-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy path11-RNN.html
More file actions
714 lines (672 loc) · 88.6 KB
/
11-RNN.html
File metadata and controls
714 lines (672 loc) · 88.6 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
<!doctype html>
<html class="no-js" lang="en" data-content_root="./">
<head><meta charset="utf-8"/>
<meta name="viewport" content="width=device-width,initial-scale=1"/>
<meta name="color-scheme" content="light dark"><meta name="viewport" content="width=device-width, initial-scale=1" />
<link rel="index" title="Index" href="genindex.html" /><link rel="search" title="Search" href="search.html" /><link rel="next" title="Memory" href="12-Memory.html" /><link rel="prev" title="Experiment Tracking Interface Specification" href="10-Experiment%20Tracking.html" />
<!-- Generated with Sphinx 8.1.3 and Furo 2024.08.06 -->
<title>Recurrent Neural Networks (RNNs) - RLtools Documentation</title>
<link rel="stylesheet" type="text/css" href="_static/pygments.css?v=8f2a1f02" />
<link rel="stylesheet" type="text/css" href="_static/styles/furo.css?v=354aac6f" />
<link rel="stylesheet" type="text/css" href="_static/nbsphinx-code-cells.css?v=2aa19091" />
<link rel="stylesheet" type="text/css" href="_static/styles/furo-extensions.css?v=302659d7" />
<link rel="stylesheet" type="text/css" href="_static/overrides.css?v=ca018f06" />
<style>
body {
--color-code-background: #f8f8f8;
--color-code-foreground: black;
--color-brand-primary: #639694;
--color-brand-content: #639694;
--color-admonition-background: orange;
--sidebar_hide_name: True;
}
@media not print {
body[data-theme="dark"] {
--color-code-background: #202020;
--color-code-foreground: #d0d0d0;
}
@media (prefers-color-scheme: dark) {
body:not([data-theme="light"]) {
--color-code-background: #202020;
--color-code-foreground: #d0d0d0;
}
}
}
</style></head>
<body>
<script>
document.body.dataset.theme = localStorage.getItem("theme") || "auto";
</script>
<svg xmlns="http://www.w3.org/2000/svg" style="display: none;">
<symbol id="svg-toc" viewBox="0 0 24 24">
<title>Contents</title>
<svg stroke="currentColor" fill="currentColor" stroke-width="0" viewBox="0 0 1024 1024">
<path d="M408 442h480c4.4 0 8-3.6 8-8v-56c0-4.4-3.6-8-8-8H408c-4.4 0-8 3.6-8 8v56c0 4.4 3.6 8 8 8zm-8 204c0 4.4 3.6 8 8 8h480c4.4 0 8-3.6 8-8v-56c0-4.4-3.6-8-8-8H408c-4.4 0-8 3.6-8 8v56zm504-486H120c-4.4 0-8 3.6-8 8v56c0 4.4 3.6 8 8 8h784c4.4 0 8-3.6 8-8v-56c0-4.4-3.6-8-8-8zm0 632H120c-4.4 0-8 3.6-8 8v56c0 4.4 3.6 8 8 8h784c4.4 0 8-3.6 8-8v-56c0-4.4-3.6-8-8-8zM115.4 518.9L271.7 642c5.8 4.6 14.4.5 14.4-6.9V388.9c0-7.4-8.5-11.5-14.4-6.9L115.4 505.1a8.74 8.74 0 0 0 0 13.8z"/>
</svg>
</symbol>
<symbol id="svg-menu" viewBox="0 0 24 24">
<title>Menu</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="feather-menu">
<line x1="3" y1="12" x2="21" y2="12"></line>
<line x1="3" y1="6" x2="21" y2="6"></line>
<line x1="3" y1="18" x2="21" y2="18"></line>
</svg>
</symbol>
<symbol id="svg-arrow-right" viewBox="0 0 24 24">
<title>Expand</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="feather-chevron-right">
<polyline points="9 18 15 12 9 6"></polyline>
</svg>
</symbol>
<symbol id="svg-sun" viewBox="0 0 24 24">
<title>Light mode</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="feather-sun">
<circle cx="12" cy="12" r="5"></circle>
<line x1="12" y1="1" x2="12" y2="3"></line>
<line x1="12" y1="21" x2="12" y2="23"></line>
<line x1="4.22" y1="4.22" x2="5.64" y2="5.64"></line>
<line x1="18.36" y1="18.36" x2="19.78" y2="19.78"></line>
<line x1="1" y1="12" x2="3" y2="12"></line>
<line x1="21" y1="12" x2="23" y2="12"></line>
<line x1="4.22" y1="19.78" x2="5.64" y2="18.36"></line>
<line x1="18.36" y1="5.64" x2="19.78" y2="4.22"></line>
</svg>
</symbol>
<symbol id="svg-moon" viewBox="0 0 24 24">
<title>Dark mode</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="icon-tabler-moon">
<path stroke="none" d="M0 0h24v24H0z" fill="none" />
<path d="M12 3c.132 0 .263 0 .393 0a7.5 7.5 0 0 0 7.92 12.446a9 9 0 1 1 -8.313 -12.454z" />
</svg>
</symbol>
<symbol id="svg-sun-with-moon" viewBox="0 0 24 24">
<title>Auto light/dark, in light mode</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round"
class="icon-custom-derived-from-feather-sun-and-tabler-moon">
<path style="opacity: 50%" d="M 5.411 14.504 C 5.471 14.504 5.532 14.504 5.591 14.504 C 3.639 16.319 4.383 19.569 6.931 20.352 C 7.693 20.586 8.512 20.551 9.25 20.252 C 8.023 23.207 4.056 23.725 2.11 21.184 C 0.166 18.642 1.702 14.949 4.874 14.536 C 5.051 14.512 5.231 14.5 5.411 14.5 L 5.411 14.504 Z"/>
<line x1="14.5" y1="3.25" x2="14.5" y2="1.25"/>
<line x1="14.5" y1="15.85" x2="14.5" y2="17.85"/>
<line x1="10.044" y1="5.094" x2="8.63" y2="3.68"/>
<line x1="19" y1="14.05" x2="20.414" y2="15.464"/>
<line x1="8.2" y1="9.55" x2="6.2" y2="9.55"/>
<line x1="20.8" y1="9.55" x2="22.8" y2="9.55"/>
<line x1="10.044" y1="14.006" x2="8.63" y2="15.42"/>
<line x1="19" y1="5.05" x2="20.414" y2="3.636"/>
<circle cx="14.5" cy="9.55" r="3.6"/>
</svg>
</symbol>
<symbol id="svg-moon-with-sun" viewBox="0 0 24 24">
<title>Auto light/dark, in dark mode</title>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round"
class="icon-custom-derived-from-feather-sun-and-tabler-moon">
<path d="M 8.282 7.007 C 8.385 7.007 8.494 7.007 8.595 7.007 C 5.18 10.184 6.481 15.869 10.942 17.24 C 12.275 17.648 13.706 17.589 15 17.066 C 12.851 22.236 5.91 23.143 2.505 18.696 C -0.897 14.249 1.791 7.786 7.342 7.063 C 7.652 7.021 7.965 7 8.282 7 L 8.282 7.007 Z"/>
<line style="opacity: 50%" x1="18" y1="3.705" x2="18" y2="2.5"/>
<line style="opacity: 50%" x1="18" y1="11.295" x2="18" y2="12.5"/>
<line style="opacity: 50%" x1="15.316" y1="4.816" x2="14.464" y2="3.964"/>
<line style="opacity: 50%" x1="20.711" y1="10.212" x2="21.563" y2="11.063"/>
<line style="opacity: 50%" x1="14.205" y1="7.5" x2="13.001" y2="7.5"/>
<line style="opacity: 50%" x1="21.795" y1="7.5" x2="23" y2="7.5"/>
<line style="opacity: 50%" x1="15.316" y1="10.184" x2="14.464" y2="11.036"/>
<line style="opacity: 50%" x1="20.711" y1="4.789" x2="21.563" y2="3.937"/>
<circle style="opacity: 50%" cx="18" cy="7.5" r="2.169"/>
</svg>
</symbol>
<symbol id="svg-pencil" viewBox="0 0 24 24">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="icon-tabler-pencil-code">
<path d="M4 20h4l10.5 -10.5a2.828 2.828 0 1 0 -4 -4l-10.5 10.5v4" />
<path d="M13.5 6.5l4 4" />
<path d="M20 21l2 -2l-2 -2" />
<path d="M17 17l-2 2l2 2" />
</svg>
</symbol>
<symbol id="svg-eye" viewBox="0 0 24 24">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor"
stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="icon-tabler-eye-code">
<path stroke="none" d="M0 0h24v24H0z" fill="none" />
<path d="M10 12a2 2 0 1 0 4 0a2 2 0 0 0 -4 0" />
<path
d="M11.11 17.958c-3.209 -.307 -5.91 -2.293 -8.11 -5.958c2.4 -4 5.4 -6 9 -6c3.6 0 6.6 2 9 6c-.21 .352 -.427 .688 -.647 1.008" />
<path d="M20 21l2 -2l-2 -2" />
<path d="M17 17l-2 2l2 2" />
</svg>
</symbol>
</svg>
<input type="checkbox" class="sidebar-toggle" name="__navigation" id="__navigation">
<input type="checkbox" class="sidebar-toggle" name="__toc" id="__toc">
