Modifying dbal_pytorch.py to work with different-sized images? #125

[michelewang]

Hi, I'm hoping to modify the code in dbal_pytorch.py to work with images that are bigger than those in MNIST (i.e. bigger than 28x28). However, I keep running into errors because the tensors produced from my larger images do not match the dimensions of the CNN provided as an example. Do you know how to adjust that neural network's parameters to accommodate different image sizes (e.g. 1024x1024 or 512x512)?

For context, this is the current CNN being used:

Thank you so much!! This is an incredible library you're using and implementation of the seminal Gal paper, and I hope to use your work in my undergrad senior thesis! I am just having trouble getting your code to work with my data.