ToolLoop improves tool-use data synthesis with dynamic self-feedback
ToolLoop: Closed-Loop Tool-Use Data Synthesis via Decomposed Generation and Dynamic Self-Feedback
Computation and Language
Summary
Training language models to use tools well needs lots of good example data, but current methods make this data in a way that can be slow and uneven. The authors created ToolLoop, which breaks down making examples into smaller steps and uses feedback during each step to improve quality. Their method helps models learn better, showing strong performance on benchmark tests. This approach also works well even when using less training data.
What this means in practice
- •For ai developers: Generate high-quality training data for language models that interact with external tools more effectively.
- •For software engineers: Improve automation in tool interaction systems by using models trained with dynamically refined synthetic data.
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.