Closed Prasanna1991 closed 7 years ago
Yes this is just the KL divergence, and Wikipedia shows the general form for multivariate Gaussians, which can be simplified for the prior of N(0, 1). Unfortunately I'm not sure I'm able to give much more of an explanation, but perhaps the Tutorial on Variational Autoencoders will be helpful.
Closing (due to a month of inactivity).
Hello,
Can you explain the way you calculated KLLoss in VAE? Isn't this a KL Divergence?
Thanks.