Image Classification on ImageNet (val) (Accuracy and Calibration)
0.1984Top-1 ErrorInfoPro+MAN++
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| InfoPro+MAN++Network=ResNeXt-101, 32 x 8d, K=22025.07 | 0.1984 | 0.0501 | — | — | — | — | — | — | |
| InfoPro+MANNetwork=ResNeXt-101, 32 x 8d, K=22025.07 | 0.2011 | 0.0518 | — | — | — | — | — | — | |
| InfoPro+MAN++Network=ResNeXt-101, 32 x 8d, K=42025.07 | 0.2016 | 0.0517 | — | — | — | — | — | — | |
| InfoProNetwork=ResNeXt-101, 32 x 8d, K=22025.07 | 0.2035 | 0.0528 | — | — | — | — | — | — | |
| InfoPro+MANNetwork=ResNeXt-101, 32 x 8d, K=42025.07 | 0.2037 | 0.0534 | — | — | — | — | — | — | |
| E2ENetwork=ResNeXt-101, 32 x 8d2025.07 | 0.2064 | 0.054 | — | — | — | — | — | — | |
| PS-KDBackbone=ResNet-152, Augmentation=CutMix2020.06 | 0.2076 | 0.0534 | 0.8 | 0.54 | 58.25 | — | — | — | |
| LSBackbone=ResNet-152, Augmentation=CutMix2020.06 | 0.2077 | 0.0536 | 0.85 | 1.9 | 63.45 | — | — | — | |
| InfoPro+MAN++Network=ResNet-152, K=22025.07 | 0.2081 | 0.0541 | — | — | — | — | — | — | |
| CutMixBackbone=ResNet-1522020.06 | 0.2104 | 0.0556 | 0.81 | 2.19 | 58.43 | — | — | — | |
| CS-KDBackbone=ResNet-152, Augmentation=CutMix2020.06 | 0.2108 | 0.0553 | 0.83 | 1.56 | 59.02 | — | — | — | |
| ResNeXt-101Model Type=ResNeXt-1012020.06 | 0.212 | 0.056 | — | — | — | — | — | — | |
| InfoPro+MANNetwork=ResNet-152, K=22025.07 | 0.2123 | 0.0553 | — | — | — | — | — | — | |
| InfoPro+MAN++Network=ResNet-152, K=42025.07 | 0.2128 | 0.0572 | — | — | — | — | — | — | |
| PS-KDBackbone=ResNet-1522020.06 | 0.2141 | 0.0585 | 0.84 | 2.51 | 61.01 | — | — | — | |
| InfoProNetwork=ResNet-152, K=22025.07 | 0.2145 | 0.0584 | — | — | — | — | — | — | |
| InfoPro+MAN++Network=ResNet-101, K=22025.07 | 0.2154 | 0.0541 | — | — | — | — | — | — | |
| InfoPro+MANNetwork=ResNet-152, K=42025.07 | 0.2159 | 0.0589 | — | — | — | — | — | — | |
| E2ENetwork=ResNet-1522025.07 | 0.216 | 0.0592 | — | — | — | — | — | — | |
| CS-KDBackbone=ResNet-1522020.06 | 0.2161 | 0.0592 | 0.9 | 5.79 | 62.12 | — | — | — | |
| InfoPro+MANNetwork=ResNet-101, K=22025.07 | 0.2165 | 0.0549 | — | — | — | — | — | — | |
| InfoPro+MAN++Network=ResNet-101, K=42025.07 | 0.2166 | 0.0572 | — | — | — | — | — | — | |
| InfoProNetwork=ResNeXt-101, 32 x 8d, K=42025.07 | 0.2169 | 0.0611 | — | — | — | — | — | — | |
| LSBackbone=ResNet-1522020.06 | 0.2173 | 0.0585 | 0.92 | 3.91 | 68.24 | — | — | — | |
| InfoPro+MANNetwork=ResNet-101, K=42025.07 | 0.2173 | 0.0581 | — | — | — | — | — | — | |
| InfoProNetwork=ResNet-101, K=22025.07 | 0.2185 | 0.0589 | — | — | — | — | — | — | |
| TF-KDBackbone=ResNet-152, Augmentation=CutMix2020.06 | 0.22 | 0.0593 | 0.85 | 2.18 | 62.57 | — | — | — | |
| E2ENetwork=ResNet-1012025.07 | 0.2203 | 0.0593 | — | — | — | — | — | — | |
| DenseNet-264Model Type=DenseNet-2642020.06 | 0.2215 | 0.0612 | — | — | — | — | — | — | |
| ResNet-152Backbone=ResNet-1522020.06 | 0.2219 | 0.0619 | 0.88 | 3.84 | 61.79 | — | — | — | |
| TF-KDBackbone=ResNet-1522020.06 | 0.2276 | 0.0643 | 0.91 | 4.7 | 65.28 | — | — | — | |
| InfoProNetwork=ResNet-101, K=42025.07 | 0.2281 | 0.0654 | — | — | — | — | — | — | |
| InfoProNetwork=ResNet-152, K=42025.07 | 0.2293 | 0.0671 | — | — | — | — | — | — | |
| E2ENetwork=VGG-132025.07 | 0.2841 | 0.0963 | — | — | — | — | — | — | |
