List your company
ResearchFormerly WizWand

Image Classification on Caltech-101 (Top-1 Accuracy)

84.6Top-1 AccuracyITO
44.66455.03265.475.768Mar 3, 2026Mar 4, 2026Mar 5, 2026Mar 6, 2026Mar 7, 2026Mar 8, 2026Mar 9, 2026
Updated 7mo ago

Evaluation results

MethodLinks
ITOBackbone=ViT-L/16, Pre-training Dataset=DataComp-1B, Training Epochs=1, Evaluation Protocol=Zero-shot2026.0384.6
ITO sub2Backbone=ViT-L/16, Pre-training Dataset=DataComp-1B, Training Epochs=1, Evaluation Protocol=Zero-shot2026.0383.4
CLIPBackbone=ViT-L/16, Pre-training Dataset=DataComp-1B, Training Epochs=1, Evaluation Protocol=Zero-shot2026.0383.2
ITOPre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0379.1
ITO sub2Pre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0377.4
GIFBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0376.11
Diff-MixBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0375.8
FLAIRPre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0375.4
SLIPPre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0374.9
Diff-AugBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0372.35
CLIPPre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0372
SigLIPPre-training Dataset=CC12M, Backbone=ViT-B/16, Epochs=30, Evaluation protocol=Zero-shot2026.0371.4
DA-FusionBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0367.65
Real GuidanceBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0367.2
ITO sub2Pre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0366.1
ITOPre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0362.9
DiffuseMixBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0362.74
FLAIRPre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0361.6
SLIPPre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0356.6
MixupBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0351.75
BaselineBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0349.95
CLIPPre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0349.8
CutMixBackbone=ResNet-50, Training Protocol=Trained from scratch2026.0347.65
SigLIPPre-train dataset=CC3M, Backbone=ViT-B/16, Training Epochs=30, Evaluation Protocol=Zero-shot2026.0346.2