List your company
ResearchFormerly WizWand

Image Classification on ImageNet & Variants

73.23AccuracyTuneCLIP
12.3928.18543.9859.775Nov 11, 2025Nov 21, 2025Dec 2, 2025Dec 13, 2025Dec 23, 2025Jan 3, 2026Jan 14, 2026
Updated 8mo ago

Evaluation results

MethodLinks
TuneCLIPBackbone=ViT-H/14-quickgelu2026.0173.23
BaselineBackbone=ViT-H/14-quickgelu2026.0171.8
NeuCLIPTraining Dataset=DFN-14M2025.1133.79
AmorLIPTraining Dataset=DFN-14M2025.1133.35
FastCLIPTraining Dataset=DFN-14M2025.1133.03
SigLIPTraining Dataset=DFN-14M2025.1132.89
OpenCLIPTraining Dataset=DFN-14M2025.1132.65
NeuCLIPTraining Dataset=CC12M2025.1125.09
FastCLIPTraining Dataset=CC12M2025.1124.61
AmorLIPTraining Dataset=CC12M2025.1123.48
SigLIPTraining Dataset=CC12M2025.1122.67
OpenCLIPTraining Dataset=CC12M2025.1121.21
NeuCLIPTraining dataset=CC3M, Evaluation protocol=Zero-shot, Number of training runs=32025.1119.85
FastCLIPTraining dataset=CC3M, Evaluation protocol=Zero-shot, Number of training runs=32025.1119.09
AmorLIPTraining dataset=CC3M, Evaluation protocol=Zero-shot, Number of training runs=32025.1117.78
SigLIPTraining dataset=CC3M, Evaluation protocol=Zero-shot, Number of training runs=32025.1116.08
OpenCLIPTraining dataset=CC3M, Evaluation protocol=Zero-shot, Number of training runs=32025.1114.73