Image Classification on ImageNet 1k (Inference Latency & Efficiency)
86.1Top-1 AccuracyViT-L CLIP In12k In1k
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ViT-L CLIP In12k In1k2025.10 | 86.1 | — | — | — | — | — | — | — | — | |
| DeiT3-L In22k In1k2025.10 | 85.7 | — | — | — | — | — | — | — | — | |
| BEiTV2 FT In1k2025.10 | 85.5 | — | — | — | — | — | — | — | — | |
| ViT CLIP In12k In1k2025.10 | 85.4 | — | — | — | — | — | — | — | — | |
| BEiTV2 FT In21k2025.10 | 85.1 | — | — | — | — | — | — | — | — | |
| DeiT3 In21k In1k2025.10 | 85 | — | — | — | — | — | — | — | — | |
| EVA02-L FT In22k In1k2025.10 | 84.8 | — | — | — | — | — | — | — | — | |
| ViT CLIP In1k2025.10 | 84.7 | — | — | — | — | — | — | — | — | |
| EVA02 FT In1k2025.10 | 84.2 | — | — | — | — | — | — | — | — | |
| DeiT3 FB In22k In1k2025.10 | 83.8 | — | — | — | — | — | — | — | — | |
| ViT-L In21K In1k2025.10 | 83.6 | — | — | — | — | — | — | — | — | |
| MAE FT In1k2025.10 | 83.5 | — | — | — | — | — | — | — | — | |
| DeiT32025.10 | 83.5 | — | — | — | — | — | — | — | — | |
| T3CFamily=Vision Trf., Backbone=ViT-L/16, Budget=Max2026.01 | 83 | 4.36 | 5.55 | 49.8 | 62 | 79.3 | 100.1 | 145 | 0.23 | |
| DINOV22025.10 | 83 | — | — | — | — | — | — | — | — | |
| T3CFamily=Vision Trf., Backbone=ViT-L/16, Budget=Med2026.01 | 82.8 | 4.22 | 5.39 | 48 | 60.2 | 77.1 | 97.7 | 131 | 0.2 | |
| T3CFamily=Vision Trf., Backbone=ViT-L/16, Budget=Tiny2026.01 | 82.5 | 4.1 | 5.25 | 46.2 | 58 | 74.9 | 95.5 | 122 | 0.22 | |
| LR+FTFamily=Vision Trf., Backbone=ViT-L/162026.01 | 82.3 | 4.78 | 6.1 | 57.6 | 71 | 91.9 | 116.3 | 212 | 0.34 | |
| EVA02-S FT In22k In1k2025.10 | 82.2 | — | — | — | — | — | — | — | — | |
| EVA02 FT In21k In1k2025.10 | 82.2 | — | — | — | — | — | — | — | — | |
| MovePruneFamily=Vision Trf., Backbone=ViT-L/162026.01 | 82.1 | 4.66 | 5.98 | 55.5 | 68.3 | 89.2 | 112 | 195 | 0.36 | |
| EVA022025.10 | 82 | — | — | — | — | — | — | — | — | |
| T3CFamily=Vision Trf., Backbone=ViT-B/16, Budget=Max2026.01 | 81.7 | 2.45 | 3.1 | 27.8 | 35 | 44 | 56.2 | 71 | 0.19 | |
| T3CFamily=Vision Trf., Backbone=ViT-B/16, Budget=Med2026.01 | 81.5 | 2.38 | 3.04 | 26.5 | 33.4 | 42.6 | 54.8 | 64 | 0.16 | |
| T3CFamily=Vision Trf., Backbone=Swin-T, Budget=Max2026.01 | 81.5 | 1.95 | 2.38 | 22.6 | 28.4 | 34.9 | 45 | 73 | 0.19 | |
| MP-QATFamily=Vision Trf., Backbone=ViT-B/162026.01 | 81.3 | 2.58 | 3.32 | 33 | 41.7 | 52.9 | 67.1 | 134 | 0.25 | |
| T3CFamily=Vision Trf., Backbone=Swin-T, Budget=Med2026.01 | 81.3 | 1.82 | 2.26 | 21 | 26.6 | 33.1 | 42.7 | 66 | 0.16 | |
| T3CFamily=Vision Trf., Backbone=ViT-B/16, Budget=Tiny2026.01 | 81.2 | 2.3 | 2.92 | 26 | 33 | 41.8 | 53.9 | 59 | 0.18 | |
| MP-QATFamily=Vision Trf., Backbone=Swin-T2026.01 | 81.2 | 2.02 | 2.46 | 27.1 | 33.2 | 41.6 | 53.2 | 128 | 0.27 | |
| T3CFamily=Vision Trf., Backbone=Swin-T, Budget=Tiny2026.01 | 81.1 | 1.72 | 2.14 | 19.8 | 25.2 | 31.6 | 40.9 | 62 | 0.17 | |
| PTQ-8bFamily=Vision Trf., Backbone=ViT-B/162026.01 | 81 | 2.72 | 3.48 | 34.9 | 44.1 | 55.6 | 70.5 | 131 | 0.28 | |
| AutoMixBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.98 | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.89 | — | — | — | — | — | — | — | — | |
| PTQ-4bFamily=Vision Trf., Backbone=Swin-T2026.01 | 80.7 | 2.14 | 2.62 | 29.4 | 36.5 | 45.1 | 57.8 | 96 | 0.39 | |
| PuzzleMixBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.67 | — | — | — | — | — | — | — | — | |
| ManifoldMixBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.64 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.6 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.59 | — | — | — | — | — | — | — | — | |
| ResizeMix*Backbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.55 | — | — | — | — | — | — | — | — | |
| PuzzleMixBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.53 | — | — | — | — | — | — | — | — | |
| ResizeMix*Backbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.52 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.45 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.42 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.27 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 80.2 | — | — | — | — | — | — | — | — | |
| EVA02 FT In21k2025.10 | 80.2 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 80.06 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 79.98 | — | — | — | — | — | — | — | — | |
| ManifoldMixBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 79.93 | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.87 | — | — | — | — | — | — | — | — | |
| PuzzleMixBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.43 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.32 | — | — | — | — | — | — | — | — | |
| ResizeMix*Backbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.27 | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 79.25 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.09 | — | — | — | — | — | — | — | — | |
| ManifoldMixBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 79.02 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 78.97 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 78.96 | — | — | — | — | — | — | — | — | |
| VanillaBackbone=R-101, Training procedure=PyTorch 300 epochs2021.03 | 78.91 | — | — | — | — | — | — | — | — | |
| ResizeMix*Backbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.91 | — | — | — | — | — | — | — | — | |
| PuzzleMixBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.86 | — | — | — | — | — | — | — | — | |
| VanillaBackbone=RX-101, Training procedure=PyTorch 100 epochs2021.03 | 78.71 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.69 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.51 | — | — | — | — | — | — | — | — | |
| ViT In21K In1k2025.10 | 78.5 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.46 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.44 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-101, Budget=Max2026.01 | 78.4 | 1.8 | 2.22 | 20.1 | 25.4 | 33.4 | 42 | 70 | 0.18 | |
| ManifoldMixBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 78.21 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-101, Budget=Med2026.01 | 78.2 | 1.72 | 2.12 | 18.9 | 24 | 32.1 | 40.6 | 63 | 0.14 | |
| VanillaBackbone=R-101, Training procedure=PyTorch 100 epochs2021.03 | 78.18 | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.91 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-101, Budget=Tiny2026.01 | 77.8 | 1.64 | 2.05 | 17.8 | 22.6 | 30.8 | 39 | 58 | 0.16 | |
| LR+FTFamily=CNN, Backbone=ResNet-1012026.01 | 77.6 | 1.98 | 2.46 | 23.8 | 29.5 | 38.9 | 49.5 | 98 | 0.28 | |
| PuzzleMixBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.54 | — | — | — | — | — | — | — | — | |
| MagPruneFamily=CNN, Backbone=ResNet-1012026.01 | 77.5 | 1.92 | 2.4 | 22.9 | 28.3 | 37.5 | 48.1 | 84 | 0.31 | |
| ResizeMix*Backbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.42 | — | — | — | — | — | — | — | — | |
| VanillaBackbone=R-50, Training procedure=PyTorch 300 epochs2021.03 | 77.35 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.19 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.17 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.14 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.12 | — | — | — | — | — | — | — | — | |
| ViT2025.10 | 77.1 | — | — | — | — | — | — | — | — | |
| ManifoldMixBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 77.01 | — | — | — | — | — | — | — | — | |
| VanillaBackbone=R-50, Training procedure=PyTorch 100 epochs2021.03 | 76.83 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-50, Budget=Max2026.01 | 76.1 | 1.34 | 1.62 | 14.9 | 18.2 | 24.7 | 32.1 | 47 | 0.15 | |
| AutoMixBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 76.1 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-50, Budget=Med2026.01 | 76 | 1.26 | 1.58 | 13.8 | 17.3 | 23.6 | 31 | 42 | 0.12 | |
| QAT-8bFamily=CNN, Backbone=ResNet-502026.01 | 75.9 | 1.36 | 1.76 | 17.4 | 22.1 | 27.6 | 36 | 90 | 0.19 | |
| PuzzleMixBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.84 | — | — | — | — | — | — | — | — | |
| ViT-S In21K In1k2025.10 | 75.8 | — | — | — | — | — | — | — | — | |
| MixUpBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.73 | — | — | — | — | — | — | — | — | |
| PTQ-8bFamily=CNN, Backbone=ResNet-502026.01 | 75.7 | 1.44 | 1.84 | 18.5 | 23.8 | 29.2 | 38.9 | 88 | 0.22 | |
| ResizeMix*Backbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.64 | — | — | — | — | — | — | — | — | |
| T3CFamily=CNN, Backbone=ResNet-50, Budget=Tiny2026.01 | 75.6 | 1.18 | 1.45 | 13 | 16.5 | 22.4 | 29.3 | 38 | 0.14 | |
| ManifoldMixBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.44 | — | — | — | — | — | — | — | — | |
| VanillaBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.29 | — | — | — | — | — | — | — | — | |
| CutMixBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.16 | — | — | — | — | — | — | — | — | |
| FMix*Backbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.12 | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=R-34, Training procedure=PyTorch 300 epochs2021.03 | 75.01 | — | — | — | — | — | — | — | — |