<label class="overlay sidebar-overlay" for="__navigation">
<div class="visually-hidden">Hide navigation sidebar</div>
</label>
<label class="overlay toc-overlay" for="__toc">
<div class="visually-hidden">Hide table of contents sidebar</div>
</label>
<a class="skip-to-content muted-link" href="#furo-main-content">Skip to content</a>
<div class="page">
<header class="mobile-header">
<div class="header-left">
<label class="nav-overlay-icon" for="__navigation">
<div class="visually-hidden">Toggle site navigation sidebar</div>
<i class="icon"><svg><use href="#svg-menu"></use></svg></i>
</label>
</div>
<div class="header-center">
<a href="index.html"><div class="brand">RLtools Documentation</div></a>
</div>
<div class="header-right">
<div class="theme-toggle-container theme-toggle-header">
<button class="theme-toggle">
<div class="visually-hidden">Toggle Light / Dark / Auto color theme</div>
<svg class="theme-icon-when-auto-light"><use href="#svg-sun-with-moon"></use></svg>
<svg class="theme-icon-when-auto-dark"><use href="#svg-moon-with-sun"></use></svg>
<svg class="theme-icon-when-dark"><use href="#svg-moon"></use></svg>
<svg class="theme-icon-when-light"><use href="#svg-sun"></use></svg>
</button>
</div>
<label class="toc-overlay-icon toc-header-icon no-toc" for="__toc">
<div class="visually-hidden">Toggle table of contents sidebar</div>
<i class="icon"><svg><use href="#svg-toc"></use></svg></i>
</label>
</div>
</header>
<aside class="sidebar-drawer">
<div class="sidebar-container">
<div class="sidebar-sticky"><a class="sidebar-brand" href="index.html">
<div class="sidebar-logo-container">
<img class="sidebar-logo" src="_static/banner.svg" alt="Logo"/>
</div>
<span class="sidebar-brand-text">RLtools Documentation</span>
</a><form class="sidebar-search-container" method="get" action="search.html" role="search">
<input class="sidebar-search" placeholder="Search" name="q" aria-label="Search">
<input type="hidden" name="check_keywords" value="yes">
<input type="hidden" name="area" value="default">
</form>
<div id="searchbox"></div><div class="sidebar-scroll"><div class="sidebar-tree">
<p class="caption" role="heading"><span class="caption-text">Contents:</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="overview.html">Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="getting_started.html">Getting Started</a></li>
<li class="toctree-l1"><a class="reference internal" href="01-Containers.html">Containers</a></li>
<li class="toctree-l1"><a class="reference internal" href="02-Multiple%20Dispatch.html">Multiple Dispatch</a></li>
<li class="toctree-l1"><a class="reference internal" href="03-Deep%20Learning.html">Deep Learning</a></li>
<li class="toctree-l1"><a class="reference internal" href="04-CPU%20Acceleration.html">CPU Acceleration</a></li>
<li class="toctree-l1"><a class="reference internal" href="05-MNIST%20Classification.html">MNIST Classification</a></li>
<li class="toctree-l1"><a class="reference internal" href="06-Deep%20Reinforcement%20Learning.html">Deep Reinforcement Learning</a></li>
<li class="toctree-l1"><a class="reference internal" href="07-The%20Loop%20Interface.html">The Loop Interface</a></li>
<li class="toctree-l1"><a class="reference internal" href="08-Custom%20Environment.html">Custom Environment</a></li>
<li class="toctree-l1"><a class="reference internal" href="09-Python%20Interface.html">Python Interface</a></li>
<li class="toctree-l1"><a class="reference internal" href="09-Python%20Interface.html#Embedding-Exported-Checkpoints">Embedding Exported Checkpoints</a></li>
<li class="toctree-l1"><a class="reference internal" href="10-Experiment%20Tracking.html">Experiment Tracking</a></li>
<li class="toctree-l1 current current-page"><a class="current reference internal" href="#">Recurrent Neural Networks (RNNs)</a></li>
<li class="toctree-l1"><a class="reference internal" href="12-Memory.html">Memory</a></li>
</ul>
</div>
</div>
</div>
</div>
</aside>
<div class="main">
<div class="content">
<div class="article-container">
<a href="#" class="back-to-top muted-link">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24">
<path d="M13 20h-2V8l-5.5 5.5-1.42-1.42L12 4.16l7.92 7.92-1.42 1.42L13 8v12z"></path>
</svg>
<span>Back to top</span>
</a>
<div class="content-icon-container">
<div class="view-this-page">
<a class="muted-link" href="_sources/11-RNN.ipynb.txt" title="View this page">
<svg><use href="#svg-eye"></use></svg>
<span class="visually-hidden">View this page</span>
</a>
</div>
<div class="theme-toggle-container theme-toggle-content">
<button class="theme-toggle">
<div class="visually-hidden">Toggle Light / Dark / Auto color theme</div>
<svg class="theme-icon-when-auto-light"><use href="#svg-sun-with-moon"></use></svg>
<svg class="theme-icon-when-auto-dark"><use href="#svg-moon-with-sun"></use></svg>
<svg class="theme-icon-when-dark"><use href="#svg-moon"></use></svg>
<svg class="theme-icon-when-light"><use href="#svg-sun"></use></svg>
</button>
</div>
<label class="toc-overlay-icon toc-content-icon no-toc" for="__toc">
<div class="visually-hidden">Toggle table of contents sidebar</div>
<i class="icon"><svg><use href="#svg-toc"></use></svg></i>
</label>
</div>
<article role="main" id="furo-main-content">
<section id="Recurrent-Neural-Networks-(RNNs)">
<h1>Recurrent Neural Networks (RNNs)<a class="headerlink" href="#Recurrent-Neural-Networks-(RNNs)" title="Link to this heading">¶</a></h1>
<p>RLtools initially only supports the <a class="reference external" href="https://en.wikipedia.org/wiki/Gated_recurrent_unit">GRU (Gated Recurrent Unit)</a>, a widely used and time-tested RNN architecture.</p>
<p>In this example, we show the supervised training of a simple sequence model that learns to do the set operation <code class="docutils literal notranslate"><span class="pre">output</span> <span class="pre">=</span> <span class="pre">max(inputs)</span></code>. As in the previous examples, we import the required datastructures and models first:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[1]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="cp">#include</span><span class="w"> </span><span class="cpf"><vector></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><algorithm></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><numeric></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><iostream></span>
<span class="cp">#define RL_TOOLS_BACKEND_ENABLE_OPENBLAS</span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/operations/cpu_mux.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn/optimizers/adam/instance/operations_generic.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn/operations_cpu_mux.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn/layers/gru/operations_generic.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn_models/sequential/operations_generic.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn/optimizers/adam/operations_generic.h></span>
<span class="cp">#include</span><span class="w"> </span><span class="cpf"><rl_tools/nn/loss_functions/mse/operations_generic.h></span>
