List your company

Object Hallucination Probing on GQA Adversarial

83.27AccuracyCAI
66.765271.050175.33579.6199Dec 22, 2025Jan 22, 2026Feb 23, 2026Mar 26, 2026Apr 27, 2026May 28, 2026Jun 29, 2026

Evaluation results

MethodLinks
CAIBackbone=Qwen-VL2026.0683.2783.51
MESAModel=Qwen-VL2026.0482.7381.93
SchroMindModel=LLaVA-1.5-7B2026.0281.7682.31
VCDBackbone=Qwen-VL2026.0681.782.74
ICTModel=Qwen-VL2026.0481.6180.75
ONLYBackbone=Qwen-VL2026.0681.4381.67
ScalpelBase Model=LLaVA-1.5-7B2026.028182.09
ICDModel=Qwen-VL2026.048180.06
PAIBackbone=Qwen-VL2026.0680.8780.41
VTIModel=Qwen-VL2026.0480.4778.44
VAFModel=Qwen-VL2026.0480.0379.75
VCDModel=Qwen-VL2026.0480.0180.75
MESAModel=LLaVA-v1.52026.0479.681.35
RegularBackbone=Qwen-VL2026.0679.278.96
ICTModel=LLaVA-v1.52026.0479.1381.07
VAFModel=LLaVA-v1.52026.0479.180.41
HDDModel=LLaVA-NeXT2025.1278.480.5
VTIModel=LLaVA-v1.52026.0478.3379.78
CAIBackbone=LLaVA-1.52026.0678.3380.46
ICTBase Model=LLaVA-1.5-7B2026.0277.480.11
ICTModel=LLaVA-1.5-7B2026.0277.480.11
RITUALModel=LLaVA-NeXT2025.1277.278.3
HDDModel=LLaVA-1.52025.1276.478.1
VCDBase Model=LLaVA-1.5-7B2026.0276.0978.78
VCDModel=LLaVA-1.5-7B2026.0276.0978.78
VCDModel=LLaVA-v1.52026.0476.0978.78
CAIBackbone=InstructBLIP2026.067677.65
ICDModel=LLaVA-v1.52026.0475.976.89
VCDModel=LLaVA-NeXT2025.1275.876.5
SamplingModel=LLaVA-NeXT2025.1275.474.7
RITUALModel=LLaVA-1.52025.1275.376.1
VCDModel=LLaVA-1.52025.1275.277.7
VanillaBase Model=LLaVA-1.5-7B2026.0275.0876.06
RegularModel=LLaVA-1.5-7B2026.0275.0876.06
OPERABase Model=LLaVA-1.5-7B2026.027578.71
OPERAModel=LLaVA-1.5-7B2026.027578.71
SamplingModel=LLaVA-1.52025.1274.875.8
NulluModel=Qwen-VL2026.0473.6165.59
VanillaModel=Qwen-VL2026.0473.5365.53
HDDModel=InstructBLIP2025.1273.576
VCDBackbone=InstructBLIP2026.0673.476.38
VCDModel=InstructBLIP2025.1273.176.1
RITUALModel=InstructBLIP2025.1272.875.5
NulluModel=LLaVA-v1.52026.0472.176.37
ONLYBackbone=LLaVA-1.52026.0671.576.75
SamplingModel=InstructBLIP2025.1270.474.1
ONLYBackbone=InstructBLIP2026.0669.3774.92
AVISCModel=LLaVA-1.5-7B2026.0269.275.58
PAIBackbone=InstructBLIP2026.0669.174.19
PAIBackbone=LLaVA-1.52026.066975.34
VCDBackbone=LLaVA-1.52026.0668.6375.41
RegularBackbone=InstructBLIP2026.0668.3373.25
M3IDModel=LLaVA-1.5-7B2026.0268.1374.78
RegularBackbone=LLaVA-1.52026.0667.9374.35
VanillaModel=LLaVA-v1.52026.0467.474.31