ToolLoop improves synthetic data for training tool-using AI models
ToolLoop: Closed-Loop Tool-Use Data Synthesis via Decomposed Generation and Dynamic Self-Feedback
Computation and Language
Summary
Getting good example data of how to use computer tools is important for teaching AI models to work with those tools. The authors found that traditional methods create data by making examples then checking them afterward, which can miss important details. They built ToolLoop, a system that makes tool use data step-by-step while constantly checking and improving its examples as it goes. This approach helps create better and more balanced training data, leading to AI that can understand and call tools more accurately even when tested on new challenges.
synthetic datalanguage modelstool usedata synthesisself-feedbackfunction callinggenerate-verify-refinebenchmarkgeneralization
Authors
Min Zeng, Yuzhou Liu, Zhenyu Cao, Hanxiu Chen, Heng Li, Caiquan Liu, Yafei Wen, Xiaoxin Chen
Abstract
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.