1 paper · 1 filter
Rong Wu, Daocheng Fu, Yu Yang +9
On-policy distillation (OPD) trains a target policy on its own rollouts using token-level supervision from a stronger teacher, but typically relies on the teacher being substantial…