willykeenan's picture
Publish reproducible BANKING77 model and agent-evaluation evidence
438f1e0 verified
Raw History Blame Contribute Delete
673 Bytes
cff-version: 1.2.0
message: "If you use this reference implementation, please cite it using this metadata."
title: "Waggle + Kea: Certified Task-Sufficient Machine Handoffs"
type: software
version: 0.3.0
date-released: 2026-08-08
authors:
- family-names: Keenan
given-names: William
repository-code: "https://github.com/willykeenan/waggle-kea"
url: "https://github.com/willykeenan/waggle-kea"
license: MIT
keywords:
- machine learning evaluation
- intent classification
- uncertainty calibration
- decision sufficiency
- selective prediction
- multi-agent systems
- machine coordination
- AI observability
- auditability
- deterministic protocols