List your company

Hallucination Assessment on AMBER (test)

5.6CHAIRDyn. CEI
5.3127.2569.211.144Jan 9, 2026Feb 1, 2026Feb 25, 2026Mar 21, 2026Apr 13, 2026May 7, 2026May 31, 2026

Evaluation results

MethodLinks
Dyn. CEILVLM=InstructBLIP2026.015.63051.9—
Dyn. CEILVLM=LLaVA-1.52026.015.92548.1—
CAACLVLM=LLaVA-1.52026.0162548.7—
Stat. CEILVLM=InstructBLIP2026.016.131.753.4—
Stat. CEILVLM=LLaVA-1.52026.016.427.348.6—
AdaVBoostBackbone=InternVL3.52026.026.550.2472.32.78
AdaVBoostBackbone=Qwen3-VL2026.026.7451.372.123
AdaVBoostBackbone=LLaVA-NeXT2026.026.9739.8561.213.41
CAACLVLM=InstructBLIP2026.01730.951.9—
M3IDLVLM=LLaVA-1.52026.017.241.457.3—
OPERALVLM=LLaVA-1.52026.017.329.547.5—
VAFBackbone=InternVL3.52026.027.353.271.643.99
VGABackbone=InternVL3.52026.027.3662.0973.995.49
VAFBackbone=Qwen3-VL2026.027.4156.6472.753.38
VanillaBackbone=InternVL3.52026.027.4962.274.245.95
MGAPBackbone=LLaVA v1.5 7B2026.057.635.151.73.8
VAFBackbone=LLaVA-NeXT2026.027.6447.4163.243.7
VGABackbone=LLaVA-NeXT2026.027.6540.7462.393.59
VanillaBackbone=Qwen3-VL2026.027.6659.2473.593.75
VanillaBackbone=LLaVA-NeXT2026.027.8349.263.874.33
VGABackbone=Qwen3-VL2026.027.9760.3173.323.34
HalTrapperBackbone=LLaVA v1.5 7B2026.05836.351.53.8
VCDLVLM=LLaVA-1.52026.018.237.351.9—
PAIBackbone=Qwen3-VL2026.028.2363.3974.113.5
PAIBackbone=LLaVA-NeXT2026.028.2950.964.744.43
PAIBackbone=InternVL3.52026.028.516674.885.29
Dyn. CEILVLM=LLaVA-NeXT2026.018.646.660.4—
ICDBackbone=LLaVA v1.5 7B2026.058.637.351.13.9
CAACLVLM=LLaVA-NeXT2026.018.847.560.5—
VCDBackbone=LLaVA v1.5 7B2026.058.938.151.24.4
CODEBackbone=LLaVA v1.5 7B2026.05939.551.14.3
AvisCLVLM=LLaVA-NeXT2026.019.250.461.1—
LLaVA-NeXTLVLM=LLaVA-NeXT2026.019.351.360.6—
OPERALVLM=InstructBLIP2026.019.740.551.2—
Stat. CEILVLM=LLaVA-NeXT2026.011047.457.8—
AvisCLVLM=InstructBLIP2026.0110.146.851.2—
M3IDLVLM=InstructBLIP2026.0110.447.351.7—
VCDLVLM=LLaVA-NeXT2026.0110.557.263.5—
VCDLVLM=InstructBLIP2026.0110.846.653.4—
AvisCLVLM=LLaVA-1.52026.01114852.5—
LLaVA v1.5 7BBackbone=LLaVA v1.5 7B2026.0511.247.950.24.6
LLaVA-1.5LVLM=LLaVA-1.52026.0111.348.150.4—
M3IDLVLM=LLaVA-NeXT2026.0112.459.861.4—
InstructBLIPLVLM=InstructBLIP2026.0112.853.552.7—