List your company

Visual Misinterpretation Hallucination Evaluation on CXR-VisHal (test)

89.3Accuracy (A)MedQwen
31.89246.79661.776.604Apr 1, 2026

Evaluation results

MethodLinks
MedQwenevaluation_mode=close-ended2026.0489.370.283.991.682.5
GPT-4oevaluation_mode=close-ended2026.048859.578.892.179.4
LLaVA-Med-1.5evaluation_mode=close-ended2026.048449.465.184.568.4
LLaVA-NeXT 7Bevaluation_mode=close-ended2026.0481.74347.436.251.8
CheXagentevaluation_mode=close-ended2026.0478.257.673.985.173.9
LLaVA-NeXT 13Bevaluation_mode=close-ended2026.0477.639.148.656.353.4
LLaVA-Medevaluation_mode=close-ended2026.0469.845.272.58069.8
LLM-CXRevaluation_mode=close-ended2026.0468.150.474.340.367.5
Med-Flamingoevaluation_mode=close-ended2026.0436.132.457.633.248.9
MiniGPT-4evaluation_mode=close-ended2026.0434.130.157.335.448.3