构建复杂多轮工具调用的图结构数据集，提升智能体推理能力。
OrchDAG: Complex Tool Orchestration in Multi-Turn Interactions with Plan DAGs
- 将工具调用建模为可控制复杂度的有向无环图
- 在多轮任务中实现92%的成功率，优于基线模型
- 适合研究智能体规划与复杂任务执行的学者
智能体工具调用受到广泛关注，但现有工作大多忽略多轮交互中的复杂性。本文提出OrchDAG，一个将工具执行建模为有向无环图（DAG）的合成数据生成管道，支持可控复杂度。利用该数据集，我们评估模型性能并提出一种基于图结构的奖励机制，用于增强RLVR训练。实验表明，该数据集构成具有挑战性但可解的基准，结合GRPO类算法时，所提奖励有效提升了性能，凸显了拓扑结构与数据复杂度在多轮工具使用中的重要性。
原文摘要 · Abstract (English)
Agentic tool use has gained traction with the rise of agentic tool calling, yet most existing work overlooks the complexity of multi-turn tool interactions. We introduce OrchDAG, a synthetic data generation pipeline that models tool execution as directed acyclic graphs (DAGs) with controllable complexity. Using this dataset, we benchmark model performance and propose a graph-based reward to enhance RLVR training. Experiments show that the dataset presents a challenging but solvable benchmark, and the proposed reward is effective when combined with GRPO-style algorithms, highlighting the importance of leveraging topological structure and data complexity in multi-turn tool use.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。