Dataset · Hugging Face

holland202/sovereign-evidence-bench

Name: sovereign-evidence-bench
Creator: holland202
License: mit
Keywords: task_categories:text-classification, language:en, license:mit, size_categories:n<1K, format:json, modality:tabular, modality:text, library:datasets, library:dask, library:polars, library:mlcroissant, region:us, evaluation, llm-safety, prompt-injection

evidence_bench — evidence-boundary benchmark for small local models Can a small local language model hold an evidence boundary — admissibility, provenance, freshness, injection resistance — or does it just pattern-match on whether the evidence mentions the claim? 39 hand-built cases, 11 categories, three verdicts (SUPPORTED / REFUTED / NOT_SUPPORTED). Gold labels are derived mechanically by oracle.py from hidden structure, and the run is preregistered (PREREG.md, committed… See the full description on the dataset page: https://huggingface.co/datasets/holland202/sovereign-evidence-bench.

View on Hugging FaceReport an issue

Author / organization	holland202
License (as declared)	mit
Downloads	137 (as reported by Hugging Face)
Likes	1
Created	2 October 2026
Last updated at source	2 October 2026

evaluationllm-safetyprompt-injectionprovenanceon-devicellama.cppgguf

Embed this entry

<iframe src="https://silytics.si/embed/item/655cd50b-f082-4ccc-81dd-001c1c9f5c7e" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 4 days

137 now. Counts as reported by the source, snapshotted daily by Silytics.

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 19 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.