-
Notifications
You must be signed in to change notification settings - Fork 0
/
ODM.html
668 lines (552 loc) · 23.7 KB
/
ODM.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<!-- saved from url=(0041)https://trajectory-transformer.github.io/ -->
<html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async="" src="./ODM/js"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-GDXSC5Y2BD');
</script>
<script src="./ODM/head.js.下载"></script> <meta name="viewport" content="width=device-width, initial-scale=1"> <link rel="shortcut icon" href="https://trajectory-transformer.github.io/undefinedimages/favicon.ico">
<meta name="viewport" content="width=device-width, initial-scale=1">
<script src="./ODM/polyfill.min.js.下载"></script>
<script id="MathJax-script" async="" src="./ODM/tex-mml-chtml.js.下载"></script>
<meta name="keywords" content="Berkeley,Deep,Reinforcement,Learning,Computer Science,Machine,Artificial,Intelligence">
<style type="text/css">.CtxtMenu_InfoClose { top:.2em; right:.2em;}
.CtxtMenu_InfoContent { overflow:auto; text-align:left; font-size:80%; padding:.4em .6em; border:1px inset; margin:1em 0px; max-height:20em; max-width:30em; background-color:#EEEEEE; white-space:normal;}
.CtxtMenu_Info.CtxtMenu_MousePost {outline:none;}
.CtxtMenu_Info { position:fixed; left:50%; width:auto; text-align:center; border:3px outset; padding:1em 2em; background-color:#DDDDDD; color:black; cursor:default; font-family:message-box; font-size:120%; font-style:normal; text-indent:0; text-transform:none; line-height:normal; letter-spacing:normal; word-spacing:normal; word-wrap:normal; white-space:nowrap; float:none; z-index:201; border-radius: 15px; /* Opera 10.5 and IE9 */ -webkit-border-radius:15px; /* Safari and Chrome */ -moz-border-radius:15px; /* Firefox */ -khtml-border-radius:15px; /* Konqueror */ box-shadow:0px 10px 20px #808080; /* Opera 10.5 and IE9 */ -webkit-box-shadow:0px 10px 20px #808080; /* Safari 3 & Chrome */ -moz-box-shadow:0px 10px 20px #808080; /* Forefox 3.5 */ -khtml-box-shadow:0px 10px 20px #808080; /* Konqueror */ filter:progid:DXImageTransform.Microsoft.dropshadow(OffX=2, OffY=2, Color="gray", Positive="true"); /* IE */}
</style><style type="text/css">.CtxtMenu_MenuClose { position:absolute; cursor:pointer; display:inline-block; border:2px solid #AAA; border-radius:18px; -webkit-border-radius: 18px; /* Safari and Chrome */ -moz-border-radius: 18px; /* Firefox */ -khtml-border-radius: 18px; /* Konqueror */ font-family: "Courier New", Courier; font-size:24px; color:#F0F0F0}
.CtxtMenu_MenuClose span { display:block; background-color:#AAA; border:1.5px solid; border-radius:18px; -webkit-border-radius: 18px; /* Safari and Chrome */ -moz-border-radius: 18px; /* Firefox */ -khtml-border-radius: 18px; /* Konqueror */ line-height:0; padding:8px 0 6px /* may need to be browser-specific */}
.CtxtMenu_MenuClose:hover { color:white!important; border:2px solid #CCC!important}
.CtxtMenu_MenuClose:hover span { background-color:#CCC!important}
.CtxtMenu_MenuClose:hover:focus { outline:none}
</style><style type="text/css">.CtxtMenu_Menu { position:absolute; background-color:white; color:black; width:auto; padding:5px 0px; border:1px solid #CCCCCC; margin:0; cursor:default; font: menu; text-align:left; text-indent:0; text-transform:none; line-height:normal; letter-spacing:normal; word-spacing:normal; word-wrap:normal; white-space:nowrap; float:none; z-index:201; border-radius: 5px; /* Opera 10.5 and IE9 */ -webkit-border-radius: 5px; /* Safari and Chrome */ -moz-border-radius: 5px; /* Firefox */ -khtml-border-radius: 5px; /* Konqueror */ box-shadow:0px 10px 20px #808080; /* Opera 10.5 and IE9 */ -webkit-box-shadow:0px 10px 20px #808080; /* Safari 3 & Chrome */ -moz-box-shadow:0px 10px 20px #808080; /* Forefox 3.5 */ -khtml-box-shadow:0px 10px 20px #808080; /* Konqueror */}
.CtxtMenu_MenuItem { padding: 1px 2em; background:transparent;}
.CtxtMenu_MenuArrow { position:absolute; right:.5em; padding-top:.25em; color:#666666; font-family: null; font-size: .75em}
