Self-Retrieval: End-to-End Information Retrieval with One Large Language Model

TL;DR

Self-Retrieval introduces an end-to-end LLM-based IR system that internalizes retrieval, transforming it into passage generation and relevance assessment, significantly improving retrieval and downstream task performance.

Contribution

It presents a novel architecture unifying IR functions within a single LLM, enabling deep integration and improved performance over traditional IR systems.

Findings

Outperforms existing retrieval methods significantly.

Enhances downstream applications like retrieval-augmented generation.

Internalizes retrieval corpus through self-supervised learning.

Abstract

The rise of large language models (LLMs) has significantly transformed both the construction and application of information retrieval (IR) systems. However, current interactions between IR systems and LLMs remain limited, with LLMs merely serving as part of components within IR systems, and IR systems being constructed independently of LLMs. This separated architecture restricts knowledge sharing and deep collaboration between them. In this paper, we introduce Self-Retrieval, a novel end-to-end LLM-driven information retrieval architecture. Self-Retrieval unifies all essential IR functions within a single LLM, leveraging the inherent capabilities of LLMs throughout the IR process. Specifically, Self-Retrieval internalizes the retrieval corpus through self-supervised learning, transforms the retrieval process into sequential passage generation, and performs relevance assessment for reranking. Experimental results demonstrate that Self-Retrieval not only outperforms existing retrieval approaches by a significant margin, but also substantially enhances the performance of LLM-driven downstream applications like retrieval-augmented generation.