Convert 3dcnn to pytorch 2dcnn

[ibro45 (Ibrahim Hadzic)]

Seems good! I would use batchnorm, just remember to set bias=False in the convolutional layer before it Any purpose to set bias=False in DenseNet torchvision?