation</p></li>
<li><p><strong>min_val</strong> (<em>Tensor</em>) – Minimum values for the clamp operation.</p></li>
<li><p><strong>max_val</strong> (<em>Tensor</em>) – Maximum values for the clamp operation.</p></li>
</ul>
</dd>
</dl>
<p class="rubric">Notes</p>
<p>x, min_val, max_val need to be broadcastable.</p>
<dl class="field-list simple">
<dt class="field-odd">Returns</dt>
<dd class="field-odd"><p>Input <cite>x</cite> clamped between the provided minimum and maximum tensors.</p>
</dd>
<dt class="field-even">Return type</dt>
<dd class="field-even"><p>Tensor</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">tensor_clamp</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([</span><span class="mf">1.7</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]),</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mf">0.0</span><span class="p">),</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mf">1.0</span><span class="p">))</span>
<span class="go">tensor([1.0000, 0.0000, 0.1000])</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops.tensor_clamp_">
<code class="sig-prename descclassname">brevitas.function.ops.</code><code class="sig-name descname">tensor_clamp_</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em>, <em class="sig-param"><span class="n">min_val</span></em>, <em class="sig-param"><span class="n">max_val</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops.tensor_clamp_" title="Permalink to this definition">¶</a></dt>
<dd><p>In-place variant of <a class="reference internal" href="#brevitas.function.ops.tensor_clamp" title="brevitas.function.ops.tensor_clamp"><code class="xref py py-func docutils literal notranslate"><span class="pre">tensor_clamp()</span></code></a>.</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
</div>
<div class="section" id="module-brevitas.function.ops_ste">
<span id="brevitas-function-ops-ste-module"></span><h2>brevitas.function.ops_ste module<a class="headerlink" href="#module-brevitas.function.ops_ste" title="Permalink to this headline">¶</a></h2>
<p>Implementation of various functions with a straight-through gradient estimator, dispatched to
either a native just-in-time compiled backend (when env BREVITAS_JIT=1) or to a torch.autograd.Function
implemented in <a class="reference internal" href="#module-brevitas.function.autograd_ste_ops" title="brevitas.function.autograd_ste_ops"><code class="xref py py-obj docutils literal notranslate"><span class="pre">autograd_ste_ops</span></code></a> (when env BREVITAS_JIT=0).
The native backend is enabled when BREVITAS_JIT is enabled to allow for end-to-end compilation of
the built-in quantizers, since as of Pytorch 1.7.0 a torch.autograd.Function is not supported by the compiler.</p>
<dl class="py function">
<dt id="brevitas.function.ops_ste.abs_binary_sign_grad">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">abs_binary_sign_grad</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.abs_binary_sign_grad" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.abs_binary_sign_grad_impl" title="brevitas.function.autograd_ste_ops.abs_binary_sign_grad_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">abs_binary_sign_grad_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.binary_sign_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">binary_sign_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.binary_sign_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.binary_sign_ste_impl" title="brevitas.function.autograd_ste_ops.binary_sign_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">binary_sign_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.ceil_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">ceil_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.ceil_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.ceil_ste_impl" title="brevitas.function.autograd_ste_ops.ceil_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">ceil_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.floor_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">floor_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.floor_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.floor_ste_impl" title="brevitas.function.autograd_ste_ops.floor_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">floor_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.round_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">round_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.round_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.round_ste_impl" title="brevitas.function.autograd_ste_ops.round_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">round_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.round_to_zero_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">round_to_zero_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.round_to_zero_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.round_to_zero_ste_impl" title="brevitas.function.autograd_ste_ops.round_to_zero_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">round_to_zero_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.scalar_clamp_min_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">scalar_clamp_min_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em>, <em class="sig-param"><span class="n">min_val</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.scalar_clamp_min_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.scalar_clamp_min_ste_impl" title="brevitas.function.autograd_ste_ops.scalar_clamp_min_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">scalar_clamp_min_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.scalar_clamp_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">scalar_clamp_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em>, <em class="sig-param"><span class="n">min_val</span></em>, <em class="sig-param"><span class="n">max_val</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.scalar_clamp_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.scalar_clamp_ste_impl" title="brevitas.function.autograd_ste_ops.scalar_clamp_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">scalar_clamp_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.tensor_clamp_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">tensor_clamp_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em>, <em class="sig-param"><span class="n">min_val</span></em>, <em class="sig-param"><span class="n">max_val</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.tensor_clamp_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.tensor_clamp_ste_impl" title="brevitas.function.autograd_ste_ops.tensor_clamp_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">tensor_clamp_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.ops_ste.ternary_sign_ste">
