sourced-evidence 0.1.1

Per-claim grounded/contradicted/unverified checking of LLM output against its own source context -- never a blended trust score.

aireleases

Sources