1 paper
Chenghui Zou, Ning Wang, Tiesunlong Shen +5
Large language models (LLMs) have advanced emotional support conversation, but existing alignment methods rely mainly on sparse preferences at the response level or outcomes at the…