大语言模型的代理推理
A Survey of Agentic Reasoning for Large Language Models: Towards Recursively Self-Improving and Collective Agents
AI总结：
本文探讨了大语言模型在开放和动态环境中的代理推理方法，通过三个层次的框架提升单体、自我进化和集体多代理推理能力，并提出了未来研究方向。
摘要：
推理是支撑推理、问题解决和决策的基本认知过程。尽管大语言模型（LLMs）在封闭世界环境中表现出强大的推理能力，但在开放性和动态环境中的表现却令人沮丧。代理推理标志着一种范式转变，通过将LLMs重新定位为能够自主行动的智能体，使它们能够通过持续互动进行计划、行动和学习。在本文的综述中，我们从三个互补的维度组织代理推理。首先，我们通过三个层次来表征环境动态：基础代理推理，它建立核心单体代理能力，包括在稳定环境中进行规划、工具使用和搜索；自我进化代理推理，它研究如何通过反馈、记忆和适应来完善这些能力；以及集体多代理推理，它将智能扩展到协作设置中，涉及协调、知识共享和共享目标。在这些层次中，我们区分了上下文推理，它通过结构化编排扩展测试时间交互，与训练后推理，后者通过强化学习和监督微调优化行为。我们进一步回顾了现实世界应用和基准测试中的代表性代理推理框架，包括科学、机器人、医疗、自主研究和数学领域。本文将代理推理方法综合成一个统一的路线图，连接思想与行动，并概述了开放挑战和未来方向，包括个性化、长周期交互、世界建模、可扩展的多代理训练以及为现实世界部署的治理。
英文摘要：
Reasoning is a fundamental cognitive process underlying inference, problem-solving, and decision-making. While large language models (LLMs) demonstrate strong reasoning capabilities in closed-world settings, they struggle in open-ended and dynamic environments. Agentic reasoning marks a paradigm shift by reframing LLMs as autonomous agents that plan, act, and learn through continual interaction. In this survey, we organize agentic reasoning along three complementary dimensions. First, we characterize environmental dynamics through three layers: foundational agentic reasoning, which establishes core single-agent capabilities including planning, tool use, and search in stable environments; self-evolving agentic reasoning, which studies how agents refine these capabilities through feedback, memory, and adaptation; and collective multi-agent reasoning, which extends intelligence to collaborative settings involving coordination, knowledge sharing, and shared goals. Across these layers, we distinguish in-context reasoning, which scales test-time interaction through structured orchestration, from post-training reasoning, which optimizes behaviors via reinforcement learning and supervised fine-tuning. We further review representative agentic reasoning frameworks across real-world applications and benchmarks, including science, robotics, healthcare, autonomous research, and mathematics. This survey synthesizes agentic reasoning methods into a unified roadmap bridging thought and action, and outlines open challenges and future directions, including personalization, long-horizon interaction, world modeling, scalable multi-agent training, and governance for real-world deployment.
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。
