-
Notifications
You must be signed in to change notification settings - Fork 71
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.
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
Autoencoder implementation and training #79
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some nits
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good overall! The only bug I noticed was in the n layer discriminator and unlikely for us to hit it.
Most of my other comments are about where specific parameter settings came from. I'm sure for most of these you already thought about them and made a decision, so feel free to ignore my comments if they aren't important
b1bac49
to
7699ffd
Compare
This PR adds an implementation of the latent diffusion autoencoder, loss function, and training setup for training the autoencoder in composer.
diffusion/models/layers.py
diffusion/models/autoencoder.py
diffusion/callbacks/log_diffusion_images.py
diffusion/algorithms/discriminator_schedule.py
contains a composer algorithm for switching on the discriminator.diffusion/train.py
also contains some modifications to configure the param groups for training the autoencoder.