<span class="k">namespace</span><span class="w"> </span><span class="nn">rlt</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="nn">rl_tools</span><span class="p">;</span>
<span class="cp">#pragma cling load("openblas")</span>
</pre></div>
</div>
</div>
<p>Then setup the environment:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[2]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">using</span><span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="kt">float</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TYPE_POLICY</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">numeric_types</span><span class="o">::</span><span class="n">Policy</span><span class="o"><</span><span class="n">T</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">DEVICE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">devices</span><span class="o">::</span><span class="n">DEVICE_FACTORY</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">devices</span><span class="o">::</span><span class="n">DefaultCPUSpecification</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">RNG</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">DEVICE</span><span class="o">::</span><span class="n">SPEC</span><span class="o">::</span><span class="n">RANDOM</span><span class="o">::</span><span class="n">ENGINE</span><span class="o"><></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="k">typename</span><span class="w"> </span><span class="nc">DEVICE</span><span class="o">::</span><span class="n">index_t</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="kt">bool</span><span class="w"> </span><span class="n">DYNAMIC_ALLOCATION</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="nb">true</span><span class="p">;</span>
</pre></div>
</div>
</div>
<p>Now we can configure the sequence model. Here we use a GRU that directly takes the input and transforms it into its latent space. This latent space is then decoded by the <code class="docutils literal notranslate"><span class="pre">OUTPUT_LAYER</span></code> to predict the outputs:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[3]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">10</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">10</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">INPUT_DIM</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">1</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">HIDDEN_DIM</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">8</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">OUTPUT_DIM</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">1</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">INPUT_SHAPE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">INPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">GRU_CONFIG</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">layers</span><span class="o">::</span><span class="n">gru</span><span class="o">::</span><span class="n">Configuration</span><span class="o"><</span><span class="n">TYPE_POLICY</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">HIDDEN_DIM</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">parameters</span><span class="o">::</span><span class="n">groups</span><span class="o">::</span><span class="n">Normal</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">GRU</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">layers</span><span class="o">::</span><span class="n">gru</span><span class="o">::</span><span class="n">BindConfiguration</span><span class="o"><</span><span class="n">GRU_CONFIG</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">OUTPUT_LAYER_CONFIG</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">layers</span><span class="o">::</span><span class="n">dense</span><span class="o">::</span><span class="n">Configuration</span><span class="o"><</span><span class="n">TYPE_POLICY</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">OUTPUT_DIM</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">activation_functions</span><span class="o">::</span><span class="n">IDENTITY</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">OUTPUT_LAYER</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">layers</span><span class="o">::</span><span class="n">dense</span><span class="o">::</span><span class="n">BindConfiguration</span><span class="o"><</span><span class="n">OUTPUT_LAYER_CONFIG</span><span class="o">></span><span class="p">;</span>
</pre></div>
</div>
</div>
<p>As usual, we assemble these layers into a <code class="docutils literal notranslate"><span class="pre">nn_models::sequential</span></code> which is a sequence of layers and implements compile-time autodiff:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[4]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">template</span><span class="w"> </span><span class="o"><</span><span class="k">typename</span><span class="w"> </span><span class="nc">T_CONTENT</span><span class="p">,</span><span class="w"> </span><span class="k">typename</span><span class="w"> </span><span class="nc">T_NEXT_MODULE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn_models</span><span class="o">::</span><span class="n">sequential</span><span class="o">::</span><span class="n">OutputModule</span><span class="o">></span>
<span class="k">using</span><span class="w"> </span><span class="n">Module</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="k">typename</span><span class="w"> </span><span class="nc">rlt</span><span class="o">::</span><span class="n">nn_models</span><span class="o">::</span><span class="n">sequential</span><span class="o">::</span><span class="n">Module</span><span class="o"><</span><span class="n">T_CONTENT</span><span class="p">,</span><span class="w"> </span><span class="n">T_NEXT_MODULE</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">MODULE_CHAIN</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">Module</span><span class="o"><</span><span class="n">GRU</span><span class="p">,</span><span class="w"> </span><span class="n">Module</span><span class="o"><</span><span class="n">OUTPUT_LAYER</span><span class="o">>></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">CAPABILITY</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">capability</span><span class="o">::</span><span class="n">Gradient</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">parameters</span><span class="o">::</span><span class="n">Adam</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">MODEL</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn_models</span><span class="o">::</span><span class="n">sequential</span><span class="o">::</span><span class="n">Build</span><span class="o"><</span><span class="n">CAPABILITY</span><span class="p">,</span><span class="w"> </span><span class="n">MODULE_CHAIN</span><span class="p">,</span><span class="w"> </span><span class="n">INPUT_SHAPE</span><span class="o">></span><span class="p">;</span>
</pre></div>
</div>
</div>
