forked from pytorch/cppdocs
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathtensor_creation.html
More file actions
926 lines (740 loc) · 60.1 KB
/
Copy pathtensor_creation.html
File metadata and controls
926 lines (740 loc) · 60.1 KB
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
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Tensor Creation API — PyTorch master documentation</title>
<link rel="canonical" href="https://pytorch.org/docs/stable/notes/tensor_creation.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="../_static/cpp_theme.css" type="text/css" />
<link rel="stylesheet" href="../_static/collapsible-lists/css/tree_view.css" type="text/css" />
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="Tensor CUDA Stream API" href="tensor_cuda_stream.html" />
<link rel="prev" title="Tensor Basics" href="tensor_basics.html" />
<script src="../_static/js/modernizr.min.js"></script>
<!-- Preload the theme fonts -->
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-book.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-Medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium-italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-SemiBold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<!-- Preload the katex fonts -->
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.2/css/all.css" integrity="sha384-vSIIfh2YWi9wW0r9iZe7RJPrKwp6bG+s9QZMoITbCckVJqGCCRhc+ccxNcdpHuYu" crossorigin="anonymous">
</head>
<div class="container-fluid header-holder tutorials-header" id="header-holder">
<div class="container">
<div class="header-container">
<a class="header-logo" href="https://pytorch.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/stable/index.html">
<span class="dropdown-title">PyTorch</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/audio/stable/index.html">
<span class="dropdown-title">torchaudio</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/text/stable/index.html">
<span class="dropdown-title">torchtext</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/vision/stable/index.html">
<span class="dropdown-title">torchvision</span>
<p></p>
</a>
<a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/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">
master
</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>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../installing.html">Installing C++ Distributions of PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../frontend.html">The C++ Frontend</a></li>
<li class="toctree-l1"><a class="reference internal" href="../api/library_root.html">Library API</a></li>
</ul>
<p class="caption"><span class="caption-text">Notes</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="faq.html">FAQ</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_basics.html">Tensor Basics</a></li>
<li class="toctree-l1 current"><a class="current reference internal" href="#">Tensor Creation API</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_cuda_stream.html">Tensor CUDA Stream API</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_indexing.html">Tensor Indexing API</a></li>
<li class="toctree-l1"><a class="reference internal" href="versioning.html">Library Versioning</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>Tensor Creation API</li>
<li class="pytorch-breadcrumbs-aside">
<a href="../_sources/notes/tensor_creation.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="tensor-creation-api">
<h1>Tensor Creation API<a class="headerlink" href="#tensor-creation-api" title="Permalink to this headline">¶</a></h1>
<p>This note describes how to create tensors in the PyTorch C++ API. It highlights
the available factory functions, which populate new tensors according to some
algorithm, and lists the options available to configure the shape, data type,
device and other properties of a new tensor.</p>
<div class="section" id="factory-functions">
<h2>Factory Functions<a class="headerlink" href="#factory-functions" title="Permalink to this headline">¶</a></h2>
<p>A <em>factory function</em> is a function that produces a new tensor. There are many
factory functions available in PyTorch (both in Python and C++), which differ
in the way they initialize a new tensor before returning it. All factory
functions adhere to the following general “schema”:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::<</span><span class="n">function</span><span class="o">-</span><span class="n">name</span><span class="o">></span><span class="p">(</span><span class="o"><</span><span class="n">function</span><span class="o">-</span><span class="n">specific</span><span class="o">-</span><span class="n">options</span><span class="o">></span><span class="p">,</span> <span class="o"><</span><span class="n">sizes</span><span class="o">></span><span class="p">,</span> <span class="o"><</span><span class="n">tensor</span><span class="o">-</span><span class="n">options</span><span class="o">></span><span class="p">)</span>
</pre></div>
</div>
<p>Let’s bisect the various parts of this “schema”:</p>
