-
Notifications
You must be signed in to change notification settings - Fork 2
/
index.html
2908 lines (2784 loc) · 111 KB
/
index.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
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<html>
<head>
<title>CIF-Bench</title>
<style>
.hidden {
display: none;
}
</style>
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
<script src="https://kit.fontawesome.com/f8ddf9854a.js" crossorigin="anonymous"></script>
<meta charset="utf-8">
<meta name="description"
content="CIF-Bench: A Chinese Instruction-Following Benchmark for Evaluating the Generalizability of Large Language Models">
<meta name="keywords" content="CIF-Bench, LLM, LLM Evaluation, Instruction Following, Large Language Model, artificial intelligence, AI, AGI, artificial general intelligence">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title> CIF-Bench: A Chinese Instruction-Following Benchmark for Evaluating the Generalizability of Large Language Models</title>
<link rel="icon" href="./c_static/images/cif_bench_logo.png">
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="stylesheet" href="./static/css/leaderboard.css">
<!-- <link href="https://unpkg.com/tabulator-tables@5.5.2/dist/css/tabulator_bulma.min.css" rel="stylesheet">
<script type="text/javascript" src="https://unpkg.com/tabulator-tables@5.5.2/dist/js/tabulator.min.js"></script> -->
<script type="text/javascript" src="static/js/sort-table.js" defer></script>
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
<script src="./static/js/question_card.js"></script>
<script src="./data/results/data_setting.js" defer></script>
<script src="./data/results/model_scores.js" defer></script>
<script src="./visualizer/data/data_public.js" defer></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.7/MathJax.js?config=TeX-MML-AM_CHTML" async></script>
</head>
<body class="cmmmu-container">
<nav class="navbar" role="navigation" aria-label="main navigation">
<div class="navbar-brand">
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
<!-- <div class="navbar-menu">
<div class="navbar-start" style="flex-grow: 1; justify-content: center;">
<div class="navbar-item has-dropdown is-hoverable">
<a class="navbar-link">
More Research
</a>
<div class="navbar-dropdown">
<a class="navbar-item" href="https://tiger-ai-lab.github.io/MAmmoTH/">
<b>MAmmoTH</b> <p style="font-size:18px; display: inline; margin-left: 5px;">🔥</p>
</a>
<a class="navbar-item" href="https://osu-nlp-group.github.io/TableLlama/">
TableLlama
<a class="navbar-item" href="https://osu-nlp-group.github.io/MagicBrush/">
MagicBrush
</a>
<a class="navbar-item" href="https://osu-nlp-group.github.io/Mind2Web/">
Mind2Web
</a>
</a>
</a>
</div>
</div>
</div>
</div> -->
</nav>
<section class="hero cmmmu-green">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title is-bold">
<img src="c_static/images/cif_bench_logo.png" style="width:1em;vertical-align: middle" alt="Logo"/>
<span class="cmmmu" style="vertical-align: middle">CIF-Bench</span>
</h1>
<h2 class="subtitle is-3 publication-subtitle">
A Chinese Instruction-Following Benchmark for Evaluating the Generalizability of Large Language Models
</h2>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://yizhilll.github.io/" style="text-decoration: none; color: inherit;">Yizhi Li*<sup style="color:#6fbf73;"></sup></a>,
</span>
<span class="author-block">
<a href="https://twitter.com/GeZhang86038849/" style="text-decoration: none; color: inherit;">Ge Zhang*<sup style="color:#6fbf73;"></sup></a>
,
</span>
<span class="author-block">
<a href="https://scholar.google.com/citations?user=hty-MWIAAAAJ&hl=en" style="text-decoration: none; color: inherit;">Xingwei Qu*<sup style="color:#6fbf73;"></sup></a>
,
</span>
<br>
