发表机构
vivo AI Lab（vivo人工智能实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
ToolLoop提出闭环工具使用数据合成框架，通过分解生成与动态自我反馈，以少量数据显著提升模型工具调用能力。
AI 中文摘要
高质量的工具使用数据对于训练语言模型有效与外部工具交互至关重要。然而，现有的合成方法通常遵循“先生成后过滤”的范式，并采用静态的事后验证，往往产生效率低下且特征分布不平衡的数据。我们提出ToolLoop，一个闭环框架，将合成过程分解为三个渐进阶段：（1）采样函数名称组合作为真实标签；（2）反向推导用户查询；（3）正向推导工具调用。在每个阶段，动态自我反馈迭代地引导模型生成高质量数据，实现了从“先生成后过滤”到“生成-验证-精炼”的转变。在伯克利函数调用排行榜（BFCL）上，使用我们的11K合成示例训练的4B参数模型在非推理模式下达到86.40%的准确率，而移除BFCL重叠候选函数的Isolate变体仍达到86.07%。在ACEBench上的跨基准评估进一步证明了强大的泛化能力，仅使用基线训练数据的18.3%即达到72.1%的总体准确率。
英文摘要
High-quality tool-use data is critical for training language models to interact effectively with external tools. However, existing synthetic approaches typically follow a generate-then-filter paradigm with static post-hoc verification, often yielding inefficient data with imbalanced feature distributions. We propose ToolLoop, a closed-loop framework that decomposes synthesis into three progressive stages: (1) sampling function name combinations as ground truth; (2) backward derivation of user queries; and (3) forward derivation of tool calls. At each stage, dynamic self-feedback iteratively guides the model toward high-quality generation, realizing a transition from generate-then-filter to generate-verify-refine. On the Berkeley Function Calling Leaderboard (BFCL), a 4B parameter model trained with our 11K synthetic examples achieves 86.40% accuracy in non-reasoning mode, while an Isolate variant that removes BFCL-overlapping candidate functions still reaches 86.07\%. Cross-benchmark evaluation on ACEBench further demonstrates strong generalization, with 72.1% overall accuracy using only 18.3% of baseline training data.
CommentsAccepted at the EMNLP 2026 Main Conference
