MobileNetV3-PyTorch
MobileNetV3-PyTorch copied to clipboard
dropout
The paper mentions training with dropout of 0.8. How would you train with dropout with this model?