Playing around with CAEs on mnist to better understand conditioning effects
The goal was to visualize the difference between adding conditional information only in the VAE bottleneck or as Input as well.
Code mostly reused from: https://github.com/lyeoni/pytorch-mnist-VAE