Yuxuan Wan

Group Fairness Meets the Black Box: Enabling Fair Algorithms on Closed LLMs via Post-Processing

Pessimistic Reward Modeling in RLHF against Reward Hacking