-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathevaluation.html
840 lines (680 loc) · 26.4 KB
/
evaluation.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
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>基于前提的跨模态推理评测</title>
<style>
/* cspell:disable-file */
/* webkit printing magic: print all background colors */
html {
-webkit-print-color-adjust: exact;
}
* {
box-sizing: border-box;
-webkit-print-color-adjust: exact;
}
html,
body {
margin: 0;
padding: 0;
}
@media only screen {
body {
margin: 2em auto;
max-width: 900px;
color: rgb(55, 53, 47);
}
}
body {
line-height: 1.5;
white-space: pre-wrap;
}
a,
a.visited {
color: inherit;
text-decoration: underline;
}
.pdf-relative-link-path {
font-size: 80%;
color: #444;
}
h1,
h2,
h3 {
letter-spacing: -0.01em;
line-height: 1.2;
font-weight: 600;
margin-bottom: 0;
}
.page-title {
font-size: 2.5rem;
font-weight: 700;
margin-top: 0;
margin-bottom: 0.75em;
}
h1 {
font-size: 1.875rem;
margin-top: 1.875rem;
}
h2 {
font-size: 1.5rem;
margin-top: 1.5rem;
}
h3 {
font-size: 1.25rem;
margin-top: 1.25rem;
}
.source {
border: 1px solid #ddd;
border-radius: 3px;
padding: 1.5em;
word-break: break-all;
}
.callout {
border-radius: 3px;
padding: 1rem;
}
figure {
margin: 1.25em 0;
page-break-inside: avoid;
}
figcaption {
opacity: 0.5;
font-size: 85%;
margin-top: 0.5em;
}
mark {
background-color: transparent;
}
.indented {
padding-left: 1.5em;
}
hr {
background: transparent;
display: block;
width: 100%;
height: 1px;
visibility: visible;
border: none;
border-bottom: 1px solid rgba(55, 53, 47, 0.09);
}
img {
max-width: 100%;
}
@media only print {
img {
max-height: 100vh;
object-fit: contain;
}
}
@page {
margin: 1in;
}
.collection-content {
font-size: 0.875rem;
}
.column-list {
display: flex;
justify-content: space-between;
}
.column {
padding: 0 1em;
}
.column:first-child {
padding-left: 0;
}
.column:last-child {
padding-right: 0;
}
.table_of_contents-item {
display: block;
font-size: 0.875rem;
line-height: 1.3;
padding: 0.125rem;
}
.table_of_contents-indent-1 {
margin-left: 1.5rem;
}
.table_of_contents-indent-2 {
margin-left: 3rem;
}
.table_of_contents-indent-3 {
margin-left: 4.5rem;
}
.table_of_contents-link {
text-decoration: none;
opacity: 0.7;
border-bottom: 1px solid rgba(55, 53, 47, 0.18);
}
table,
th,
td {
border: 1px solid rgba(55, 53, 47, 0.09);
border-collapse: collapse;
}
table {
border-left: none;
border-right: none;
}
th,
td {
font-weight: normal;
padding: 0.25em 0.5em;
line-height: 1.5;
min-height: 1.5em;
text-align: left;
}
th {
color: rgba(55, 53, 47, 0.6);
}
ol,
ul {
margin: 0;
margin-block-start: 0.6em;
margin-block-end: 0.6em;
}
li>ol:first-child,
li>ul:first-child {
margin-block-start: 0.6em;
}
ul>li {
list-style: disc;
}
ul.to-do-list {
text-indent: -1.7em;
}
ul.to-do-list>li {
list-style: none;
}
.to-do-children-checked {
text-decoration: line-through;
opacity: 0.375;
}