<p>We need an optimizer as well, of course:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[5]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">struct</span><span class="w"> </span><span class="nc">ADAM_PARAMS</span><span class="o">:</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">optimizers</span><span class="o">::</span><span class="n">adam</span><span class="o">::</span><span class="n">DEFAULT_PARAMETERS_TENSORFLOW</span><span class="o"><</span><span class="n">TYPE_POLICY</span><span class="o">></span><span class="p">{</span>
<span class="w"> </span><span class="k">static</span><span class="w"> </span><span class="k">constexpr</span><span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">ALPHA</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mf">0.003</span><span class="p">;</span>
<span class="p">};</span>
<span class="k">using</span><span class="w"> </span><span class="n">ADAM_SPEC</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">optimizers</span><span class="o">::</span><span class="n">adam</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">TYPE_POLICY</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">ADAM_PARAMS</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">OPTIMIZER</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">optimizers</span><span class="o">::</span><span class="n">Adam</span><span class="o"><</span><span class="n">ADAM_SPEC</span><span class="o">></span><span class="p">;</span>
</pre></div>
</div>
</div>
<p>Now we can instantiate, allocate and initialize the data structures:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[6]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">DATASET_SIZE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">1000</span><span class="p">;</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">100</span><span class="p">;</span>
<span class="n">DEVICE</span><span class="w"> </span><span class="n">device</span><span class="p">;</span>
<span class="n">RNG</span><span class="w"> </span><span class="n">rng</span><span class="p">;</span>
<span class="n">MODEL</span><span class="w"> </span><span class="n">model</span><span class="p">;</span>
<span class="n">MODEL</span><span class="o">::</span><span class="n">Buffer</span><span class="o"><></span><span class="w"> </span><span class="n">buffer</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TEST_MODEL_TMP</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">MODEL</span><span class="o">::</span><span class="k">template</span><span class="w"> </span><span class="n">CHANGE_BATCH_SIZE</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="o">></span><span class="p">;</span><span class="w"> </span><span class="c1">// inference only model with test set as batch size</span>
<span class="k">using</span><span class="w"> </span><span class="n">TEST_MODEL</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">TEST_MODEL_TMP</span><span class="o">::</span><span class="k">template</span><span class="w"> </span><span class="n">CHANGE_CAPABILITY</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">capability</span><span class="o">::</span><span class="n">Forward</span><span class="o"><>></span><span class="p">;</span>
<span class="n">TEST_MODEL</span><span class="w"> </span><span class="n">test_model</span><span class="p">;</span>
<span class="n">TEST_MODEL</span><span class="o">::</span><span class="n">Buffer</span><span class="o"><></span><span class="w"> </span><span class="n">test_buffer</span><span class="p">;</span>
<span class="n">MODEL</span><span class="o">::</span><span class="n">State</span><span class="o"><></span><span class="w"> </span><span class="n">state</span><span class="p">;</span>
<span class="n">OPTIMIZER</span><span class="w"> </span><span class="n">optimizer</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">MODEL</span><span class="o">::</span><span class="n">INPUT_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">input</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">MODEL</span><span class="o">::</span><span class="n">OUTPUT_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">output_target</span><span class="p">,</span><span class="w"> </span><span class="n">d_output</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">DATASET_SHAPE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">DATASET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">INPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">DATASET_TARGET_SHAPE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">DATASET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">OUTPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">DATASET_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">dataset_X</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">DATASET_TARGET_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">dataset_y</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TESTSET_SHAPE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">INPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TESTSET_TARGET_SHAPE</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">OUTPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">testset_X</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_TARGET_SHAPE</span><span class="o">>></span><span class="w"> </span><span class="n">testset_y</span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TESTSET_SHAPE_PERMUTED</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">INPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="k">using</span><span class="w"> </span><span class="n">TESTSET_TARGET_SHAPE_PERMUTED</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Shape</span><span class="o"><</span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SIZE</span><span class="p">,</span><span class="w"> </span><span class="n">OUTPUT_DIM</span><span class="o">></span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_SHAPE_PERMUTED</span><span class="o">>></span><span class="w"> </span><span class="n">testset_X_permuted</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_TARGET_SHAPE_PERMUTED</span><span class="o">>></span><span class="w"> </span><span class="n">testset_y_permuted</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">Tensor</span><span class="o"><</span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">Specification</span><span class="o"><</span><span class="n">T</span><span class="p">,</span><span class="w"> </span><span class="n">TI</span><span class="p">,</span><span class="w"> </span><span class="n">TESTSET_TARGET_SHAPE_PERMUTED</span><span class="o">>></span><span class="w"> </span><span class="n">testset_output_permuted</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="n">device</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">);</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">SEED</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">,</span><span class="w"> </span><span class="n">SEED</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">test_model</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">buffer</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">test_buffer</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">state</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">optimizer</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">input</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output_target</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">d_output</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X_permuted</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y_permuted</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">malloc</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_output_permuted</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">optimizer</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">init_weights</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">reset_optimizer_state</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">optimizer</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">);</span>
