Image Classification on CIFAR100 (Top-1 Accuracy)
86.5Top-1 AccuracyViC-MAE
Evaluation results
| Method | Links | |
|---|---|---|
| ViC-MAEBackbone=ViT/L-16, Pre-training=K710+MiT+IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 86.5 | |
| ViC-MAEBackbone=ViT/L-16, Pre-training=IN1K+K400, Evaluation Protocol=Linear Evaluation2023.03 | 85.4 | |
| MAEBackbone=ViT/L-16, Pre-training=IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 82.9 | |
| B-12/Wi-1024 + DA + TTABlocks=12, Width=1024, Data Augmentation=true, Test-Time Augmentation=true, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 82.6 | |
| OmniMAEBackbone=ViT/L-16, Pre-training=SSv2+IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 82.2 | |
| SCEEvaluation protocol=Linear classifier2021.11 | 80.4 | |
| B-12/Wi-1024 + DABlocks=12, Width=1024, Data Augmentation=true, Test-Time Augmentation=false, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 80 | |
| ViC-MAEBackbone=ViT/B-16, Pre-training=MiT, Evaluation Protocol=Linear Evaluation2023.03 | 79.88 | |
| MAEBackbone=ViT/B-16, Pre-training=MiT, Evaluation Protocol=Linear Evaluation2023.03 | 79.5 | |
| MAEBackbone=ViT/B-16, Pre-training=K400, Evaluation Protocol=Linear Evaluation2023.03 | 79.49 | |
| NNCLREvaluation protocol=Linear classifier2021.11 | 79 | |
| ViC-MAEBackbone=ViT/B-16, Pre-training=K400, Evaluation Protocol=Linear Evaluation2023.03 | 78.8 | |
| BYOLEvaluation protocol=Linear classifier2021.11 | 78.4 | |
| SupervisedEvaluation protocol=Linear classifier2021.11 | 78.3 | |
| xCLIPLinear probing=true, Backbone=ViT-B/16, Pre-train=IT35M2022.10 | 77.8 | |
| B-6/Wi-1024 + DABlocks=6, Width=1024, Data Augmentation=true, Test-Time Augmentation=false, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 76.4 | |
| nCLIPLinear probing=true, Backbone=ViT-B/16, Pre-train=IT35M2022.10 | 76.3 | |
| UncompressedModel=Wide-ResNet, M.S. (MB)=139.42021.04 | 76.2 | |
| CLIPLinear probing=true, Backbone=ViT-B/16, Pre-train=IT35M2022.10 | 75.6 | |
| DIFFQModel=Wide-ResNet, M.S. (MB)=4.72021.04 | 75.6 | |
| SimCLREvaluation protocol=Linear classifier2021.11 | 74.4 | |
| CLIP-PGS0.3Linear probing=true, Training epochs=102025.03 | 72.3 | |
| CLIP-PGS0.3Image Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 72.3 | |
| CLIP-PGS0.5Linear probing=true, Training epochs=102025.03 | 70.3 | |
| CLIP-PGS0.5Image Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 70.3 | |
| E-CLIPLinear probing=true, Training epochs=102025.03 | 69.7 | |
| E-CLIPImage Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 69.7 | |
| CLIP-PGS0.3Image Encoder=ViT-S/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 68.7 | |
| CLIP-PGS0.5Image Encoder=ViT-S/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 68.6 | |
| CLIPLinear probing=true, Training epochs=102025.03 | 67.4 | |
| CLIPImage Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 67.4 | |
| A-CLIPLinear probing=true, Training epochs=102025.03 | 66.1 | |
| A-CLIPImage Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 66.1 | |
| FLIPLinear probing=true, Training epochs=102025.03 | 65.5 | |
| FLIPImage Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 65.5 | |
| OURS_SepProtocol=Linear probing2023.03 | 61.19 | |
| FNG-CEBackbone=ResNet-182026.03 | 60.04 | |
| CEBackbone=ResNet-182026.03 | 59.72 | |
| OURS_GCProtocol=Linear probing2023.03 | 58.89 | |
| CE+L2Backbone=ResNet-182026.03 | 57.95 | |
| OURS_BrProtocol=Linear probing2023.03 | 57.72 | |
| CLIPProtocol=Linear probing2023.03 | 56.36 | |
| CyCLIPProtocol=Linear probing2023.03 | 54.23 | |
| DiffQuantModel=Wide-ResNet, M.S. (MB)=7.92021.04 | 49.3 | |
| B-6/Wi-1024Blocks=6, Width=1024, Data Augmentation=false, Test-Time Augmentation=false, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 43 |