List your company

Hallucination detection on MSCOCO

91.7AUC ROCHALLUSHIFT++
48.12459.43770.7582.063Dec 8, 2025

Evaluation results

MethodLinks
HALLUSHIFT++Model=Llama-3.2-11B-Vision2025.1291.78371.8
HALLUSHIFT++Model=paligemma-3b-mix-2242025.128777.569
HALLUSHIFT++Model=Llama-3.2-11B-Vision-Instruct2025.1286.177.462.6
HALLUSHIFT++Model=kosmos-2-patch14-2242025.1285.467.257.6
HALLUSHIFT++Model=Phi-3.5-vision-instruct2025.1285.371.456.9
HALLUSHIFT++Model=instructblip-vicuna-7b2025.128567.756.4
HALLUSHIFT++Model=llava-1.5-7b-hf2025.1284.767.258.4
HALLUSHIFT++Model=Qwen2.5-VL-3B-Instruct2025.1283.672.958.5
HALLUSHIFTModel=Llama-3.2-11B-Vision2025.1265.852.560.9
HALLUSHIFTModel=paligemma-3b-mix-2242025.1261.765.464.2
ITIModel=paligemma-3b-mix-2242025.1258.462.161.5
HALLUSHIFTModel=Qwen2.5-VL-3B-Instruct2025.1255.961.156.2
HALLUSHIFTModel=Phi-3.5-vision-instruct2025.1254.945.954.3
HALLUSHIFTModel=kosmos-2-patch14-2242025.125443.741
HALLUSHIFTModel=llava-1.5-7b-hf2025.1253.142.143.1
HALLUSHIFTModel=Llama-3.2-11B-Vision-Instruct2025.1252.152.160.5
HALLUSHIFTModel=instructblip-vicuna-7b2025.1251.734.343.3
ITIModel=kosmos-2-patch14-2242025.1250.340.838.2
ITIModel=Llama-3.2-11B-Vision-Instruct2025.1249.848.357.2