ul.toggle>li {
list-style: none;
}
ul {
padding-inline-start: 1.7em;
}
ul>li {
padding-left: 0.1em;
}
ol {
padding-inline-start: 1.6em;
}
ol>li {
padding-left: 0.2em;
}
.mono ol {
padding-inline-start: 2em;
}
.mono ol>li {
text-indent: -0.4em;
}
.toggle {
padding-inline-start: 0em;
list-style-type: none;
}
/* Indent toggle children */
.toggle>li>details {
padding-left: 1.7em;
}
.toggle>li>details>summary {
margin-left: -1.1em;
}
.selected-value {
display: inline-block;
padding: 0 0.5em;
background: rgba(206, 205, 202, 0.5);
border-radius: 3px;
margin-right: 0.5em;
margin-top: 0.3em;
margin-bottom: 0.3em;
white-space: nowrap;
}
.collection-title {
display: inline-block;
margin-right: 1em;
}
.simple-table {
margin-top: 1em;
font-size: 0.875rem;
empty-cells: show;
}
.simple-table td {
height: 29px;
min-width: 120px;
}
.simple-table th {
height: 29px;
min-width: 120px;
}
.simple-table-header-color {
background: rgb(247, 246, 243);
color: black;
}
.simple-table-header {
font-weight: 500;
}
time {
opacity: 0.5;
}
.icon {
display: inline-block;
max-width: 1.2em;
max-height: 1.2em;
text-decoration: none;
vertical-align: text-bottom;
margin-right: 0.5em;
}
img.icon {
border-radius: 3px;
}
.user-icon {
width: 1.5em;
height: 1.5em;
border-radius: 100%;
margin-right: 0.5rem;
}
.user-icon-inner {
font-size: 0.8em;
}
.text-icon {
border: 1px solid #000;
text-align: center;
}
.page-cover-image {
display: block;
object-fit: cover;
width: 100%;
max-height: 30vh;
}
.page-header-icon {
font-size: 3rem;
margin-bottom: 1rem;
}
.page-header-icon-with-cover {
margin-top: -0.72em;
margin-left: 0.07em;
}
.page-header-icon img {
border-radius: 3px;
}
.link-to-page {
margin: 1em 0;
padding: 0;
border: none;
font-weight: 500;
}
p>.user {
opacity: 0.5;
}
td>.user,
td>time {
white-space: nowrap;
}
input[type="checkbox"] {
transform: scale(1.5);
margin-right: 0.6em;
vertical-align: middle;
}
p {
margin-top: 0.5em;
margin-bottom: 0.5em;
}
.image {
border: none;
margin: 1.5em 0;
padding: 0;
border-radius: 0;
text-align: center;
}
.code,
code {
background: rgba(135, 131, 120, 0.15);
border-radius: 3px;
padding: 0.2em 0.4em;
border-radius: 3px;
font-size: 85%;
tab-size: 2;
}
code {
color: #eb5757;
}
.code {
padding: 1.5em 1em;
}
.code-wrap {
white-space: pre-wrap;
word-break: break-all;
}
.code>code {
background: none;
padding: 0;
font-size: 100%;
color: inherit;
}
blockquote {
font-size: 1.25em;
margin: 1em 0;
padding-left: 1em;
border-left: 3px solid rgb(55, 53, 47);
}
.bookmark {
text-decoration: none;
max-height: 8em;
padding: 0;
display: flex;
width: 100%;
align-items: stretch;
}
.bookmark-title {
font-size: 0.85em;
overflow: hidden;
text-overflow: ellipsis;
height: 1.75em;
white-space: nowrap;
}
.bookmark-text {
display: flex;
flex-direction: column;
}
.bookmark-info {
flex: 4 1 180px;
padding: 12px 14px 14px;
display: flex;
flex-direction: column;
justify-content: space-between;