<span class="author-block">Jiali Li<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Zhaoqun Li<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Zekun Wang<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Hao Li<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Ruibin Yuan<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Yinghao Ma<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Kai Zhang<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Wangchunshu Zhou<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Yiming Liang<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Lei Zhang<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Lei Ma<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Jiajun Zhang<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Zuowen Li<sup style="color:#007bff;"></sup>,</span>
<span class="author-block">Stephen W. Huang<sup style="color:#007bff;"></sup>,</span>
<br>
<span class="author-block">
<a href="https://chenghualin.wordpress.com/" style="text-decoration: none; color: inherit;">Chenghua Lin<sup style="color:#ffac33;">†</sup></a>
,
</span>
<span class="author-block">
<a href="https://wenhuchen.github.io/" style="text-decoration: none; color: inherit;">Wenhu Chen<sup style="color:#ffac33;">†</sup></a>
,
</span>
<span class="author-block">
<a href="https://bigaidream.github.io/" style="text-decoration: none; color: inherit;">Jie Fu<sup style="color:#ffac33;">†</sup></a>
</span>
</div>
<!-- <br>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup style="color:#6fbf73;">1</sup>IN.AI Research,</span>
<span class="author-block"><sup style="color:#ffac33;">2</sup>University of Waterloo,</span>
<span class="author-block"><sup style="color:#ed4b82;">3</sup>The Ohio State University,</span>
<span class="author-block"><sup style="color:#007bff;">4</sup>Independent,</span></br>
<span class="author-block"><sup style="color:#ffac33;">5</sup>Carnegie Mellon University,</span>
<span class="author-block"><sup style="color:#ed4b82;">6</sup>University of Victoria,</span>
<span class="author-block"><sup style="color:#9b51e0;">7</sup>Princeton University</span>
</div>
<br>
<div class="is-size-5 publication-authors">
<span class="author-block">*Core Contributors</span><br>
<span class="author-block">†Corresponding to:</span>
<span class="author-block"><a href="mailto:yizhi.li@hotmail.com">yizhi.li@hotmail.com</a>,</span>
<span class="author-block"><a href="mailto:su.809@osu.edu">su.809@osu.edu</a>,</span>
<span class="author-block"><a href="mailto:wenhuchen@uwaterloo.ca">wenhuchen@uwaterloo.ca</a></span>
</div> -->
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<!-- @PAN TODO: change links -->
<a href="https://arxiv.org/abs/2402.13109"
class="external-link button is-normal is-rounded">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>arXiv</span>
</a>
</span>
<!-- <span class="link-block">
<a href="https://huggingface.co/papers"
class="external-link button is-normal is-rounded">
<span class="icon">
<p style="font-size:18px">🤗</p>
</span>
<span>HF Paper</span>
</a>
</span> -->
<!-- <span class="link-block">
<a href="https://huggingface.co/datasets"
class="external-link button is-normal is-rounded">
<span class="icon">
<p style="font-size:18px">🤗</p>
</span>
<span>Dataset</span>
</a>
</span> -->
<span class="link-block">
<a href="https://github.com/yizhilll/CIF-Bench"
class="external-link button is-normal is-rounded">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
<!-- Dataset Link. -->
<!-- Twitter Link. -->
<!-- <span class="link-block">
<a href="https://twitter.com/GeZhang86038849/status/1749660947223044325"
class="external-link button is-normal is-rounded">
<span class="icon has-text-white">
<i class="fa-brands fa-x-twitter"></i>
</span>
<span>Twitter</span>
</a>
</span>
</div> -->
</div>
</div>
</div>
</div>
</div>
</section>
<style>
.center {
display: block;
margin-left: auto;
margin-right: auto;
width: 80%;
}
</style>
<section class="hero teaser">
<div class="container is-max-desktop">
<!-- <div class="hero-body">
<img src="static/images/tease_scores.png" alt="Examples from the dataset"/>
<h2 class="subtitle has-text-centered">
<span class="dnerf">Nerfies</span> turns selfie videos from your phone into
free-viewpoint
portraits.
