composing-pretrained-models-web/index.html at main · energy-based-model/composing-pretrained-models-web · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>

<script src="bootstrap.js"></script>
<script type="text/javascript" charset="utf-8" src="https://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js"></script>
<script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>

<style type="text/css">
body {
    font-family: "Titillium Web", "HelveticaNeue-Light", "Helvetica Neue Light", "Helvetica Neue", Helvetica, Arial, "Lucida Grande", sans-serif;
    font-weight: 300;
    font-size: 17px;
    margin-left: auto;
    margin-right: auto;
}

@media screen and (min-width: 980px){
    body {
        width: 980px;
    }
}

h1 {
    font-weight:300;
    line-height: 1.15em;
}

h2 {
    font-size: 1.75em;
}
a:link,a:visited {
    color: #5364cc;
    text-decoration: none;
}
a:hover {
    color: #208799;
}
h1 {
    text-align: center;
}
h2,h3 {
    text-align: left;
}

h1 {
    font-size: 40px;
    font-weight: 500;
}
h2 {
    font-weight: 400;
    margin: 16px 0px 4px 0px;
}
h3 {
    font-weight: 600;
    margin: 16px 0px 4px 0px;
}

.paper-title {
    padding: 1px 0px 1px 0px;
}
section {
    margin: 32px 0px 32px 0px;
    text-align: justify;
    clear: both;
}
.col-5 {
     width: 20%;
     float: left;
}
.col-4 {
     width: 25%;
     float: left;
}
.col-3 {
     width: 33%;
     float: left;
}
.col-2 {
     width: 50%;
     float: left;
}
.col-1 {
     width: 100%;
     float: left;
}

.author-row, .affil-row {
    font-size: 26px;
}

.author-row-new {
    text-align: center;
}

.author-row-new a {
    display: inline-block;
    font-size: 20px;
    padding: 4px;
}

.author-row-new sup {
    color: #313436;
    font-size: 12px;
}

.affiliations-new {
    font-size: 18px;
    text-align: center;
    width: 80%;
    margin: 0 auto;
    margin-bottom: 20px;
}

.row {
    margin: 16px 0px 16px 0px;
}
.authors {
    font-size: 26px;
}
.affiliatons {
    font-size: 18px;
}
.affil-row {
    margin-top: 18px;
}
.teaser {
    max-width: 100%;
}
.text-center {
    text-align: center;
}
.screenshot {
    width: 256px;
    border: 1px solid #ddd;
}
.screenshot-el {
    margin-bottom: 16px;
}
hr {
    height: 1px;
    border: 0;
    border-top: 1px solid #ddd;
    margin: 0;
}
.material-icons {
    vertical-align: -6px;
}
p {
    line-height: 1.25em;
}
.caption {
    font-size: 16px;
    color: #666;
    margin-top: 4px;
    margin-bottom: 10px;
}


video {
    display: block;
    margin: auto;
}


figure {
    display: block;
    margin: auto;
    margin-top: 10px;
    margin-bottom: 10px;
}
#bibtex pre {
    font-size: 14px;
    background-color: #eee;
    padding: 16px;
}
.blue {
    color: #2c82c9;
    font-weight: bold;
}
.orange {
    color: #d35400;
    font-weight: bold;
}
.flex-row {
    display: flex;
    flex-flow: row wrap;
    padding: 0;
    margin: 0;
    list-style: none;
}

.paper-btn-coming-soon {
    position: relative;
    top: 0;
    left: 0;
}

.coming-soon {
    position: absolute;
    top: -15px;
    right: -15px;
}

.paper-btn {
  position: relative;
  text-align: center;

  display: inline-block;
  margin: 8px;
  padding: 8px 8px;

  border-width: 0;
  outline: none;
  border-radius: 2px;

  background-color: #5364cc;
  color: white !important;
  font-size: 20px;
  width: 100px;
  font-weight: 600;
}
.paper-btn-parent {
    display: flex;
    justify-content: center;
    margin: 16px 0px;
}

.paper-btn:hover {
    opacity: 0.85;
}

.container {
    margin-left: auto;
    margin-right: auto;
    padding-left: 16px;
    padding-right: 16px;
}

