2 papers
cs.AI2026
Are Benchmarks Reliable? Toward Structural Diagnosis via Item-Level Capability Boundaries
Haiquan Hu, Yuzhu Liang, Weicheng Tang +4
Evaluating large language models (LLMs) relies heavily on benchmark scores, yet aggregate metrics can obscure whether individual items serve their intended measurement purpose. BSD…
cs.CV2026
PROVE: Proof-guided Regime-aware Operator Verification for Hallucination Detection in Medical Visual Question Answering
Keyang Zhou, Siyi Li, Zhongnan Shi +3
In medical visual question answering (VQA), hallucinations of vision-language models (VLMs) may lead to confident but incorrect responses, raising the risk of diagnostic errors. Ex…