</h2>
</div> -->
<!-- <div class="box m-5"> -->
<div class="content has-text-centered">
<img src="c_static/images/overview_cifbench.png" alt="Task Category Distribution" width="50%"/>
<p> Overview of the CIF-Bench. It is distinguished by its <b>comprehensiveness</b>. The radii have three groups, determined by the number of tasks contained (≤ 10, ≤ 20, and > 20).</p>
</div>
<!-- </div> -->
</div>
</div>
</section>
<section class="section">
<div class="container" style="margin-bottom: 2vh;">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<!-- <h2 class="title ">🔔News</h2> -->
<div class="content has-text-justified">
<!-- <p>
<b>🔥[2023-12-04]: Our evaluation server for the test set is now available on <a href="https://eval.ai/web/challenges/challenge-page/2179/overview">EvalAI</a>. We welcome all submissions and look forward to your participation! 😆</b>
</p> -->
</div>
<h2 class="title is-3">Abstract </h2>
<div class="content has-text-justified">
<p>
The advancement of large language models (LLMs) has enhanced the ability to generalize across a wide range of unseen natural language processing (NLP) tasks through instruction-following. Yet, their effectiveness often diminishes in low-resource languages like Chinese, exacerbated by biased evaluations from data leakage, casting doubt on their true generalizability to new linguistic territories. In response, we introduce the Chinese Instruction-Following Benchmark (CIF-Bench), designed to evaluate the zero-shot generalizability of LLMs to the Chinese language. CIF-Bench comprises 150 tasks and 15,000 input-output pairs, developed by native speakers to test complex reasoning and Chinese cultural nuances across 20 categories. To mitigate evaluation bias, we release only half of the dataset publicly, with the remainder kept private, and introduce diversified instructions to minimize score variance, totaling 45,000 data instances. Our evaluation of 28 selected LLMs reveals a noticeable performance gap, with the best model scoring only 52.9%, highlighting the limitations of LLMs in less familiar language and task contexts. This work aims to uncover the current limitations of LLMs in handling Chinese tasks, pushing towards the development of more culturally informed and linguistically diverse models with the released data and benchmark.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
</section>
<!-- DATASET SECTION -->
<section class="hero is-light is-small">
<div class="hero-body has-text-centered">
<h1 class="title is-1 mmmu">
<img src="c_static/images/cif_bench_logo.png" style="width:1em;vertical-align: middle" alt="Logo"/>
<span class="mmmu" style="vertical-align: middle">CIF-Bench</span>
</h1>
</div>
</section>
<!-- <section class="section">
<div class="container" style="margin-bottom: 2vh;">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Introduction</h2>
<div class="content has-text-justified">
<p> -->
<section class="section">
<div class="container">
<div class="columns is-centered has-text-centered">
<!-- <div class="column is-full-width has-text-centered"> -->
<div class="column is-four-fifths">
<h2 class="title is-3">Overview</h2>
<div class="content has-text-justified">
<p>
we introduce the Chinese Instruction-Following Benchmark (CIF-Bench), a novel benchmark designed for the zero-shot generalizability evaluation of LLMs, with Chinese serving as an insightful example for multilingual transferred instruction-following tasks. Our benchmark comprises 150 tasks and 15, 000 input-output pairs, with the assistance of native speaker annotators, ensuring the inclusion of human-authored tasks that are not only challenging but also naturally expressed. A significant portion (38.7%) of these tasks are designed to test a model’s complex natural language inference (NLI) and reasoning capabilities, as well as drawing upon Chinese culture spread across 20 distinct categories. In an effort to mitigate future evaluation biases from data leakage, we decide to publicly release only half of the data instances, reserving the rest as a private dataset to maintain an impartial benchmark. Furthermore, CIF-Bench enhances its robustness by introducing 5 variations of instructions per task, using these to diminish score variance in private split evaluations. CIF-Bench also pioneers a model-based automatic pipeline designed to tackle the inherent challenges of evaluating open-ended natural language generation outputs. </p>
<img src="c_static/images/teaser.png" alt="A large language model can tackle English task translated to Chinese, but fail to respond to instruction originally in Chinese." class="center">
<br>
<p>
