Rlad Knowledge Distillation

RLAD performs selective distillation during RL, using trust-region ratio objectives to balance exploration, exploitation, and teacher guidance for efficient reasoning model training.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit b4fb2b13f3

Frequently asked questions

npx get-skill add adu2021/rlad-knowledge-distillation