Bridging Relevance and Reasoning: Rationale Distillation in Retrieval-Augmented Generation

TL;DR

This paper introduces RADIO, a framework that extracts rationales using LLMs to better align rerankers with generator needs in retrieval-augmented generation, improving response quality.

Contribution

RADIO proposes a rationale distillation method to enhance the alignment between rerankers and generators in RAG systems, addressing a key gap in relevance and reasoning.

Findings

Improved reranking accuracy over baseline methods

Enhanced response relevance and reasoning quality

Effective across multiple datasets and tasks

Abstract

The reranker and generator are two critical components in the Retrieval-Augmented Generation (i.e., RAG) pipeline, responsible for ranking relevant documents and generating responses. However, due to differences in pre-training data and objectives, there is an inevitable gap between the documents ranked as relevant by the reranker and those required by the generator to support answering the query. To address this gap, we propose RADIO, a novel and practical preference alignment framework with RAtionale DIstillatiOn. Specifically, we first propose a rationale extraction method that leverages the reasoning capabilities of Large Language Models (LLMs) to extract the rationales necessary for answering the query. Subsequently, a rationale-based alignment process is designed to rerank the documents based on the extracted rationales, and fine-tune the reranker to align the preferences. We conduct extensive experiments on two tasks across three datasets to demonstrate the effectiveness of our approach compared to baseline methods. Our code is released online to ease reproduction.