使用PyTorch的实现
火车
通过更改train.py data_dir来修改imagenet路径
python train.py
Number of parameters: 23941296
Time per operator type:
1636.49 ms. 79.33%. Conv
247.179 ms. 11.9822%. Sigmoid
141.509 ms. 6.85977%. Mul
17.3817 ms. 0.842592%. Add
12.8334 ms. 0.622111%. FC
7.49557 ms. 0.363354%. ReduceMean
2062.88 ms in Total
FLOP per operato