<code class="sig-prename descclassname">brevitas.function.ops_ste.</code><code class="sig-name descname">ternary_sign_ste</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.ops_ste.ternary_sign_ste" title="Permalink to this definition">¶</a></dt>
<dd><p>Wrapper for either <a class="reference internal" href="#brevitas.function.autograd_ste_ops.ternary_sign_ste_impl" title="brevitas.function.autograd_ste_ops.ternary_sign_ste_impl"><code class="xref py py-func docutils literal notranslate"><span class="pre">ternary_sign_ste_impl()</span></code></a> (with env
BREVITAS_JIT=0) or its native just-in-time compiled variant (with BREVITAS_JIT=1).</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">Tensor</span></code></p>
</dd>
</dl>
</dd></dl>
</div>
<div class="section" id="module-brevitas.function.shape">
<span id="brevitas-function-shape-module"></span><h2>brevitas.function.shape module<a class="headerlink" href="#module-brevitas.function.shape" title="Permalink to this headline">¶</a></h2>
<p>Implementation of various functions to compute shapes that induce flattening along certain
dimensions of a tensor.</p>
<dl class="py function">
<dt id="brevitas.function.shape.over_batch_over_output_channels">
<code class="sig-prename descclassname">brevitas.function.shape.</code><code class="sig-name descname">over_batch_over_output_channels</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.shape.over_batch_over_output_channels" title="Permalink to this definition">¶</a></dt>
<dd><p>Returns a shape s such that x.view(s) is a 3-dim tensor with batches
at dimension 0, output channels at dimension 1, and any other feature at dimension 2.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – Input tensor with batches at dimension 0 and output channels at dimension 1.</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>A tuple containing the 3-dim shape.</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">over_batch_over_output_channels</span><span class="p">(</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="mi">3</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="go">(2, 3, -1)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.shape.over_batch_over_tensor">
<code class="sig-prename descclassname">brevitas.function.shape.</code><code class="sig-name descname">over_batch_over_tensor</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.shape.over_batch_over_tensor" title="Permalink to this definition">¶</a></dt>
<dd><p>Computes the shape s such that x.view(s) is a 2-dim tensor with batches
at dimension 0 and any other feature at dimension 1.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – Input tensor with batches at dimension 0.</p>
</dd>
<dt class="field-even">Return type</dt>
<dd class="field-even"><p><code class="xref py py-data docutils literal notranslate"><span class="pre">Tuple</span></code>[<code class="xref py py-class docutils literal notranslate"><span class="pre">int</span></code>, <code class="xref py py-class docutils literal notranslate"><span class="pre">int</span></code>]</p>
</dd>
<dt class="field-odd">Returns</dt>
<dd class="field-odd"><p>A tuple containing the 2-dim shape.</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">over_batch_over_tensor</span><span class="p">(</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="mi">3</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="go">(2, -1)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.shape.over_output_channels">
<code class="sig-prename descclassname">brevitas.function.shape.</code><code class="sig-name descname">over_output_channels</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.shape.over_output_channels" title="Permalink to this definition">¶</a></dt>
<dd><p>Computes the shape s such that x.view(s) is a 2-dim tensor with output channels
at dimension 0 and any other feature at dimension 1.</p>
<p>Args:
x (Tensor): Input tensor with output channels at dimension 0.</p>
<dl class="field-list simple">
<dt class="field-odd">Return type</dt>
<dd class="field-odd"><p><code class="xref py py-data docutils literal notranslate"><span class="pre">Tuple</span></code>[<code class="xref py py-class docutils literal notranslate"><span class="pre">int</span></code>, <code class="xref py py-class docutils literal notranslate"><span class="pre">int</span></code>]</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>A tuple containing the 2-dim shape.</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">over_output_channels</span><span class="p">(</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="mi">3</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="go">(2, -1)</span>
</pre></div>
</div>
</dd></dl>
<dl class="py function">
<dt id="brevitas.function.shape.over_tensor">
<code class="sig-prename descclassname">brevitas.function.shape.</code><code class="sig-name descname">over_tensor</code><span class="sig-paren">(</span><em class="sig-param"><span class="n">x</span></em><span class="sig-paren">)</span><a class="headerlink" href="#brevitas.function.shape.over_tensor" title="Permalink to this definition">¶</a></dt>
<dd><p>Computes the shape s such that x.view(s) is a flat tensor.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>x</strong> (<em>Tensor</em>) – Input tensor.</p>
</dd>
<dt class="field-even">Return type</dt>
<dd class="field-even"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">int</span></code></p>
</dd>
<dt class="field-odd">Returns</dt>
<dd class="field-odd"><p>The number -1 corresponding to a flat shape.</p>
</dd>
</dl>
<p class="rubric">Examples</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">over_tensor</span><span class="p">(</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="mi">3</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="go">-1</span>
</pre></div>
</div>
</dd></dl>
</div>
<div class="section" id="module-brevitas.function">
<span id="module-contents"></span><h2>Module contents<a class="headerlink" href="#module-brevitas.function" title="Permalink to this headline">¶</a></h2>
</div>
</div>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<p>
© Copyright 2021 - Xilinx, Inc.
</p>
</div>
Built with <a href="https://www.sphinx-doc.org/">Sphinx</a> using a
<a href="https://github.com/readthedocs/sphinx_rtd_theme">theme</a>
provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
</div>
</div>
</section>
</div>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
</body>
</html>