One-Shot Safety Alignment for Large Language Models via Optimal Dualization

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.