<ol class="arabic simple">
<li><p><code class="docutils literal notranslate"><span class="pre"><function-name></span></code> is the name of the function you would like to invoke,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre"><functions-specific-options></span></code> are any required or optional parameters a particular factory function accepts,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre"><sizes></span></code> is an object of type <code class="docutils literal notranslate"><span class="pre">IntArrayRef</span></code> and specifies the shape of the resulting tensor,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre"><tensor-options></span></code> is an instance of <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> and configures the data type, device, layout and other properties of the resulting tensor.</p></li>
</ol>
<div class="section" id="picking-a-factory-function">
<h3>Picking a Factory Function<a class="headerlink" href="#picking-a-factory-function" title="Permalink to this headline">¶</a></h3>
<p>The following factory functions are available at the time of this writing (the
hyperlinks lead to the corresponding Python functions, since they often have
more eloquent documentation – the options are the same in C++):</p>
<ul class="simple">
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.arange">arange</a>: Returns a tensor with a sequence of integers,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.empty">empty</a>: Returns a tensor with uninitialized values,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.eye">eye</a>: Returns an identity matrix,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.full">full</a>: Returns a tensor filled with a single value,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.linspace">linspace</a>: Returns a tensor with values linearly spaced in some interval,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.logspace">logspace</a>: Returns a tensor with values logarithmically spaced in some interval,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.ones">ones</a>: Returns a tensor filled with all ones,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.rand">rand</a>: Returns a tensor filled with values drawn from a uniform distribution on <code class="docutils literal notranslate"><span class="pre">[0,</span> <span class="pre">1)</span></code>.</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.randint">randint</a>: Returns a tensor with integers randomly drawn from an interval,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.randn">randn</a>: Returns a tensor filled with values drawn from a unit normal distribution,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.randperm">randperm</a>: Returns a tensor filled with a random permutation of integers in some interval,</p></li>
<li><p><a class="reference external" href="https://pytorch.org/docs/stable/torch.html#torch.zeros">zeros</a>: Returns a tensor filled with all zeros.</p></li>
</ul>
</div>
<div class="section" id="specifying-a-size">
<h3>Specifying a Size<a class="headerlink" href="#specifying-a-size" title="Permalink to this headline">¶</a></h3>
<p>Functions that do not require specific arguments by nature of how they fill the
tensor can be invoked with just a size. For example, the following line creates
a vector with 5 components, initially all set to 1:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">5</span><span class="p">);</span>
</pre></div>
</div>
<p>What if we wanted to instead create a <code class="docutils literal notranslate"><span class="pre">3</span> <span class="pre">x</span> <span class="pre">5</span></code> matrix, or a <code class="docutils literal notranslate"><span class="pre">2</span> <span class="pre">x</span> <span class="pre">3</span> <span class="pre">x</span> <span class="pre">4</span></code>
tensor? In general, an <code class="docutils literal notranslate"><span class="pre">IntArrayRef</span></code> – the type of the size parameter of factory
functions – is constructed by specifying the size along each dimension in
curly braces. For example, <code class="docutils literal notranslate"><span class="pre">{2,</span> <span class="pre">3}</span></code> for a tensor (in this case matrix) with
two rows and three columns, <code class="docutils literal notranslate"><span class="pre">{3,</span> <span class="pre">4,</span> <span class="pre">5}</span></code> for a three-dimensional tensor, and
<code class="docutils literal notranslate"><span class="pre">{2}</span></code> for a one-dimensional tensor with two components. In the one
dimensional case, you can omit the curly braces and just pass the single
integer like we did above. Note that the squiggly braces are just one way of
constructing an <code class="docutils literal notranslate"><span class="pre">IntArrayRef</span></code>. You can also pass an <code class="docutils literal notranslate"><span class="pre">std::vector<int64_t></span></code> and
a few other types. Either way, this means we can construct a three-dimensional
