xLAM Function-Calling 60k (APIGen)
- Type
- dataset
- Venue
- Hugging Face / Salesforce AI Research
- Year
- 2026
- Source
- huggingface
- Access
- free
- Language
- English
- Added
- 2026-07-17T20:18:03.652742+00:00
- Verified
- 2026-07-17T20:18:03.652742+00:00
Summary
xlam-function-calling-60k is a dataset of 60,000 function-calling examples collected via APIGen, an automated data-generation pipeline that produces verifiable high-quality data for function-calling applications. Each example includes a natural-language query, available tools (APIs with names, descriptions, parameters), and answers (tool calls with arguments), verified through three hierarchical stages: format checking, actual function executions, and semantic verification (human evaluation showed >95% correctness). Data was generated by DeepSeek-V2-Chat (first 33,659 entries) and Mixtral-8x22B-Instruct across 3,673 executable APIs in 21 categories.
Keywords
function-calling agents apigen synthetic tool-use salesforce xlam llm-agent verified
Topics
NLP / Function Calling / Agents
Research notes
- Gated (requires accepting license and citing APIGen). Paper: arxiv 2406.18518. 1,896 affected data points were regenerated in July 2024 after community-identified issues.