Image Classification on ImageNet-Sketch (test)
0.7337Top-1 AccPaLI-X
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PaLI-XResolution=756, Evaluation Protocol=fine-tuned2023.05 | 0.7337 | — | — | — | — | |
| PaLI-XResolution=224, Evaluation Protocol=fine-tuned2023.05 | 0.7256 | — | — | — | — | |
| PaLI-17BResolution=224, Evaluation Protocol=fine-tuned2023.05 | 0.7121 | — | — | — | — | |
| RobustViTBackbone=AR-L, Annotated segmentation=true2022.06 | 0.542 | 0.758 | — | — | — | |
| RobustViTBackbone=AR-L, Annotated segmentation=false2022.06 | 0.541 | 0.758 | — | — | — | |
| DPEBackbone=CLIP-ViT-B/162025.01 | 0.5226 | — | — | — | — | |
| FGABackbone=CLIP-ViT-B/162025.01 | 0.5181 | — | — | — | — | |
| OriginalBackbone=AR-L, Annotated segmentation=false2022.06 | 0.518 | 0.736 | — | — | — | |
| TDABackbone=CLIP-ViT-B/162025.01 | 0.5054 | — | — | — | — | |
| PromptAlign⋆Backbone=CLIP-ViT-B/162025.01 | 0.5023 | — | — | — | — | |
| MTA+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4996 | — | — | — | — | |
| DiffTPT+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.495 | — | — | — | — | |
| TPT+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.494 | — | — | — | — | |
| STSS+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4936 | — | — | — | — | |
| SAPT+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4931 | — | — | — | — | |
| Tip-AdapterBackbone=CLIP-ViT-B/162025.01 | 0.4888 | — | — | — | — | |
| CoCoOpBackbone=CLIP-ViT-B/162025.01 | 0.4875 | — | — | — | — | |
| SAR†+CoOpBackbone=CLIP-ViT-B/16, Evaluation Protocol=single-sample setting2025.01 | 0.4865 | — | — | — | — | |
| C-TPT+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4863 | — | — | — | — | |
| MTABackbone=CLIP-ViT-B/162025.01 | 0.4858 | — | — | — | — | |
| ZERO+CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4852 | — | — | — | — | |
| C-TPTBackbone=CLIP-ViT-B/162025.01 | 0.485 | — | — | — | — | |
| ZEROBackbone=CLIP-ViT-B/162025.01 | 0.4829 | — | — | — | — | |
| CLIP-ViT-B/16Backbone=CLIP-ViT-B/162025.01 | 0.4824 | — | — | — | — | |
| CoOpBackbone=CLIP-ViT-B/162025.01 | 0.4799 | — | — | — | — | |
| DiffTPTBackbone=CLIP-ViT-B/162025.01 | 0.468 | — | — | — | — | |
| TPTBackbone=CLIP-ViT-B/162025.01 | 0.464 | — | — | — | — | |
| RobustViTBackbone=AR-B, Annotated segmentation=false2022.06 | 0.447 | 0.674 | — | — | — | |
| RobustViTBackbone=AR-B, Annotated segmentation=true2022.06 | 0.446 | 0.674 | — | — | — | |
| APE + FD-AlignShots=16, Base Method=APE, Alignment=FD-Align2023.10 | 0.4423 | — | — | — | — | |
| APE-T + FD-AlignShots=16, Base Method=APE-T, Alignment=FD-Align2023.10 | 0.4404 | — | — | — | — | |
| Tip + FD-AlignShots=16, Base Method=Tip-adapter, Alignment=FD-Align2023.10 | 0.4384 | — | — | — | — | |
| Tip-F + FD-AlignShots=16, Base Method=Tip-adapter-F, Alignment=FD-Align2023.10 | 0.4367 | — | — | — | — | |
