Image Classification on ImageNet 1k (val) (DeiT Model Results)
79.85Top-1 Acc (DeiT-S)Pretrained
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PretrainedW/A=32/322022.11 | 79.85 | 81.8 | 83.11 | |
| NoisyQuant-PTQ4ViTW/A=8/82022.11 | 79.51 | 81.45 | 82.49 | |
| PTQ4ViTW/A=8/82022.11 | 79.47 | 81.48 | 82.97 | |
| FQ-ViTW/A=8/82022.11 | 79.17 | 81.2 | — | |
| NoisyQuant-LinearW/A=8/82022.11 | 79.11 | 81.3 | 82.23 | |
| EasyQuantW/A=8/82022.11 | 78.98 | 81.19 | 82.1 | |
| Liu et al.W/A=8 MP2022.11 | 78.09 | 81.29 | — | |
| PTQ4ViT-LinearW/A=8/82022.11 | 77.65 | 80.94 | 82.33 | |
| Liu et al.W/A=8/82022.11 | 77.47 | 80.48 | — | |
| NoisyQuant-PTQ4ViTW/A=6/62022.11 | 77.43 | 80.7 | 81.65 | |
| Bit-SplitW/A=8/82022.11 | 76.39 | 79.42 | — | |
| NoisyQuant-LinearW/A=6/62022.11 | 76.37 | 79.77 | 81.4 | |
| PTQ4ViTW/A=6/62022.11 | 76.28 | 80.25 | 81.55 | |
| EasyQuantW/A=6/62022.11 | 75.27 | 79.47 | 81.26 | |
| Liu et al.W/A=6 MP2022.11 | 75.1 | 77.47 | — | |
| Bit-SplitW/A=6/62022.11 | 74.58 | 76.39 | — | |
| Liu et al.W/A=6/62022.11 | 74.58 | 77.02 | — | |
| PercentileW/A=8/82022.11 | 73.98 | 75.21 | 80.02 | |
| PTQ4ViT-LinearW/A=6/62022.11 | 72.26 | 78.78 | 68.44 | |
| PercentileW/A=6/62022.11 | 70.49 | 73.99 | 78.24 |