Note Wholly synthetic set of 130 operational safety scenarios across 13 categories. Useful for structured review and regression testing; it is not a safety certification.

Note Balanced synthetic set of 192 operational requests with deterministic 128/32/32 splits. Useful for transparent baselines, not representative production traffic.

nwhite-systems/nwhite-ai-operations-intent-classifier

Text Classification • Updated 1 day ago • 10

Note Genuine CPU-friendly TF-IDF and logistic-regression baseline with reloadable artefacts. Perfect synthetic scores do not establish real-world accuracy or fairness.

Note Free Static Space connecting synthetic use cases, safety scenarios and browser-local classification. Educational only; no paid endpoint or high-stakes decision use.

ibm-granite/granite-guardian-hap-38m

Text Classification • 38.5M • Updated Dec 19, 2024 • 673 • 48

Note Compact 38M English HAP classifier illustrating a narrowly scoped guard component. It does not cover privacy, permissions, factuality or broader agent safety.

ibm-granite/granite-docling-258M

Image-Text-to-Text • 0.3B • Updated Sep 23, 2025 • 490k • 1.24k

Note Compact document-conversion model with evaluation and responsible-use notes. Generated structure or text should retain source provenance and human review.