Picture for Xinhong Xie

Xinhong Xie

Learning from Response not Preference: A Stackelberg Approach for LLM Detoxification using Non-parallel Data

Add code
Oct 27, 2024
Viaarxiv icon