tensor filled with values from a unit normal distribution by writing:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</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">3</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="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">sizes</span><span class="p">()</span> <span class="o">==</span> <span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="kt">int64_t</span><span class="o">></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">5</span><span class="p">});</span>
</pre></div>
</div>
<p><code class="docutils literal notranslate"><span class="pre">tensor.sizes()</span></code> returns an <code class="docutils literal notranslate"><span class="pre">IntArrayRef</span></code> which can be compared against an
<code class="docutils literal notranslate"><span class="pre">std::vector<int64_t></span></code>, and we can see that it contains the sizes we passed
to the tensor. You can also write <code class="docutils literal notranslate"><span class="pre">tensor.size(i)</span></code> to access a single dimension,
which is equivalent to but preferred over <code class="docutils literal notranslate"><span class="pre">tensor.sizes()[i]</span></code>.</p>
</div>
<div class="section" id="passing-function-specific-parameters">
<h3>Passing Function-Specific Parameters<a class="headerlink" href="#passing-function-specific-parameters" title="Permalink to this headline">¶</a></h3>
<p>Neither <code class="docutils literal notranslate"><span class="pre">ones</span></code> nor <code class="docutils literal notranslate"><span class="pre">randn</span></code> accept any additional parameters to change their
behavior. One function which does require further configuration is <code class="docutils literal notranslate"><span class="pre">randint</span></code>,
which takes an upper bound on the value for the integers it generates, as well
as an optional lower bound, which defaults to zero. Here we create a <code class="docutils literal notranslate"><span class="pre">5</span> <span class="pre">x</span> <span class="pre">5</span></code>
square matrix with integers between 0 and 10:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">randint</span><span class="p">(</span><span class="cm">/*high=*/</span><span class="mi">10</span><span class="p">,</span> <span class="p">{</span><span class="mi">5</span><span class="p">,</span> <span class="mi">5</span><span class="p">});</span>
</pre></div>
</div>
<p>And here we raise the lower bound to 3:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">randint</span><span class="p">(</span><span class="cm">/*low=*/</span><span class="mi">3</span><span class="p">,</span> <span class="cm">/*high=*/</span><span class="mi">10</span><span class="p">,</span> <span class="p">{</span><span class="mi">5</span><span class="p">,</span> <span class="mi">5</span><span class="p">});</span>
</pre></div>
</div>
<p>The inline comments <code class="docutils literal notranslate"><span class="pre">/*low=*/</span></code> and <code class="docutils literal notranslate"><span class="pre">/*high=*/</span></code> are not required of course,
but aid readability just like keyword arguments in Python.</p>
<div class="admonition tip">
<p class="admonition-title">Tip</p>
<p>The main take-away is that the size always follows the function specific
arguments.</p>
</div>
<div class="admonition attention">
<p class="admonition-title">Attention</p>
<p>Sometimes a function does not need a size at all. For example, the size of
the tensor returned by <code class="docutils literal notranslate"><span class="pre">arange</span></code> is fully specified by its function-specific
arguments – the lower and upper bound of a range of integers. In that case
the function does not take a <code class="docutils literal notranslate"><span class="pre">size</span></code> parameter.</p>
</div>
</div>
<div class="section" id="configuring-properties-of-the-tensor">
<h3>Configuring Properties of the Tensor<a class="headerlink" href="#configuring-properties-of-the-tensor" title="Permalink to this headline">¶</a></h3>
<p>The previous section discussed function-specific arguments. Function-specific
arguments can only change the values with which tensors are filled, and
sometimes the size of the tensor. They never change things like the data type
(e.g. <code class="docutils literal notranslate"><span class="pre">float32</span></code> or <code class="docutils literal notranslate"><span class="pre">int64</span></code>) of the tensor being created, or whether it
lives in CPU or GPU memory. The specification of these properties is left to
the very last argument to every factory function: a <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object,
discussed below.</p>
<p><code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> is a class that encapsulates the construction axes of a
Tensor. With <em>construction axis</em> we mean a particular property of a Tensor that
can be configured before its construction (and sometimes changed afterwards).
These construction axes are:</p>
<ul class="simple">
<li><p>The <code class="docutils literal notranslate"><span class="pre">dtype</span></code> (previously “scalar type”), which controls the data type of the
elements stored in the tensor,</p></li>
<li><p>The <code class="docutils literal notranslate"><span class="pre">layout</span></code>, which is either strided (dense) or sparse,</p></li>
<li><p>The <code class="docutils literal notranslate"><span class="pre">device</span></code>, which represents a compute device on which a tensor is stored (like a CPU or CUDA GPU),</p></li>
<li><p>The <code class="docutils literal notranslate"><span class="pre">requires_grad</span></code> boolean to enable or disable gradient recording for a tensor,</p></li>
</ul>
<p>If you are used to PyTorch in Python, these axes will sound very familiar. The
allowed values for these axes at the moment are:</p>
<ul class="simple">
<li><p>For <code class="docutils literal notranslate"><span class="pre">dtype</span></code>: <code class="docutils literal notranslate"><span class="pre">kUInt8</span></code>, <code class="docutils literal notranslate"><span class="pre">kInt8</span></code>, <code class="docutils literal notranslate"><span class="pre">kInt16</span></code>, <code class="docutils literal notranslate"><span class="pre">kInt32</span></code>, <code class="docutils literal notranslate"><span class="pre">kInt64</span></code>, <code class="docutils literal notranslate"><span class="pre">kFloat32</span></code> and <code class="docutils literal notranslate"><span class="pre">kFloat64</span></code>,</p></li>
<li><p>For <code class="docutils literal notranslate"><span class="pre">layout</span></code>: <code class="docutils literal notranslate"><span class="pre">kStrided</span></code> and <code class="docutils literal notranslate"><span class="pre">kSparse</span></code>,</p></li>
<li><p>For <code class="docutils literal notranslate"><span class="pre">device</span></code>: Either <code class="docutils literal notranslate"><span class="pre">kCPU</span></code>, or <code class="docutils literal notranslate"><span class="pre">kCUDA</span></code> (which accepts an optional device index),</p></li>
<li><p>For <code class="docutils literal notranslate"><span class="pre">requires_grad</span></code>: either <code class="docutils literal notranslate"><span class="pre">true</span></code> or <code class="docutils literal notranslate"><span class="pre">false</span></code>.</p></li>
</ul>
<div class="admonition tip">
<p class="admonition-title">Tip</p>
<p>There exist “Rust-style” shorthands for dtypes, like <code class="docutils literal notranslate"><span class="pre">kF32</span></code> instead of
<code class="docutils literal notranslate"><span class="pre">kFloat32</span></code>. See <a class="reference external" href="https://github.com/pytorch/pytorch/blob/master/torch/csrc/api/include/torch/types.h">here</a>
for the full list.</p>
</div>
<p>An instance of <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> stores a concrete value for each of these
axes. Here is an example of creating a <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object that represents
a 64-bit float, strided tensor that requires a gradient, and lives on CUDA
device 1:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="k">auto</span> <span class="n">options</span> <span class="o">=</span>
<span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">()</span>
<span class="p">.</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">)</span>
<span class="p">.</span><span class="n">layout</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kStrided</span><span class="p">)</span>
<span class="p">.</span><span class="n">device</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span>
<span class="p">.</span><span class="n">requires_grad</span><span class="p">(</span><span class="nb">true</span><span class="p">);</span>
</pre></div>
</div>
<p>Notice how we use the ‘“builder”-style methods of <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> to
construct the object piece by piece. If we pass this object as the last
argument to a factory function, the newly created tensor will have these
properties:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">full</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="cm">/*value=*/</span><span class="mi">123</span><span class="p">,</span> <span class="n">options</span><span class="p">);</span>
<span class="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">dtype</span><span class="p">()</span> <span class="o">==</span> <span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">);</span>
<span class="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">layout</span><span class="p">()</span> <span class="o">==</span> <span class="n">torch</span><span class="o">::</span><span class="n">kStrided</span><span class="p">);</span>
<span class="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">device</span><span class="p">().</span><span class="n">type</span><span class="p">()</span> <span class="o">==</span> <span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">);</span> <span class="c1">// or device().is_cuda()</span>
<span class="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">device</span><span class="p">().</span><span class="n">index</span><span class="p">()</span> <span class="o">==</span> <span class="mi">1</span><span class="p">);</span>
<span class="n">assert</span><span class="p">(</span><span class="n">tensor</span><span class="p">.</span><span class="n">requires_grad</span><span class="p">());</span>
</pre></div>
</div>
<p>Now, you may be thinking: do I really need to specify each axis for every new
tensor I create? Fortunately, the answer is “no”, as <strong>every axis has a default
value</strong>. These defaults are:</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">kFloat32</span></code> for the dtype,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">kStrided</span></code> for the layout,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">kCPU</span></code> for the device,</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">false</span></code> for <code class="docutils literal notranslate"><span class="pre">requires_grad</span></code>.</p></li>