By selecting a range of popular LLMs that support Chinese for evaluation, we aim to depict the limits of current instruction-following capabilities in language transfer contexts as the many models follow an English-oriented pre-training paradigm (Huang et al., 2023b). Our findings reveal that even the best-performing model achieves a score of only 52.9% on CIF-Bench, underscoring the gap that exists when LLMs are confronted with tasks in a less-familiar language and unseen data instances. We find that this performance decrement is particularly noticeable in scenarios involving unseen tasks and unseen input-output pairs, contrasting with the models’ performance on existing Chinese datasets and translated English-language tasks. Such results suggest that while LLMs exhibit impressive generalizability in a context more aligned with observed data, their effectiveness diminishes when faced with the dual challenges of unacquainted languages and novel tasks. </p>
<p>
To summarize our contributions, we: (1) Present a new benchmark that addresses a critical gap in existing NLP research by focusing on the generalizability of LLMs to an underrepresented language in terms of training and evaluation resources; (2) Construct an instruction-following evaluation dataset with 150 tasks and 45, 000 data samples, and release half of the input-output pairs for future LLM evaluation research; (3) Provide an in-depth analysis of 28 LLMs, revealing their limitations in adapting to less familiar languages and task contexts, offering insights into where improvements are needed for instruction-following generalizability. </p>
</div>
</div>
</div>
<!-- <div class="columns is-centered m-6">
<div class="column is-max-desktop has-text-centered">
<h2 class="title is-3" id="visualization">Visualization</h2>
<iframe src="visualizer/explore.html" style="width: 100%;min-height: 100vh; border-radius: 20px;"></iframe>
</div>
</div> -->
</div>
</section>
<!-- RESULTS SECTION -->
<section class="hero is-light is-small">
<div class="hero-body has-text-centered">
<h1 class="title is-1 mmmu">Experiment Results</h1>
</div>
</section>
<section class="section">
<div class="container">
<!-------------------------------------------------------------------- RESULTS SECTION -------------------------------------------------------------------->
<div class="columns is-centered m-6">
<div class="column is-full has-text-centered content">
<h2 class="title is-3" id="leaderboard">Leaderboard</h2>
<div class="content">
<div class="content has-text-justified">
<p>
As the CIF-Bench aims to provide a comprehensive evaluation of the LLM instruction-following capability, we argue that the metrics should be designed case by case in task granularity to evaluate the open-ended textual outputs, rather than simply reformatting all tasks into choice questions and using the conditional probability to approximate the models' predictions.
</p>
<p>
After a thorough review of the task instructions, we categorize the output requirements into the four following types and design corresponding task-level metrics. Multi-class Classification: We use accuracy as the metric if the task requires the model to predict one label from 2 or more classes in the output. Multi-label Classification: We use F1 score as the metric if the task requires the model to predict one label from 2 or more classes in the output. Creative Generation: Regarding the tasks that have no absolute criteria of the standard answer, we require a model-based evaluator to provide information regarding a given output, including creativity, fluency, the level of instruction-following, and the confidence of the evaluator. Semantic Similarity: For the remaining tasks that can be evaluated by the semantic similarity between the golden reference and model output, we use a pre-trained language. All scores used in CIF-Bench either naturally range from 0 to 1, or are normalized to the same range.
</p>
<p>
One core dilemma in evaluating the open-ended instruction-following capabilities of LLMs is that model predictions are hard to verify even with reference answers. For instance, it is intractable to handcraft regex rules to extract the predictions from LLMs for the extensive number of tasks, since the answers could be expressed in various formats, or drowned in redundant contexts like reasoning progress. Inspired by G-Eval, we leverage OpenAI's GPT-4 as a relatively reliable evaluator for multi-class classification, multi-label classification, and creative generation tasks, to overcome such issues. The GPT-4 evaluator is prompted to assess the outputs according to the given task instruction and the input-output reference. For the answers that can be evaluated with semantic similarity, we use a lightweight multilingual encoder, BLEURT, to measure the relevance between the reference and LLM output.
</p>
Given a set of task instructions <script type="math/tex">I</script>, we denote the performance score of model <script type="math/tex">m</script> on task <script type="math/tex">t</script> as: <script type="math/tex">S^{m}_{t}=\frac{1}{|D_t|}\sum_{d \in D_t}\frac{1}{|I|}\sum_{i \in I}{{s^{m}_{t}(i,d)}}</script>, where <script type="math/tex">D_t</script> refers to the set of data samples for task <script type="math/tex">t</script>. In the case of the public split, the instruction set <script type="math/tex">I</script> is reduced to one single element. In we take the average of task-level scores <script type="math/tex">\overline{S^m}</script> as the indicator of overall performance for a model <script type="math/tex">m</script>.
