Image Classification on ImageNet (val) (Original and R-CLR Accuracy)
76.43Original AccuracyTaylor Pruning
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Taylor PruningModel=ResNet-50 91%, Params ↓ %=11.4, FLOPs ↓ %=20.02021.05 | 76.43 | — | |
| Taylor PruningModel=ResNet-50 81%, Params ↓ %=30.1, FLOPs ↓ %=35.02021.05 | 75.48 | — | |
| Taylor PruningModel=ResNet-50 72%, Params ↓ %=44.5, FLOPs ↓ %=45.02021.05 | 74.5 | — | |
| Random Pruning + CLRModel=ResNet-50 72%, Params ↓ %=44.5, FLOPs ↓ %=45.02021.05 | — | 74.91 | |
| Random Pruning + CLRModel=ResNet-50 81%, Params ↓ %=30.1, FLOPs ↓ %=35.02021.05 | — | 75.54 | |
| Random Pruning + CLRModel=ResNet-50 91%, Params ↓ %=11.4, FLOPs ↓ %=20.02021.05 | — | 75.93 |