</ul>
<p>What this means is that any axis you omit during the construction of a
<code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object will take on its default value. For example, this is
our previous <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object, but with the <code class="docutils literal notranslate"><span class="pre">dtype</span></code> and <code class="docutils literal notranslate"><span class="pre">layout</span></code>
defaulted:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="k">auto</span> <span class="n">options</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">().</span><span class="n">device</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">,</span> <span class="mi">1</span><span class="p">).</span><span class="n">requires_grad</span><span class="p">(</span><span class="nb">true</span><span class="p">);</span>
</pre></div>
</div>
<p>In fact, we can even omit all axes to get an entirely defaulted
<code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="k">auto</span> <span class="n">options</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">();</span> <span class="c1">// or `torch::TensorOptions options;`</span>
</pre></div>
</div>
<p>A nice consequence of this is that the <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object we just spoke
so much about can be entirely omitted from any tensor factory call:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="c1">// A 32-bit float, strided, CPU tensor that does not require a gradient.</span>
<span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">tensor</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">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">});</span>
<span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">range</span> <span class="o">=</span> <span class="n">torch</span><span class="o">::</span><span class="n">arange</span><span class="p">(</span><span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">);</span>
</pre></div>
</div>
<p>But the sugar gets sweeter: In the API presented here so far, you may have
noticed that the initial <code class="docutils literal notranslate"><span class="pre">torch::TensorOptions()</span></code> is quite a mouthful to
write. The good news is that for every construction axis (dtype, layout, device
and <code class="docutils literal notranslate"><span class="pre">requires_grad</span></code>), there is one <em>free function</em> in the <code class="docutils literal notranslate"><span class="pre">torch::</span></code>
namespace which you can pass a value for that axis. Each function then returns
a <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> object preconfigured with that axis, but allowing even
further modification via the builder-style methods shown above. For example,</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">().</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">))</span>
</pre></div>
</div>
<p>is equivalent to</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">))</span>
</pre></div>
</div>
<p>and further instead of</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">().</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">).</span><span class="n">layout</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kStrided</span><span class="p">))</span>
</pre></div>
</div>
<p>we can just write</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">).</span><span class="n">layout</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kStrided</span><span class="p">))</span>
</pre></div>
</div>
<p>which saves us quite a bit of typing. What this means is that in practice, you
should barely, if ever, have to write out <code class="docutils literal notranslate"><span class="pre">torch::TensorOptions</span></code>. Instead use
the <code class="docutils literal notranslate"><span class="pre">torch::dtype()</span></code>, <code class="docutils literal notranslate"><span class="pre">torch::device()</span></code>, <code class="docutils literal notranslate"><span class="pre">torch::layout()</span></code> and
<code class="docutils literal notranslate"><span class="pre">torch::requires_grad()</span></code> functions.</p>
<p>A final bit of convenience is that <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> is implicitly
constructible from individual values. This means that whenever a function has a
parameter of type <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code>, like all factory functions do, we can
directly pass a value like <code class="docutils literal notranslate"><span class="pre">torch::kFloat32</span></code> or <code class="docutils literal notranslate"><span class="pre">torch::kStrided</span></code> in place
of the full object. Therefore, when there is only a single axis we would like
to change compared to its default value, we can pass only that value. As such,
what was</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">TensorOptions</span><span class="p">().</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">))</span>
</pre></div>
</div>
<p>became</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">))</span>
</pre></div>
</div>
<p>and can finally be shortened to</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">ones</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">)</span>
</pre></div>
</div>
<p>Of course, it is not possible to modify further properties of the
<code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> instance with this short syntax, but if all we needed was to
change one property, this is quite practical.</p>
<p>In conclusion, we can now compare how <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> defaults, together with
the abbreviated API for creating <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> using free functions, allow
