The authors designed two complementary data pipelines to populate a unified execution-grounded substrate.
Executable environments enable scalable tool-use training data and verification
Davinci-Tool combines two complementary data pipelines to produce 30,395 verified trajectories across 2,260 tools, outperforming a 16× larger model.
Academic
Xiaoyang Yuan · Ji Zhao · Xingyu Hu · Wenhan Wang · Yutao Hu · Qiaosheng Zhang · +4 more
Tongji University · Shanghai Innovation Institute · University of Science and Technology of China · Southeast University
Research Digest··3 min read
The authors introduce Davinci-Tool, a 30-billion-parameter model for reliable multi-step tool use, and Davinci-ToolData, an execution-grounded dataset.
Why this paper
From Tongji University and 3 others
In one line
Davinci-Tool, trained on execution-grounded verified data, outperforms a model with 16 times its parameters on three tool-use benchmarks.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks (3 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§