Reproduce “Cause-Aware Empathetic Response Generation via Chain-of-Thought Fine-Tuning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xinhao Chen, Chong Yang, Man Lan, Li Cai, Yang Chen, Tu Hu, Xinlin Zhuang, Aimin Zhou
A new cause-aware Chain-of-Thought fine-tuning method, CFEG, is presented for empathetic response generation using Large Language Models. This approach integrates explicit emotion cause reasoning and listener-aware prompts with cause-oriented common-sense knowledge, achieving state-of-the-art performance and outperforming larger models like ChatGPT in empathy and informativeness.
Key insights
- Explicitly reasoning about the cause of an emotion significantly enhances an LLM's capacity for deeper affective understanding and cognitive inference, yielding more pertinent empathetic responses.
- Aligning external common-sense knowledge with specific emotional causes, rather than general dialogue context, effectively boosts response diversity and consistency.
- Strategic prompt engineering and fine-tuning on moderately sized LLMs can enable powerful empathetic reasoning and performance, surpassing larger, less specifically tuned models like ChatGPT.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.