New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[WIP] Bits per dim for generative models #144

Open

kashif opened this issue Jun 8, 2019 · 1 comment

Collaborator

kashif commented Jun 8, 2019 •

edited

Model	MNIST	Fashion MNIST
NICE	4.36	-
MADE	2.04	-
IAF-VAE	-	-
MAF	1.89	-
Real NVP	1.06	-
Glow	1.05	-
FFJORD	0.99	-
i-ResNET	1.05	-
Residual Flow	0.97	-
Flow++	-	-
Convolutional DRAW	-	-
PixelRNN	-	-
PixelCNN	-	-
Gated PixelCNN	-	-
Image Transformer	-	-
PixelSNAIL	-	-
MintNet	0.98	-
SOS	1.81	-
NAF	-	-
TAN	-	-

kashif mentioned this issue

Replicating the bits / dim results in the original VQ-VAE paper zalandoresearch/pytorch-vq-vae#1

Open

gitlabspy commented Feb 22, 2021

Hi, may I ask what is the correct way of calculating bit/dim ( for different n-bits images) from NLL ?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment