In Machine-Learning-Collection/ML/Pytorch/CNN_architectures
/pytorch_efficientnet.py,
I believe the correct implementation of channel expansion in the EfficientNet MBBlock should use a kernel size of 1.
if self.expand: self.expand_conv = CNNBlock( in_channels, hidden_dim, kernel_size=3, stride=1, padding=1, )
In Machine-Learning-Collection/ML/Pytorch/CNN_architectures
/pytorch_efficientnet.py,
I believe the correct implementation of channel expansion in the EfficientNet MBBlock should use a kernel size of 1.
if self.expand: self.expand_conv = CNNBlock( in_channels, hidden_dim, kernel_size=3, stride=1, padding=1, )