Preferences

Overall Progress 0%

Bradley-Terry from pairwise comparisons; train policy with PPO.

Go to Chapter 80: RL from Human Feedback (RLHF) Basics →