List your company

Generative Hallucination on AMBER generative subset

10.9CHAIRBaseline
4.2445.9727.79.428Jan 5, 2026

Evaluation results

MethodLinks
BaselineModel=Shikra2026.0110.949.550.7
VCDModel=Shikra2026.019.746.950.2
OPERAModel=Shikra2026.018.942.851
ICTModel=Shikra2026.018.742.550.8
AFTER (w/o QAO)Model=Shikra2026.017.938.250.6
VTIModel=Shikra2026.017.538.548.6
BaselineModel=InstructBLIP2026.017.435.453.5
HACLModel=LLaVA-v1.52026.017.131.449.6
BaselineModel=LLaVA-v1.52026.016.931.648.9
VCDModel=InstructBLIP2026.016.932.353.8
AFTERModel=Shikra2026.016.933.250.4
OPERAModel=InstructBLIP2026.016.631.453.5
ICTModel=InstructBLIP2026.016.230.853.6
AFTER (w/o QAO)Model=InstructBLIP2026.015.828.653.7
ICTModel=LLaVA-v1.52026.015.426.648.8
VTIModel=InstructBLIP2026.015.326.753
AFTER (w/o QAO)Model=LLaVA-v1.52026.015.222.348.6
AFTERModel=InstructBLIP2026.015.225.153.6
VCDModel=LLaVA-v1.52026.015.127.648.6
VTIModel=LLaVA-v1.52026.015.123.747.8
OPERAModel=LLaVA-v1.52026.014.927.949
AFTERModel=LLaVA-v1.52026.014.520.548.7