Hexa: Self-Improving for Knowledge-Grounded Dialogue System

TL;DR

Hexa introduces a self-improving approach for knowledge-grounded dialogue systems that enhances intermediate step generation without requiring ground truth data, leading to improved overall dialogue quality.

Contribution

The paper presents a novel bootstrapping scheme with guided prompts and a modified loss to improve intermediate step diversity without labeled data.

Findings

Improves performance on benchmark datasets

Enhances diversity of self-generated responses

Leverages self-improvement mechanism effectively

Abstract

A common practice in knowledge-grounded dialogue generation is to explicitly utilize intermediate steps (e.g., web-search, memory retrieval) with modular approaches. However, data for such steps are often inaccessible compared to those of dialogue responses as they are unobservable in an ordinary dialogue. To fill in the absence of these data, we develop a self-improving method to improve the generative performances of intermediate steps without the ground truth data. In particular, we propose a novel bootstrapping scheme with a guided prompt and a modified loss function to enhance the diversity of appropriate self-generated responses. Through experiments on various benchmark datasets, we empirically demonstrate that our method successfully leverages a self-improving mechanism in generating intermediate and final responses and improves the performances on the task of knowledge-grounded dialogue generation.