FS-Researcher:基于文件系统的长周期研究任务测试时扩展
FS-Researcher: Test-Time Scaling for Long-Horizon Research Tasks with File-System-Based Agents
AI总结:
本文提出FS-Researcher框架,通过持久化工作空间实现长周期研究任务的测试时扩展,采用双代理结构提升研究质量与扩展性。
摘要:
深度研究正成为大型语言模型(LLM)代理的代表性长周期任务。然而,深度研究中的长轨迹常超出模型上下文限制,压缩了证据收集和报告写作的token预算,阻碍了测试时扩展。我们引入FS-Researcher,一种基于文件系统的双代理框架,通过持久化工作空间扩展深度研究。具体而言,一个上下文构建代理扮演图书馆员角色,浏览互联网、撰写结构化笔记并归档原始资料到分层知识库中,该知识库可远超上下文长度。另一个报告撰写代理则分阶段生成最终报告,将知识库视为事实来源。在该框架中,文件系统充当持久化外部内存和跨代理与会话的共享协调媒介,使迭代优化超越上下文窗口。在两个开放性基准(DeepResearch Bench和DeepConsult)上的实验表明,FS-Researcher在不同基础模型上均达到最先进的报告质量。进一步分析显示,最终报告质量与分配给上下文构建代理的计算量呈正相关,验证了文件系统范式下的有效测试时扩展。代码和数据已开源至https://github.com/Ignoramus0817/FS-Researcher。
英文摘要:
Deep research is emerging as a representative long-horizon task for large language model (LLM) agents. However, long trajectories in deep research often exceed model context limits, compressing token budgets for both evidence collection and report writing, and preventing effective test-time scaling. We introduce FS-Researcher, a file-system-based, dual-agent framework that scales deep research beyond the context window via a persistent workspace. Specifically, a Context Builder agent acts as a librarian which browses the internet, writes structured notes, and archives raw sources into a hierarchical knowledge base that can grow far beyond context length. A Report Writer agent then composes the final report section by section, treating the knowledge base as the source of facts. In this framework, the file system serves as a durable external memory and a shared coordination medium across agents and sessions, enabling iterative refinement beyond the context window. Experiments on two open-ended benchmarks (DeepResearch Bench and DeepConsult) show that FS-Researcher achieves state-of-the-art report quality across different backbone models. Further analyses demonstrate a positive correlation between final report quality and the computation allocated to the Context Builder, validating effective test-time scaling under the file-system paradigm. The code and data are open-sourced at https://github.com/Ignoramus0817/FS-Researcher.
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。