Why output for 3-rd channel is unused in Logistic mixture?

NVlabs / NVAE

The Official PyTorch Implementation of "NVAE: A Deep Hierarchical Variational Autoencoder" (NeurIPS 2020 spotlight paper)

https://arxiv.org/abs/2007.03898

Other

999 stars 163 forks source link

Open univanxx opened 1 year ago

univanxx commented 1 year ago

Hello! Can anyone explain me one thing: when counting mean3 for 3-rd channel (blue, I suppose), why don't we use samples[:, 2, :, :, :]: https://github.com/NVlabs/NVAE/blob/9fc1a288fb831c87d93a4e2663bc30ccf9225b29/distributions.py#L139-L152

Also, why do we need to update means with samples when counting log prob? For example, in Tacotron-2 code the are no updates of means.

SerezD commented 7 months ago

Hi, did you find an answer to that? I have the same doubt