Skip to content

Implementation of a straight-through gradient wrapper to allow for discrete latent representations. Provides binary discretizer which maps hidden representations to {0, 1} and a learnable multi-value discretizer, which maps hidden activations to their closest value in a set of given size.

Notifications You must be signed in to change notification settings

LGirrbach/Discrete-Latent-Representation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

Discrete-Latent-Representation

Implementation of a straight-through gradient wrapper to allow for discrete latent representations. Provides binary discretizer which maps hidden representations to {0, 1} and a learnable multi-value discretizer, which maps hidden activations to their closest value in a set of given size.

About

Implementation of a straight-through gradient wrapper to allow for discrete latent representations. Provides binary discretizer which maps hidden representations to {0, 1} and a learnable multi-value discretizer, which maps hidden activations to their closest value in a set of given size.

Topics

Resources

Stars

Watchers

Forks