From Ranking to Selection: A Simple but Efficient Dynamic Passage Selector for Retrieval Augmented Generation

TL;DR

The paper introduces the Dynamic Passage Selector (DPS), a novel reranking framework for retrieval-augmented generation that dynamically selects relevant passages, significantly improving performance on complex multi-hop queries without altering existing pipelines.

Contribution

DPS is a supervised learning-based reranker that captures inter-passage dependencies and adaptively selects passages, outperforming state-of-the-art methods across multiple benchmarks.

Findings

DPS improves F1-score by 30.06% on MuSiQue dataset.

DPS outperforms strong baselines like Qwen3-reranker and RankingGPT.

DPS enhances reasoning in complex RAG scenarios.

Abstract

Retrieval-augmented generation (RAG) systems are often bottlenecked by their reranking modules, which typically score passages independently and select a fixed Top-K size. This approach struggles with complex multi-hop queries that require synthesizing evidence across multiple documents, creating a trade-off where small K values omit crucial information and large K values introduce noise. To address this, we introduce the Dynamic Passage Selector (DPS), a novel reranking framework that treats passage selection as a supervised learning problem. Unlike traditional point-wise or list-wise methods, DPS is fine-tuned to capture inter-passage dependencies and dynamically select the most relevant set of passages for generation. As a seamless plug-and-play module, DPS requires no modifications to the standard RAG pipeline. Comprehensive evaluations on five benchmarks show that DPS consistently outperforms state-of-the-art rerankers and fine-tuning methods. Notably, on the challenging MuSiQue dataset, DPS improves the F1-score by 30.06% and 15.4% over strong baselines like Qwen3-reranker and RankingGPT, respectively. Our results demonstrate that by enabling adaptive evidence selection, DPS substantially enhances reasoning capabilities in complex RAG scenarios.