MOPD: Multi-Teacher On-Policy Distillation for Capability Integration in LLM Post-Training
Open paper
Benchmark
Model
Rank
Results
instruction-following-on-ifbench
Qwen3-30B-A3B + MOPD
–
Accuracy: 77.89
mathematical-reasoning-on-aime-2025
MiMo-V2-Flash + MOPD
–
Accuracy: 94.1
mathematical-reasoning-on-aime-2026
Qwen3-30B-A3B + MOPD
–
Accuracy: 65.31
Rank counts only results with a code link.