# VQ-VAE

A variational autoencoder that quantizes its latent space into a learned codebook of discrete tokens.

## What it is used for

Tokenizing images and audio for autoregressive and diffusion models.