Image Classification on ImageNet-1k (val) (Top-1 Accuracy, MACs, Parameters, and Accuracy Retention)
0.94MACs (G)VBP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VBPBackbone=DeiT-T, Pruning Rate=45%2025.07 | 0.94 | 4.12 | 49.77 | 70.08 | |
| DeiT-TPruning Status=Full baseline2025.07 | 1.26 | 5.72 | — | 72.02 | |
| VBPBackbone=DeiT-S, Pruning Rate=50%2025.07 | 3.21 | 14.96 | 64.44 | 78.62 | |
| VBPBackbone=Swin-T, Pruning Rate=45%2025.07 | 3.23 | 21.31 | 55.12 | 79.41 | |
| Swin-TPruning Status=Full baseline2025.07 | 4.5 | 28.29 | — | 80.91 | |
| DeiT-SPruning Status=Full baseline2025.07 | 4.61 | 22.05 | — | 79.7 | |
| VBPBackbone=Swin-S, Pruning Rate=50%2025.07 | 5.63 | 35.02 | 67.01 | 81.86 | |
| Swin-SPruning Status=Full baseline2025.07 | 8.76 | 49.61 | — | 83.04 | |
| VBPBackbone=Swin-B, Pruning Rate=55%2025.07 | 10.22 | 56.29 | 62.61 | 83.61 | |
| VBPBackbone=DeiT-B, Pruning Rate=55%2025.07 | 11.44 | 55.4 | 57.58 | 80.67 | |
| VBPBackbone=Swin-B, Pruning Rate=20%2025.07 | 13.58 | 76.42 | 83.9 | 84.67 | |
| VBPBackbone=DeiT-B, Pruning Rate=20%2025.07 | 15.35 | 75.24 | 80.87 | 81.76 | |
| Swin-BPruning Status=Full baseline2025.07 | 15.46 | 87.77 | — | 84.71 | |
| DeiT-BPruning Status=Full baseline2025.07 | 17.58 | 86.57 | — | 81.73 |