</pre></div>
</div>
</div>
<p>The toy task we are facing here is <code class="docutils literal notranslate"><span class="pre">output</span> <span class="pre">=</span> <span class="pre">max(inputs)</span></code> hence we sample random numbers from a Gaussian and then calculate the max for the target values:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[7]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="k">template</span><span class="w"> </span><span class="o"><</span><span class="k">typename</span><span class="w"> </span><span class="nc">DATASET_X</span><span class="p">,</span><span class="w"> </span><span class="k">typename</span><span class="w"> </span><span class="nc">DATASET_Y</span><span class="o">></span>
<span class="kt">void</span><span class="w"> </span><span class="n">max_dataset</span><span class="p">(</span><span class="n">DATASET_X</span><span class="o">&</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="n">DATASET_Y</span><span class="o">&</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">){</span>
<span class="w"> </span><span class="k">static_assert</span><span class="p">(</span><span class="n">DATASET_X</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="n">FIRST</span><span class="w"> </span><span class="o">==</span><span class="w"> </span><span class="n">DATASET_Y</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="n">FIRST</span><span class="p">);</span>
<span class="w"> </span><span class="k">static_assert</span><span class="p">(</span><span class="n">DATASET_X</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="k">template</span><span class="w"> </span><span class="n">GET</span><span class="o"><</span><span class="mi">1</span><span class="o">></span><span class="w"> </span><span class="o">==</span><span class="w"> </span><span class="n">DATASET_Y</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="k">template</span><span class="w"> </span><span class="n">GET</span><span class="o"><</span><span class="mi">1</span><span class="o">></span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">randn</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">);</span>
<span class="w"> </span><span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">sample_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">sample_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">DATASET_X</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="n">FIRST</span><span class="p">;</span><span class="w"> </span><span class="n">sample_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">max</span><span class="p">;</span>
<span class="w"> </span><span class="kt">bool</span><span class="w"> </span><span class="n">max_set</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="nb">false</span><span class="p">;</span>
<span class="w"> </span><span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">step_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">step_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">DATASET_X</span><span class="o">::</span><span class="n">SHAPE</span><span class="o">::</span><span class="k">template</span><span class="w"> </span><span class="n">GET</span><span class="o"><</span><span class="mi">1</span><span class="o">></span><span class="p">;</span><span class="w"> </span><span class="n">step_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">el</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">get</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="n">sample_i</span><span class="p">,</span><span class="w"> </span><span class="n">step_i</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span>
<span class="w"> </span><span class="k">if</span><span class="p">(</span><span class="o">!</span><span class="n">max_set</span><span class="w"> </span><span class="o">||</span><span class="w"> </span><span class="n">el</span><span class="w"> </span><span class="o">></span><span class="w"> </span><span class="n">max</span><span class="p">){</span>
<span class="w"> </span><span class="n">max_set</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="nb">true</span><span class="p">;</span>
<span class="w"> </span><span class="n">max</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">el</span><span class="p">;</span>
<span class="w"> </span><span class="p">}</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">set</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">,</span><span class="w"> </span><span class="n">max</span><span class="p">,</span><span class="w"> </span><span class="n">sample_i</span><span class="p">,</span><span class="w"> </span><span class="n">step_i</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span>
<span class="w"> </span><span class="p">}</span>
<span class="w"> </span><span class="p">}</span>
<span class="p">}</span>
</pre></div>
</div>
</div>
<p>We want to generate a training and test set. We created a <code class="docutils literal notranslate"><span class="pre">test_model</span></code> that natively operates on <code class="docutils literal notranslate"><span class="pre">BATCH_SIZE</span> <span class="pre">=</span> <span class="pre">TESTSET_SIZE</span></code> so we can directly feed the testset into it without creating an additional, batched loop. The standard input format in RLtools is <code class="docutils literal notranslate"><span class="pre">(SEQUENCE_STEPS</span> <span class="pre">x</span> <span class="pre">BATCH_SAMPLES</span> <span class="pre">x</span> <span class="pre">FEATURES)</span></code> hence we permute the dataset generated with the previous function:</p>
<div class="nbinput nblast docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[8]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="n">max_dataset</span><span class="p">(</span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">);</span>
<span class="n">max_dataset</span><span class="p">(</span><span class="n">testset_X</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y</span><span class="p">);</span>
<span class="c1">// models operate on (SEQUENCE_STEP x BATCH_SIZE x FEATURE_DIM) for performance reasons:</span>