| FT + FD-AlignShots=16, Strategy=Fully Fine-tuning, Alignment=FD-Align2023.10 | 0.435 | — | — | — | — | |
| APEShots=16, Base Method=APE2023.10 | 0.4328 | — | — | — | — | |
| APE-TShots=16, Base Method=APE-T2023.10 | 0.4323 | — | — | — | — | |
| Right for the Right Reason (RRR)Backbone=AR-B, Annotated segmentation=true2022.06 | 0.432 | 0.657 | — | — | — | |
| OriginalBackbone=AR-B, Annotated segmentation=false2022.06 | 0.431 | 0.657 | — | — | — | |
| GradMaskBackbone=AR-B, Annotated segmentation=true2022.06 | 0.431 | 0.657 | — | — | — | |
| Tip-FShots=16, Base Method=Tip-adapter-F2023.10 | 0.4254 | — | — | — | — | |
| TipShots=16, Base Method=Tip-adapter2023.10 | 0.4248 | — | — | — | — | |
| CLIPZero-shot=true2023.10 | 0.4231 | — | — | — | — | |
| FTShots=16, Strategy=Fully Fine-tuning2023.10 | 0.4224 | — | — | — | — | |
| RobustViTBackbone=ViT-L, Annotated segmentation=false2022.06 | 0.418 | 0.648 | — | — | — | |
| CLIP-RefineZero-shot=true2025.04 | 0.4146 | — | — | — | — | |
| RobustViTBackbone=ViT-L, Annotated segmentation=true2022.06 | 0.412 | 0.643 | — | — | — | |
| OriginalBackbone=ViT-L, Annotated segmentation=false2022.06 | 0.404 | 0.634 | — | — | — | |
| Self-KDZero-shot=true2025.04 | 0.3852 | — | — | — | — | |
| Pre-trainedZero-shot=true2025.04 | 0.3838 | — | — | — | — | |
| ConvNeXt-SNumber of Parameters=50.0M, FLOPs=8.7G2023.03 | 0.371 | — | — | — | — | |
| NEXcepTion-SFLOPs=8.5G, Params=43.4M2022.12 | 0.366 | — | — | — | — | |
| FastViT-MA36Number of Parameters=42.7M, FLOPs=7.9G2023.03 | 0.366 | — | — | — | — | |
| GLAShots=16, Backbone=ResNet-502023.10 | 0.364 | — | — | — | — | |
| RobustViTBackbone=ViT-B, Annotated segmentation=true2022.06 | 0.362 | 0.585 | — | — | — | |
| GradMaskBackbone=ViT-B, Annotated segmentation=true2022.06 | 0.358 | 0.578 | — | — | — | |
| Right for the Right Reason (RRR)Backbone=ViT-B, Annotated segmentation=true2022.06 | 0.358 | 0.578 | — | — | — | |
| RobustViTBackbone=ViT-B, Annotated segmentation=false2022.06 | 0.358 | 0.582 | — | — | — | |
| FastViT-SA36Number of Parameters=30.4M, FLOPs=5.6G2023.03 | 0.358 | — | — | — | — | |
| m2-mixZero-shot=true2025.04 | 0.3559 | — | — | — | — | |
| ContrastiveZero-shot=true2025.04 | 0.3557 | — | — | — | — | |
| CLIPShots=0, Backbone=ResNet-50, Mode=Zero-shot2023.10 | 0.355 | — | — | — | — | |
| OriginalBackbone=ViT-B, Annotated segmentation=false2022.06 | 0.354 | 0.574 | — | — | — | |
| RVT-SNumber of Parameters=22.1M, FLOPs=4.7G2023.03 | 0.35 | — | — | — | — | |
| NEXcepTion-TFLOPs=4.7G, Params=24.5M2022.12 | 0.3473 | — | — | — | — | |
| HyCD+LAlignZero-shot=true2025.04 | 0.3473 | — | — | — | — | |
| NEXcepTion-TPFLOPs=4.5G, Params=26.6M2022.12 | 0.347 | — | — | — | — | |
| RVT-SFLOPs=4.7G, Params=23.3M2022.12 | 0.347 | — | — | — | — | |
| EfficientNet-B4Number of Parameters=19.3M, FLOPs=4.2G2023.03 | 0.341 | — | — | — | — | |
