List your company

Math Reasoning on Mean of Six Math Benchmarks (Pass@K)

43.8Pass@1GRPO
14.26421.93229.637.268Oct 21, 2025

Evaluation results

MethodLinks
GRPOModel=Qwen2.5-Math-7B2025.1043.861.871.659.1
GRPOModel=Qwen2.5-7B-Instruct2025.1039.656.367.554.5
SePTModel=Qwen2.5-Math-7B2025.1039.557.767.955
GRPOModel=Qwen2.5-7B2025.1039.256.666.154
BaselineModel=Qwen2.5-7B-Instruct2025.1036.856.667.353.6
SePTModel=Qwen2.5-7B-Instruct2025.1036.656.86954.1
SePTModel=Qwen2.5-7B2025.1032.354.665.350.7
BaselineModel=Qwen2.5-Math-7B2025.1022.747.36143.7
BaselineModel=Qwen2.5-7B2025.1021.348.762.244.1
GRPOModel=DeepSeek-Math-7B-Instruct2025.1018.237.75135.6
SePTModel=DeepSeek-Math-7B-Instruct2025.1015.935.550.333.9
BaselineModel=DeepSeek-Math-7B-Instruct2025.1015.434.648.132.7