Instruct Once, Chat Consistently in Multiple Rounds: An Efficient Tuning Framework for Dialogue

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Jian, Leong, Chak Tou, Wang, Jiashuo, Lin, Dongding, Li, Wenjie, Wei, Xiao-Yong
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910463666683904
author Wang, Jian
Leong, Chak Tou
Wang, Jiashuo
Lin, Dongding
Li, Wenjie
Wei, Xiao-Yong
author_facet Wang, Jian
Leong, Chak Tou
Wang, Jiashuo
Lin, Dongding
Li, Wenjie
Wei, Xiao-Yong
contents Tuning language models for dialogue generation has been a prevalent paradigm for building capable dialogue agents. Yet, traditional tuning narrowly views dialogue generation as resembling other language generation tasks, ignoring the role disparities between two speakers and the multi-round interactive process that dialogues ought to be. Such a manner often leads to unsatisfactory chat consistency for the built agent. In this work, we emphasize the interactive, communicative nature of dialogue and argue that it is more feasible to model the speaker roles of agent and user separately, enabling the agent to adhere to its role consistently. With this in mind, we propose an efficient Multi-round Interactive Dialogue Tuning (Midi-Tuning) framework. It models the agent and user individually with two adapters built upon large language models. The adapters make use of respective utterances round by round in alternating order and they are tuned via a round-level memory caching mechanism. Extensive experiments demonstrate that, our framework performs superior to traditional fine-tuning and harbors the tremendous potential for improving dialogue consistency.
format Preprint
id arxiv_https___arxiv_org_abs_2402_06967
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Instruct Once, Chat Consistently in Multiple Rounds: An Efficient Tuning Framework for Dialogue
Wang, Jian
Leong, Chak Tou
Wang, Jiashuo
Lin, Dongding
Li, Wenjie
Wei, Xiao-Yong
Computation and Language
Artificial Intelligence
Tuning language models for dialogue generation has been a prevalent paradigm for building capable dialogue agents. Yet, traditional tuning narrowly views dialogue generation as resembling other language generation tasks, ignoring the role disparities between two speakers and the multi-round interactive process that dialogues ought to be. Such a manner often leads to unsatisfactory chat consistency for the built agent. In this work, we emphasize the interactive, communicative nature of dialogue and argue that it is more feasible to model the speaker roles of agent and user separately, enabling the agent to adhere to its role consistently. With this in mind, we propose an efficient Multi-round Interactive Dialogue Tuning (Midi-Tuning) framework. It models the agent and user individually with two adapters built upon large language models. The adapters make use of respective utterances round by round in alternating order and they are tuned via a round-level memory caching mechanism. Extensive experiments demonstrate that, our framework performs superior to traditional fine-tuning and harbors the tremendous potential for improving dialogue consistency.
title Instruct Once, Chat Consistently in Multiple Rounds: An Efficient Tuning Framework for Dialogue
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2402.06967