용어

RLHF

이 정의는 영어입니다.

Reinforcement learning from human feedback: a training step that uses people’s preferences between outputs.