arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2605.26952cs.CL
收藏

基于策略内在知识边界增强的高效智能体强化学习

Efficient Agentic Reinforcement Learning with On-Policy Intrinsic Knowledge Boundary Enhancement

Tencent Inc(腾讯公司);The Chinese University of Hong Kong(香港中文大学);Shenzhen MSU-BIT University(深圳MSU-BIT大学)

Dingwei Chen, Zefang Zong, Zhipeng Ma, Leo Luo, Yang Li, Chengming Li, Peng Chen, Jie Jiang

AI总结:

提出AKBE方法,通过双路径(有工具和无工具)在线策略训练动态探测模型内在知识边界,构建针对性监督信号,在保持准确率的同时减少工具调用。

摘要:

智能体强化学习已被证明对于训练具有外部工具使用能力的基于LLM的智能体是有效的。然而,我们发现智能体强化学习训练会导致冗余工具调用增加,并模糊模型的内在知识边界,即模型无法区分何时需要工具以及何时参数化知识足够。现有的基于奖励塑形的解决方案创建了粗粒度的优化目标,倾向于激励不加区分的工具调用抑制,导致奖励黑客行为。在本文中,我们提出AKBE(智能体知识边界增强),一种在线策略方法,通过在训练期间进行双路径(有工具和无工具)展开来动态探测模型的内在知识边界。我们将知识边界定义为每个实例是否需要工具以及所需的最小工具调用次数。通过比较各路径的正确性,AKBE对轨迹进行分类并构建针对性的监督信号,为每个问题引导高效的工具使用模式。这些信号无缝集成到智能体强化学习训练循环中。在七个QA基准上的实验表明,与标准智能体强化学习相比,AKBE平均任务准确率提高+1.85,工具调用减少18%,工具生产率提高25%,且没有任何准确率-效率权衡。进一步分析表明其在不同RL算法上的即插即用兼容性以及每个信号类别的机制。我们的代码可在https://github.com/CuSO4-Chen/AKBE获取。

英文摘要:

Agentic reinforcement learning (RL) has proven effective for training LLM-based agents with external tool-use capabilities. However, we identify that agentic RL training induces increasing redundant tool calls and blurs the model's intrinsic knowledge boundary, where the model fails to distinguish when tools are needed versus when parametric knowledge suffices. Existing solutions based on reward shaping create coarse-grained optimization targets that tend to incentivize indiscriminate tool-call suppression, leading to reward hacking. In this paper, we propose AKBE (Agentic Knowledge Boundary Enhancement), an on-policy method that dynamically probes the model's intrinsic knowledge boundary through dual-path (with-tool and no-tool) rollouts during training. We define the knowledge boundary as the per-instance determination of whether tools are required and the minimum tool calls necessary. By comparing correctness across paths, AKBE categorizes trajectories and constructs targeted supervisory signals that guide efficient tool-use patterns for each question. These signals are integrated seamlessly into the agentic RL training loop. Experiments on seven QA benchmarks demonstrate that AKBE improves task accuracy by +1.85 on average and reduces tool calls by 18% over standard agentic RL, yielding 25% higher tool productivity without any accuracy-efficiency trade-off. Further analysis suggests its plug-and-play compatibility across different RL algorithms and the mechanism of each signal category. Our code is available at https://github.com/CuSO4-Chen/AKBE.

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。

↑