Image Classification on CIFAR10 (Accuracy)
99.1AccuracyDeiT-B
Evaluation Results
| Method | Links | |
|---|---|---|
| DeiT-BParams (M)=86.62022.02 | 99.1 | |
| MLCDPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 98.9 | |
| ViTAE-SParams (M)=23.62022.02 | 98.8 | |
| EfficientNet-L2Resolution=800px, Evaluation Protocol=Linear Probing2021.11 | 98.7 | |
| UNICOMPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 98.5 | |
| T2T-ViT-14Params (M)=21.52022.02 | 98.3 | |
| EfficientNet-B5Params (M)=302022.02 | 98.1 | |
| ViT-B/16Params (M)=86.52022.02 | 98.1 | |
| CLIP+Pre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 98.1 | |
| MMFTModel=ViT, Method=Ours2026.01 | 98.04 | |
| CLIPPre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 98 | |
| ViT-L/16Resolution=384px, Evaluation Protocol=Linear Probing2021.11 | 97.9 | |
| CLIP-ViT-L/14Resolution=336px, Evaluation Protocol=Linear Probing2021.11 | 97.9 | |
| ViT-L/16Params (M)=304.32022.02 | 97.9 | |
| OpenCLIPPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 97.9 | |
| BYOLEvaluation Protocol=Fine-tuned, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 97.8 | |
| SimCLR v1Evaluation Protocol=Fine-tuned, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 97.7 | |
| FNCEvaluation Protocol=Fine-tuned, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 97.7 | |
| FLYPModel=ViT, Method=FLYP2026.01 | 97.64 | |
| Florence-CoSwin-HResolution=384px, Backbone=CoSwin-H, Evaluation Protocol=Linear Probing2021.11 | 97.6 | |
| SimCLR v2Evaluation Protocol=Fine-tuned, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 97.5 | |
| FTModel=ViT, Method=FT2026.01 | 97.45 | |
| X-FM_baseLinear evaluation=true, Model size=Base, Patch size=16*16, Resolution=224*2242023.01 | 97.4 | |
| ViTAE-TParams (M)=4.82022.02 | 97.3 | |
| FLIPPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Zero-shot classification2024.07 | 97.2 | |
| SimCLRv2Backbone=ResNet-152x3, Evaluation Protocol=Linear Probing2021.11 | 96.8 | |
| ViC-MAEBackbone=ViT/L-16, Pre-training=K710+MiT+IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 96.8 | |
| MMFTModel=RN50, Method=Ours2026.01 | 96.64 | |
| CLIPLinear evaluation=true, Model size=Base, Patch size=16*16, Resolution=224*2242023.01 | 96.2 | |
| CLIP (reported)Pre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Zero-shot classification2024.07 | 96.2 | |
| CLIP-ViT-B/16Img-text pairs=400M, Evaluation protocol=Linear probing2022.09 | 96.2 | |
| OmniVLImg-text pairs=14M*, Evaluation protocol=Linear probing2022.09 | 96.2 | |
| FLYPModel=RN50, Method=FLYP2026.01 | 95.93 | |
| FTModel=RN50, Method=FT2026.01 | 95.79 | |
| CLIP-ViT-L/14Backbone=ViT-L/14, Resolution=336px, Evaluation Protocol=Zero-shot2021.11 | 95.7 | |
| FLIP-ViT-L/14Backbone=ViT-L/14, Evaluation Protocol=Zero-shot2021.11 | 95.7 | |
| ViC-MAEBackbone=ViT/L-16, Pre-training=IN1K+K400, Evaluation Protocol=Linear Evaluation2023.03 | 95.6 | |
| ALBEFImg-text pairs=14M, Evaluation protocol=Linear probing2022.09 | 95.6 | |
| B-12/Wi-1024 + DA + TTABlocks=12, Width=1024, Data Augmentation=true, Test-Time Augmentation=true, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 95.5 | |
| MLCDPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Zero-shot classification2024.07 | 95.3 | |
| CLIPPre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Zero-shot classification2024.07 | 95.2 | |
| BLIPImg-text pairs=14M, Evaluation protocol=Linear probing2022.09 | 95.2 | |
| MAEBackbone=ViT/L-16, Pre-training=IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 95 | |
| ViC-MAEBackbone=ViT/B-16, Pre-training=MiT, Evaluation Protocol=Linear Evaluation2023.03 | 94.92 | |
| OpenCLIPImage Encoder=ViT-B/16, Zero-shot=true2026.02 | 94.9 | |
| MAEBackbone=ViT/B-16, Pre-training=K400, Evaluation Protocol=Linear Evaluation2023.03 | 94.86 | |
| EuclideanEmbedding Space=Euclidean2021.07 | 94.81 | |
| C-HyperbolicEmbedding Space=Clipped Hyperbolic2021.07 | 94.76 | |
| Florence-CoSwin-HBackbone=CoSwin-H, Resolution=384px, Evaluation Protocol=Zero-shot2021.11 | 94.6 | |