}
.bookmark-image {
width: 33%;
flex: 1 1 180px;
display: block;
position: relative;
object-fit: cover;
border-radius: 1px;
}
.bookmark-description {
color: rgba(55, 53, 47, 0.6);
font-size: 0.75em;
overflow: hidden;
max-height: 4.5em;
word-break: break-word;
}
.bookmark-href {
font-size: 0.75em;
margin-top: 0.25em;
}
.sans {
font-family: ui-sans-serif, -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, "Apple Color Emoji", Arial, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol";
}
.code {
font-family: "SFMono-Regular", Menlo, Consolas, "PT Mono", "Liberation Mono", Courier, monospace;
}
.serif {
font-family: Lyon-Text, Georgia, ui-serif, serif;
}
.mono {
font-family: iawriter-mono, Nitti, Menlo, Courier, monospace;
}
.pdf .sans {
font-family: Inter, ui-sans-serif, -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, "Apple Color Emoji", Arial, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol", 'Twemoji', 'Noto Color Emoji', 'Noto Sans CJK JP';
}
.pdf:lang(zh-CN) .sans {
font-family: Inter, ui-sans-serif, -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, "Apple Color Emoji", Arial, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol", 'Twemoji', 'Noto Color Emoji', 'Noto Sans CJK SC';
}
.pdf:lang(zh-TW) .sans {
font-family: Inter, ui-sans-serif, -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, "Apple Color Emoji", Arial, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol", 'Twemoji', 'Noto Color Emoji', 'Noto Sans CJK TC';
}
.pdf:lang(ko-KR) .sans {
font-family: Inter, ui-sans-serif, -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, "Apple Color Emoji", Arial, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol", 'Twemoji', 'Noto Color Emoji', 'Noto Sans CJK KR';
}
.pdf .code {
font-family: Source Code Pro, "SFMono-Regular", Menlo, Consolas, "PT Mono", "Liberation Mono", Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK JP';
}
.pdf:lang(zh-CN) .code {
font-family: Source Code Pro, "SFMono-Regular", Menlo, Consolas, "PT Mono", "Liberation Mono", Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK SC';
}
.pdf:lang(zh-TW) .code {
font-family: Source Code Pro, "SFMono-Regular", Menlo, Consolas, "PT Mono", "Liberation Mono", Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK TC';
}
.pdf:lang(ko-KR) .code {
font-family: Source Code Pro, "SFMono-Regular", Menlo, Consolas, "PT Mono", "Liberation Mono", Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK KR';
}
.pdf .serif {
font-family: PT Serif, Lyon-Text, Georgia, ui-serif, serif, 'Twemoji', 'Noto Color Emoji', 'Noto Serif CJK JP';
}
.pdf:lang(zh-CN) .serif {
font-family: PT Serif, Lyon-Text, Georgia, ui-serif, serif, 'Twemoji', 'Noto Color Emoji', 'Noto Serif CJK SC';
}
.pdf:lang(zh-TW) .serif {
font-family: PT Serif, Lyon-Text, Georgia, ui-serif, serif, 'Twemoji', 'Noto Color Emoji', 'Noto Serif CJK TC';
}