<span class="k">auto</span><span class="w"> </span><span class="n">permuted_X</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">permute</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">PermutationSpec</span><span class="o"><</span><span class="mi">0</span><span class="p">,</span><span class="w"> </span><span class="mi">1</span><span class="o">></span><span class="p">{});</span>
<span class="k">auto</span><span class="w"> </span><span class="n">permuted_y</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">permute</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">PermutationSpec</span><span class="o"><</span><span class="mi">0</span><span class="p">,</span><span class="w"> </span><span class="mi">1</span><span class="o">></span><span class="p">{});</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">copy</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">permuted_X</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X_permuted</span><span class="p">);</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">copy</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">permuted_y</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y_permuted</span><span class="p">);</span>
</pre></div>
</div>
</div>
<p>Here is an example of an input sequence and the expected output:</p>
<div class="nbinput docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[9]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">"Input: </span><span class="se">\n</span><span class="s">"</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">print</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">));</span>
<span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">"Expected output: "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">endl</span><span class="p">;</span>
<span class="n">rlt</span><span class="o">::</span><span class="n">print</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">));</span>
</pre></div>
</div>
</div>
<div class="nboutput nblast docutils container">
<div class="prompt empty docutils container">
</div>
<div class="output_area docutils container">
<div class="highlight"><pre>
Input:
-1.507621e+00
1.071986e+00
8.269271e-01
1.601774e+00
-1.074195e+00
-5.420533e-01
-6.830205e-01
1.492320e+00
6.583855e-02
9.513746e-01
Expected output:
-1.507621e+00
1.071986e+00
1.071986e+00
1.601774e+00
1.601774e+00
1.601774e+00
1.601774e+00
1.601774e+00
1.601774e+00
1.601774e+00
</pre></div></div>
</div>
<p>Now we have everything in place to train the model:</p>
<div class="nbinput docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[10]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="n">TI</span><span class="o">></span><span class="w"> </span><span class="n">indices</span><span class="p">(</span><span class="n">DATASET_SIZE</span><span class="p">);</span>
<span class="n">std</span><span class="o">::</span><span class="n">iota</span><span class="p">(</span><span class="n">indices</span><span class="p">.</span><span class="n">begin</span><span class="p">(),</span><span class="w"> </span><span class="n">indices</span><span class="p">.</span><span class="n">end</span><span class="p">(),</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span><span class="w"> </span><span class="c1">// fill with range 0..DATASET_SIZE</span>
<span class="k">constexpr</span><span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">N_EPOCH</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">100</span><span class="p">;</span>
<span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">epoch_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">epoch_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">N_EPOCH</span><span class="p">;</span><span class="w"> </span><span class="n">epoch_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">epoch_loss</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">shuffle</span><span class="p">(</span><span class="n">indices</span><span class="p">.</span><span class="n">begin</span><span class="p">(),</span><span class="w"> </span><span class="n">indices</span><span class="p">.</span><span class="n">end</span><span class="p">(),</span><span class="w"> </span><span class="n">rng</span><span class="p">.</span><span class="n">engine</span><span class="p">);</span>
<span class="w"> </span><span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">batch_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">batch_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">DATASET_SIZE</span><span class="w"> </span><span class="o">/</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="p">;</span><span class="w"> </span><span class="n">batch_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">sequence_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">sequence_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="p">;</span><span class="w"> </span><span class="n">sequence_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">TI</span><span class="w"> </span><span class="n">index</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="w"> </span><span class="o">*</span><span class="w"> </span><span class="n">batch_i</span><span class="w"> </span><span class="o">+</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">;</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">input_sample</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">input</span><span class="p">,</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">ViewSpec</span><span class="o"><</span><span class="mi">1</span><span class="o">></span><span class="p">{});</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">output_sample</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output_target</span><span class="p">,</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">,</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">ViewSpec</span><span class="o"><</span><span class="mi">1</span><span class="o">></span><span class="p">{});</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">dataset_input_sample</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_X</span><span class="p">,</span><span class="w"> </span><span class="n">indices</span><span class="p">[</span><span class="n">index</span><span class="p">],</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">ViewSpec</span><span class="o"><</span><span class="mi">0</span><span class="o">></span><span class="p">{});</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">dataset_output_sample</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_y</span><span class="p">,</span><span class="w"> </span><span class="n">indices</span><span class="p">[</span><span class="n">index</span><span class="p">],</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">tensor</span><span class="o">::</span><span class="n">ViewSpec</span><span class="o"><</span><span class="mi">0</span><span class="o">></span><span class="p">{});</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">copy</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_input_sample</span><span class="p">,</span><span class="w"> </span><span class="n">input_sample</span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">copy</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">dataset_output_sample</span><span class="p">,</span><span class="w"> </span><span class="n">output_sample</span><span class="p">);</span>