tensor creation in C++ with the same convenience as in Python. Compare this
call in Python:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">.</span><span class="n">randn</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="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float32</span><span class="p">,</span> <span class="n">device</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">'cuda'</span><span class="p">,</span> <span class="mi">1</span><span class="p">),</span> <span class="n">requires_grad</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
</pre></div>
</div>
<p>with the equivalent call in C++:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">randn</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="n">torch</span><span class="o">::</span><span class="n">dtype</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">).</span><span class="n">device</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">,</span> <span class="mi">1</span><span class="p">).</span><span class="n">requires_grad</span><span class="p">(</span><span class="nb">true</span><span class="p">))</span>
</pre></div>
</div>
<p>Pretty close!</p>
</div>
</div>
<div class="section" id="conversion">
<h2>Conversion<a class="headerlink" href="#conversion" title="Permalink to this headline">¶</a></h2>
<p>Just as we can use <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> to configure how new tensors should be
created, we can also use <code class="docutils literal notranslate"><span class="pre">TensorOptions</span></code> to convert a tensor from one set of
properties to a new set of properties. Such a conversion usually creates a new
tensor and does not occur in-place. For example, if we have a <code class="docutils literal notranslate"><span class="pre">source_tensor</span></code>
created with</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">source_tensor</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="mi">3</span><span class="p">},</span> <span class="n">torch</span><span class="o">::</span><span class="n">kInt64</span><span class="p">);</span>
</pre></div>
</div>
<p>we can convert it from <code class="docutils literal notranslate"><span class="pre">int64</span></code> to <code class="docutils literal notranslate"><span class="pre">float32</span></code>:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">float_tensor</span> <span class="o">=</span> <span class="n">source_tensor</span><span class="p">.</span><span class="n">to</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kFloat32</span><span class="p">);</span>
</pre></div>
</div>
<div class="admonition attention">
<p class="admonition-title">Attention</p>
<p>The result of the conversion, <code class="docutils literal notranslate"><span class="pre">float_tensor</span></code>, is a new tensor pointing to
new memory, unrelated to the source <code class="docutils literal notranslate"><span class="pre">source_tensor</span></code>.</p>
</div>
<p>We can then move it from CPU memory to GPU memory:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">gpu_tensor</span> <span class="o">=</span> <span class="n">float_tensor</span><span class="p">.</span><span class="n">to</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">);</span>
</pre></div>
</div>
<p>If you have multiple CUDA devices available, the above code will copy the
tensor to the <em>default</em> CUDA device, which you can configure with a
<code class="docutils literal notranslate"><span class="pre">torch::DeviceGuard</span></code>. If no <code class="docutils literal notranslate"><span class="pre">DeviceGuard</span></code> is in place, this will be GPU
1. If you would like to specify a different GPU index, you can pass it to
the <code class="docutils literal notranslate"><span class="pre">Device</span></code> constructor:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">gpu_two_tensor</span> <span class="o">=</span> <span class="n">float_tensor</span><span class="p">.</span><span class="n">to</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">Device</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCUDA</span><span class="p">,</span> <span class="mi">1</span><span class="p">));</span>
</pre></div>
</div>
<p>In the case of CPU to GPU copy and reverse, we can also configure the memory
copy to be <em>asynchronous</em> by passing <code class="docutils literal notranslate"><span class="pre">/*non_blocking=*/false</span></code> as the last
argument to <code class="docutils literal notranslate"><span class="pre">to()</span></code>:</p>
<div class="highlight-cpp notranslate"><div class="highlight"><pre><span></span><span class="n">torch</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">async_cpu_tensor</span> <span class="o">=</span> <span class="n">gpu_tensor</span><span class="p">.</span><span class="n">to</span><span class="p">(</span><span class="n">torch</span><span class="o">::</span><span class="n">kCPU</span><span class="p">,</span> <span class="cm">/*non_blocking=*/</span><span class="nb">true</span><span class="p">);</span>
</pre></div>
</div>
</div>
<div class="section" id="conclusion">
<h2>Conclusion<a class="headerlink" href="#conclusion" title="Permalink to this headline">¶</a></h2>
<p>This note hopefully gave you a good understanding of how to create and convert
tensors in an idiomatic fashion using the PyTorch C++ API. If you have any
further questions or suggestions, please use our <a class="reference external" href="https://discuss.pytorch.org/">forum</a> or <a class="reference external" href="https://github.com/pytorch/pytorch/issues">GitHub issues</a> to get in touch.</p>
</div>
</div>
</article>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="tensor_cuda_stream.html" class="btn btn-neutral float-right" title="Tensor CUDA Stream API" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
<a href="tensor_basics.html" class="btn btn-neutral" title="Tensor Basics" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
</div>
<hr>
<div role="contentinfo">
<p>
© Copyright 2019, Torch Contributors.
