11,357 synthetic multi-turn Korean conversations in which an assistant calls tools, reads their responses, and completes a user's task. Across three domains (commerce, travel, calendar) with 30 mock tools.
Every trajectory was produced by simulating three roles — a user, the assistant, and a deterministic mock tool environment — and then verified by replaying the recorded calls against that environment.
This is… See the full description on the dataset page:
https://huggingface.co/datasets/drlee1/kotraj.