| DGL+MAN++Network=VGG-13, K=102025.07 | 0.3093 | 0.1024 | — | — | — | — | — | — | |
| DGL+MANNetwork=VGG-13, K=102025.07 | 0.3199 | 0.1084 | — | — | — | — | — | — | |
| MAN++Network=ViT-Tiny, K=42025.07 | 0.3543 | 0.1461 | — | — | — | — | — | — | |
| MAN++Network=ViT-Tiny, K=62025.07 | 0.3551 | 0.1464 | — | — | — | — | — | — | |
| E2ENetwork=ViT-Tiny2025.07 | 0.3559 | 0.1469 | — | — | — | — | — | — | |
| DGLNetwork=VGG-132025.07 | 0.356 | 0.142 | — | — | — | — | — | — | |
| MAN++Network=ViT-Small, K=42025.07 | 0.3693 | 0.1695 | — | — | — | — | — | — | |
| MAN++Network=ViT-Small, K=62025.07 | 0.3715 | 0.1719 | — | — | — | — | — | — | |
| E2ENetwork=ViT-Small2025.07 | 0.3738 | 0.1732 | — | — | — | — | — | — | |
| MAN++Network=ViT-Base, K=42025.07 | 0.389 | 0.1929 | — | — | — | — | — | — | |
| MAN++Network=ViT-Base, K=62025.07 | 0.3903 | 0.1952 | — | — | — | — | — | — | |
| E2ENetwork=ViT-Base2025.07 | 0.3918 | 0.1955 | — | — | — | — | — | — | |
| CaRotBackbone=RN502026.05 | — | — | — | 4.71 | — | — | — | 76.12 | |
| CaRotBackbone=ViT-B/162026.05 | — | — | — | 4.7 | — | — | — | 83.15 | |
| CaRotBackbone=ViT-L/142026.05 | — | — | — | 3.49 | — | — | — | 86.95 | |
| CLIPEvaluation Protocol=Zero-shot2026.06 | — | — | — | — | — | — | — | 66.7 | |
| CrossViT-18-DaggerAttention=QUEST, Epochs=3002026.03 | — | — | 0.78 | — | — | 82.8 | 96.1 | — | |
| CrossViT-9-DaggerAttention=Standard, Epochs=3002026.03 | — | — | 1.009 | — | — | 76.4 | 93.4 | — | |
| CrossViT-9-DaggerAttention=QUEST, Epochs=3002026.03 | — | — | 0.991 | — | — | 77 | 93.6 | — | |
| CrossViT-SmallAttention=Standard, Epochs=3002026.03 | — | — | 0.834 | — | — | 80.5 | 95.5 | — | |
| CrossViT-SmallAttention=QUEST, Epochs=3002026.03 | — | — | 0.826 | — | — | 80.9 | 95.6 | — | |
| FLYPBackbone=RN502026.05 | — | — | — | 5.16 | — | — | — | 76.16 | |
| FLYPBackbone=ViT-B/162026.05 | — | — | — | 6.4 | — | — | — | 82.72 | |
| FLYPBackbone=ViT-L/142026.05 | — | — | — | 7.29 | — | — | — | 86.19 | |
| GPUAEvaluation Protocol=Zero-shot2026.06 | — | — | — | — | — | — | — | 76.5 | |
| LP-FTBackbone=RN502026.05 | — | — | — | 10.42 | — | — | — | 76.25 | |
| LP-FTBackbone=ViT-B/162026.05 | — | — | — | 5.1 | — | — | — | 82.44 | |
| LP-FTBackbone=ViT-L/142026.05 | — | — | — | 10.56 | — | — | — | 84.74 | |
| TRACERBackbone=RN502026.05 | — | — | — | 4.7 | — | — | — | 76.48 | |
| TRACERBackbone=ViT-B/162026.05 | — | — | — | 4.5 | — | — | — | 82.76 | |
| TRACERBackbone=ViT-L/142026.05 | — | — | — | 5.07 | — | — | — | 86.27 | |
| ViT-2B/16Parameters=2.4B, Attention=Standard, Epochs=102026.03 | — | — | 6.153 | — | — | 3.1 | 9.4 | — | |
| ViT-2B/16Parameters=2.4B, Attention=QKNorm-DS, Epochs=102026.03 | — | — | 5.74 | — | — | 5.2 | 14.9 | — | |
| ViT-2B/16Parameters=2.4B, Attention=QUEST, Epochs=102026.03 | — | — | 5.011 | — | — | 11.9 | 27.8 | — | |
| ViT-H/14Parameters=600M, Attention=Standard, Epochs=1002026.03 | — | — | 1.545 | — | — | 68.3 | 88.8 | — | |
| ViT-H/14Parameters=600M, Attention=QUEST, Epochs=1002026.03 | — | — | 1.495 | — | — | 69.2 | 89.4 | — | |
| ViT-H/14Parameters=600M, Attention=Standard, Epochs=100 + 202026.03 | — | — | 1.042 | — | — | 76.2 | 93.3 | — | |
| ViT-H/14Parameters=600M, Attention=QUEST, Epochs=100 + 202026.03 | — | — | 1.002 | — | — | 76.8 | 93.6 | — |