HIRAG: Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jiao, YiHan, Tan, ZheHao, Yang, Dan, Sun, DuoLin, Feng, Jie, Shen, Yue, Wang, Jian, Wei, Peng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915488207994880
author Jiao, YiHan
Tan, ZheHao
Yang, Dan
Sun, DuoLin
Feng, Jie
Shen, Yue
Wang, Jian
Wei, Peng
author_facet Jiao, YiHan
Tan, ZheHao
Yang, Dan
Sun, DuoLin
Feng, Jie
Shen, Yue
Wang, Jian
Wei, Peng
contents Retrieval-augmented generation (RAG) has become a fundamental paradigm for addressing the challenges faced by large language models in handling real-time information and domain-specific problems. Traditional RAG systems primarily rely on the in-context learning (ICL) capabilities of the large language model itself. Still, in-depth research on the specific capabilities needed by the RAG generation model is lacking, leading to challenges with inconsistent document quality and retrieval system imperfections. Even the limited studies that fine-tune RAG generative models often \textit{lack a granular focus on RAG task} or \textit{a deeper utilization of chain-of-thought processes}. To address this, we propose that RAG models should possess three progressively hierarchical abilities (1) Filtering: the ability to select relevant information; (2) Combination: the ability to combine semantic information across paragraphs; and (3) RAG-specific reasoning: the ability to further process external knowledge using internal knowledge. Thus, we introduce our new RAG instruction fine-tuning method, Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation (HIRAG) incorporates a "think before answering" strategy. This method enhances the model's open-book examination capability by utilizing multi-level progressive chain-of-thought. Experiments show that the HIRAG training strategy significantly improves the model's performance on datasets such as RGB, PopQA, MuSiQue, HotpotQA, and PubmedQA.
format Preprint
id arxiv_https___arxiv_org_abs_2507_05714
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HIRAG: Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation
Jiao, YiHan
Tan, ZheHao
Yang, Dan
Sun, DuoLin
Feng, Jie
Shen, Yue
Wang, Jian
Wei, Peng
Computation and Language
Artificial Intelligence
Retrieval-augmented generation (RAG) has become a fundamental paradigm for addressing the challenges faced by large language models in handling real-time information and domain-specific problems. Traditional RAG systems primarily rely on the in-context learning (ICL) capabilities of the large language model itself. Still, in-depth research on the specific capabilities needed by the RAG generation model is lacking, leading to challenges with inconsistent document quality and retrieval system imperfections. Even the limited studies that fine-tune RAG generative models often \textit{lack a granular focus on RAG task} or \textit{a deeper utilization of chain-of-thought processes}. To address this, we propose that RAG models should possess three progressively hierarchical abilities (1) Filtering: the ability to select relevant information; (2) Combination: the ability to combine semantic information across paragraphs; and (3) RAG-specific reasoning: the ability to further process external knowledge using internal knowledge. Thus, we introduce our new RAG instruction fine-tuning method, Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation (HIRAG) incorporates a "think before answering" strategy. This method enhances the model's open-book examination capability by utilizing multi-level progressive chain-of-thought. Experiments show that the HIRAG training strategy significantly improves the model's performance on datasets such as RGB, PopQA, MuSiQue, HotpotQA, and PubmedQA.
title HIRAG: Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2507.05714