.venue {
    font-size: 23px;
}

.topnav {
    background-color: #EEEEEE;
    overflow: hidden;
}

.topnav div {
    max-width: 1070px;
    margin: 0 auto;
}

.topnav a {
    display: inline-block;
    color: black;
    text-align: center;
    vertical-align: middle;
    padding: 16px 16px;
    text-decoration: none;
    font-size: 18px;
}

.topnav img {
    padding: 2px 0px;
    width: 100%;
    margin: 0.2em 0px 0.3em 0px;
    vertical-align: middle;
}

pre {
    font-size: 0.9em;
    padding-left: 7px;
    padding-right: 7px;
    padding-top: 3px;
    padding-bottom: 3px;
    border-radius: 3px;
    background-color: rgb(235, 235, 235);
    overflow-x: auto;
}

.download-thumb {
    display: flex;
}

@media only screen and (max-width: 620px) {
    .download-thumb {
        display: none;
    }
}

.paper-stuff {
    width: 50%;
    font-size: 20px;
}

@media only screen and (max-width: 620px) {
    .paper-stuff {
        width: 100%;
    }
}
* {
  box-sizing: border-box;
}

.column {
  text-align: center;
  float: left;
  width: 16.666%;
  padding: 5px;
}
.column3 {
  text-align: center;
  float: left;
  width: 33.333%;
  padding: 5px;
}
.column4 {
  text-align: center;
  float: left;
  width: 50%;
  padding: 5px;
}
.column5 {
  text-align: center;
  float: left;
  width: 20%;
  padding: 5px;
}
.border-right {
    border-right: 1px solid black;
}
.border-bottom{
    border-bottom: 1px solid black;
}


/* Clearfix (clear floats) */
.row::after {
  content: "";
  clear: both;
  display: table;
}
.img-fluid {
  max-width: 100%;
  height: auto;
}
.figure-img {
  margin-bottom: 0.5rem;
  line-height: 1;
}


.rounded-circle {
  border-radius: 50% !important;
}


/* Responsive layout - makes the three columns stack on top of each other instead of next to each other */
@media screen and (max-width: 500px) {
  .column {
    width: 100%;
  }
}
@media screen and (max-width: 500px) {
  .column3 {
    width: 100%;
  }
}

</style>
<link rel="stylesheet" href="bootstrap-grid.css">

<script type="text/javascript" src="../js/hidebib.js"></script>
    <link href='https://fonts.googleapis.com/css?family=Titillium+Web:400,600,400italic,600italic,300,300italic' rel='stylesheet' type='text/css'>
    <head>
        <title> Composing Ensembles of Pre-trained Models via Iterative Consensus</title>
        <meta name="viewport" content="width=device-width, initial-scale=1">
        <meta property="og:description" content="Composing Ensembles of Pre-trained Models via Iterative Consensus"/>
        <link href="https://fonts.googleapis.com/css2?family=Material+Icons" rel="stylesheet">
        <meta name="twitter:card" content="summary_large_image">
        <meta name="twitter:creator" content="@karsten_kreis">
        <meta name="twitter:title" content="Composing Ensembles of Pre-trained Models via Iterative Consensus">
        <meta name="twitter:description" content="">
        <meta name="twitter:image" content="">
    </head>

 <body>


<div class="container">
    <div class="paper-title">
    <h1>
        Composing Ensembles of Pre-trained Models via Iterative Consensus
    </div>

    <div id="authors">
        <center>
            <div class="author-row-new">
                <a href="https://shuangli59.github.io/">Shuang Li*<sup>1</sup></a>,
                <a href="https://yilundu.github.io/">Yilun Du*<sup>1</sup></a>,
                <a href="https://scholar.google.com/citations?user=rRJ9wTJMUB8C&hl=en">Joshua B. Tenenbaum<sup>1</sup></a>,
                <a href="https://groups.csail.mit.edu/vision/torralbalab/">Antonio Torralba<sup>1</sup></a>
                <a href="https://scholar.google.com/citations?user=Vzr1RukAAAAJ&hl=en">Igor Mordatch<sup>2</sup></a>
            </div>
        </center>
        <center>
        <div class="affiliations">
            <span><sup>1</sup> MIT</span>
            <span><sup>2</sup> Google Brain</span><br/>
        </div>

        <br>*indicates equal contribution. Shuang Li did all the experiments on image generation, video question answering, and mathematical reasoning. Yilun Du did all the experiments on robot manipulation.