| MAEBackbone=ViT/B-16, Pre-training=MiT, Evaluation Protocol=Linear Evaluation2023.03 | 94.47 | |
| B-12/Wi-1024 + DABlocks=12, Width=1024, Data Augmentation=true, Test-Time Augmentation=false, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 94.2 | |
| OmniMAEBackbone=ViT/L-16, Pre-training=SSv2+IN1K, Evaluation Protocol=Linear Evaluation2023.03 | 94.2 | |
| BLMBackbone=CLIP (ViT-B32)2024.10 | 94.2 | |
| CLIP-ResNet-50x64Evaluation Protocol=Linear Probing2021.11 | 94.1 | |
| One-to-one MappingBackbone=CLIP (ViT-B32), Mapping=One-to-one Mapping2024.10 | 94.1 | |
| BLM+Backbone=CLIP (ViT-B32)2024.10 | 94.1 | |
| OpenCLIPPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Zero-shot classification2024.07 | 94.1 | |
| DaVinciLinear evaluation=true, Model size=Base, Patch size=16*16, Resolution=224*2242023.01 | 94 | |
| NoneBackbone=CLIP (ViT-B32), Mapping=None2024.10 | 94 | |
| R-ExplaiNet26-64Parameter Size=0.89MP2024.10 | 93.8 | |
| ViC-MAEBackbone=ViT/B-16, Pre-training=K400, Evaluation Protocol=Linear Evaluation2023.03 | 93.64 | |
| DISSLEvaluation Protocol=Linear probe, Backbone=ResNet-50, Pre-training Dataset=ImageNet, Multi-crop Augmentation=2 x 160 + 4 x 962022.09 | 93.6 | |
| ResNet26-64Parameter Size=0.89MP2024.10 | 93.41 | |
| FNCEvaluation Protocol=Linear eval, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 93 | |
| FLAVALinear evaluation=true, Model size=Base, Patch size=16*16, Resolution=224*2242023.01 | 92.9 | |
| FLAVAImg-text pairs=70M, Evaluation protocol=Linear probing2022.09 | 92.9 | |
| FFF-ViT-B/32Pre-train dataset=Open70M, Backbone=ViT-B/32, Zero-shot=true2024.05 | 92.7 | |
| SimCLR v2Evaluation Protocol=Linear eval, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 92.4 | |
| FFF-ViT-B/32Pre-train dataset=Open30M, Backbone=ViT-B/32, Zero-shot=true2024.05 | 92.4 | |
| SwAVEvaluation Protocol=Linear probe, Backbone=ResNet-50, Pre-training Dataset=ImageNet, Multi-crop Augmentation=2 x 160 + 4 x 962022.09 | 92 | |
| METER-CLIP-B/16Img-text pairs=400M+4M, Evaluation protocol=Linear probing2022.09 | 91.8 | |
| ZSModel=MLLM, Method=ZS2026.01 | 91.56 | |
| B-6/Wi-1024 + DABlocks=6, Width=1024, Data Augmentation=true, Test-Time Augmentation=false, Pre-training=ImageNet21k, Evaluation Protocol=Fine-tuning2023.06 | 91.5 | |
| CLIP-MapbaseImage Encoder=ViT-39M/16, Zero-shot=true2026.02 | 91.4 | |
| BYOLEvaluation Protocol=Linear eval, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 91.3 | |
| DLMEEvaluation Protocol=linear-test2022.07 | 91.3 | |
| TinyCLIPImage Encoder=ViT-39M/16, Zero-shot=true2026.02 | 91.3 | |
| DLME-A1Evaluation Protocol=linear-test2022.07 | 91 | |
| UNICONNoise Rate (%)=90%2022.03 | 90.81 | |
| SimCLR v1Evaluation Protocol=Linear eval, Backbone=ResNet, Pre-training Dataset=ImageNet2020.11 | 90.6 | |
| PHResNet152Params=26.6M, Storage Memory=103MB, n=22021.10 | 90.54 | |
| PHResNet50Params=7.6M, Storage Memory=29MB, n=32021.10 | 90.423 | |
| FLAVAImg-text pairs=14M, Evaluation protocol=Linear probing2022.09 | 90.4 | |
| CLIP2022.01 | 90.3 | |
| ActiveDCSampling Ratio=2%2023.11 | 90.3 | |
| DLME-A2Evaluation Protocol=linear-test2022.07 | 90.2 | |
| ActiveFTSampling Ratio=2%2023.11 | 90.1 | |
| PHResNet152Params=17.8M, Storage Memory=70MB, n=32021.10 | 90.077 | |
| MAE2022.01 | 90 | |
| CLIP-PGS0.3Linear probing=true, Training epochs=102025.03 | 90 | |
| CLIP-PGS0.3Image Encoder=ViT-B/16, Evaluation Protocol=Linear Probing, Training Epochs=102025.03 | 90 | |
| PHResNet152Params=13.4M, Storage Memory=53MB, n=42021.10 | 89.897 | |
| Complex ResNet152Params=26.3M, Storage Memory=101MB2021.10 | 89.792 | |
| PHResNet50Params=11.1M, Storage Memory=43MB, n=22021.10 | 89.75 | |
| Complex ResNet50Params=11.1M, Storage Memory=43MB2021.10 | 89.699 | |
| Quaternion ResNet50Params=5.7M, Storage Memory=22MB2021.10 | 89.67 | |
| ResNet50Params=22.5M, Storage Memory=86MB2021.10 | 89.63 | |
| PHResNet18Params=3.6M, Storage Memory=13MB, n=32021.10 | 89.603 | |
| ViT2022.01 | 89.6 | |
| K-MeansSampling Ratio=2%2023.11 | 89.6 |