Temporal Self Rewarding Lm

Improves self-rewarding language models through temporal framework that prevents representational collapse by anchoring rejected responses and guiding chosen responses across training phases.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit 1137ef0a42

Frequently asked questions

npx get-skill add adu2021/temporal-self-rewarding-lm