</p>
</div>
<!-- <div class="model-labels-container"> -->
<!-- <span class="leaderboard-label" style="background-color: #f8fffe;">Open-Source</span>
<span class="leaderboard-label" style="background-color: #f9f2f8;">Closed</span> -->
<!-- <span class="leaderboard-label" style="background-color: #e4efdc;">Multimodal</span>
<span class="leaderboard-label" style="background-color: #e0ebf3;">Language-Only</span>
<span class="leaderboard-label" style="background-color: #def9cb;">Proprietary</span>
</div> -->
<!-- <table id="table1" class="js-sort-table">
<tr>
<td class="js-sort-number"><strong>Reset</strong></td>
<td class="js-sort-number"><strong>Test Overall</strong></td>
<td class="js-sort-number"><strong>Validation Overall</strong></td>
<td class="js-sort-number"><strong>Art & Design</strong></td>
<td class="js-sort-number"><strong>Business</strong></td>
<td class="js-sort-number"><strong>Science</strong></td>
<td class="js-sort-number"><strong>Health & Medicine</strong></td>
<td class="js-sort-number"><strong>Human & Social Sci.</strong></td>
<td class="js-sort-number"><strong>Tech & Eng.</strong></td>
</tr>
<tr style="background-color: #def9cb;">
<td style="text-align: left;">
<b> GPT-4V </b>
</td>
<td> <b>43.7</b> </td>
<td> <b>42.5</b> </td>
<td> 61.0 </td>
<td> <b>36.3</b> </td>
<td> <b>40.9</b> </td>
<td> <b>46.8</b> </td>
<td> <b>44.2</b> </td>
<td> <b>41.5</b> </td>
</tr>
<tr style="background-color: #def9cb;">
<td style="text-align: left;">
<b>Qwen-VL-Plus </b>
</td>
<td> 36.8 </td>
<td> 39.5 </td>
<td> 61.5 </td>
<td> 23.2 </td>
<td> 32.8 </td>
<td> 40.5 </td>
<td> 43.4 </td>
<td> 33.3 </td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b> Yi-VL-34B </b>
</td>
<td style="text-decoration: underline;"> 36.5 </td>
<td style="text-decoration: underline;"> 36.2 </td>
<td style="text-decoration: underline;"> <b>62.9</b> </td>
<td> 19.1 </td>
<td> 31.5 </td>
<td style="text-decoration: underline;"> 42.1 </td>
<td style="text-decoration: underline;"> 42.5 </td>
<td style="text-decoration: underline;"> 34.5 </td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b> Yi-VL-6B </b>
</td>
<td> 35.0 </td>
<td> 35.8 </td>
<td> 58.0 </td>
<td> 19.9 </td>
<td style="text-decoration: underline;"> 32.3 </td>
<td> 39.3 </td>
<td> 40.6 </td>
<td> 32.1 </td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>InternVL-Chat-V1.1</b>
</td>
<td>34.0</td>
<td>34.7</td>
<td>56.7</td>
<td>19.7</td>
<td>28.6</td>
<td>39.2</td>
<td>39.6</td>
<td>32.3</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>Qwen-VL-Chat</b>
</td>
<td>31.3</td>
<td>30.7</td>
<td>52.6</td>
<td>18.5</td>
<td>26.9</td>
<td>33.4</td>
<td>34.1</td>
<td>31.4</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>SPHINX-MoE</b>
</td>
<td>29.5</td>
<td>29.3</td>
<td>41.7</td>
<td style="text-decoration: underline;">20.3</td>
<td>27.8</td>
