List your company

Mathematical Reasoning on MATH500 (mean@8)

84.6Mean@8 ScoreACPO
71.360874.797978.23581.6721Jul 3, 2026

Evaluation results

MethodLinks
ACPOModel=DeepSeek-R1-Distill-Qwen-7B2026.0784.6
ACPOModel=Qwen3-8B-Base2026.0783.6
SAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.0783.2
SAPOModel=Qwen3-8B-Base2026.0782.4
GTPOModel=Qwen3-8B-Base2026.0781.62
DAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.0780.4
GTPOModel=DeepSeek-R1-Distill-Qwen-7B2026.0778.85
80/20Model=DeepSeek-R1-Distill-Qwen-7B2026.0778.72
80/20Model=Qwen3-8B-Base2026.0778.22
GRPOModel=DeepSeek-R1-Distill-Qwen-7B2026.0777.95
ACPOModel=Qwen2.5-Math-7B2026.0777.8
DAPOModel=Qwen3-8B-Base2026.0776.8
SAPOModel=Qwen2.5-Math-7B2026.0776.4
GTPOModel=Qwen2.5-Math-7B2026.0775.5
DAPOModel=Qwen2.5-Math-7B2026.0775
80/20Model=Qwen2.5-Math-7B2026.0774.75
GRPOModel=Qwen3-8B-Base2026.0772.8
GRPOModel=Qwen2.5-Math-7B2026.0771.87