180k typed tool-calling decisions released so tiny models can route tools
MaziyarPanahi · x · 2026-09-29
Maziyar Panahi is releasing a dataset of 180,000 Jev-format tool-calling decisions, fully typed and labeled across three axes: which tool to call, whether to call one at all, and whether the args are complete. The data is built on NVIDIA's open agent dataset with clean train/val/test splits.
The idea: train your own small router model to make these decisions instead of paying an LLM to do it, then race it against Jev.
More from coding & agent
- An AWS cost agent kept trying to kill the DR database — so its owner taught it to hold a grudge — Gamer_Star2020 · 2026-09-30
- Why auto-saving LLM hypotheses into memory is a terrible idea: lessons from an incident triage agent — afsana08 · 2026-09-30
- A clinical AI copilot that must cite patient records — or admit it doesn't know — sritha_reddy13 · 2026-09-29
- From stateless chatbots to memory-driven sales agents: how DealMind was built — Kanvitha · 2026-09-29
- Side-by-side test: incident agent with historical memory nails the fix, stateless LLM just guesses — boorgularishvitha · 2026-09-29
- Beyond coding agents: a 6-layer roadmap to building autonomous software factories — MaryamMiradi · 2026-09-29