<td>28.9</td>
<td>31.8</td>
<td>30.9</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>InternVL-Chat-ViT-6B-Vicuna-7B</b>
</td>
<td>26.7</td>
<td>26.4</td>
<td>39.7</td>
<td>13.8</td>
<td>23.0</td>
<td>31.7</td>
<td>26.5</td>
<td>28.5</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>InternVL-Chat-ViT-6B-Vicuna-13B</b>
</td>
<td>26.1</td>
<td>27.4</td>
<td>38.5</td>
<td>13.9</td>
<td>22.1</td>
<td>30.2</td>
<td>29.8</td>
<td>27.5</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>Emu2-Chat</b>
</td>
<td>24.5</td>
<td>23.8</td>
<td>35.3</td>
<td>11.7</td>
<td>22.1</td>
<td>25.5</td>
<td>28.0</td>
<td>27.1</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>CogAgent-Chat</b>
</td>
<td>23.6</td>
<td>24.6</td>
<td>33.8</td>
<td>14.1</td>
<td>20.6</td>
<td>26.3</td>
<td>24.8</td>
<td>25.3</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>Chinese-LLaVa</b>
</td>
<td>23.4</td>
<td>25.5</td>
<td>34.4</td>
<td>11.7</td>
<td>21.6</td>
<td>25.5</td>
<td>26.3</td>
<td>24.7</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>VisCPM</b>
</td>
<td>22.7</td>
<td>25.2</td>
<td>37.7</td>
<td>11.3</td>
<td>19.1</td>
<td>26.1</td>
<td>24.0</td>
<td>23.7</td>
</tr>
<tr style="background-color: #e4efdc;">
<td style="text-align: left;">
<b>mPLUG-Owl2</b>
</td>
<td>22.2</td>
<td>20.8</td>
<td>30.4</td>
<td>13.3</td>
<td>19.6</td>
<td>25.2</td>
<td>24.7</td>
<td>23.4</td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Yi-6B + OCR </b>
</td>
<td> 26.8 </td>
<td> 28.4 </td>
<td> 33.4 </td>
<td> 16.9 </td>
<td> 24.8 </td>
<td> 32.3 </td>
<td> 33.2 </td>
<td> 25.5 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Qwen-7B + OCR </b>
</td>
<td> 26.1 </td>
<td> 27.0 </td>
<td> 44.6 </td>
<td> 14.3 </td>
<td> 22.1 </td>
<td> 29.3 </td>
<td> 29.8 </td>
<td> 25.4 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Qwen-7B </b>
</td>
<td> 25.1 </td>
<td> 24.7 </td>
<td> 43.8 </td>
<td> 12.6 </td>
<td> 20.7 </td>
<td> 30.5 </td>
<td> 26.9 </td>
<td> 24.5 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Baichuan-7B + OCR </b>
</td>
<td> 24.7 </td>
<td> 25.3 </td>
<td> 40.2 </td>
<td> 15.2 </td>
<td> 21.0 </td>
<td> 27.9 </td>
<td> 30.7 </td>
<td> 22.8 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Baichuan-7B </b>
</td>
<td> 24.3 </td>
<td> 26.0 </td>
<td> 42.7 </td>
<td> 12.6 </td>
<td> 19.6 </td>
<td> 28.0 </td>
<td> 27.8 </td>
<td> 23.9 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> Yi-6B </b>
</td>
<td> 24.2 </td>
<td> 25.6 </td>
<td> 26.3 </td>
<td> 15.0 </td>
<td> 23.4 </td>
<td> 29.1 </td>
<td> 27.0 </td>
<td> 24.7 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> DeepSeek-7B + OCR </b>
</td>
<td> 23.2 </td>
<td> 25.2 </td>
<td> 41.2 </td>
<td> 13.2 </td>
<td> 19.4 </td>
<td> 26.1 </td>
<td> 26.5 </td>
<td> 21.8 </td>
</tr>
<tr style="background-color: #e0ebf3;">