.pdf:lang(ko-KR) .serif {
font-family: PT Serif, Lyon-Text, Georgia, ui-serif, serif, 'Twemoji', 'Noto Color Emoji', 'Noto Serif CJK KR';
}
.pdf .mono {
font-family: PT Mono, iawriter-mono, Nitti, Menlo, Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK JP';
}
.pdf:lang(zh-CN) .mono {
font-family: PT Mono, iawriter-mono, Nitti, Menlo, Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK SC';
}
.pdf:lang(zh-TW) .mono {
font-family: PT Mono, iawriter-mono, Nitti, Menlo, Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK TC';
}
.pdf:lang(ko-KR) .mono {
font-family: PT Mono, iawriter-mono, Nitti, Menlo, Courier, monospace, 'Twemoji', 'Noto Color Emoji', 'Noto Sans Mono CJK KR';
}
.highlight-default {
color: rgba(55, 53, 47, 1);
}
.highlight-gray {
color: rgba(120, 119, 116, 1);
fill: rgba(120, 119, 116, 1);
}
.highlight-brown {
color: rgba(159, 107, 83, 1);
fill: rgba(159, 107, 83, 1);
}
.highlight-orange {
color: rgba(217, 115, 13, 1);
fill: rgba(217, 115, 13, 1);
}
.highlight-yellow {
color: rgba(203, 145, 47, 1);
fill: rgba(203, 145, 47, 1);
}
.highlight-teal {
color: rgba(68, 131, 97, 1);
fill: rgba(68, 131, 97, 1);
}
.highlight-blue {
color: rgba(51, 126, 169, 1);
fill: rgba(51, 126, 169, 1);
}
.highlight-purple {
color: rgba(144, 101, 176, 1);
fill: rgba(144, 101, 176, 1);
}
.highlight-pink {
color: rgba(193, 76, 138, 1);
fill: rgba(193, 76, 138, 1);
}
.highlight-red {
color: rgba(212, 76, 71, 1);
fill: rgba(212, 76, 71, 1);
}
.highlight-gray_background {
background: rgba(241, 241, 239, 1);
}
.highlight-brown_background {
background: rgba(244, 238, 238, 1);
}
.highlight-orange_background {
background: rgba(251, 236, 221, 1);
}
.highlight-yellow_background {
background: rgba(251, 243, 219, 1);
}
.highlight-teal_background {
background: rgba(237, 243, 236, 1);
}
.highlight-blue_background {
background: rgba(231, 243, 248, 1);
}
.highlight-purple_background {
background: rgba(244, 240, 247, 0.8);
}
.highlight-pink_background {
background: rgba(249, 238, 243, 0.8);
}
.highlight-red_background {
background: rgba(253, 235, 236, 1);
}
.block-color-default {
color: inherit;
fill: inherit;
}
.block-color-gray {
color: rgba(120, 119, 116, 1);
fill: rgba(120, 119, 116, 1);
}
.block-color-brown {
color: rgba(159, 107, 83, 1);
fill: rgba(159, 107, 83, 1);
}
.block-color-orange {
color: rgba(217, 115, 13, 1);
fill: rgba(217, 115, 13, 1);
}
.block-color-yellow {
color: rgba(203, 145, 47, 1);
fill: rgba(203, 145, 47, 1);
}
.block-color-teal {
color: rgba(68, 131, 97, 1);
fill: rgba(68, 131, 97, 1);
}
.block-color-blue {
color: rgba(51, 126, 169, 1);
fill: rgba(51, 126, 169, 1);
}
.block-color-purple {
color: rgba(144, 101, 176, 1);
fill: rgba(144, 101, 176, 1);
}
.block-color-pink {
color: rgba(193, 76, 138, 1);
fill: rgba(193, 76, 138, 1);
}
.block-color-red {
color: rgba(212, 76, 71, 1);
fill: rgba(212, 76, 71, 1);
}
.block-color-gray_background {
