ResNet re-implementation using PyTorch
- Configure imagenet path by changing data_dir in main.py
- bash ./main.sh $ --train for training model, $ is number of GPUs
- python main.py --test for testing
| Version | Top@1 | Top@5 | Params | FLOPS | Pretrained weights |
|---|---|---|---|---|---|
| Resnet 18 | 71.5 | 89.9 | 11.7M | 1.8G | model |
| Resnet 34 | 76.3 | 92.8 | 21.8M | 3.6G | model |
| Resnet 50 | 80.1 | 94.5 | 25.5M | 4.1G | model |
| Resnet 101 | 81.2 | 95.1 | 44.5M | 7.8G | model |
| Resnet 152 | 81.7 | 95.1 | 60.1M | 11.6G | model |