arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2512.20458cs.IR
收藏

激光:通过结构化协议和上下文寄存器治理长 horizon 的代理搜索

Laser: Governing Long-Horizon Agentic Search via Structured Protocol and Context Register

Tencent Hunyuan(腾讯混元);Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院)

Shuting Wang, Qiaolin Xia, Vich Wang, Herberttli, Bobsimons, Zhicheng Dou

AI总结:

Laser通过结构化协议和上下文寄存器稳定并扩展代理搜索,提升多跳问答性能。

摘要:

近年来,大语言模型(LLMs)和大推理模型(LRMs)的进步使能够将多步推理与外部工具使用交织在一起的代理搜索系统成为可能。然而,现有框架大多依赖于无结构的自然语言推理,并在上下文中积累原始中间轨迹,这常常导致推理轨迹不稳定、上下文溢出以及在复杂的多跳查询中性能下降。在本研究中,我们引入了Laser,一种用于稳定和扩展代理搜索的一般框架。Laser定义了一个符号动作协议,将代理行为组织到三个空间中:规划、任务解决和回顾。每个动作都明确指定了语义和确定性的执行格式,使推理过程结构化和逻辑化,并且能够可靠地解析动作。这种设计使中间决策可解释和可追溯,增强了显式的回顾和对推理轨迹的细粒度控制。与可解析的动作相结合,Laser进一步维护了一个紧凑的上下文寄存器,只存储推理过程中的必要状态,使代理能够在不不受控制地扩展上下文的情况下进行长时间的推理。在Qwen2.5/3系列模型上对具有挑战性的多跳问答数据集的实验表明,Laser在仅提示和微调设置下均一致优于现有的代理搜索基线,证明了Laser为稳健、可扩展的代理搜索提供了原则性且有效的基础。

英文摘要:

Recent advances in Large Language Models (LLMs) and Large Reasoning Models (LRMs) have enabled agentic search systems that interleave multi-step reasoning with external tool use. However, existing frameworks largely rely on unstructured natural-language reasoning and accumulate raw intermediate traces in the context, which often leads to unstable reasoning trajectories, context overflow, and degraded performance on complex multi-hop queries. In this study, we introduce Laser, a general framework for stabilizing and scaling agentic search. Laser defines a symbolic action protocol that organizes agent behaviors into three spaces: planning, task-solving, and retrospection. Each action is specified with explicit semantics and a deterministic execution format, enabling structured and logical reasoning processes and reliable action parsing. This design makes intermediate decisions interpretable and traceable, enhancing explicit retrospection and fine-grained control over reasoning trajectories. In coordination with parsable actions, Laser further maintains a compact context register that stores only essential states of the reasoning process, allowing the agent to reason over long horizons without uncontrolled context expansion. Experiments on Qwen2.5/3-series models across challenging multi-hop QA datasets show that Laser consistently outperforms existing agentic search baselines under both prompting-only and fine-tuning settings, demonstrating that Laser provides a principled and effective foundation for robust, scalable agentic search.

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。

↑