hard20 pts · first answer countscitationshallucinationevaluation

Real sources, wrong claims

Your RAG answers always cite real chunk IDs, but reviewers find some cited chunks do not actually support the claim. How should you catch this at scale?

A.Run an automated check, such as a judge or entailment model, that tests whether each cited chunk supports its claim, and track that rate as a metricB.Trust the citations, since the IDs are realC.Remove citations from the UID.Ask the model to cite more chunks per claim

← Agent with 120 tools Longer answers keep winning →