        <div class="affil-row">
            <div class="venue text-center"><b>ICLR 2023</b></div>
        </div>

        </center>

        <div style="clear: both">
            <div class="paper-btn-parent">
            <a class="paper-btn" href="https://arxiv.org/abs/2210.11522">
                <span class="material-icons"> description </span>
                 Paper
            </a>
            <div class="paper-btn-coming-soon">
                <!-- https://github.com/ShuangLI59/Composing-Ensembles-of-Pre-trained-Models-via-Iterative-Consensus -->
                <a class="paper-btn" href="">
                    <span class="material-icons"> code </span>
                    Code
                </a>
            </div>
        </div></div>
    </div>


    <!-- <section id="teaser-image">
        <center>
            <figure>
                <video class="centered" width="80%" autoplay loop muted playsinline class="video-background " >
                    <source src="assets/LION_video_v10.mp4#t=0.001" type="video/mp4">
                    Your browser does not support the video tag.
                </video>
            </figure>

        </center>
    </section>
     -->


    <br>

    <section id="teaser-image">
        <center>
            <center><p><b>A unified framework for composing pre-trained models.</b></p></center>

            <figure>
                <video width="650" loop autoplay muted>
                    <source src="materials/teaser-2.mp4" type="video/mp4">
                </video>

                <br><br>

                <video width="960" loop autoplay muted>
                    <source src="materials/all_results.mp4" type="video/mp4">
                </video>

                <!-- <br><br> -->

                <!-- <video width="800" loop autoplay muted style="border:1px solid black">
                    <source src="materials/new3.mp4" type="video/mp4">
                </video>
 -->
            </figure>

        </center>
    </section>


    <section id="abstract"/>
        <hr>
        <h2>Abstract</h2>
        <div class="flex-row">
            <p>
                Large pre-trained models exhibit distinct and complementary capabilities dependent on the data they are trained on. Language models such as GPT-3 are capable of textual reasoning but cannot understand visual information, while vision models such as DALL-E can generate photorealistic photos but fail to understand complex language descriptions.
                <br><br>
                In this work, we propose a unified framework for composing ensembles of different pre-trained models -- combining the strengths of each individual model to solve various multimodal problems in a zero-shot manner. We use pre-trained models as "generators" or "scorers" and compose them via closed-loop iterative consensus optimization. The generator constructs proposals and the scorers iteratively provide feedback to refine the generated result. Such closed-loop communication enables models to correct errors caused by other models, significantly boosting performance on downstream tasks, e.g. improving accuracy on grade school math problems by 7.5%, without requiring any model finetuning. We demonstrate that consensus achieved by an ensemble of scorers outperforms the feedback of a single scorer, by leveraging the strengths of each expert model. Results show that the proposed method can be used as a general purpose framework for a wide range of zero-shot multimodal tasks, such as image generation, video question answering, mathematical reasoning, and robotic manipulation.
            </p>
        </div>
    </section>
    <section id="method"/>
        <hr>
        <h2>Method</h2>

            <div class="mx-auto">
                <left><p>The proposed framework that composes a "generator" and an ensemble of "scorers" through iterative consensus enables zero-shot generalization across a variety of multimodal tasks.</p></left>
                <center><img class="card-img-top" src="materials/teaser.png" style="width:950px"></center>
            </div>


            <br><br><br>

            <!-- <div class="row">
                    <div class="column4">
                        <center><img class="card-img-top" src="materials/framework.png" style="width:400px"></center>
                    </div>

                    <div class="column4">
                        <video width="400" loop autoplay muted>
                            <source src="materials/teaser3.mp4" type="video/mp4">
                        </video>
                    </div>
            </div>
 -->


            <div class="flex-row">
                <div class="mx-auto">
                    <left><p><b>Overview of the proposed unified framework.</b> Dashed lines are omitted for certain tasks. Orange lines represent the components used to refine the generated result.</p></left>
                    <br>
                    <center><img class="card-img-top" src="materials/framework.png" style="width:400px"></center>