.CtxtMenu_MenuActive .CtxtMenu_MenuArrow {color:white}
.CtxtMenu_MenuArrow.CtxtMenu_RTL {left:.5em; right:auto}
.CtxtMenu_MenuCheck { position:absolute; left:.7em; font-family: null}
.CtxtMenu_MenuCheck.CtxtMenu_RTL { right:.7em; left:auto }
.CtxtMenu_MenuRadioCheck { position:absolute; left: .7em;}
.CtxtMenu_MenuRadioCheck.CtxtMenu_RTL { right: .7em; left:auto}
.CtxtMenu_MenuInputBox { padding-left: 1em; right:.5em; color:#666666; font-family: null;}
.CtxtMenu_MenuInputBox.CtxtMenu_RTL { left: .1em;}
.CtxtMenu_MenuComboBox { left:.1em; padding-bottom:.5em;}
.CtxtMenu_MenuSlider { left: .1em;}
.CtxtMenu_SliderValue { position:absolute; right:.1em; padding-top:.25em; color:#333333; font-size: .75em}
.CtxtMenu_SliderBar { outline: none; background: #d3d3d3}
.CtxtMenu_MenuLabel { padding: 1px 2em 3px 1.33em; font-style:italic}
.CtxtMenu_MenuRule { border-top: 1px solid #DDDDDD; margin: 4px 3px;}
.CtxtMenu_MenuDisabled { color:GrayText}
.CtxtMenu_MenuActive { background-color: #606872; color: white;}
.CtxtMenu_MenuDisabled:focus { background-color: #E8E8E8}
.CtxtMenu_MenuLabel:focus { background-color: #E8E8E8}
.CtxtMenu_ContextMenu:focus { outline:none}
.CtxtMenu_ContextMenu .CtxtMenu_MenuItem:focus { outline:none}
.CtxtMenu_SelectionMenu { position:relative; float:left; border-bottom: none; -webkit-box-shadow:none; -webkit-border-radius:0px; }
.CtxtMenu_SelectionItem { padding-right: 1em;}
.CtxtMenu_Selection { right: 40%; width:50%; }
.CtxtMenu_SelectionBox { padding: 0em; max-height:20em; max-width: none; background-color:#FFFFFF;}
.CtxtMenu_SelectionDivider { clear: both; border-top: 2px solid #000000;}
.CtxtMenu_Menu .CtxtMenu_MenuClose { top:-10px; left:-10px}
</style><style id="MJX-CHTML-styles">
mjx-container[jax="CHTML"] {
line-height: 0;
}
mjx-container [space="1"] {
margin-left: .111em;
}
mjx-container [space="2"] {
margin-left: .167em;
}
mjx-container [space="3"] {
margin-left: .222em;
}
mjx-container [space="4"] {
margin-left: .278em;
}
mjx-container [space="5"] {
margin-left: .333em;
}
mjx-container [rspace="1"] {
margin-right: .111em;
}
mjx-container [rspace="2"] {
margin-right: .167em;
}
mjx-container [rspace="3"] {
margin-right: .222em;
}
mjx-container [rspace="4"] {
margin-right: .278em;
}
mjx-container [rspace="5"] {
margin-right: .333em;
}
mjx-container [size="s"] {
font-size: 70.7%;
}
mjx-container [size="ss"] {
font-size: 50%;
}
mjx-container [size="Tn"] {
font-size: 60%;
}
mjx-container [size="sm"] {
font-size: 85%;
}
mjx-container [size="lg"] {
font-size: 120%;
}
mjx-container [size="Lg"] {
font-size: 144%;
}
mjx-container [size="LG"] {
font-size: 173%;
}
mjx-container [size="hg"] {
font-size: 207%;
}
mjx-container [size="HG"] {
font-size: 249%;
}
mjx-container [width="full"] {
width: 100%;
}
mjx-box {
display: inline-block;
}
mjx-block {
display: block;
}
mjx-itable {
display: inline-table;
}
mjx-row {
display: table-row;
}
mjx-row > * {
display: table-cell;
}
mjx-mtext {
display: inline-block;
}
mjx-mstyle {
display: inline-block;
}
mjx-merror {
display: inline-block;
color: red;
background-color: yellow;
}
mjx-mphantom {
visibility: hidden;
}
_::-webkit-full-page-media, _:future, :root mjx-container {
will-change: opacity;
}
mjx-assistive-mml {
position: absolute !important;
top: 0px;
left: 0px;
clip: rect(1px, 1px, 1px, 1px);
padding: 1px 0px 0px 0px !important;
border: 0px !important;
display: block !important;
width: auto !important;
overflow: hidden !important;
-webkit-touch-callout: none;
-webkit-user-select: none;
-khtml-user-select: none;
-moz-user-select: none;
-ms-user-select: none;
user-select: none;
}
mjx-assistive-mml[display="block"] {
width: 100% !important;
}
mjx-c::before {
display: block;
width: 0;
}
.MJX-TEX {
font-family: MJXZERO, MJXTEX;
}
.TEX-B {
font-family: MJXZERO, MJXTEX-B;
}
.TEX-I {
font-family: MJXZERO, MJXTEX-I;
}
.TEX-MI {
font-family: MJXZERO, MJXTEX-MI;
}
.TEX-BI {
font-family: MJXZERO, MJXTEX-BI;
