arXiv cs.AI· Chao Wan, Paul J\"unger, Beixi Du, Bodong Liu, Navaneetha Krishnan, Kilian Q. Weinberger·· 14 小时前AI 评分40
DecisionRAG：用决策模型实现无检索器的文档导航
Retriever-Free Retrieval: Navigating Documents with Decision Models
AI 导读
DecisionRAG 提出一种无需第一阶段检索器的检索方法，用预训练语言模型的决策模型在单次前向中联合评估候选分支、不生成任何 token，从宽泛主题逐层导航到具体段落。
正文 · 原文
Abstract:While language models have become dramatically more capable, retrieval still largely relies on lexical matching or similarity in fixed embedding spaces. This creates a capability asymmetry: what the retriever fails to find, even the strongest generator cannot recover. We propose DecisionRAG, which brings the capabilities of pretrained language models to retrieval using decision models that jointly evaluate candidate branches in a single forward pass, without generating any tokens, and show that judging candidates in a shared context makes their relevance estimates comparable. Like a person browsing a library, DecisionRAG navigates the corpus's inherent structure from broad topics down to individual passages, deciding at each level which branches are worth exploring. Without task-specific training or a first-stage retriever, DecisionRAG outperforms dense retrieval with an 8B reranker by 13.8 R@1 points on MultiDoc2Dial and 17.4 Hit@1 points on Natural Questions. Across four benchmarks, including a corpus of 2.68 million Wikipedia paragraphs, retrieval takes just 0.7-1.8 seconds per query.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.35868 [cs.AI] |
| (or arXiv:2610.35868v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35868 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Paul Jünger [view email]
[v1]
Mon, 26 Oct 2026 17:59:44 UTC (3,006 KB)
来源：arXiv cs.AI · arxiv.org