<span class="w"> </span><span class="p">}</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">forward</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">,</span><span class="w"> </span><span class="n">input</span><span class="p">,</span><span class="w"> </span><span class="n">buffer</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">);</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">output</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">output</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">);</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">output_matrix_view</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">matrix_view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output</span><span class="p">);</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">output_target_matrix_view</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">matrix_view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output_target</span><span class="p">);</span>
<span class="w"> </span><span class="k">auto</span><span class="w"> </span><span class="n">d_output_matrix_view</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">matrix_view</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">d_output</span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">loss_functions</span><span class="o">::</span><span class="n">mse</span><span class="o">::</span><span class="n">gradient</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output_matrix_view</span><span class="p">,</span><span class="w"> </span><span class="n">output_target_matrix_view</span><span class="p">,</span><span class="w"> </span><span class="n">d_output_matrix_view</span><span class="p">);</span>
<span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">batch_loss</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">loss_functions</span><span class="o">::</span><span class="n">mse</span><span class="o">::</span><span class="n">evaluate</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">output_matrix_view</span><span class="p">,</span><span class="w"> </span><span class="n">output_target_matrix_view</span><span class="p">);</span>
<span class="w"> </span><span class="n">epoch_loss</span><span class="w"> </span><span class="o">+=</span><span class="w"> </span><span class="n">batch_loss</span><span class="p">;</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">zero_gradient</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">backward</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">,</span><span class="w"> </span><span class="n">input</span><span class="p">,</span><span class="w"> </span><span class="n">d_output</span><span class="p">,</span><span class="w"> </span><span class="n">buffer</span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">step</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">optimizer</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">);</span>
<span class="w"> </span><span class="p">}</span>
<span class="w"> </span><span class="n">epoch_loss</span><span class="w"> </span><span class="o">/=</span><span class="w"> </span><span class="n">DATASET_SIZE</span><span class="w"> </span><span class="o">/</span><span class="w"> </span><span class="n">BATCH_SIZE</span><span class="p">;</span>
<span class="w"> </span><span class="k">if</span><span class="p">((</span><span class="n">epoch_i</span><span class="o">+</span><span class="mi">1</span><span class="p">)</span><span class="w"> </span><span class="o">%</span><span class="w"> </span><span class="mi">5</span><span class="w"> </span><span class="o">==</span><span class="w"> </span><span class="mi">0</span><span class="p">){</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">copy</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">model</span><span class="p">,</span><span class="w"> </span><span class="n">test_model</span><span class="p">);</span>
<span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">evaluate</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">test_model</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">testset_output_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">test_buffer</span><span class="p">,</span><span class="w"> </span><span class="n">rng</span><span class="p">);</span>
<span class="w"> </span><span class="n">T</span><span class="w"> </span><span class="n">test_loss</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">nn</span><span class="o">::</span><span class="n">loss_functions</span><span class="o">::</span><span class="n">mse</span><span class="o">::</span><span class="n">evaluate</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_output_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y_permuted</span><span class="p">);</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">"Epoch "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="p">(</span><span class="n">epoch_i</span><span class="o">+</span><span class="mi">1</span><span class="p">)</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">" train loss: "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">epoch_loss</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">" test loss: "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">test_loss</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">endl</span><span class="p">;</span>
<span class="w"> </span><span class="p">}</span>
<span class="p">}</span>
</pre></div>
</div>
</div>
<div class="nboutput nblast docutils container">
<div class="prompt empty docutils container">
</div>
<div class="output_area docutils container">
<div class="highlight"><pre>
Epoch 5 train loss: 4.290424e-02 test loss: 3.762348e-02
Epoch 10 train loss: 1.859014e-02 test loss: 1.703962e-02
Epoch 15 train loss: 1.132567e-02 test loss: 9.239300e-03
Epoch 20 train loss: 8.027696e-03 test loss: 5.971038e-03
Epoch 25 train loss: 6.056546e-03 test loss: 4.463931e-03
Epoch 30 train loss: 4.469625e-03 test loss: 3.213925e-03
Epoch 35 train loss: 3.648721e-03 test loss: 3.196432e-03
Epoch 40 train loss: 3.001935e-03 test loss: 1.977837e-03
Epoch 45 train loss: 2.315897e-03 test loss: 1.590888e-03
Epoch 50 train loss: 1.958636e-03 test loss: 1.422130e-03
Epoch 55 train loss: 1.633920e-03 test loss: 1.367668e-03
Epoch 60 train loss: 1.455133e-03 test loss: 2.061459e-03
Epoch 65 train loss: 1.359705e-03 test loss: 9.891201e-04
Epoch 70 train loss: 1.101305e-03 test loss: 1.248149e-03
Epoch 75 train loss: 1.034629e-03 test loss: 1.741087e-03
Epoch 80 train loss: 9.677236e-04 test loss: 8.029980e-04
Epoch 85 train loss: 9.137033e-04 test loss: 7.385706e-04
Epoch 90 train loss: 8.515422e-04 test loss: 8.916398e-04
