Network Pruning
Network Pruning is a popular approach to reduce a heavy network to obtain a light-weight form by removing redundancy in the heavy network. In this approach, a complex over-parameterized network is first trained, then pruned based on come criterions, and finally fine-tuned to achieve comparable performance with reduced parameters.
Source: Ensemble Knowledge Distillation for Learning Improved and Efficient Networks
Papers
Showing 1–10 of 534 papers
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | ResNet50-2.3 GFLOPs | Accuracy | 78.79 | — | Unverified |
| 2 | ResNet50-1.5 GFLOPs | Accuracy | 78.07 | — | Unverified |
| 3 | ResNet50 2.5 GFLOPS | Accuracy | 78 | — | Unverified |
| 4 | RegX-1.6G | Accuracy | 77.97 | — | Unverified |
| 5 | ResNet50 2.0 GFLOPS | Accuracy | 77.7 | — | Unverified |
| 6 | ResNet50-3G FLOPs | Accuracy | 77.1 | — | Unverified |
| 7 | ResNet50-2G FLOPs | Accuracy | 76.4 | — | Unverified |
| 8 | ResNet50-1G FLOPs | Accuracy | 76.38 | — | Unverified |
| 9 | TAS-pruned ResNet-50 | Accuracy | 76.2 | — | Unverified |
| 10 | ResNet50 | Accuracy | 75.59 | — | Unverified |