<td style="text-align: left;">
<b> DeepSeek-7B </b>
</td>
<td> 21.9 </td>
<td> 22.3 </td>
<td> 41.3 </td>
<td> 11.2 </td>
<td> 18.3 </td>
<td> 23.5 </td>
<td> 24.7 </td>
<td> 21.3 </td>
</tr>
<tr style="background-color: white;">
<td style="text-align: left;">
<b>Frequent Choice</b>
</td>
<td>26.0</td>
<td>24.1</td>
<td>36.2</td>
<td>11.8</td>
<td>23.9</td>
<td>30.2</td>
<td>28.5</td>
<td>27.7</td>
</tr>
<tr style="background-color: white;">
<td style="text-align: left;">
<b>Random Choice</b>
</td>
<td>21.6</td>
<td>21.6</td>
<td>32.9</td>
<td>9.1</td>
<td>18.8</td>
<td>23.8</td>
<td>23.8</td>
<td>23.9</td>
</tr> -->
<!-- <tr>
<td colspan="8" style="font-size: 18px;"><b>Large Language Models (LLMs): Only Text as Input</b></td>
</tr> -->
<!-- <tr style="background-color: #f8fffe;">
<td style="text-align: left;"><b>Llama2 7B</b></td>
<td>28.7</td>
<td>30.7</td>
<td>27.2</td>
<td>26.7</td>
<td>27.7</td>
<td>32.6</td>
<td>29.8</td>
</tr>
<tr style="background-color: #f4fdf5;">
<td style="text-align: left;"><b>FLAN-T5-XXL</b></td>
<td><b>31.2</b></td>
<td>36.8</td>
<td><b>28.9</b></td>
<td>26.7</td>
<td>32.8</td>
<td><b>44.8</b></td>
<td><b>28.3</b></td>
</tr>
<tr style="background-color: #f4fdf5;">
<td style="text-align: left;"><b>FLAN-T5-XXL + OCR</b></td>
<td><b>31.9</b></td>
<td>36.2</td>
<td>28.8</td>
<td>26.2</td>
<td>32.6</td>
<td><b>50.5</b></td>
<td><b>29.7</b></td>
</tr>
<tr style="background-color: #f4fdf5;">
<td style="text-align: left;"><b>FLAN-T5-XXL + LLaVA Caption</b></td>
<td><b>31.9</b></td>
<td><b>38.4</b></td>
<td>27.8</td>
<td><b>27.0</b></td>
<td><b>33.2</b></td>
<td>49.9</td>
<td>28.7</td>
</tr>
<tr style="background-color: #e7fde9;">
<td style="text-align: left;"><b>Vicuna-13B</b></td>
<td>31.0</td>
<td>35.1</td>
<td><b>30.1</b></td>
<td>24.7</td>
<td>31.4</td>
<td>44.8</td>
<td>30.1</td>
</tr>
<tr style="background-color: #e7fde9;">
<td style="text-align: left;"><b>Vicuna-13B + OCR</b></td>
<td>31.9</td>
<td>37.1</td>
<td>28.6</td>
<td><b>26.5</b></td>
<td>32.0</td>
<td>49.3</td>
<td>30.0</td>
</tr>
<tr style="background-color: #e7fde9;">
<td style="text-align: left;"><b>Vicuna-13B + LLaVA Caption</b></td>
<td><b>32.7</b></td>
<td><b>42.0</b></td>
<td>26.8</td>
<td>26.2</td>
<td><b>33.4</b></td>
<td><b>49.4</b></td>
<td><b>31.4</b></td>
</tr> -->
<!-- <tr style="background-color: #f8fffe;">
<td style="text-align: left;"><b>GPT-4 Text</b></td>
<td>33.8</td>
<td>32.9</td>
<td>28.5</td>
<td>30.6</td>
<td>41.3</td>
<td>53.0</td>
<td>28.4</td>
</tr> -->
<!-- </table> -->
<div class="columns is-centered m-6">
<div class="column is-full has-text-centered content">
<h2 class="title is-3">Leaderboard</h2>
<style>
.scrollable-table {
overflow-y: auto;
height: 600px; /* Adjust the height as needed */
}
table {
border-collapse: collapse;
width: 100%;
}
th, td {
border: 1px solid #dddddd;
