List your company

Object Hallucination Detection on AMBER out-of-distribution (OOD)

0.8611AUCOverthinking Score
0.4855560.5830530.680550.778047Mar 8, 2026

Evaluation results

MethodLinks
Overthinking ScoreVLM=LLaVA-1.5, Classifier Type=GB2026.030.86110.36650.7158
Overthinking ScoreVLM=LLaVA-1.5, Classifier Type=MLP2026.030.83280.30910.5976
MetaTokenVLM=LLaVA-1.5, Classifier Type=GB2026.030.82150.23480.6554
SVARVLM=LLaVA-1.52026.030.79510.12780.5687
MetaTokenVLM=LLaVA-1.5, Classifier Type=LR2026.030.7950.11550.5476
MetaTokenVLM=LLaVA-1.5, Classifier Type=MLP2026.030.79410.14040.5854
Overthinking ScoreVLM=LLaVA-1.5, Classifier Type=LR2026.030.70760.07970.5392
HalLocVLM=LLaVA-1.52026.030.50.02270.4944