EfficientNet-PyTorch
EfficientNet-PyTorch copied to clipboard
initialization
I don't see the initialization weights in similar tf code, is it not needed?
Same question here
How should we initialize, if I were to train from scratch?