text-align: left;
padding: 8px;
}
th {
background-color: #f2f2f2;
}
</style>
<div class="scrollable-table">
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th>Model Name</th>
<th>Overall</th>
<th>Chinese Culture</th>
<th>Classification</th>
<th>Code</th>
<th>Commonsense</th>
<th>Creative NLG</th>
<th>Evaluation</th>
<th>Grammar</th>
<th>Linguistic</th>
<th>Motion Detection</th>
<th>NER</th>
<th>NLI</th>
<th>QA</th>
<th>Reasoning</th>
<th>Role Playing</th>
<th>Sentiment</th>
<th>Structured Data</th>
<th>Style Transfer</th>
<th>Summarization</th>
<th>Toxic</th>
<th>Translation</th>
</tr>
</thead>
<tbody>
<tr>
<td>Baichuan2-13B-Chat</td>
<td>0.529</td>
<td>0.520</td>
<td>0.674</td>
<td>0.333</td>
<td>0.641</td>
<td>0.497</td>
<td>0.686</td>
<td>0.542</td>
<td>0.528</td>
<td>0.578</td>
<td>0.563</td>
<td>0.632</td>
<td>0.569</td>
<td>0.515</td>
<td>0.752</td>
<td>0.624</td>
<td>0.459</td>
<td>0.462</td>
<td>0.332</td>
<td>0.441</td>
<td>0.273</td>
</tr>
<tr>
<td>Qwen-72B-Chat</td>
<td>0.519</td>
<td>0.486</td>
<td>0.630</td>
<td>0.296</td>
<td>0.634</td>
<td>0.508</td>
<td>0.634</td>
<td>0.458</td>
<td>0.520</td>
<td>0.494</td>
<td>0.550</td>
<td>0.626</td>
<td>0.565</td>
<td>0.528</td>
<td>0.762</td>
<td>0.613</td>
<td>0.496</td>
<td>0.459</td>
<td>0.282</td>
<td>0.608</td>
<td>0.271</td>
</tr>
<tr>
<td>Yi-34B-Chat</td>
<td>0.512</td>
<td>0.483</td>
<td>0.606</td>
<td>0.347</td>
<td>0.623</td>
<td>0.497</td>
<td>0.598</td>
<td>0.480</td>
<td>0.490</td>
<td>0.575</td>
<td>0.525</td>
<td>0.619</td>
<td>0.554</td>
<td>0.494</td>
<td>0.757</td>
<td>0.580</td>
<td>0.472</td>
<td>0.439</td>
<td>0.346</td>
<td>0.514</td>
<td>0.259</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td>
<td>0.500</td>
<td>0.481</td>
<td>0.582</td>
<td>0.307</td>
<td>0.614</td>
<td>0.494</td>
<td>0.645</td>
<td>0.428</td>
<td>0.475</td>
<td>0.496</td>
<td>0.513</td>
<td>0.616</td>
<td>0.548</td>
<td>0.507</td>
<td>0.764</td>
<td>0.583</td>
<td>0.469</td>
<td>0.453</td>
<td>0.283</td>
<td>0.575</td>
<td>0.262</td>
</tr>
<tr>
<td>Deepseek-Llm-67B-Chat</td>
<td>0.471</td>
<td>0.467</td>
<td>0.571</td>
<td>0.259</td>
<td>0.577</td>
<td>0.486</td>
<td>0.549</td>
<td>0.442</td>
<td>0.476</td>
<td>0.475</td>
<td>0.509</td>
<td>0.566</td>
<td>0.496</td>
<td>0.439</td>
<td>0.711</td>
<td>0.546</td>
<td>0.409</td>
<td>0.436</td>
<td>0.262</td>
<td>0.570</td>
<td>0.235</td>
</tr>
<tr>
<td>Baichuan-13B-Chat</td>
<td>0.450</td>
<td>0.408</td>
<td>0.491</td>
<td>0.286</td>
<td>0.552</td>
<td>0.439</td>
<td>0.670</td>
<td>0.417</td>
<td>0.422</td>
<td>0.482</td>
<td>0.486</td>
<td>0.565</td>
<td>0.505</td>
<td>0.377</td>
<td>0.704</td>
<td>0.552</td>
<td>0.387</td>
<td>0.402</td>
<td>0.350</td>
<td>0.431</td>
<td>0.304</td>
</tr>
<tr>
<td>Chatglm3-6B</td>
<td>0.436</td>
<td>0.381</td>
<td>0.439</td>
<td>0.330</td>