KnowHal is a new benchmark that jointly tests entity, attribute, relation, and knowledge hallucinations in multimodal language models using paired true/false questions on shared images.
Proceedings of the Computer Vision and Pattern Recognition Conference , pages=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
KnowHal: A Knowledge-Driven Benchmark for Comprehensive Multimodal Hallucination Evaluation
KnowHal is a new benchmark that jointly tests entity, attribute, relation, and knowledge hallucinations in multimodal language models using paired true/false questions on shared images.