Co-Evolving Policy Distillation

Benchmark Model Rank Results
mathematical-reasoning-on-aime-2025CoPD–Accuracy: 49.58
mathematical-reasoning-on-aime24CoPD–Acc: 60.42
mathematical-reasoning-on-math-500CoPD–Accuracy: 94.5
multimodal-reasoning-on-mathvisionCoPD–Accuracy: 57.88
multimodal-reasoning-on-mathvistaCoPD–Accuracy: 76.8
multimodal-reasoning-on-mmmuCoPD–Accuracy: 67.78
multimodal-reasoning-on-mmmu-proCoPD–Accuracy: 55.12