R$^3$AG: Retriever Routing for Retrieval-Augmented Generation

TL;DR

R$^3$AG introduces a dynamic routing framework for retrieval-augmented generation that models query-specific preferences by decomposing retriever capabilities into retrieval quality and generation utility, leading to improved task performance.

Contribution

It proposes a novel dynamic routing method that explicitly models and learns query-specific retriever preferences, surpassing static routing approaches.

Findings

R$^3$AG outperforms static routing methods on multiple knowledge-intensive tasks.

Decomposing retriever capabilities improves retrieval relevance and answer correctness.

Contrastive learning effectively captures query-specific preference shifts.

Abstract

Retrieval-augmented generation (RAG) has become a cornerstone for knowledge-intensive tasks. However, the efficacy of RAG is often bottlenecked by the ``one-size-fits-all'' retrieval paradigm, as different queries exhibit distinct preferences for different retrievers. While recent routing techniques attempt to select the optimal retriever dynamically, they typically operate under a ``single and static capability'' assumption, selecting retrievers solely based on semantic relevance. This overlooks a critical distinction in RAG: a retrieved document must not only be relevant but also effectively support the generator in producing correct answers. To address this limitation, we propose R$^3$AG, a novel routing framework that explicitly models the dynamic alignment between queries and retriever capabilities. Unlike previous approaches, R$^3$AG decomposes retriever capability into two learnable dimensions: retrieval quality and generation utility. We employ a contrastive learning objective that leverages complementary supervision signals, \textit{i.e.}, document assessments and downstream answer correctness, to capture query-specific preference shifts. Extensive experiments on several knowledge-intensive tasks show that R$^3$AG consistently outperforms both the best individual retrievers and state-of-the-art static routing methods.