180K labeled tool-calling decisions released on HF to train tiny router models
MaziyarPanahi · x · 2026-09-29
- Developer Maziyar Panahi released AgentToolDecisions-180K on Hugging Face: 180,000 Jev-format tool-calling decisions, typed and labeled (which tool, whether to call one, are the args complete).
- Splits: 171,056 train / 2,713 validation / 6,231 test, with no decision group crossing splits; labels come from NVIDIA's upstream open agent data, not from Jev.
- Purpose: train your own small router/decision model to make tool-calling calls instead of a big LLM, then race it against Jev.
Related event: 180K-Example Tool-Calling Dataset Open-Sourced for Small Model Routers(3 posts)→
More from coding & agent
- GPT-6.1 Sol matches GPT-6 Astra at ~1/5 cost, ultrafast version targets 300 tok/s — haider1 · 2026-09-30
- Dots review: OpenAI's always-on agents read your Slack and flag conflicts first — every · 2026-09-30
- Developer Has Dots Drive Its Own Computer to Run Blender and Model Itself in 3D — Dimillian · 2026-09-30
- GPT-Sol 6.1 Ships Too, as Omarsar Argues Codex-Dots Combo Unlocks New Agent Workflows — omarsar0 · 2026-09-30
- LangSmith adds Trajectories: collapsing agent sessions into readable, ordered execution paths — LangChain · 2026-09-30
- Codex Cloud announced at OpenAI DevDay, available across Plus through Enterprise tiers — testingcatalog · 2026-09-30