</p>
</div>
<div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</div>
</footer>
</div>
</div>
<div class="pytorch-content-right" id="pytorch-content-right">
<div class="pytorch-right-menu" id="pytorch-right-menu">
<div class="pytorch-side-scroll" id="pytorch-side-scroll-right">
<ul>
<li><a class="reference internal" href="#">Tensor Creation API</a><ul>
<li><a class="reference internal" href="#factory-functions">Factory Functions</a><ul>
<li><a class="reference internal" href="#picking-a-factory-function">Picking a Factory Function</a></li>
<li><a class="reference internal" href="#specifying-a-size">Specifying a Size</a></li>
<li><a class="reference internal" href="#passing-function-specific-parameters">Passing Function-Specific Parameters</a></li>
<li><a class="reference internal" href="#configuring-properties-of-the-tensor">Configuring Properties of the Tensor</a></li>
</ul>
</li>
<li><a class="reference internal" href="#conversion">Conversion</a></li>
<li><a class="reference internal" href="#conclusion">Conclusion</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
</section>
</div>
<script type="text/javascript" id="documentation_options" data-url_root="../" src="../_static/documentation_options.js"></script>
<script src="../_static/jquery.js"></script>
<script src="../_static/underscore.js"></script>
<script src="../_static/doctools.js"></script>
<script src="../_static/language_data.js"></script>
<script src="../_static/collapsible-lists/js/CollapsibleLists.compressed.js"></script>
<script src="../_static/collapsible-lists/js/apply-collapsible-lists.js"></script>
<script type="text/javascript" src="../_static/js/vendor/popper.min.js"></script>
<script type="text/javascript" src="../_static/js/vendor/bootstrap.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/1.5.0/list.min.js"></script>
<script type="text/javascript" src="../_static/js/theme.js"></script>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
<!-- Begin Footer -->
<div class="container-fluid docs-tutorials-resources" id="docs-tutorials-resources">
<div class="container">
<div class="row">
<div class="col-md-4 text-center">
<h2>Docs</h2>
<p>Access comprehensive developer documentation for PyTorch</p>
<a class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html">View Docs</a>
</div>
<div class="col-md-4 text-center">
<h2>Tutorials</h2>
<p>Get in-depth tutorials for beginners and advanced developers</p>
<a class="with-right-arrow" href="https://pytorch.org/tutorials">View Tutorials</a>
</div>
<div class="col-md-4 text-center">
<h2>Resources</h2>
<p>Find development resources and get your questions answered</p>
<a class="with-right-arrow" href="https://pytorch.org/resources">View Resources</a>
</div>
</div>
</div>
</div>
<footer class="site-footer">
<div class="container footer-container">
<div class="footer-logo-wrapper">
<a href="https://pytorch.org/" class="footer-logo"></a>
</div>
<div class="footer-links-wrapper">
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/">PyTorch</a></li>
<li><a href="https://pytorch.org/get-started">Get Started</a></li>
<li><a href="https://pytorch.org/features">Features</a></li>
<li><a href="https://pytorch.org/ecosystem">Ecosystem</a></li>
<li><a href="https://pytorch.org/blog/">Blog</a></li>
<li><a href="https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md">Contributing</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/resources">Resources</a></li>
<li><a href="https://pytorch.org/tutorials">Tutorials</a></li>
<li><a href="https://pytorch.org/docs/stable/index.html">Docs</a></li>
<li><a href="https://discuss.pytorch.org" target="_blank">Discuss</a></li>
<li><a href="https://github.com/pytorch/pytorch/issues" target="_blank">Github Issues</a></li>
<li><a href="https://pytorch.org/assets/brand-guidelines/PyTorch-Brand-Guidelines.pdf" target="_blank">Brand Guidelines</a></li>
</ul>
</div>
<div class="footer-links-col follow-us-col">
<ul>
<li class="list-title">Stay Connected</li>
<li>
<div id="mc_embed_signup">
<form
action="https://twitter.us14.list-manage.com/subscribe/post?u=75419c71fe0a935e53dfa4a3f&id=91d0dccd39"
method="post"
id="mc-embedded-subscribe-form"
name="mc-embedded-subscribe-form"
class="email-subscribe-form validate"
target="_blank"
novalidate>
