List your company

Mathematical Problem Solving on AIME '25 (test)

22.71Avg@16 AccPRM-CoT (Process-Aware)
6.028410.359214.6919.0208Dec 2, 2025

Evaluation results

MethodLinks
PRM-CoT (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.1222.712036.6740
RLVR (Ground Truth)Base Policy=Qwen2.5-Math-7B2025.1215.8316.673033.33
PRM (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.1212.9216.673033.33
SFT (Baseline)Base Policy=Qwen2.5-Math-7B2025.126.673.316.6723.33