background: rgba(241, 241, 239, 1);
}
.block-color-brown_background {
background: rgba(244, 238, 238, 1);
}
.block-color-orange_background {
background: rgba(251, 236, 221, 1);
}
.block-color-yellow_background {
background: rgba(251, 243, 219, 1);
}
.block-color-teal_background {
background: rgba(237, 243, 236, 1);
}
.block-color-blue_background {
background: rgba(231, 243, 248, 1);
}
.block-color-purple_background {
background: rgba(244, 240, 247, 0.8);
}
.block-color-pink_background {
background: rgba(249, 238, 243, 0.8);
}
.block-color-red_background {
background: rgba(253, 235, 236, 1);
}
.select-value-color-pink {
background-color: rgba(245, 224, 233, 1);
}
.select-value-color-purple {
background-color: rgba(232, 222, 238, 1);
}
.select-value-color-green {
background-color: rgba(219, 237, 219, 1);
}
.select-value-color-gray {
background-color: rgba(227, 226, 224, 1);
}
.select-value-color-opaquegray {
background-color: rgba(255, 255, 255, 0.0375);
}
.select-value-color-orange {
background-color: rgba(250, 222, 201, 1);
}
.select-value-color-brown {
background-color: rgba(238, 224, 218, 1);
}
.select-value-color-red {
background-color: rgba(255, 226, 221, 1);
}
.select-value-color-yellow {
background-color: rgba(253, 236, 200, 1);
}
.select-value-color-blue {
background-color: rgba(211, 229, 239, 1);
}
.checkbox {
display: inline-flex;
vertical-align: text-bottom;
width: 16;
height: 16;
background-size: 16px;
margin-left: 2px;
margin-right: 5px;
}
.checkbox-on {
background-image: url("data:image/svg+xml;charset=UTF-8,%3Csvg%20width%3D%2216%22%20height%3D%2216%22%20viewBox%3D%220%200%2016%2016%22%20fill%3D%22none%22%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%3E%0A%3Crect%20width%3D%2216%22%20height%3D%2216%22%20fill%3D%22%2358A9D7%22%2F%3E%0A%3Cpath%20d%3D%22M6.71429%2012.2852L14%204.9995L12.7143%203.71436L6.71429%209.71378L3.28571%206.2831L2%207.57092L6.71429%2012.2852Z%22%20fill%3D%22white%22%2F%3E%0A%3C%2Fsvg%3E");
}
.checkbox-off {
background-image: url("data:image/svg+xml;charset=UTF-8,%3Csvg%20width%3D%2216%22%20height%3D%2216%22%20viewBox%3D%220%200%2016%2016%22%20fill%3D%22none%22%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%3E%0A%3Crect%20x%3D%220.75%22%20y%3D%220.75%22%20width%3D%2214.5%22%20height%3D%2214.5%22%20fill%3D%22white%22%20stroke%3D%22%2336352F%22%20stroke-width%3D%221.5%22%2F%3E%0A%3C%2Fsvg%3E");
}
</style>
</head>
<body>
<article id="112e77be-c354-4cdb-8e1f-c62be788ce47" class="page sans">
<header><img class="page-cover-image"
src="src/pmr_head.png"
style="object-position:center 100%" />
