Open bodokaiser opened 7 years ago
If we use padding='same' for the first convolution layer with kernel_size=9 we get output shape [batch_size, 16, 16, 16, 1] as intended from the original paper.
padding='same'
kernel_size=9
[batch_size, 16, 16, 16, 1]
If we use
padding='same'
for the first convolution layer withkernel_size=9
we get output shape[batch_size, 16, 16, 16, 1]
as intended from the original paper.