Image Classification on ImageNet (test) (Accuracy and Hardware Consumption)
70Accuracy (%)BOLD
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BOLDTraining Modality=KD: ResNet50, Base (mapping dimension)=2562024.05 | 70 | 38.82 | 24.45 | |
| ResNet18Training Modality=FP Baseline, Precision=Full-precision2024.05 | 69.7 | 100 | 100 | |
| BNext-18Training Modality=KD: ResNet342024.05 | 68.4 | 45.48 | 37.51 | |
| BOLDTraining Modality=Large Models, Base (mapping dimension)=2562024.05 | 66.9 | 38.82 | 24.45 | |
| BOLD + BNTraining Modality=KD: ResNet34, Base (mapping dimension)=1922024.05 | 65.9 | 26.91 | 16.91 | |
| Melius-Net-29Training Modality=Large Models2024.05 | 65.8 | — | — | |
| ReactNet-ResNet18Training Modality=KD: ResNet342024.05 | 65.5 | 45.43 | — | |
| Real2BinaryTraining Modality=KD: ResNet342024.05 | 65.4 | — | — | |
| PokeBNN-ResNet18Training Modality=KD: ResNet502024.05 | 65.2 | — | — | |
| Bi-RealNet-152Training Modality=Large Models2024.05 | 64.5 | — | — | |
| Bi-RealNet-34Training Modality=FP Shortcut2024.05 | 62.2 | 80 | 58.24 | |
| Bi-RealNet-18Training Modality=FP Shortcut2024.05 | 56.4 | 46.6 | 32.99 | |
| BOLD + BNTraining Modality=FP Baseline, Base (mapping dimension)=642024.05 | 51.8 | 8.77 | 3.87 | |
| XNOR-NetTraining Modality=FP Baseline2024.05 | 51.2 | — | — | |
| BinaryNetTraining Modality=FP Baseline2024.05 | 42.2 | — | — |