Image Classification on CIFAR-100 (test) (Precision@K)
93.36Precision@1EVA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EVAModel=BEIT-L/14, Pre-training Dataset=CLIP OpenAI, IN-22K, IN-1K, Evaluation Protocol=Linear Probing2024.11 | 93.36 | — | — | — | |
| DINOprotocol=Fine Tuning, arch.=ViT-B/162024.08 | 91.7 | — | — | — | |
| BAMprotocol=Fine Tuning, arch.=ViT-B/162024.08 | 91.4 | — | — | — | |
| MoCo-v3protocol=Fine Tuning, arch.=ViT-B/162024.08 | 90.5 | — | — | — | |
| ScaleKDModel=ViT-B/16, Pre-training Dataset=IN-1K, Evaluation Protocol=Linear Probing2024.11 | 89.11 | — | — | — | |
| CLIPModel=ViT-B/16, Pre-training Dataset=CLIP OpenAI, IN-12K, IN-1K, Evaluation Protocol=Linear Probing2024.11 | 88.92 | — | — | — | |
| CLIPModel=ViT-B/16, Pre-training Dataset=LAION-2B, IN-12K, IN-1K, Evaluation Protocol=Linear Probing2024.11 | 88.88 | — | — | — | |
| Sup-INprotocol=Fine Tuning, arch.=ViT-B/162024.08 | 87.1 | — | — | — | |
| Supervised-INEvaluation Protocol=Fine-tuned2021.06 | 86.4 | — | — | — | |
| BYOLEvaluation Protocol=Fine-tuned2021.06 | 86.1 | — | — | — | |
| SimCLREvaluation Protocol=Fine-tuned2021.06 | 85.9 | — | — | — | |
| CLIPModel=ViT-B/16, Pre-training Dataset=LAION-2B, IN-1K, Evaluation Protocol=Linear Probing2024.11 | 85.51 | — | — | — | |
| SSL-HSIC (w/o target)Evaluation Protocol=Fine-tuned2021.06 | 85.3 | — | — | — | |
| SSL-HSIC (w/ target)Evaluation Protocol=Fine-tuned2021.06 | 84.7 | — | — | — | |
| CLIPModel=ViT-B/16, Pre-training Dataset=LAION-300M, Evaluation Protocol=Linear Probing2024.11 | 84.07 | — | — | — | |
| From-scratchModel=ViT-B/16, Pre-training Dataset=IN-1K, Evaluation Protocol=Linear Probing2024.11 | 81.76 | — | — | — | |
| Random initEvaluation Protocol=Fine-tuned2021.06 | 80.2 | — | — | — | |
| BAMprotocol=Linear Eval., arch.=ResNet-502024.08 | 80.2 | — | — | — | |
| R-PolyNets152Parameters=58.5M2023.03 | 79.3 | — | — | — | |
| D-PolyNets152Parameters=54.2M2023.03 | 79.1 | — | — | — | |
| PaCoBackbone=ResNet-502021.07 | 79.1 | — | — | — | |
| NNCLRprotocol=Linear Eval., arch.=ResNet-502024.08 | 79 | — | — | — | |
| R-PolyNets34Parameters=22.6M2023.03 | 78.8 | — | — | — | |
| D-PolyNets34Parameters=13.5M2023.03 | 78.7 | — | — | — | |
| BYOLEvaluation Protocol=Linear2021.06 | 78.4 | — | — | — | |
| BYOLprotocol=Linear Eval., arch.=ResNet-502024.08 | 78.4 | — | — | — | |
| Supervised-INEvaluation Protocol=Linear2021.06 | 78.3 | — | — | — | |
| multi-taskBackbone=ResNet-502021.07 | 78 | — | — | — | |
| CE(baseline)Backbone=ResNet-502021.07 | 77.9 | — | — | — | |
| SSL-HSIC (w/ target)Evaluation Protocol=Linear2021.06 | 77 | — | — | — | |
| ResNet152Parameters=58.3M2023.03 | 76.8 | — | — | — | |
| SupconBackbone=ResNet-502021.07 | 76.5 | — | — | — | |
| ResNet34Parameters=21.3M2023.03 | 76.2 | — | — | — | |
| SSL-HSIC (w/o target)Evaluation Protocol=Linear2021.06 | 75.9 | — | — | — | |
| SimCLREvaluation Protocol=Linear2021.06 | 71.6 | — | — | — | |
| SimCLRprotocol=Linear Eval., arch.=ResNet-502024.08 | 71.6 | — | — | — | |
| NNArchitecture=feed-forward NN, Layers=3, Hidden units per layer=10002017.08 | 32.9 | 50.2 | 58.6 | — | |
| ProjectionNet + FullyConnectedT (projection functions)=60, d (bits per projection)=12, Additional layers=FullyConnected (128 units)2017.08 | 20.8 | 35.6 | 44.8 | — | |
| ProjectionNetT (projection functions)=60, d (bits per projection)=122017.08 | 17.8 | 30.1 | 39.2 | — | |
| ProjectionNetT (projection functions)=10, d (bits per projection)=122017.08 | 12.8 | 25.2 | 33.8 | — | |
| CLR2022.05 | — | — | — | 75.9 | |
| ConvNet+Aug2022.05 | — | — | — | 84.7 | |
| Dynamics 12022.05 | — | — | — | 87.48 | |
| Dynamics 22022.05 | — | — | — | 87.42 | |
| PSO gBest2022.05 | — | — | — | 87.19 | |
| PyramidNet AMP2022.05 | — | — | — | 86.64 | |
| ResNet152-SAM2022.05 | — | — | — | 87.08 | |
| SENet2022.05 | — | — | — | 84.59 | |
| ViT-B/162022.05 | — | — | — | 87.13 |