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

General Multimodal Reasoning on MMMU-Pro

38.09Score

Qwen2.5-VL-7B + SRPO

25.578828.826932.07535.3231May 8, 2026
Updated 23d ago

Evaluation Results

MethodLinks
2026.05
38.09
2026.05
36.94
2026.05
35.89
2026.05
35.37
2026.05
35.2
2026.05
35.2
2026.05
34.91
2026.05
34.73
2026.05
34.33
2026.05
33.35
2026.05
32.65
2026.05
31.33
2026.05
30.73
2026.05
29.65
2026.05
28.61
2026.05
28.15
2026.05
27.91
2026.05
27.8
2026.05
27.68
2026.05
27.45
2026.05
26.06