Attributing Response to Context: A Jensen-Shannon Divergence Driven Mechanistic Study of Context Attribution in Retrieval-Augmented Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Ruizhe, Chen, Chen, Hu, Yuchen, Gao, Yanjun, Wang, Xi, Yilmaz, Emine
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914320340746240
author Li, Ruizhe
Chen, Chen
Hu, Yuchen
Gao, Yanjun
Wang, Xi
Yilmaz, Emine
author_facet Li, Ruizhe
Chen, Chen
Hu, Yuchen
Gao, Yanjun
Wang, Xi
Yilmaz, Emine
contents Retrieval-Augmented Generation (RAG) leverages large language models (LLMs) combined with external contexts to enhance the accuracy and reliability of generated responses. However, reliably attributing generated content to specific context segments, context attribution, remains challenging due to the computationally intensive nature of current methods, which often require extensive fine-tuning or human annotation. In this work, we introduce a novel Jensen-Shannon Divergence driven method to Attribute Response to Context (ARC-JSD), enabling efficient and accurate identification of essential context sentences without additional fine-tuning, gradient-calculation or surrogate modelling. Evaluations on a wide range of RAG benchmarks, such as TyDi QA, Hotpot QA, and Musique, using instruction-tuned LLMs in different scales demonstrate superior accuracy and significant computational efficiency improvements compared to the previous surrogate-based method. Furthermore, our mechanistic analysis reveals specific attention heads and multilayer perceptron (MLP) layers responsible for context attribution, providing valuable insights into the internal workings of RAG models and how they affect RAG behaviours. Our code is available at https://github.com/ruizheliUOA/ARC_JSD.
format Preprint
id arxiv_https___arxiv_org_abs_2505_16415
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Attributing Response to Context: A Jensen-Shannon Divergence Driven Mechanistic Study of Context Attribution in Retrieval-Augmented Generation
Li, Ruizhe
Chen, Chen
Hu, Yuchen
Gao, Yanjun
Wang, Xi
Yilmaz, Emine
Computation and Language
Artificial Intelligence
Machine Learning
Retrieval-Augmented Generation (RAG) leverages large language models (LLMs) combined with external contexts to enhance the accuracy and reliability of generated responses. However, reliably attributing generated content to specific context segments, context attribution, remains challenging due to the computationally intensive nature of current methods, which often require extensive fine-tuning or human annotation. In this work, we introduce a novel Jensen-Shannon Divergence driven method to Attribute Response to Context (ARC-JSD), enabling efficient and accurate identification of essential context sentences without additional fine-tuning, gradient-calculation or surrogate modelling. Evaluations on a wide range of RAG benchmarks, such as TyDi QA, Hotpot QA, and Musique, using instruction-tuned LLMs in different scales demonstrate superior accuracy and significant computational efficiency improvements compared to the previous surrogate-based method. Furthermore, our mechanistic analysis reveals specific attention heads and multilayer perceptron (MLP) layers responsible for context attribution, providing valuable insights into the internal workings of RAG models and how they affect RAG behaviours. Our code is available at https://github.com/ruizheliUOA/ARC_JSD.
title Attributing Response to Context: A Jensen-Shannon Divergence Driven Mechanistic Study of Context Attribution in Retrieval-Augmented Generation
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2505.16415