HALLMARK: Diagnosing Three Failure Modes in LLM Citation Verifiers
HALLMARK (Hallucination benchmark): 2,526 BibTeX entries spanning 14 hallucination types, three difficulty tiers, six diagnostic sub-tests per entry, and a contamination-resistant held-out split, and across the benchmark one result is consistent: the false-positive rate, not recall, decides whether a verifier is deploy...