List your company
ResearchFormerly WizWand

Visual Reasoning on MMMU-Pro (ECE)

14.4ECEProd(VF, Contr.)
12.15227.32642.557.674Sep 30, 2025
Updated 6mo ago

Evaluation results

MethodLinks
Prod(VF, Contr.)Model=LLaVA2025.0914.4
Min(VF, Contr.)Model=LLaVA2025.0915.4
Prod(VF, Contr.)Model=Avg2025.0918.2
Prod(VF, Contr.)Model=GPT-4o2025.0920
Min(VF, Contr.)Model=Avg2025.0920.1
Prod(VF, Contr.)Model=Qwen2.52025.0920.3
Min(VF, Contr.)Model=GPT-4o2025.0921.5
InformativenessModel=LLaVA2025.0921.6
ContrastivenessModel=LLaVA2025.0922.1
Min(VF, Contr.)Model=Qwen2.52025.0923.4
ContrastivenessModel=Avg2025.0925.1
ContrastivenessModel=GPT-4o2025.0925.6
PlausibilityModel=GPT-4o2025.0927.2
ContrastivenessModel=Qwen2.52025.0927.6
Avg(VF, Contr.)Model=GPT-4o2025.0927.9
PlausibilityModel=LLaVA2025.0931.8
PlausibilityModel=Avg2025.0932.7
Avg(VF, Contr.)Model=Avg2025.0934.3
SimulatabilityModel=GPT-4o2025.0935.7
Avg(VF, Contr.)Model=LLaVA2025.0936
Avg(VF, Contr.)Model=Qwen2.52025.0938.9
PlausibilityModel=Qwen2.52025.0939.1
SimulatabilityModel=Avg2025.0940.7
Visual FidelityModel=GPT-4o2025.0941.2
SimulatabilityModel=Qwen2.52025.0941.8
SimulatabilityModel=LLaVA2025.0944.6
InformativenessModel=Avg2025.0948.7
Visual FidelityModel=Avg2025.0950.1
Visual FidelityModel=Qwen2.52025.0953.5
InformativenessModel=GPT-4o2025.0953.8
Visual FidelityModel=LLaVA2025.0955.8
InformativenessModel=Qwen2.52025.0970.6