<h1 class="page-title">基于前提的跨模态推理评测</h1></header><div class="page-body"><p id="272c03ef-aa45-4773-923d-09db05214736" class=""></p><nav id="e8670c5a-a74b-4578-abc1-c602c5f71b8b" class="block-color-blue table_of_contents"><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#b240be71-8b25-4a26-8680-bd2f0a70453b">一、任务简介</a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#b01e00dc-8099-4288-9762-aa7251f74241">二、数据介绍</a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#0659d763-aa07-4483-bcfa-118c06fb2a1f">三、评价标准</a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#bdcd9598-4204-4d31-8797-13846f3bd587">四、预计赛程</a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#d35c9ea6-8810-47ce-a186-20482cc70b01"><strong>五、报名方式</strong></a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#6c565c6e-ad5e-4159-b49a-13eabb760d9d"><strong>六、奖项设置</strong></a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#897a157e-53f4-41d3-963d-d01dddd90ebe">七、评测委员会</a></div><div class="table_of_contents-item table_of_contents-indent-0"><a class="table_of_contents-link"href="#c88c55cb-27fe-4cc8-90c7-5397f98d0be7">附录</a></div></nav><h3 id="b240be71-8b25-4a26-8680-bd2f0a70453b" class="">一、任务简介</h3><p id="8932d961-6341-40ee-b8aa-fa6a97af8114" class="">在视觉与语言的跨模态推理方向上，常见做法是给定图片形式的事件和文本形式的问题，让机器在多个备选答案中选择符合图片内容的答案。我们认为过去的这类研究属于 "无条件"的跨模态推理，即对图像没有指定任何先验知识，而在实际场景中，却往往不是如此。认知心理学中的ABC理论（Ellis,1995）表明，人类的情绪和由此产生的行为并不直接来自于特定事件，而是来自于特定前提下对事件的解读。受此启发，我们提出了一个"基于前提的跨模态推理"任务（PMR）。在这个任务中，我们为相同图片提供了不同的前提背景信息（文本形式），要求模型先根据背景信息解读图片，再正确回答文本形式的问题。PMR的数据集包含15,360个人工标注的[前提+图片+问题] —— [答案]配对，每个样本都经过了多阶段的众包标注和审核，该任务以高质量的数据为机器跨模态推理提出了更高的挑战。</p><p id="cea19b15-a9f2-4d1b-93dd-391a94af5e72" class="">基于前提的跨模态推理评测设置为单个主任务，每一张图片配有一条文本形式的前提句和问句，以及四个可能的选项，模型需要基于前提线索理解图片，再从四个选项中选择唯一符合问题和图片的答案。</p><h3 id="b01e00dc-8099-4288-9762-aa7251f74241" class="">二、数据介绍</h3><p id="19bb9a86-c367-4525-9fcd-3ad54b887b59" class="">数据样例如下，每一张图片配有一条文本形式的前提句和一个问句，以及四个可能的选项，模型需要基于前提线索理解图片，再从四个选项中选择唯一符合问题和图片的答案，图片提供元素标注信息，方便只熟悉文本的参赛者参与评测，图片中的元素标注和文本中的实体保持一致。</p><figure id="bb2a4d49-605d-48c4-b8e5-3fc167c4e578" class="image"><a href="src/sample.png"><img style="width:1230px"src="src/sample.png" /></a></figure><p id="e526db00-f662-4e22-bcad-612c18294784" class="">我们的标注过程确保了仅凭借文本形式的前提线索或者图片形式的场景线索都无法在四个选项中选出唯一正确的选项，因此，该评测要求模型必须在重复利用前提、图片、问题的基础上做出分类。</p><p id="be75da79-894c-4b0a-8aed-ca444249642e" class="">数据规模上，该评测任务包含15,360个人工标注的[前提+图片+问题] —— [答案]配对，以及15,360个对抗匹配自动生成的前[前提+图片+问题] —— [答案]配对，前提句和答案句的平均长度分别是9.49和14.31词。</p><p id="ca151fb0-ff64-4aa6-aedd-b1a0ffe479e7" class="">将文本形式的条件引入跨模态推理是该评测的主要动机之一，因此前提句的设计至关重要。数据集中的前提句包含六个不同大类，具体分布情况如下图：</p><figure id="37d6e44f-d789-441c-bf6d-928e1aba0319" class="image"><a href="src/distribution.png"><img style="width:622px"src="src/distribution.png" /></a>】</figure><p id="a7d3adc7-20cf-4335-a447-4aaf2e8399c5" class="">更多数据样例及demo展示见<a href="https://2030nlp.github.io/PMR/">https://2030nlp.github.io/PMR/</a>。</p><p id="09afd530-f232-41f4-819c-d2fafe1e7842" class="">数据集论文：</p><pre id="52262f40-c06f-47d1-96f7-0d908bd9653f" class="code"><code>@inproceedings{dong-etal-2022-premise,
title = "Premise-based Multimodal Reasoning: Conditional Inference on Joint Textual and Visual Clues",