<div id="mc_embed_signup_scroll" class="email-subscribe-form-fields-wrapper">
<div class="mc-field-group">
<label for="mce-EMAIL" style="display:none;">Email Address</label>
<input type="email" value="" name="EMAIL" class="required email" id="mce-EMAIL" placeholder="Email Address">
</div>
<div id="mce-responses" class="clear">
<div class="response" id="mce-error-response" style="display:none"></div>
<div class="response" id="mce-success-response" style="display:none"></div>
</div> <!-- real people should not fill this in and expect good things - do not remove this or risk form bot signups-->
<div style="position: absolute; left: -5000px;" aria-hidden="true"><input type="text" name="b_75419c71fe0a935e53dfa4a3f_91d0dccd39" tabindex="-1" value=""></div>
<div class="clear">
<input type="submit" value="" name="subscribe" id="mc-embedded-subscribe" class="button email-subscribe-button">
</div>
</div>
</form>
</div>
</li>
</ul>
<div class="footer-social-icons">
<a href="https://www.facebook.com/pytorch" target="_blank" class="facebook"></a>
<a href="https://twitter.com/pytorch" target="_blank" class="twitter"></a>
<a href="https://www.youtube.com/pytorch" target="_blank" class="youtube"></a>
</div>
</div>
</div>
</div>
</footer>
<div class="cookie-banner-wrapper">
<div class="container">
<p class="gdpr-notice">To analyze traffic and optimize your experience, we serve cookies on this site. By clicking or navigating, you agree to allow our usage of cookies. As the current maintainers of this site, Facebook’s Cookies Policy applies. Learn more, including about available controls: <a href="https://www.facebook.com/policies/cookies/">Cookies Policy</a>.</p>
<img class="close-button" src="../_static/images/pytorch-x.svg">
</div>
</div>
<!-- End Footer -->
<!-- Begin Mobile Menu -->
<div class="mobile-main-menu">
<div class="container-fluid">
<div class="container">
<div class="mobile-main-menu-header-container">
<a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
<a class="main-menu-close-button" href="#" data-behavior="close-mobile-menu"></a>
</div>
</div>
</div>
<div class="mobile-main-menu-links-container">
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.org/get-started">Get Started</a>
</li>
<li>
<a href="https://pytorch.org/ecosystem">Ecosystem</a>
</li>
<li>
<a href="https://pytorch.org/mobile">Mobile</a>
</li>
<li>
<a href="https://pytorch.org/hub">PyTorch Hub</a>
</li>
<li>
<a href="https://pytorch.org/blog/">Blog</a>
</li>
<li>
<a href="https://pytorch.org/tutorials">Tutorials</a>
</li>
<li class="resources-mobile-menu-title" class="active">
Docs
</li>
<ul class="resources-mobile-menu-items">
<li>
<a href="https://pytorch.org/docs/stable/index.html">PyTorch</a>
</li>
<li>
<a href="https://pytorch.org/audio/stable/index.html">torchaudio</a>
</li>
<li>
<a href="https://pytorch.org/text/stable/index.html">torchtext</a>
</li>
<li>
<a href="https://pytorch.org/vision/stable/index.html">torchvision</a>
</li>
<li>
<a href="https://pytorch.org/elastic/">TorchElastic</a>
</li>
<li>
<a href="https://pytorch.org/serve/">TorchServe</a>
</li>
<li>
<a href="https://pytorch.org/xla">PyTorch on XLA Devices</a>
</li>
</ul>
<li class="resources-mobile-menu-title">
Resources
</li>
<ul class="resources-mobile-menu-items">
<li>
<a href="https://pytorch.org/resources">Developer Resources</a>
</li>
<li>
<a href="https://pytorch.org/features">About</a>
</li>
<li>
<a href="https://pytorch.org/hub">Models (Beta)</a>
</li>
<li>
<a href="https://pytorch.org/#community-module">Community</a>
</li>
<li>
<a href="https://discuss.pytorch.org/">Forums</a>
</li>
</ul>
<li>
<a href="https://github.com/pytorch/pytorch">Github</a>
</li>
</ul>
</div>
</div>
</div>
<!-- End Mobile Menu -->
<script type="text/javascript" src="../_static/js/vendor/anchor.min.js"></script>
<script type="text/javascript">
$(document).ready(function() {
mobileMenu.bind();
mobileTOC.bind();
pytorchAnchors.bind();
sideMenus.bind();
scrollToAnchor.bind();
highlightNavigation.bind();
mainMenuDropdown.bind();
filterTags.bind();
// Add class to links that have code blocks, since we cannot create links in code blocks
$("article.pytorch-article a span.pre").each(function(e) {
$(this).closest("a").addClass("has-code");
});
})
</script>
</body>
</html>