List your company

Mathematical Reasoning on AMC23 (Pass@1 accuracy)

70Pass@1 AccuracyPUM+GRPO
51.856.52561.2565.975Jun 5, 2026

Evaluation results

MethodLinks
PUM+GRPOSetting=Outcome-Anchored2026.0670
PUMSetting=Process-Only2026.0665
PRM+GRPOSetting=Outcome-Anchored2026.0660
PQM+GRPOSetting=Outcome-Anchored2026.0660
PRMSetting=Process-Only2026.0657.5
PQMSetting=Process-Only2026.0657.5
Vanilla GRPOSetting=Outcome-Anchored2026.0652.5