From the source
Lead story
Ask your AI
Top stories
Models & availability
Latest
ToolGrad: Efficient tool-use dataset generation with textual "gradients"
Framework generates tool-use datasets more efficiently with higher pass rates than prior methods.
Google Research introduces ToolGrad, a data generation framework that reverses the traditional paradigm by first generating tool-use answers before user queries, enabling LLMs to achieve better tool-use performance.
The framework was presented at ACL 2026.
Experiments show that fine-tuning Gemma-3 models on ToolGrad-500 data improves tool-use performance across all tested parameter sizes, with ToolGrad-12B achieving a score of 83.1 on the Berkeley Function Calling Leaderboard.
From the source
ToolGrad is a data generation framework that reverses the traditional paradigm by first generating tool-use answers before user queries. We show this design enables LLMs to achieve better tool-use performance.
research.google