List your company

Science Reasoning on MMLU Pro

66.49AccuracyEAPO
35.75843.736551.71559.6935Jun 29, 2026

Evaluation results

MethodLinks
EAPOModel=Qwen-3-8B-Base2026.0666.49
MOPDModel=Qwen-3-8B-Base2026.0665.99
OPDModel=Qwen-3-8B-Base2026.0665.91
DAPOModel=Qwen-3-8B-Base2026.0665.29
GRPOModel=Qwen-3-8B-Base2026.0664.71
EAPOModel=Qwen-2.5-Math-7B2026.0647.57
EAPO w/o sISModel=Qwen-2.5-Math-7B2026.0647.35
Trajectory ReplayModel=Qwen-2.5-Math-7B2026.0646.5
DAPOModel=Qwen-2.5-Math-7B2026.0644.09
MOPDModel=Qwen-2.5-Math-7B2026.0643.81
OPDModel=Qwen-2.5-Math-7B2026.0643.34
GRPOModel=Qwen-2.5-Math-7B2026.0636.94