List your company

Mathematical Problem Solving on AIME'24 (test)

26.67Average@16 AccuracyPRM-CoT (Process-Aware)
8.896413.510718.12522.7393Dec 2, 2025

Evaluation results

MethodLinks
PRM-CoT (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.1226.67304050
RLVR (Ground Truth)Base Policy=Qwen2.5-Math-7B2025.1222.7126.6736.6750
PRM (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.1218.1226.674046.67
SFT (Baseline)Base Policy=Qwen2.5-Math-7B2025.129.583.323.3326.67