CogniBench is a sentence-level benchmark that labels LLM inferences, explanations, and opinions as faithful or hallucinated, expanding hallucination evaluation beyond verbatim factual claims.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CogniBench: A Legal-inspired Framework and Dataset for Assessing Cognitive Faithfulness of Large Language Models
CogniBench is a sentence-level benchmark that labels LLM inferences, explanations, and opinions as faithful or hallucinated, expanding hallucination evaluation beyond verbatim factual claims.