Size of the bottleneck layers

[sam-cms]

It seems that the bottleneck layers do not match the actual Paper where the number of filters are supposed to be 1024. making the filters across the Model look like this :
filters = [64, 128, 256, 512, 1024] which is obviously not the case in here.
Was there a case specific reason behind your choice ?
Because eventually, it seems that the number of the constitutional layers does not match the 23 layers mentioned in the original Paper.

[Hsankesara]

Hey @med-cms,

Sorry for the late response. As you've pointed, the number of filters is 512 in the bottleneck layer. It is my bad that I didn't write it more clearly and thank you for pointing it out.

self.bottleneck = torch.nn.Sequential(
                            torch.nn.Conv2d(kernel_size=3, in_channels=256, out_channels=512),
                            torch.nn.ReLU(),
                            torch.nn.BatchNorm2d(512),
                            torch.nn.Conv2d(kernel_size=3, in_channels=512, out_channels=512),
                            torch.nn.ReLU(),
                            torch.nn.BatchNorm2d(512),
                            torch.nn.ConvTranspose2d(in_channels=512, out_channels=256, kernel_size=3, stride=2, padding=1, output_padding=1)
                            )

I have decreased the size mostly because of the memory reasons and it was slowing it down. I did not follow the proper exactly but followed the concept mentioned in the paper and tried to show some results using it. Thanks for letting me know and let me know of you have any question.