arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2502.06205cs.CLcs.AIcs.LG
收藏

C-3PO:紧凑即插即用代理优化以实现类人检索增强生成

C-3PO: Compact Plug-and-Play Proxy Optimization to Achieve Human-like Retrieval-Augmented Generation

Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学人工智能学院);Tongyi Lab(通义实验室);Soochow University(苏州大学);University of Oxford(牛津大学);Tsinghua University(清华大学)

Guoxin Chen, Minpeng Liao, Peiying Yu, Dingmin Wang, Zile Qiao, Chao Yang, Xin Zhao, Kai Fan

AI总结:

针对RAG系统中检索器与LLM对齐难题,提出C-3PO框架,通过轻量级多代理系统协作优化检索流程,无需修改组件,显著提升性能并保持即插即用与泛化能力。

摘要:

检索增强生成(RAG)系统在协调独立开发的检索器与大型语言模型(LLMs)方面面临根本性挑战。现有方法通常涉及修改任一组件或引入简单的中间模块,导致实际限制和次优性能。受人类搜索行为(通常涉及提出搜索查询和审查文档的来回过程)的启发,我们提出C-3PO,一个以代理为中心的框架,通过轻量级多代理系统促进检索器与LLMs之间的通信。我们的框架实现了三个专门代理,它们协作优化整个RAG流程,而无需修改检索器和LLMs。这些代理共同评估检索需求、生成有效查询,并选择适合LLMs的信息。为了实现有效的多代理协调,我们开发了一种树结构展开方法,用于强化学习中的奖励信用分配。在领域内和分布外场景中的大量实验表明,C-3PO显著提升了RAG性能,同时保持了即插即用的灵活性和卓越的泛化能力。

英文摘要:

Retrieval-augmented generation (RAG) systems face a fundamental challenge in aligning independently developed retrievers and large language models (LLMs). Existing approaches typically involve modifying either component or introducing simple intermediate modules, resulting in practical limitations and sub-optimal performance. Inspired by human search behavior -- typically involving a back-and-forth process of proposing search queries and reviewing documents, we propose C-3PO, a proxy-centric framework that facilitates communication between retrievers and LLMs through a lightweight multi-agent system. Our framework implements three specialized agents that collaboratively optimize the entire RAG pipeline without altering the retriever and LLMs. These agents work together to assess the need for retrieval, generate effective queries, and select information suitable for the LLMs. To enable effective multi-agent coordination, we develop a tree-structured rollout approach for reward credit assignment in reinforcement learning. Extensive experiments in both in-domain and out-of-distribution scenarios demonstrate that C-3PO significantly enhances RAG performance while maintaining plug-and-play flexibility and superior generalization capabilities.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