List your company

Hallucination Robustness on Hallusion-Bench

65.35aAccSceneAlign
25.268435.674246.0856.4858Jan 9, 2026

Evaluation results

MethodLinks
SceneAlignModel=Qwen3-VL-4B2026.0165.3541.0541.78
SceneAlignModel=Qwen2.5-VL-7B2026.0162.8837.5738.02
SFTModel=Qwen3-VL-4B2026.0162.139.4539.2
SceneAlignModel=InternVL3-8B2026.0161.7236.8835.01
SFTModel=Qwen2.5-VL-7B2026.0160.3436.7235.84
SceneAlignModel=Qwen2.5-VL-3B2026.0158.2534.5933.64
SFTModel=InternVL3-8B2026.0157.8432.9631.42
BaseModel=Qwen3-VL-4B2026.0157.6534.2834.2
BaseModel=Qwen2.5-VL-7B2026.0157.3133.2432.09
SFTModel=Qwen2.5-VL-3B2026.0153.2828.1528.36
BaseModel=InternVL3-8B2026.0149.5826.0324.45
SceneAlignModel=LLaVA-Next-8B2026.0146.8924.5718.02
LLaVA-ReasonerModel=LLaVA-Next-8B2026.0141.8520.8112.53
AoTModel=LLaVA-Next-8B2026.0141.6420.2314.95
SFTModel=LLaVA-Next-8B2026.0139.3520.0214.74
BaseModel=Qwen2.5-VL-3B2026.0134.2817.0520.66
BaseModel=LLaVA-Next-8B2026.0126.8118.7912.75