Epoch 95 train loss: 7.121030e-04 test loss: 6.450592e-04
Epoch 100 train loss: 6.727659e-04 test loss: 9.201005e-04
</pre></div></div>
</div>
<p>Now we check if the predictions of the model are plausible:</p>
<div class="nbinput docutils container">
<div class="prompt highlight-none notranslate"><div class="highlight"><pre><span></span>[11]:
</pre></div>
</div>
<div class="input_area highlight-c++ notranslate"><div class="highlight"><pre><span></span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">fixed</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">setprecision</span><span class="p">(</span><span class="mi">2</span><span class="p">);</span><span class="w"> </span><span class="c1">// fixed point printing</span>
<span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">sequence_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">sequence_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="mi">5</span><span class="p">;</span><span class="w"> </span><span class="n">sequence_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">"Test sequence "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">sequence_i</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">endl</span><span class="p">;</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">"Input => Target ~ Predicted"</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">endl</span><span class="p">;</span>
<span class="w"> </span><span class="k">for</span><span class="p">(</span><span class="n">TI</span><span class="w"> </span><span class="n">step_i</span><span class="w"> </span><span class="o">=</span><span class="w"> </span><span class="mi">0</span><span class="p">;</span><span class="w"> </span><span class="n">step_i</span><span class="w"> </span><span class="o"><</span><span class="w"> </span><span class="n">SEQUENCE_LENGTH</span><span class="p">;</span><span class="w"> </span><span class="n">step_i</span><span class="o">++</span><span class="p">){</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">" "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">setw</span><span class="p">(</span><span class="mi">5</span><span class="p">)</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">get</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_X_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">step_i</span><span class="p">,</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">" => "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">setw</span><span class="p">(</span><span class="mi">5</span><span class="p">)</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">get</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_y_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">step_i</span><span class="p">,</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="s">" ~ "</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">setw</span><span class="p">(</span><span class="mi">5</span><span class="p">)</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">rlt</span><span class="o">::</span><span class="n">get</span><span class="p">(</span><span class="n">device</span><span class="p">,</span><span class="w"> </span><span class="n">testset_output_permuted</span><span class="p">,</span><span class="w"> </span><span class="n">step_i</span><span class="p">,</span><span class="w"> </span><span class="n">sequence_i</span><span class="p">,</span><span class="w"> </span><span class="mi">0</span><span class="p">);</span>
<span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">cout</span><span class="w"> </span><span class="o"><<</span><span class="w"> </span><span class="n">std</span><span class="o">::</span><span class="n">endl</span><span class="p">;</span>
<span class="w"> </span><span class="p">}</span>
<span class="p">}</span>
</pre></div>
</div>
</div>
<div class="nboutput nblast docutils container">
<div class="prompt empty docutils container">
</div>
<div class="output_area docutils container">
<div class="highlight"><pre>
Test sequence 0
Input => Target ~ Predicted
0.65 => 0.65 ~ 0.66
0.89 => 0.89 ~ 0.88
1.38 => 1.38 ~ 1.37
0.74 => 1.38 ~ 1.37
-1.00 => 1.38 ~ 1.37
0.06 => 1.38 ~ 1.38
0.07 => 1.38 ~ 1.39
-0.98 => 1.38 ~ 1.39
1.20 => 1.38 ~ 1.39
0.12 => 1.38 ~ 1.40
Test sequence 1
Input => Target ~ Predicted
-0.08 => -0.08 ~ -0.09
0.21 => 0.21 ~ 0.21
-0.19 => 0.21 ~ 0.21
-1.47 => 0.21 ~ 0.21
-0.16 => 0.21 ~ 0.19
-1.06 => 0.21 ~ 0.21
-0.36 => 0.21 ~ 0.20
0.62 => 0.62 ~ 0.59
0.31 => 0.62 ~ 0.60
-0.30 => 0.62 ~ 0.63
Test sequence 2
Input => Target ~ Predicted
1.29 => 1.29 ~ 1.32
0.08 => 1.29 ~ 1.30
0.76 => 1.29 ~ 1.30
0.14 => 1.29 ~ 1.31
0.43 => 1.29 ~ 1.32
-1.59 => 1.29 ~ 1.31
-0.13 => 1.29 ~ 1.32
-0.30 => 1.29 ~ 1.33
-0.64 => 1.29 ~ 1.32
1.25 => 1.29 ~ 1.35
Test sequence 3
Input => Target ~ Predicted
-0.44 => -0.44 ~ -0.45
0.62 => 0.62 ~ 0.63
-0.06 => 0.62 ~ 0.64
1.33 => 1.33 ~ 1.34
1.23 => 1.33 ~ 1.37
0.40 => 1.33 ~ 1.37
0.32 => 1.33 ~ 1.38
-0.11 => 1.33 ~ 1.39
-0.75 => 1.33 ~ 1.38
0.25 => 1.33 ~ 1.39
Test sequence 4
Input => Target ~ Predicted
-0.10 => -0.10 ~ -0.11
0.17 => 0.17 ~ 0.17
0.01 => 0.17 ~ 0.19
0.66 => 0.66 ~ 0.64
0.79 => 0.79 ~ 0.80
-0.81 => 0.79 ~ 0.81
-0.85 => 0.79 ~ 0.82
-0.19 => 0.79 ~ 0.83
0.05 => 0.79 ~ 0.83
0.55 => 0.79 ~ 0.82
</pre></div></div>
</div>
</section>
</article>
</div>
<footer>
<div class="related-pages">
<a class="next-page" href="12-Memory.html">
<div class="page-info">
<div class="context">
<span>Next</span>
</div>
<div class="title">Memory</div>
</div>
<svg class="furo-related-icon"><use href="#svg-arrow-right"></use></svg>
</a>
<a class="prev-page" href="10-Experiment%20Tracking.html">
<svg class="furo-related-icon"><use href="#svg-arrow-right"></use></svg>
<div class="page-info">
<div class="context">
<span>Previous</span>
</div>
<div class="title">Experiment Tracking Interface Specification</div>
</div>
</a>
</div>
<div class="bottom-of-page">
<div class="left-details">
<div class="copyright">
Copyright © 2023
</div>
Made with <a href="https://www.sphinx-doc.org/">Sphinx</a> and <a class="muted-link" href="https://pradyunsg.me">@pradyunsg</a>'s
<a href="https://github.com/pradyunsg/furo">Furo</a>
</div>
<div class="right-details">
</div>
</div>
</footer>
</div>
<aside class="toc-drawer no-toc">
</aside>
</div>
</div><script src="_static/documentation_options.js?v=5929fcd5"></script>
<script src="_static/doctools.js?v=9bcbadda"></script>
<script src="_static/sphinx_highlight.js?v=dc90522c"></script>
<script src="_static/scripts/furo.js?v=5fa4622c"></script>
<script crossorigin="anonymous" integrity="sha256-Ae2Vz/4ePdIu6ZyI/5ZGsYnb+m0JlOmKPjt6XZ9JJkA=" src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.4/require.min.js"></script>
<script>window.MathJax = {"tex": {"inlineMath": [["$", "$"], ["\\(", "\\)"]], "processEscapes": true}, "options": {"ignoreHtmlClass": "tex2jax_ignore|mathjax_ignore|document", "processHtmlClass": "tex2jax_process|mathjax_process|math|output_area"}}</script>
<script defer="defer" src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>
</body>
</html>