List your company

Mathematical Reasoning on HMMT F25 (avg@16, pass@16)

11.25Avg@16 ScoreRollout-level |Ai| replay
-0.452.58755.6258.6625Jun 3, 2026

Evaluation results

MethodLinks
Rollout-level |Ai| replayModel=Qwen3-4B2026.0611.2536.67
uniformModel=Qwen3-4B2026.067.8122.15
GRPOModel=Qwen3-4B2026.067.2916.67
σgModel=Qwen3-4B2026.067.0830
Rollout-level |Ai| replayModel=Qwen3-1.7B2026.060.626.67
uniformModel=Qwen3-1.7B2026.060.526.29
GRPOModel=Qwen3-0.6B2026.060.423.33
Rollout-level |Ai| replayModel=Qwen3-0.6B2026.060.423.33
σgModel=Qwen3-0.6B2026.060.213.33
GRPOModel=Qwen3-1.7B2026.060.213.33
σgModel=Qwen3-1.7B2026.060.213.33
uniformModel=Qwen3-0.6B2026.0600