                    <br><br>

                    <!-- <video width="800" loop autoplay muted style="border:1px solid black"> -->
                    <video width="850" loop autoplay muted controls>
                        <source src="materials/new3-2.mp4" type="video/mp4">
                    </video>
                </div>

                <br><br>
                <p><b>Image generation: </b> A pre-trained diffusion model is used as the generator, and multiple scorers, such as CLIP and image classifiers, are used to provide feedback to the generator.</p>
                <p><b>Video question answering: </b> GPT-2 is used as the generator, and a set of CLIP models are used as scorers.</p>
                <p><b>Grade school math: </b> GPT-2 is used as the generator, and a set of question-solution classifiers are used as scorers.</p>
                <p><b>Robot manipulation: </b> MPC+World model is used as the generator, and a pre-trained image segmentation model is used to compute the scores from multiple camera views to select the best action.</p>

            </div>

    </section>


    <section id="results">
        <hr>
        <h2>Video Question Answering Results</h2>
            <center>
            <figure>
                <video class="centered" width="100%" autoplay loop muted playsinline class="video-background " >
                    <source src="materials/vqa.mp4#t=0.001" type="video/mp4">
                    Your browser does not support the video tag.
                </video>
            </figure>
            </center>

        <hr>


        <h2>Grade School Math Results</h2>

            <center>
            <figure>
                <video class="centered" width="100%" autoplay loop muted playsinline class="video-background " >
                    <source src="materials/math.mp4#t=0.001" type="video/mp4">
                    Your browser does not support the video tag.
                </video>
            </figure>
            </center>

        <hr>

        <h2>Image Generation</h2>

            <center>
            <figure>
                <video class="centered" width="100%" autoplay loop muted playsinline class="video-background " >
                    <source src="materials/image_generation.mp4#t=0.001" type="video/mp4">
                    Your browser does not support the video tag.
                </video>
            </figure>
            </center>

        <hr>


        <h2>Robot Manipulation Results</h2>

            <center>
            <figure>
                <video class="centered" width="100%" autoplay loop muted playsinline class="video-background " >
                    <source src="materials/robot_faster.mp4#t=0.001" type="video/mp4">
                    Your browser does not support the video tag.
                </video>
            </figure>
            </center>

        <!-- <hr> -->


    </section>


    <section id="related_projects">
        <hr>
        <h2>Related Projects</h2>

          <br>
          Check out a list of our related papers on compositionality. A full list can be found <a href="https://energy-based-model.github.io/Energy-based-Model-MIT/">here</a>!
          <br>

        <div class="row vspace-top">
        <div class="col-sm-3">
            <video width="100%" playsinline="" autoplay="" loop="" preload="" muted="">
                <source src="materials/related/teaser_glide.mp4" type="video/mp4">
            </video>
        </div>
        <div class="col-sm-9">
          <div class="paper-title">
            <a href="https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/">Compositional Visual Generation with Composable Diffusion Models</a>
        </div>
        <div>
            We present a method to compose different diffusion models together, drawing on the close connection of
            diffusion models with EBMs. We illustrate how compositional operators enable
            the ability to composing multiple sets of objects together as well as generate images subject to
            complex text prompts.
        </div>
        </div>
        </div>

        <div class="row vspace-top">
        <div class="col-sm-3">
            <video width="100%" playsinline="" autoplay="" loop="" preload="" muted="">
                <source src="materials/related/clevr_teaser.mp4" type="video/mp4">
            </video>
        </div>
        <div class="col-sm-9">
          <div class="paper-title">
            <a href="https://composevisualrelations.github.io/">Learning to Compose Visual Relations</a>
          </div>
          <div>
            The visual world around us can be described as a structured set of objects and their associated relations. In this work, we propose to represent each relation as an unnormalized density (an energy-based model), enabling us to compose separate relations in a factorized manner. We show that such a factorized decomposition allows the model to both generate and edit scenes that have multiple sets of relations more faithfully.
          </div>
        </div>
        </div>

