baseline-v0

Centroid retrieval router over 265 tool names from public agent benchmarks (tau-bench, Hermes, ToolACE). TF-IDF only, no torch, no GPU, no API key. Smallest of the four pretrained baselines.

Quick start

Install the SDK directly from GitHub (PyPI publish pending):

pip install git+https://github.com/dalek-ai/agent-tool-router.git
from agent_tool_router import Router
r = Router.from_pretrained("baseline-v0")
r.route("cancel my pending order and refund the credit", k=3)

Router.from_pretrained("<name>") resolves the bare name to dalek-ai/<name> on HuggingFace Hub and caches locally on first call.

Numbers

Cross-corpus, held-out 2 041 tasks, vocab = 265 tools (≥ 3 occurrences):

metric model random ratio
top-1 33.0% 0.38% 87.5×
top-3 63.8% 1.13% 56.4×
top-5 83.0% 1.89% 44.0×

For the long-tail catalog (18 671 tools) prefer baseline-v1-desc or baseline-v1-desc-hybrid.

Repo & demo

github.com/dalek-ai/agent-tool-router · MIT. Live demo: dalek-ai/agent-tool-router-demo (gradio Space, FR/EN).

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support