I've found that autoencoder and VAE predictions tend to converge. This might be a similar thing. If anyone has advice on preventing that let me know.. All I

I've found that autoencoder and VAE predictions tend to converge. This might be a similar thing. If anyone has advice on preventing that let me know.. All I can think of is add some random noise into the input or making it more GAN like... hey that's an idea.