ShareGPT Tool Calls: Structured Function-Calling Conversations for AI Fine-Tuning

A clean, structured tool-call dataset formatted in the canonical ShareGPT schema used by models like OpenHermes and UltraChat. It is derived from the glaiveai/glaive-function-calling-v2 dataset and was created by Guilherme34. The dataset was last updated on April 26, 2026.

Use Cases

• Fine-tune language models for tool-use based on structured conversation examples.
• Benchmark model performance on function-calling tasks using a standardized format.
• Train instruction-following assistants to parse user requests and invoke appropriate tools.
• Study the structure of human-AI conversations involving external API calls.

Strengths

• Follows the canonical ShareGPT schema, ensuring compatibility with popular fine-tuning pipelines.
• Derived from a known source dataset (glaiveai/glaive-function-calling-v2), suggesting a structured origin.
• Last update timestamp (2026-04-26) is provided, indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality and scope require manual inspection.

Provenance

Source

Derived from the glaiveai/glaive-function-calling-v2 dataset.

Collection Method

Processed and formatted into the ShareGPT schema.

Freshness

Last updated 2026-04-26 19:10:40.