List your company

Mathematical Reasoning on Geo3k

52AccuracyNoisyRollout-7B
17.47226.43635.444.364Jun 7, 2026

Evaluation results

MethodLinks
NoisyRollout-7BModel Scale=7B, Rollout=82026.0652
PRPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0646.9
VPPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0644.4
PAPO-DModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0644.1
R1-ShareVL-7BModel Scale=7B, Rollout=82026.0641.2
VL-Rethinker-7BModel Scale=7B, Rollout=82026.0640.7
MM-Eureka-7BModel Scale=7B, Rollout=82026.0640.3
PAPO-GModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0640.3
GRPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0640.2
PRPOModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0637.3
Qwen2.5-VL-7BModel Scale=7B, Rollout=82026.0636.7
DAPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0635.9
VPPOModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0635.8
PAPO-DModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0635.7
DAPOModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0631.2
PAPO-GModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0631.1
GRPOModel Scale=3B, Base Model=Qwen2.5-VL-3B, Rollout=82026.0628.7
Qwen2.5-VL-3BModel Scale=3B, Rollout=82026.0618.8