List your company

Mathematical Reasoning on AMC (Acc, PCE, ECE)

79.8AccuracyGRPO
66.38469.86773.3576.833Apr 25, 2026

Evaluation results

MethodLinks
GRPOConf.=probe2026.0479.80.0080.045
C^2GSPGConf.=logits2026.0479.70.0590.075
RLCMConf.=probe2026.0478.50.010.07
RLCRConf.=verbal2026.0476.40.1110.111
BaseConf.=probe2026.0466.90.0550.067