Closed nmichlo closed 2 years ago
https://github.com/google-research/disentanglement_lib/blob/86a644d4ed35c771560dc3360756363d35477357/disentanglement_lib/methods/shared/architectures.py#L148-L150
The Beta-VAE paper states that all the convolutional kernels are of size 4x4. I am not sure if it is intended that the 3rd and 4th convolutional layers in the conv_encoder function reduce these kernel sizes to 2x2?
conv_encoder
https://github.com/google-research/disentanglement_lib/blob/86a644d4ed35c771560dc3360756363d35477357/disentanglement_lib/methods/shared/architectures.py#L184-L201
An open issue already exists. Sorry. #23
https://github.com/google-research/disentanglement_lib/blob/86a644d4ed35c771560dc3360756363d35477357/disentanglement_lib/methods/shared/architectures.py#L148-L150
The Beta-VAE paper states that all the convolutional kernels are of size 4x4. I am not sure if it is intended that the 3rd and 4th convolutional layers in the
conv_encoder
function reduce these kernel sizes to 2x2?https://github.com/google-research/disentanglement_lib/blob/86a644d4ed35c771560dc3360756363d35477357/disentanglement_lib/methods/shared/architectures.py#L184-L201