ERPO: Advancing Safety Alignment via Ex-Ante Reasoning Preference Optimization

Add code
Apr 03, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: