Disentangled-Sequential-Autoencoder
Disentangled-Sequential-Autoencoder copied to clipboard
Is the time scale of f the correct in loss?
Hi,
I understand z has time index and f doesn't have time index. So I think that kl divergence of f does not match the time scale unless you multiply the length of time. Is it right?
https://github.com/yatindandi/Disentangled-Sequential-Autoencoder/blob/master/trainer.py#L26