}
.TEX-S1 {
font-family: MJXZERO, MJXTEX-S1;
}
.TEX-S2 {
font-family: MJXZERO, MJXTEX-S2;
}
.TEX-S3 {
font-family: MJXZERO, MJXTEX-S3;
}
.TEX-S4 {
font-family: MJXZERO, MJXTEX-S4;
}
.TEX-A {
font-family: MJXZERO, MJXTEX-A;
}
.TEX-C {
font-family: MJXZERO, MJXTEX-C;
}
.TEX-CB {
font-family: MJXZERO, MJXTEX-CB;
}
.TEX-FR {
font-family: MJXZERO, MJXTEX-FR;
}
.TEX-FRB {
font-family: MJXZERO, MJXTEX-FRB;
}
.TEX-SS {
font-family: MJXZERO, MJXTEX-SS;
}
.TEX-SSB {
font-family: MJXZERO, MJXTEX-SSB;
}
.TEX-SSI {
font-family: MJXZERO, MJXTEX-SSI;
}
.TEX-SC {
font-family: MJXZERO, MJXTEX-SC;
}
.TEX-T {
font-family: MJXZERO, MJXTEX-T;
}
.TEX-V {
font-family: MJXZERO, MJXTEX-V;
}
.TEX-VB {
font-family: MJXZERO, MJXTEX-VB;
}
mjx-stretchy-v mjx-c, mjx-stretchy-h mjx-c {
font-family: MJXZERO, MJXTEX-S1, MJXTEX-S4, MJXTEX, MJXTEX-A ! important;
}
@font-face /* 0 */ {
font-family: MJXZERO;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Zero.woff") format("woff");
}
@font-face /* 1 */ {
font-family: MJXTEX;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Main-Regular.woff") format("woff");
}
@font-face /* 2 */ {
font-family: MJXTEX-B;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Main-Bold.woff") format("woff");
}
@font-face /* 3 */ {
font-family: MJXTEX-I;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Math-Italic.woff") format("woff");
}
@font-face /* 4 */ {
font-family: MJXTEX-MI;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Main-Italic.woff") format("woff");
}
@font-face /* 5 */ {
font-family: MJXTEX-BI;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Math-BoldItalic.woff") format("woff");
}
@font-face /* 6 */ {
font-family: MJXTEX-S1;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Size1-Regular.woff") format("woff");
}
@font-face /* 7 */ {
font-family: MJXTEX-S2;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Size2-Regular.woff") format("woff");
}
@font-face /* 8 */ {
font-family: MJXTEX-S3;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Size3-Regular.woff") format("woff");
}
@font-face /* 9 */ {
font-family: MJXTEX-S4;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Size4-Regular.woff") format("woff");
}
@font-face /* 10 */ {
font-family: MJXTEX-A;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_AMS-Regular.woff") format("woff");
}
@font-face /* 11 */ {
font-family: MJXTEX-C;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Calligraphic-Regular.woff") format("woff");
}
@font-face /* 12 */ {
font-family: MJXTEX-CB;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Calligraphic-Bold.woff") format("woff");
}
@font-face /* 13 */ {
font-family: MJXTEX-FR;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Fraktur-Regular.woff") format("woff");
}
@font-face /* 14 */ {
font-family: MJXTEX-FRB;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Fraktur-Bold.woff") format("woff");
}
@font-face /* 15 */ {
font-family: MJXTEX-SS;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_SansSerif-Regular.woff") format("woff");
}
@font-face /* 16 */ {
font-family: MJXTEX-SSB;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_SansSerif-Bold.woff") format("woff");
}
@font-face /* 17 */ {
font-family: MJXTEX-SSI;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_SansSerif-Italic.woff") format("woff");
}
@font-face /* 18 */ {
font-family: MJXTEX-SC;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Script-Regular.woff") format("woff");
}
@font-face /* 19 */ {
font-family: MJXTEX-T;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Typewriter-Regular.woff") format("woff");
}
@font-face /* 20 */ {
font-family: MJXTEX-V;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Vector-Regular.woff") format("woff");
}
@font-face /* 21 */ {
font-family: MJXTEX-VB;
src: url("https://cdn.jsdelivr.net/npm/mathjax@3/es5/output/chtml/fonts/woff-v2/MathJax_Vector-Bold.woff") format("woff");
}
</style></head><body><article class="post-content">
