P²-MOPD Speeds Up Multi-Teacher On-Policy Distillation

StepFun released P²-MOPD, a new method that uses expert weights as parameter priors to accelerate multi-teacher on-policy distillation, roughly halving the number of distillation steps while preserving performance.

2026-11-09 ~ 2026-11-10 · 3 related posts

• StepFun Releases P²-MOPD: Parameter Priors Speed Up Multi-Teacher On-Policy Distillation — stepfun-ai · 2026-11-09
• P²-MOPD uses expert weights as parameter priors to halve multi-teacher distillation steps — _akhaliq · 2026-11-10
• P²-MOPD paper uses parameter priors to speed up multi-teacher on-policy distillation — _akhaliq · 2026-11-10