Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Visual Mathematical Reasoning on MMK12 (val)
Loading...
58.86
Accuracy
Qwen2.5-VL-3B-Instruct + VEPO
44.8096
48.4573
52.105
55.7527
Jun 2, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
Qwen2.5-VL-3B-Instruct + VEPO
Base Model Size=3B, Fi...
2026.06
58.86
PAPO-DAPO-3B
Base Model Size=3B
2026.06
58.85
VPPO-3B
Base Model Size=3B
2026.06
58.24
NoisyRollout-3B
Base Model Size=3B
2026.06
58.12
Qwen2.5-VL-3B-Instruct + GRPO
Base Model Size=3B, Fi...
2026.06
57.37
Qwen2.5-VL-3B-Instruct + GRPO (Top 20% high entropy)
Base Model Size=3B, Fi...
2026.06
56.81
R1-ShareVL
Base Model Size=3B
2026.06
56.26
Qwen2.5-VL-3B-Instruct + GRPO (Top 40% high entropy)
Base Model Size=3B, Fi...
2026.06
54.52
Qwen2.5-VL-3B-Instruct
Base Model Size=3B
2026.06
45.35
Feedback
Search any
task
Search any
task