Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Counterfactual Reasoning on CVQA

71.41Accuracy

Blind-VLM

26.263637.984349.70561.4257Feb 7, 2026
Updated 3mo ago

Evaluation Results

MethodLinks
2026.02
71.41
2026.02
70.44
2026.02
69.47
2026.02
69.29
2026.02
68.85
2026.02
68.23
2026.02
66.64
2026.02
66.37
2026.02
65.75
2026.02
65.57
2026.02
64.07
2026.02
63.72
2026.02
63.72
2026.02
62.39
2026.02
60.44
2026.02
59.56
2026.02
57.61
2026.02
57.17
2026.02
55.66
2026.02
53.54
2026.02
45.74
2026.02
42.34
2026.02
42.09
2026.02
42
2026.02
41.83
2026.02
41.39
2026.02
41.13
2026.02
41.04
2026.02
39.39
2026.02
39.13
2026.02
38.26
2026.02
38.09
2026.02
36.96
2026.02
36.35
2026.02
35.74
2026.02
35.13
2026.02
34.52
2026.02
33.83
2026.02
32.69
2026.02
28