Agentic-R: Learning to Retrieve for Agentic Search

Source: arXiv:2601.11888 · Code: https://github.com/8421BCD/Agentic-R

Overview

Agentic search has recently emerged as a powerful paradigm, where an agent interleaves multi-step reasoning with on-demand retrieval to solve complex questions. Despite its success, how to design a retriever for agentic search remains largely underexplored. Existing search agents typically rely on similarity-based retrievers, while similar passages are not always useful for final answer generation. This paper proposes a novel retriever training framework tailored for agentic search.

Key Contributions

• Dual-signal passage utility: Unlike retrievers designed for single-turn retrieval-augmented generation (RAG) that only rely on local passage utility, Agentic-R measures passage utility in multi-turn agentic search using both local query-passage relevance and global answer correctness.
• Iterative, bidirectional training: The search agent and the retriever are optimized bidirectionally and iteratively. Rather than training a RAG retriever once with fixed questions, the Agentic-R retriever is continuously improved using evolving, higher-quality queries produced by the agent itself.
• Strong empirical results: Extensive experiments on seven single-hop and multi-hop QA benchmarks show that Agentic-R consistently outperforms strong baselines across different search agents.

Method Pipeline (as summarized in the original report)

1. Input & representation: encode queries, documents, and context into dense/sparse representations or structured prompts; 2. Core modules: retriever, reranker, planner, memory, and tool interfaces, composed per task; 3. Learning strategy: supervised fine-tuning, contrastive learning, distillation, and reinforcement learning with process rewards; 4. Inference strategy: single-turn or iterative retrieval, parallel sub-queries, early stopping, and budget control.

Original Abstract

> Agentic search has recently emerged as a powerful paradigm, where an agent interleaves multi-step reasoning with on-demand retrieval to solve complex questions. Despite its success, how to design a retriever for agentic search remains largely underexplored. Existing search agents typically rely on similarity-based retrievers, while similar passages are not always useful for final answer generation. In this paper, we propose a novel retriever training framework tailored for agentic search. Unlike retrievers designed for single-turn retrieval-augmented generation (RAG) that only rely on local passage utility, we propose to use both local query-passage relevance and global answer correctness to measure passage utility in a multi-turn agentic search. We further introduce an iterative training strategy, where the search agent and the retriever are optimized bidirectionally and iteratively. Different from RAG retrievers that are only trained once with fixed questions, our retriever is continuously improved using evolving and higher-quality queries from the agent. Extensive experiments on seven single-hop and multi-hop QA benchmarks demonstrate that our retriever, termed Agentic-R, consistently outperforms strong baselines across different search agents.

Related Items

• A Systematic Framework for Enterprise Knowledge Retrieval (arXiv 2512.05411)
• Retrieval Augmented Generation and Understanding in Vision: A Survey (arXiv 2503.18016)
• Synergizing RAG and Reasoning: A Systematic Review (arXiv 2504.15909)
• AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforcement Learning (arXiv 2509.24193)
• AgentX: Agent-Driven Self-Iteration of Industrial Recommender Systems (arXiv 2606.26859)
• Agentic Information Retrieval (arXiv 2410.09713)

*Note: the surrounding commentary sections in the source post are editorial context from a paper-tracking list; quantitative results should be verified against the original PDF.*