Saved in:
Bibliographic Details
Main Authors: Zhu, Junda, Yan, Lingyong, Shi, Haibo, Yin, Dawei, Sha, Lei
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.18111
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914967564845056
author Zhu, Junda
Yan, Lingyong
Shi, Haibo
Yin, Dawei
Sha, Lei
author_facet Zhu, Junda
Yan, Lingyong
Shi, Haibo
Yin, Dawei
Sha, Lei
contents Large language models (LLMs) are proven to benefit a lot from retrieval-augmented generation (RAG) in alleviating hallucinations confronted with knowledge-intensive questions. RAG adopts information retrieval techniques to inject external knowledge from semantic-relevant documents as input contexts. However, since today's Internet is flooded with numerous noisy and fabricating content, it is inevitable that RAG systems are vulnerable to these noises and prone to respond incorrectly. To this end, we propose to optimize the retrieval-augmented Generator with an Adversarial Tuning Multi-agent system (ATM). The ATM steers the Generator to have a robust perspective of useful documents for question answering with the help of an auxiliary Attacker agent through adversarially tuning the agents for several iterations. After rounds of multi-agent iterative tuning, the Generator can eventually better discriminate useful documents amongst fabrications. The experimental results verify the effectiveness of ATM and we also observe that the Generator can achieve better performance compared to the state-of-the-art baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2405_18111
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle ATM: Adversarial Tuning Multi-agent System Makes a Robust Retrieval-Augmented Generator
Zhu, Junda
Yan, Lingyong
Shi, Haibo
Yin, Dawei
Sha, Lei
Computation and Language
Large language models (LLMs) are proven to benefit a lot from retrieval-augmented generation (RAG) in alleviating hallucinations confronted with knowledge-intensive questions. RAG adopts information retrieval techniques to inject external knowledge from semantic-relevant documents as input contexts. However, since today's Internet is flooded with numerous noisy and fabricating content, it is inevitable that RAG systems are vulnerable to these noises and prone to respond incorrectly. To this end, we propose to optimize the retrieval-augmented Generator with an Adversarial Tuning Multi-agent system (ATM). The ATM steers the Generator to have a robust perspective of useful documents for question answering with the help of an auxiliary Attacker agent through adversarially tuning the agents for several iterations. After rounds of multi-agent iterative tuning, the Generator can eventually better discriminate useful documents amongst fabrications. The experimental results verify the effectiveness of ATM and we also observe that the Generator can achieve better performance compared to the state-of-the-art baselines.
title ATM: Adversarial Tuning Multi-agent System Makes a Robust Retrieval-Augmented Generator
topic Computation and Language
url https://arxiv.org/abs/2405.18111