Sam Babichenko

Learning from human preferences

Reward models, KL-regularized fine-tuning and direct preference optimization.

InstructGPTOuyang et al., 20228 questions Direct preference optimizationKorbak et al., 2022; Rafailov et al., 2023; Azar et al., 20248 questions Reward model overoptimizationGao et al., 20238 questions