<meta name="twitter:title" content="Offline Reinforcement Learning as One Big Sequence Modeling Problem">
<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:image" content="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/humanoid_padded.png">
<article class="post-content">
<title>Reinforcement Learning as One Big Sequence Modeling Problem</title>
<link rel="stylesheet" href="./Reinforcement Learning as One Big Sequence Modeling Problem_files/font.css">
<link rel="stylesheet" href="./Reinforcement Learning as One Big Sequence Modeling Problem_files/main.css">
<link rel="stylesheet" type="text/css" href="https://cdn.rawgit.com/dreampulse/computer-modern-web-font/master/fonts.css">
<style>
body {
font-family: "Computer Modern Serif", serif;
font-size: 14pt;
}
* {padding:0;margin:0;box-sizing:border-box;}
#video {
position: relative;
padding-bottom: 45%; /* 16:9 */
height: 0;
}
#video iframe {
position: absolute;
top: 0;
left: 0;
width: 80%;
height: 100%;
transform: translateX(12.5%);
}
</style>
<style type="text/css">/**
* Style sheet used by new LibX tooltip code
*/
/* We insert a <div> with libx-tooltip style under the body.
* This will inherit body's style - we can't afford to inherit undesirable
* styles and we must redefine what we need. OTOH, some things, e.g.
* font-size, might be ok to be inherited to stay within the page's tone.
*/
.libx-tooltip {
display: none;
overflow: visible;
padding: 5px;
z-index: 100;
background-color: #eee;
color: #000;
font-weight: normal;
font-style: normal;
text-align: left;
border: 2px solid #666;
border-radius: 5px;
-webkit-border-radius: 5px;
-moz-border-radius: 5px;
}
.libx-tooltip p {
/* override default 1em margin to keep paragraphs inside a tooltip closer together. */
margin: .2em;
}
</style><style type="text/css">/**
* Style sheet used by LibX autolinking code
*/
.libx-autolink {
}
</style>
<div class="outercontainer">
<div class="container">
<div class="content project_title">
<br>
<h2>Offline Reinforcement Learning as <br> One Big Sequence Modeling Problem</h2>
<!-- <div class="authors">
<a href="https://people.eecs.berkeley.edu/~janner/">Michael Janner</a>,
<a href="https://scholar.google.com/citations?user=qlwwdfEAAAAJ&hl=en">Qiyang (Colin) Li</a>, and
<a href="https://people.eecs.berkeley.edu/~svlevine/">Sergey Levine</a>
</div> -->
<!-- <br> -->
<!-- <a href="https://arxiv.org/abs/2106.02039">Paper</a> -->
<!-- <a href="./trajectory-transformer-neurips-2021.pdf">Paper</a> -->
<div>
<span class="venue"><a href="https://neurips.cc/">NeurIPS 2021 (spotlight)</a></span>
<span class="tag">
<a href="https://arxiv.org/abs/2106.02039">Paper</a>
<!-- <a href="./trajectory-transformer-neurips-2021.pdf">Paper</a> -->
<a href="https://github.com/JannerM/trajectory-transformer">Code</a>
<a href="https://bair.berkeley.edu/blog/2021/11/19/trajectory-transformer/">Blog</a>
<a href="https://trajectory-transformer.github.io/files/bib.txt">BibTex</a>
</span>
</div>
</div>
<br><br>
<center>
<b><font size="4">Trajectory Transformer</font></b>
<video width="100%" height="auto" autoplay="" playsinline="" muted="">
<source src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/gif_transformer.mp4" type="video/mp4">
</video>
<b><font size="4">Single-Step Model</font></b>
<video width="100%" height="auto" autoplay="" playsinline="" muted="">
<source src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/gif_single_step.mp4" type="video/mp4">
</video>
<br>
<i>Long-horizon predictions of the Trajectory Transformer compared <br>to those of a feedforward single-step dynamics model.</i>
</center>
<p>
<br><br>
</p><div class="content">
<div class="text">
<p>
</p><div class="title"><b>Summary</b></div>
<!-- <b>
<font size="5">Summary</font>
</b> -->
<!-- -->
We view reinforcement learning as a generic sequence modeling problem and investigate how much of the usual machinery of reinforcement learning algorithms can be replaced with the tools that have found widespread use in large-scale language modeling.
