List your company

Mathematical Reasoning on Math (Acc, Eng)

80AccuracyMAH-DPO
71.752873.893976.03578.1761Oct 1, 2025

Evaluation results

MethodLinks
MAH-DPOTest-time guidance=Accuracy Value2025.108085.53
MODPO2025.1072.873.67
Single-Head DPO2025.1072.5371.6
MAH-DPOTest-time guidance=None2025.1072.4787.33
MAH-DPOTest-time guidance=Engaging PRM2025.1072.0790.6