List your company

Hallucination Evaluation on MME Color

170Score (Chinese)Baseline
76.4100.7125149.3Jun 3, 2025

Evaluation results

MethodLinks
BaselineBackbone=Qwen-VL-Chat2025.06170180150153.3——103.3165148.3
CLAIMBackbone=Qwen-VL-Chat2025.06165170158.390——165—149.7
CLAIMBackbone=LLaVA-1.52025.06155120125150108.3———131.7
VCDBackbone=Qwen-VL-Chat2025.06150—165146.7——93.3160143.7
VCDBackbone=LLaVA-1.52025.0695—1458513088.3——108.7
BaselineBackbone=LLaVA-1.52025.06801651357511080——96