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

Vision-Language Evaluation on SEED-Bench

74.74Accuracy

FP16

58.04862.381566.71571.0485Apr 2, 2026
Updated 13d ago

Evaluation Results

MethodLinks
2026.04
74.74-
2026.04
74.68-
2026.04
74.61-
2026.04
74.52-
2026.04
74.48-
2026.04
74.47-
2026.04
74.46-
2026.04
74.44-
2026.04
74.42-
2026.04
74.41-
2026.04
74.38-
2026.04
74.37-
2026.04
74.23-
2026.04
74.21-
2026.04
74.11-
2026.04
73.73-
2026.04
73.62-
2026.04
63.68-
2026.04
63.6-
2026.04
63.53-
2026.04
63.52-
2026.04
63.32-
2026.04
63.25-
2026.04
63-
2026.04
62.99-
2026.04
62.8-
2026.04
62.69-
2026.04
62.51-
2026.04
62.06-
2026.04
61.67-
2026.04
61.26-
2026.04
61-
2026.04
59.94-
2026.04
58.69-
2024.05
-46.4
-53.4
2024.05
-45
-58.2
2024.05
-57.8
2024.05
-58.6
2024.05
-60.6
2024.05
-61.1
2024.05
-62.5