Enregistré dans:
Détails bibliographiques
Auteurs principaux: Yan, Shi-Qi, Liu, Quan, Ling, Zhen-Hua
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2501.13726
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914084642881536
author Yan, Shi-Qi
Liu, Quan
Ling, Zhen-Hua
author_facet Yan, Shi-Qi
Liu, Quan
Ling, Zhen-Hua
contents While Retrieval-Augmented Generation (RAG) has exhibited promise in utilizing external knowledge, its generation process heavily depends on the quality and accuracy of the retrieved context. Large language models (LLMs) struggle to evaluate the correctness of non-parametric knowledge retrieved externally when it differs from internal memorization, leading to knowledge conflicts during response generation. To this end, we introduce the Retrieval Preference Optimization (RPO), a lightweight and effective alignment method to adaptively leverage multi-source knowledge based on retrieval relevance. An implicit representation of retrieval relevance is derived and incorporated into the reward model to integrate retrieval evaluation and response generation into a single model, solving the problem that previous methods necessitate the additional procedure to assess the retrieval quality. Notably, RPO is the only RAG-dedicated alignment approach that quantifies the awareness of retrieval relevance in training, overcoming mathematical obstacles. Experiments on four datasets demonstrate that RPO outperforms RAG by 4-10% in accuracy without any extra component, exhibiting its robust generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2501_13726
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RPO: Retrieval Preference Optimization for Robust Retrieval-Augmented Generation
Yan, Shi-Qi
Liu, Quan
Ling, Zhen-Hua
Computation and Language
While Retrieval-Augmented Generation (RAG) has exhibited promise in utilizing external knowledge, its generation process heavily depends on the quality and accuracy of the retrieved context. Large language models (LLMs) struggle to evaluate the correctness of non-parametric knowledge retrieved externally when it differs from internal memorization, leading to knowledge conflicts during response generation. To this end, we introduce the Retrieval Preference Optimization (RPO), a lightweight and effective alignment method to adaptively leverage multi-source knowledge based on retrieval relevance. An implicit representation of retrieval relevance is derived and incorporated into the reward model to integrate retrieval evaluation and response generation into a single model, solving the problem that previous methods necessitate the additional procedure to assess the retrieval quality. Notably, RPO is the only RAG-dedicated alignment approach that quantifies the awareness of retrieval relevance in training, overcoming mathematical obstacles. Experiments on four datasets demonstrate that RPO outperforms RAG by 4-10% in accuracy without any extra component, exhibiting its robust generalization.
title RPO: Retrieval Preference Optimization for Robust Retrieval-Augmented Generation
topic Computation and Language
url https://arxiv.org/abs/2501.13726