List your company

Logic Reasoning on CommonsenseQA

69.8Pass@1MIG
68.96869.18469.469.616Feb 1, 2026

Evaluation results

MethodLinks
MIGEvaluation Mode=Out-of-Distribution (OOD), Training Strategy=Transferred from MATH (300 steps)2026.0269.893.40.4
GRPOEvaluation Mode=Out-of-Distribution (OOD), Training Strategy=Transferred from MATH (300 steps)2026.0269.492.8—
Base ModelEvaluation Mode=Out-of-Distribution (OOD), Training Strategy=Transferred from MATH (300 steps)2026.026992.8—