The core of our approach is the Trajectory Transformer, trained on sequences of states, actions, and rewards treated interchangeably, and a set of beam-search-based planners.
<p></p>
</div>
</div>
<br>
<br>
<center>
<img width="70%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/blog/architecture.png">
<br>
<br>
</center>
<div class="content">
<div class="text">
<p>
</p><div class="title"><b>Transformers as dynamics models</b></div>
<!-- <b>
<font size="5">Transformers as dynamics models</font>
</b> -->
<!-- -->
Predictive dynamics models often have excellent single-step error, but poor long-horizon accuracy due to compounding errors.
We show that Transformers are more reliable long-horizon predictors than state-of-the-art single-step models, even in continuous Markovian domains.
<p></p>
<br>
<br>
<center>
<img width="25%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/markov.png">
<img width="25%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/striated.png">
<br>
<i>Attention patterns of the Trajectory Transformer, showing (left) a discovered <br>Markovian stratetgy and (right) an approach with action smoothing.</i>
</center>
<br>
</div>
</div>
<br>
<br>
<div class="content">
<div class="text">
<p>
</p><div class="title"><b>Beam search as trajectory optimizer</b></div>
<!-- <b>
<font size="5.0">
Beam search as trajectory optimizer
</font>
</b> -->
<!-- . -->
<!-- Various control settings can be reduced to slight modifications of beam search with a sequence model. -->
<ul>
<li>Decoding a Trajectory Transformer with unmodified beam search gives rise to a model-based imitative method that optimizes for entire predicted trajectories to match those of an expert policy.</li>
<li>Conditioning trajectories on a future desired state alongside previously-encountered states yields a goal-reaching method.</li>
<br>
<center>
<img width="25%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/0.png">
<img width="25%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/1.png">
<img width="25%" src="https://people.eecs.berkeley.edu/~janner/trajectory-transformer/images/2.png">
<br>
<img width="2%" src="./ODM/rolloutblack-1.png">
Start
<img width="2%" src="./ODM/rolloutblue-1.png">
Goal
</center>
<br>
<li>Replacing log-probabilities from the sequence model with reward predictions yields a model-based planning method, surprisingly effective despite lacking the details usually required to make planning with learned models effective.</li>
<br>
<center>
<img width="80%" src="./ODM/bar.png">
</center>
<br>
</ul>
<p></p>
</div>
</div>
<div class="content">
<div class="text">
<div class="title"><b>Offline Reinforcement Learning as One Big Sequence Modeling Problem</b></div>
<div class="authors">
<a href="https://jannerm.github.io/">Michael Janner</a>,
<a href="https://scholar.google.com/citations?user=qlwwdfEAAAAJ&hl=en">Qiyang (Colin) Li</a>, and
<a href="https://people.eecs.berkeley.edu/~svlevine/">Sergey Levine</a>
</div>
<div>
<span class="venue"><a href="https://neurips.cc/">NeurIPS 2021</a></span>
<span class="tag">
<a href="https://arxiv.org/abs/2106.02039">Paper</a>
<!-- <a href="./trajectory-transformer-neurips-2021.pdf">Paper</a> -->
<a href="https://github.com/JannerM/trajectory-transformer">Code</a>
<a href="https://bair.berkeley.edu/blog/2021/11/19/trajectory-transformer/">Blog</a>
<a href="https://trajectory-transformer.github.io/files/bib.txt">BibTex</a>
</span>
</div>
</div>
</div>
<br><br><br>
<div class="content">
<div class="text">
<div class="title"><b>Related Publication</b></div>
Chen et al concurrently proposed another sequence modeling approach to reinforcement learning. At a high-level, ours is more model-based in spirit and theirs is more model-free, which allows us to evaluate Transformers as long-horizon dynamics models (<i>e.g.</i>, in the humanoid predictions above) and allows them to evaluate their policies in image-based environments (<i>e.g.</i>, Atari).
<a href="https://sites.google.com/berkeley.edu/decision-transformer">We encourage you to check out their work as well.</a>
</div>
</div>
</div>
</div>
<br><br><br><br>
</article></article></body></html>