OC-OPD: Observation-coordinated Supervision for On-Policy Distillation
Reverse and forward Kullback--Leibler (KL) divergences provide complementary supervision in on-policy distillation: reverse KL refines the student's active token rankings, and forward KL allocates mass to teacher-supported candidates. At student-generated prefixes, we form a four-observation matrix by treating the student and teacher as both candidate sources and evaluators. The diagonal…
We haven't written up this one. arXiv cs.AI has the full story — the link below goes straight to it.
