Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Visual Reasoning on Geo3K
Loading...
56.74
Accuracy
Qwen2.5-VL-32B-Instruct + NoisyRollout
25.2488
33.4244
41.6
49.7756
Mar 11, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
Qwen2.5-VL-32B-Instruct + NoisyRollout
Student Model=Qwen2.5-...
2026.03
56.74
REOPOLD
Teacher Model=Qwen2.5-...
2026.03
53.58
RKL
Teacher Model=Qwen2.5-...
2026.03
51.75
GRPO
Teacher Model=Qwen2.5-...
2026.03
51.4
REOPOLD
Teacher Model=Qwen2.5-...
2026.03
50.58
NoisyRollout
Teacher Model=Qwen2.5-...
2026.03
50.08
RKL
Teacher Model=Qwen2.5-...
2026.03
48.09
Qwen2.5-VL-7B-Instruct
Student Model=Qwen2.5-...
2026.03
39.77
PAPO
Teacher Model=Qwen2.5-...
2026.03
32.95
Qwen2.5-VL-3B-Instruct
Student Model=Qwen2.5-...
2026.03
26.46
Feedback
Search any
task
Search any
task