| FastViT-SA24Number of Parameters=20.6M, FLOPs=3.8G2023.03 | 0.34 | — | — | — | — | |
| RobustViTBackbone=AR-S, Annotated segmentation=false2022.06 | 0.338 | 0.555 | — | — | — | |
| ConvNeXt-TFLOPs=4.5G, Params=28.6M2022.12 | 0.338 | — | — | — | — | |
| ConvNeXt-TNumber of Parameters=28.6M, FLOPs=4.5G2023.03 | 0.338 | — | — | — | — | |
| RobustViTBackbone=AR-S, Annotated segmentation=true2022.06 | 0.335 | 0.554 | — | — | — | |
| ConViT-SNumber of Parameters=27.8M, FLOPs=5.4G2023.03 | 0.331 | — | — | — | — | |
| ProGradShots=16, Backbone=ResNet-502023.10 | 0.331 | — | — | — | — | |
| PLOTShots=16, Backbone=ResNet-502023.10 | 0.33 | — | — | — | — | |
| Right for the Right Reason (RRR)Backbone=AR-S, Annotated segmentation=true2022.06 | 0.329 | 0.542 | — | — | — | |
| OriginalBackbone=AR-S, Annotated segmentation=false2022.06 | 0.328 | 0.542 | — | — | — | |
| GradMaskBackbone=AR-S, Annotated segmentation=true2022.06 | 0.325 | 0.537 | — | — | — | |
| CoOpShots=16, Backbone=ResNet-502023.10 | 0.325 | — | — | — | — | |
| RobustViTBackbone=DeiT-B, Annotated segmentation=false2022.06 | 0.315 | 0.499 | — | — | — | |
| ResNet-50Number of Parameters=25.6M, FLOPs=4.1G, AugMix=true2023.03 | 0.315 | — | — | — | — | |
| RILSBackbone=ViT-B/16, Zero-shot=true, Training Epochs=25, Training Data=L-20M2023.01 | 0.314 | — | — | — | — | |
| OriginalBackbone=DeiT-B, Annotated segmentation=false2022.06 | 0.312 | 0.486 | — | — | — | |
| GradMaskBackbone=DeiT-B, Annotated segmentation=true2022.06 | 0.312 | 0.491 | — | — | — | |
| MAE+CLIPBackbone=ViT-B/16, Zero-shot=true, Training Epochs=25, Training Data=L-20M2023.01 | 0.311 | — | — | — | — | |
| Right for the Right Reason (RRR)Backbone=DeiT-B, Annotated segmentation=true2022.06 | 0.309 | 0.488 | — | — | — | |
| RobustViTBackbone=DeiT-B, Annotated segmentation=true2022.06 | 0.309 | 0.492 | — | — | — | |
| RVT-TiNumber of Parameters=8.6M, FLOPs=1.3G2023.03 | 0.3 | — | — | — | — | |
| XceptionFLOPs=8.4G, Params=43.4M2022.12 | 0.2988 | — | — | — | — | |
| FastViT-SA12Number of Parameters=10.9M, FLOPs=1.9G2023.03 | 0.297 | — | — | — | — | |
| Causal TransportabilityAlgorithm=SimCLR2022.04 | 0.2951 | — | — | — | — | |
| RobustViTBackbone=DeiT-S, Annotated segmentation=false2022.06 | 0.294 | 0.48 | — | — | — | |
| ResNeXt50-32x4dNumber of Parameters=25.0M, FLOPs=4.3G2023.03 | 0.293 | — | — | — | — | |
| RobustViTBackbone=DeiT-S, Annotated segmentation=true2022.06 | 0.291 | 0.478 | — | — | — | |
| Swin-TFLOPs=4.5G, Params=28.3M2022.12 | 0.291 | — | — | — | — | |
| Swin-TNumber of Parameters=28.3M, FLOPs=4.5G2023.03 | 0.291 | — | — | — | — | |
| OriginalBackbone=DeiT-S, Annotated segmentation=false2022.06 | 0.288 | 0.467 | — | — | — | |
| CLIPBackbone=ViT-B/16, Zero-shot=true, Training Epochs=25, Training Data=L-20M2023.01 | 0.281 | — | — | — | — | |
| ERMAlgorithm=SimCLR2022.04 | 0.2743 | — | — | — | — |