List your company
ResearchFormerly WizWand

Open-ended Vision-Language Generation on GPT-4V Hallucination Evaluation

6.75Correctness ScoreACE
4.59725.15615.7156.2739May 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
ACEModel=LLaVA-NeXT, Comparison Baseline=SID2026.056.756.32
ACEModel=LLaVA-NeXT, Comparison Baseline=OPERA2026.056.586.27
ACEModel=LLaVA-NeXT, Comparison Baseline=Greedy2026.056.515.89
ACEModel=LLaVA-1.5, Comparison Baseline=SID2026.056.256.02
ACEModel=LLaVA-NeXT, Comparison Baseline=VCD2026.056.195.84
OPERAModel=LLaVA-1.52026.056.185.52
SIDModel=LLaVA-NeXT2026.056.175.85
ACEModel=LLaVA-1.5, Comparison Baseline=OPERA2026.056.135.99
ACEModel=LLaVA-1.5, Comparison Baseline=VCD2026.056.125.97
ACEModel=Shikra, Comparison Baseline=SID2026.056.075.96
OPERAModel=LLaVA-NeXT2026.056.075.29
ACEModel=LLaVA-1.5, Comparison Baseline=Greedy2026.055.936.08
ACEModel=Shikra, Comparison Baseline=OPERA2026.055.915.92
VCDModel=LLaVA-NeXT2026.055.885.53
SIDModel=LLaVA-1.52026.055.865.68
ACEModel=InstructBLIP, Comparison Baseline=SID2026.055.845.51
ACEModel=Shikra, Comparison Baseline=Greedy2026.055.825.31
ACEModel=InstructBLIP, Comparison Baseline=OPERA2026.055.715.38
ACEModel=Shikra, Comparison Baseline=VCD2026.055.685.44
ACEModel=InstructBLIP, Comparison Baseline=Greedy2026.055.575.24
VCDModel=LLaVA-1.52026.055.515.58
ACEModel=InstructBLIP, Comparison Baseline=VCD2026.055.415.49
SIDModel=Shikra2026.055.415.38
OPERAModel=Shikra2026.055.394.82
GreedyModel=LLaVA-NeXT2026.055.395.61
SIDModel=InstructBLIP2026.055.355.09
VCDModel=Shikra2026.055.265.31
OPERAModel=InstructBLIP2026.055.224.91
GreedyModel=LLaVA-1.52026.055.125.84
VCDModel=InstructBLIP2026.055.045.26
GreedyModel=Shikra2026.054.975.22
GreedyModel=InstructBLIP2026.054.685.15