List your company
ResearchFormerly WizWand

Object Hallucination Detection on POPE (average over three sampling strategies)

83.94AUROCInsLen
44.88855.026565.16575.3035May 12, 2026
Updated 6mo ago

Evaluation results

MethodLinks
InsLenModel=LLaVA-1.5-7B2026.0583.9496.14
InsLenModel=mPLUG-Owl3-8B2026.0579.4296.45
InsLenModel=LLaVA-OneVision1.5-8B2026.0576.595.9
InsLenModel=Qwen3-VL-8B2026.0575.4293.7
InsLenModel=InstructBLIP-7B2026.0574.0994.59
EntropyModel=Qwen3-VL-8B2026.0573.5292.89
GLSIMModel=LLaVA-1.5-7B2026.0570.1387.32
NLLModel=LLaVA-OneVision1.5-8B2026.0568.4392.27
GLSIMModel=mPLUG-Owl3-8B2026.0568.3889.4
GLSIMModel=Qwen3-VL-8B2026.0568.2488.44
SVARModel=mPLUG-Owl3-8B2026.0567.1690.02
NLLModel=LLaVA-1.5-7B2026.0564.2688.84
Internal Conf.Model=LLaVA-OneVision1.5-8B2026.0563.7486.01
SVARModel=LLaVA-OneVision1.5-8B2026.0563.6285
SVARModel=Qwen3-VL-8B2026.0561.4788.23
Contextual LensModel=Qwen3-VL-8B2026.0561.4286.97
EntropyModel=LLaVA-OneVision1.5-8B2026.0560.7384.69
GLSIMModel=LLaVA-OneVision1.5-8B2026.0560.687.89
NLLModel=mPLUG-Owl3-8B2026.0559.5188.46
GLSIMModel=InstructBLIP-7B2026.0558.288.46
SVARModel=InstructBLIP-7B2026.0558.0787.18
Contextual LensModel=InstructBLIP-7B2026.0556.0783.94
Internal Conf.Model=mPLUG-Owl3-8B2026.0555.4689.63
EntropyModel=mPLUG-Owl3-8B2026.0555.3587.35
NLLModel=Qwen3-VL-8B2026.0555.1992.59
EntropyModel=InstructBLIP-7B2026.0554.6587.55
Internal Conf.Model=LLaVA-1.5-7B2026.0554.4185.87
Contextual LensModel=LLaVA-OneVision1.5-8B2026.0554.2587.35
Contextual LensModel=LLaVA-1.5-7B2026.0553.4683.17
NLLModel=InstructBLIP-7B2026.0553.286.64
EntropyModel=LLaVA-1.5-7B2026.0552.5985.66
Internal Conf.Model=InstructBLIP-7B2026.055281.77
SVARModel=LLaVA-1.5-7B2026.0551.6285.21
Contextual LensModel=mPLUG-Owl3-8B2026.0546.6486.75
Internal Conf.Model=Qwen3-VL-8B2026.0546.3987.54