Investigating Context-Faithfulness in Large Language Models: The Roles of Memory Strength and Evidence Style

TL;DR

This paper explores how memory strength and evidence style affect large language models' ability to stay faithful to external context, revealing that paraphrased evidence enhances reliance on external information especially when internal memory is weak.

Contribution

It introduces a novel measurement of memory strength in LLMs and demonstrates how evidence style influences their context-faithfulness, providing insights for improving retrieval-augmented generation.

Findings

High memory strength leads to reliance on internal memory.

Paraphrased evidence increases receptiveness more than repetition.

Insights for enhancing context-aware LLMs.

Abstract

Retrieval-augmented generation (RAG) improves Large Language Models (LLMs) by incorporating external information into the response generation process. However, how context-faithful LLMs are and what factors influence LLMs' context faithfulness remain largely unexplored. In this study, we investigate the impact of memory strength and evidence presentation on LLMs' receptiveness to external evidence. We quantify the memory strength of LLMs by measuring the divergence in LLMs' responses to different paraphrases of the same question, which is not considered by previous works. We also generate evidence in various styles to examine LLMs' behavior. Our results show that for questions with high memory strength, LLMs are more likely to rely on internal memory. Furthermore, presenting paraphrased evidence significantly increases LLMs' receptiveness compared to simple repetition or adding details. These findings provide key insights for improving retrieval-augmented generation and context-aware LLMs. Our code is available at https://github.com/liyp0095/ContextFaithful.