Thinksafe Safety Alignment

Align reasoning models to be safe without external supervision by prepending refusal instructions to harmful queries and training on self-generated safe responses. Unlocks latent safety knowledge within the model's native distribution.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit c01c8f0d97

Frequently asked questions

npx get-skill add adu2021/thinksafe-safety-alignment