        <div class="row vspace-top">
        <div class="col-sm-3">
            <img src="materials/related/comp_cartoon.png" class="img-fluid" alt="comp_carton" style="width:100%">
        </div>
        <div class="col-sm-9">
          <div class="paper-title">
            <a href="https://energy-based-model.github.io/compositional-generation-inference/">Compositional Visual Generation with Energy Based Models</a>
        </div>
        <div>
            We present a set of compositional operators that enable EBMs to exhibit <b>zero-shot compositional</b> visual generation, enabling us to compose visual concepts
            (through operators of conjunction, disjunction, or negation) together in a zero-shot manner.
            Our approach enables us to generate faces given a  description
            ((Smiling AND Female) OR (NOT Smiling AND Male)) or to combine several different objects together.
        </div>
        </div>

    </section>


    <section id="paper">
        <h2>Team</h2>
        <div class="row">
            <div class="column5">
                <a href='https://shuangli59.github.io/'>
                    <img  src=./materials/people/lishuang.jpg class="figure-img img-fluid rounded-circle" height=200px width=200px>
                </a>
                <p class=profname>Shuang Li</p>
                <p class=institution>MIT</p>
            </div>

            <div class="column5">
                <a href='https://yilundu.github.io/'>
                    <img  src=./materials/people/yilun3.png class="figure-img img-fluid rounded-circle" height=200px width=200px>
                </a>
                <p class=profname> Yilun Du </p>
                <p class=institution>MIT</p>
            </div>

            <div class="column5">
                <a href='https://scholar.google.com/citations?user=rRJ9wTJMUB8C&hl=en'>
                    <img  src=./materials/people/josh2.jpg class="figure-img img-fluid rounded-circle" height=200px width=200px>
                </a>
                <p class=profname> Joshua Tenenbaum </p>
                <p class=institution>MIT</p>
            </div>

            <div class="column5">
                <a href="https://groups.csail.mit.edu/vision/torralbalab/">
                    <img src=./materials/people/antonioTorralba.jpg class="figure-img img-fluid rounded-circle" height=200px width=200px>
                </a>
                <p class=profname> Antonio Torralba </p>
                <p class=institution>MIT</p>
            </div>

            <div class="column5">
                <a href='https://scholar.google.com/citations?user=Vzr1RukAAAAJ&hl=en'>
                    <img  src=./materials/people/xuusrug0_400x400.jpeg class="figure-img img-fluid rounded-circle" height=200px width=200px>
                </a>
                <p class=profname> Igor Mordatch </p>
                <p class=institution>Google Brain</p>
            </div>
    </section>


<!--
    <section id="paper">
        <h2>Paper</h2>
        <hr>
        <div class="flex-row">
            <div class="download-thumb">
            <div style="box-sizing: border-box; padding: 16px; margin: auto;">
                <a href="https://energy-based-model.github.io/composing-pretrained-models/"><img class="screenshot" src="materials/thumb_finger.png"></a>
            </div>
        </div>
            <div class="paper-stuff">
                <p><b>Composing Ensembles of Pre-trained Models via Iterative Consensus</b></p>
                <p>Shuang Li, Yilun Du, Joshua B. Tenenbaum, Antonio Torralba, Igor Mordatch</p>
                <div><span class="material-icons"> description </span><a href="https://arxiv.org/abs/2210.06978"> arXiv version</a></div>
                <div><span class="material-icons"> integration_instructions </span><a href="https://github.com/nv-tlabs/LION"> Code</a></div>
            </div>
            </div>
        </div>
    </section>

 -->

    <!-- <section id="bibtex">
        <h2>Citation</h2>
        <hr>
        <pre><code>@inproceedings{zeng2022lion,
            title={LION: Latent Point Diffusion Models for 3D Shape Generation},
            author={Xiaohui Zeng and Arash Vahdat and Francis Williams and Zan Gojcic and Or Litany and Sanja Fidler and Karsten Kreis},
            booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
            year={2022}
        }</code></pre>
    </section> -->

    <section>
        This webpage template was recycled from <a href='https://nv-tlabs.github.io/LION/'>here</a>.
        <center><p><a href='https://accessibility.mit.edu/'><b>Accessibility</b></a></p></center>
    </section>


</div>
</body>
</html>