ImpRAG: Retrieval-Augmented Generation with Implicit Queries

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Wenzheng, Lin, Xi Victoria, Stratos, Karl, Yih, Wen-tau, Chen, Mingda
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918143574671360
author Zhang, Wenzheng
Lin, Xi Victoria
Stratos, Karl
Yih, Wen-tau
Chen, Mingda
author_facet Zhang, Wenzheng
Lin, Xi Victoria
Stratos, Karl
Yih, Wen-tau
Chen, Mingda
contents Retrieval-Augmented Generation (RAG) systems traditionally treat retrieval and generation as separate processes, requiring explicit textual queries to connect them. This separation can limit the ability of models to generalize across diverse tasks. In this work, we propose a query-free RAG system, named ImpRAG, which integrates retrieval and generation into a unified model. ImpRAG allows models to implicitly express their information needs, eliminating the need for human-specified queries. By dividing pretrained decoder-only language models into specialized layer groups, ImpRAG optimizes retrieval and generation tasks simultaneously. Our approach employs a two-stage inference process, using the same model parameters and forward pass for both retrieval and generation, thereby minimizing the disparity between retrievers and language models. Experiments on 8 knowledge-intensive tasks demonstrate that ImpRAG achieves 3.6-11.5 improvements in exact match scores on unseen tasks with diverse formats, highlighting its effectiveness in enabling models to articulate their own information needs and generalize across tasks. Our analysis underscores the importance of balancing retrieval and generation parameters and leveraging generation perplexities as retrieval training objectives for enhanced performance.
format Preprint
id arxiv_https___arxiv_org_abs_2506_02279
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ImpRAG: Retrieval-Augmented Generation with Implicit Queries
Zhang, Wenzheng
Lin, Xi Victoria
Stratos, Karl
Yih, Wen-tau
Chen, Mingda
Computation and Language
Retrieval-Augmented Generation (RAG) systems traditionally treat retrieval and generation as separate processes, requiring explicit textual queries to connect them. This separation can limit the ability of models to generalize across diverse tasks. In this work, we propose a query-free RAG system, named ImpRAG, which integrates retrieval and generation into a unified model. ImpRAG allows models to implicitly express their information needs, eliminating the need for human-specified queries. By dividing pretrained decoder-only language models into specialized layer groups, ImpRAG optimizes retrieval and generation tasks simultaneously. Our approach employs a two-stage inference process, using the same model parameters and forward pass for both retrieval and generation, thereby minimizing the disparity between retrievers and language models. Experiments on 8 knowledge-intensive tasks demonstrate that ImpRAG achieves 3.6-11.5 improvements in exact match scores on unseen tasks with diverse formats, highlighting its effectiveness in enabling models to articulate their own information needs and generalize across tasks. Our analysis underscores the importance of balancing retrieval and generation parameters and leveraging generation perplexities as retrieval training objectives for enhanced performance.
title ImpRAG: Retrieval-Augmented Generation with Implicit Queries
topic Computation and Language
url https://arxiv.org/abs/2506.02279