English

Training Dialogue Systems by AI Feedback for Improving Overall Dialogue Impression

Computation and Language 2025-01-28 v2

Abstract

To improve user engagement during conversations with dialogue systems, we must improve individual dialogue responses and dialogue impressions such as consistency, personality, and empathy throughout the entire dialogue. While such dialogue systems have been developing rapidly with the help of large language models (LLMs), reinforcement learning from AI feedback (RLAIF) has attracted attention to align LLM-based dialogue models for such dialogue impressions. In RLAIF, a reward model based on another LLM is used to create a training signal for an LLM-based dialogue model using zero-shot/few-shot prompting techniques. However, evaluating an entire dialogue only by prompting LLMs is challenging. In this study, the supervised fine-tuning (SFT) of LLMs prepared reward models corresponding to 12 metrics related to the impression of the entire dialogue for evaluating dialogue responses. We tuned our dialogue models using the reward model signals as feedback to improve the impression of the system. The results of automatic and human evaluations showed that tuning the dialogue model using our reward model corresponding to dialogue impression improved the evaluation of individual metrics and the naturalness of the dialogue response.

Keywords

Cite

@article{arxiv.2501.12698,
  title  = {Training Dialogue Systems by AI Feedback for Improving Overall Dialogue Impression},
  author = {Kai Yoshida and Masahiro Mizukami and Seiya Kawano and Canasai Kruengkrai and Hiroaki Sugiyama and Koichiro Yoshino},
  journal= {arXiv preprint arXiv:2501.12698},
  year   = {2025}
}

Comments

Accepted to ICASSP 2025

R2 v1 2026-06-28T21:13:16.691Z