author = "Dong, Qingxiu and Qin, Ziwei and Xia, Heming and Feng, Tian and
Tong, Shoujie and
Meng, Haoran and
Xu, Lin and
Wei, Zhongyu and
Zhan, Weidong and
Chang, Baobao and
Li, Sujian and
Liu, Tianyu and
Sui, Zhifang",
booktitle = "ACL (Volume 1: Long Papers)",
month = may,
year = "2022",
address = "Dublin, Ireland",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.acl-long.66",
pages = "932--946",
}</code></pre>
<p id="09afd530-f232-41f4-819c-d2fafe2e7842" class="">数据集标注原始图片库来源于VCR论文，该论文收集并标注了110k来自movieclip的图片：</p>
<pre id="52262f40-c06f-47d1-96f7-0d908bd9654f" class="code"><code>@inproceedings{zellers2019vcr,
author = {Zellers, Rowan and Bisk, Yonatan and Farhadi, Ali and Choi, Yejin},
title = {From Recognition to Cognition: Visual Commonsense Reasoning},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}
}</code></pre><p id="729b21c0-2ae0-4f8f-855e-fb1ad05ee7c8" class=""></p><h3 id="0659d763-aa07-4483-bcfa-118c06fb2a1f" class="">三、评价标准</h3><p id="b1f1b7f8-c62d-4890-ab00-7cff7479dab0" class="">该评测任务为分类任务，每个样例有四个备选答案，其中有唯一正确答案，我们用模型选中正确答案的Accuracy作为评测指标。最终的评测成绩计算比较简单，直接采用每一组最后一次提交结果的Accuracy作为最终成绩。</p><h3 id="bdcd9598-4204-4d31-8797-13846f3bd587" class="">四、预计赛程</h3><ul id="d91c0bf6-6c6e-49dc-b30c-a66b913c7637" class="bulleted-list"><li style="list-style-type:disc">评测开始（2022年6月1日）<ul id="76781a37-7dcc-486b-bad9-5e3e052f1f29"class="bulleted-list"><li style="list-style-type:circle">发布报名系统，开放报名 （截止日期2022年6月20日）</li></ul><ul id="12a68878-3206-41c3-b215-3e0378a76c2d" class="bulleted-list"><li style="list-style-type:circle">发布训练集以及验证集</li></ul><ul id="9d2c2ce4-eb82-4925-bbc2-4ccf723af3e3" class="bulleted-list"><li style="list-style-type:circle">发布基线模型、排行榜页面、结果提交页面</li></ul></li></ul><ul id="f79c7776-eebe-44ad-8672-a9c949c8252e" class="bulleted-list"><li style="list-style-type:disc">第一阶段（2022年6月1日—2022年8月20日，近3个月）：<ulid="7dfc32bd-1ae4-4361-9c1a-bd528d8b9965" class="bulleted-list"><li style="list-style-type:circle">参赛者需要提交验证集结果，排行榜按验证集成绩排序</li></ul><ul id="5f1566e3-38d6-4fb0-b013-6b90dc64d5d4" class="bulleted-list"><li style="list-style-type:circle">参赛者每周可提交3次模型，排行榜每周日更新</li></ul><ul id="81b7e777-caa1-4ddd-9e01-ba9a3a34150c" class="bulleted-list"><li style="list-style-type:circle">此阶段可以不提交模型</li></ul></li></ul><ul id="ff9022ad-7dc0-40c2-b35a-b1d3e55cb1b2" class="bulleted-list"><li style="list-style-type:disc">第二阶段（2022年8月20日—2022年9月20日，约1个月）：<ulid="e27ac5b3-febf-4bd7-9ae5-66af0bfd3cb2" class="bulleted-list"><li style="list-style-type:circle">此阶段开始的同时，报名截止</li></ul><ul id="40fea03d-80bf-47dd-aa30-57e877ac3537" class="bulleted-list"><li style="list-style-type:circle">发布不带答案的测试集</li></ul><ul id="13c745f5-2e98-41a3-b0ae-39fd0a5710f9" class="bulleted-list"><li style="list-style-type:circle">参赛者需要提交测试集结果，排行榜按测试集成绩排序</li></ul><ul id="a68dfc9c-75d7-48aa-b0cb-0acf56b4496b" class="bulleted-list"><li style="list-style-type:circle">参赛者每周可提交3次模型，排行榜每天更新</li></ul><ul