Model Internals-based Answer Attribution for Trustworthy Retrieval-Augmented Generation

TL;DR

MIRAGE introduces a model internals-based method for faithful answer attribution in retrieval-augmented generation, improving transparency and control over source citation in question answering systems.

Contribution

The paper proposes MIRAGE, a novel plug-and-play approach that uses model internals and saliency methods for accurate answer attribution in RAG, addressing limitations of self-citation prompting.

Findings

High agreement with human answer attribution.

Comparable citation quality and efficiency to self-citation.

Enhanced control over attribution parameters.

Abstract

Ensuring the verifiability of model answers is a fundamental challenge for retrieval-augmented generation (RAG) in the question answering (QA) domain. Recently, self-citation prompting was proposed to make large language models (LLMs) generate citations to supporting documents along with their answers. However, self-citing LLMs often struggle to match the required format, refer to non-existent sources, and fail to faithfully reflect LLMs' context usage throughout the generation. In this work, we present MIRAGE --Model Internals-based RAG Explanations -- a plug-and-play approach using model internals for faithful answer attribution in RAG applications. MIRAGE detects context-sensitive answer tokens and pairs them with retrieved documents contributing to their prediction via saliency methods. We evaluate our proposed approach on a multilingual extractive QA dataset, finding high agreement with human answer attribution. On open-ended QA, MIRAGE achieves citation quality and efficiency comparable to self-citation while also allowing for a finer-grained control of attribution parameters. Our qualitative evaluation highlights the faithfulness of MIRAGE's attributions and underscores the promising application of model internals for RAG answer attribution.