Aligning Language Models with Offline Learning from Human Feedback

Jian Hu, Li Tao, June Yang +1

Learning from human preferences is crucial for language models (LMs) to effectively cater to human needs and societal values. Previous research has made notable progress by leverag…