Panacea: Pareto Alignment via Preference Adaptation for LLMs

Yifan Zhong, Chengdong Ma, Xiaoyuan Zhang +5

Current methods for large language model alignment typically use scalar human preference labels. However, this convention tends to oversimplify the multi-dimensional and heterogene…