id="d8148f10-3555-40c2-8847-4c017aba17a2" class="bulleted-list"><li style="list-style-type:circle">此阶段不提交模型视为弃权</li></ul></li></ul><ul id="9992581d-0a4d-46c7-b77c-3f514df540e3" class="bulleted-list"><li style="list-style-type:disc">第三阶段（2022年9月20日—2022年9月25日，共5天）：<ulid="afe3db13-3538-41b0-bc97-4e15f9281bce" class="bulleted-list"><li style="list-style-type:circle">参赛者选择1个模型作为最终参赛模型提交</li></ul><ul id="f8ca5cd8-449d-4e31-8ac6-f32705f0592d" class="bulleted-list"><li style="list-style-type:circle">此阶段不提交模型视为弃权</li></ul></li></ul><ul id="af793b62-36c4-4fe3-a455-9d1fd6996567" class="bulleted-list"><li style="list-style-type:disc">总结回顾阶段（2022年10月1日—2022年10月16日）<ulid="b33f2eac-89aa-4361-abb3-c1a5988ba422" class="bulleted-list"><li style="list-style-type:circle">公布结果并邀请优秀评测单位撰写技术报告（2022年10月1日）</li></ul><ul id="30e66194-8878-444b-938f-60689cfc623e" class="bulleted-list"><li style="list-style-type:circle">技术报告收取截止（2022年10月10日）</li></ul><ul id="c2589561-f7f1-44fd-a8be-43dc60f3c6f8" class="bulleted-list"><li style="list-style-type:circle">CCL 2022评测研讨会（2022年10月14日—2022年10月16日）</li></ul></li></ul><h3 id="d35c9ea6-8810-47ce-a186-20482cc70b01" class=""><strong>五、报名方式</strong></h3><p id="f7242c31-e2b7-40ca-8d10-41c00794ce6d" class="block-color-red">👉<a href="https://docs.google.com/forms/d/e/1FAIpQLSfpe6Dvi7eedexmDjRlbo0O6e-2PvnMNtCaBHNk9LiXwlBXfw/viewform?usp=sf_link">点击填写报名表</a>（截止日期2022年6月20日，如果无法打开，则填写<a href="src/table.pdf">PDF格式报名表</a>并发送到[email protected]）。</p><h3 id="6c565c6e-ad5e-4159-b49a-13eabb760d9d" class=""><strong>六、奖项设置</strong></h3><p id="320e038e-e80a-4ce8-9a58-ed2acc21b3e6" class=""><strong>评测奖金由华为公司赞助</strong>，奖池共计30000元：</p><p id="3e6aebc8-115e-4f62-83cc-b5f0934a3a86" class="">一等奖（1名），奖金10000元；</p><p id="9ff43082-a03c-41bd-b922-5cf8af5d7483" class="">二等奖（2名），各奖5000元；</p><p id="1d946077-670a-4b24-88f8-4f8d166de029" class="">三等奖（4名），各奖2500元。</p><p id="a3bbd089-271f-4c1b-8e96-88172c25c8a4" class=""></p><h3 id="897a157e-53f4-41d3-963d-d01dddd90ebe" class="">七、评测委员会</h3><p id="a5c4e363-c622-42be-ad89-8bf0d3e312cc" class="">单位：北京大学，复旦大学</p><p id="70fcfcce-9ca4-4e8c-aa98-9c1d39427465" class="">主席：穗志方，詹卫东，常宝宝，李素建，魏忠钰</p><p id="cc8f138e-7fb8-4c4f-a8ad-d9b7c47b4f5b" class="">成员：董青秀，秦梓巍，夏鹤明 等</p><p id="cc8f138e-7fb8-4c4f-a8ad-d9b7c47b4f5c" class="">联系邮箱：[email protected]</p><p id="d943b740-a48d-48e5-ad9c-879049d1f880" class=""></p><h3 id="c88c55cb-27fe-4cc8-90c7-5397f98d0be7" class="">附录</h3></p></p><figure id="bfbcda87-d704-4c89-8224-a85a3f59ce5c" class="link-to-page"><a href="src/agreement.html">CCL2022基于前提的跨模态推理评测选手参赛协议</a></figure><figure id="320e926d-6a23-40aa-8b98-a6d085954fc2" class="link-to-page"><a href="src/LICENSE.html">CCL2022基于前提的跨模态推理评测数据集使用许可</a></figure><figure id="320e926d-6a23-40aa-8b98-a6d085954fc1" class="block-color-red"><a href="src/notice.html">🎙️评测最新通知站点</a></figure></div></article></body></html>