Likelihood-Based Generative Models

VQ-VAE-2

Introduced by Razavi et al. in Generating Diverse High-Fidelity Images with VQ-VAE-2

VQ-VAE-2 is a type of variational autoencoder that combines a a two-level hierarchical VQ-VAE with a self-attention autoregressive model (PixelCNN) as a prior. The encoder and decoder architectures are kept simple and light-weight as in the original VQ-VAE, with the only difference that hierarchical multi-scale latent maps are used for increased resolution.

Source: Generating Diverse High-Fidelity Images with VQ-VAE-2

Papers


Paper Code Results Date Stars

Tasks


Task Papers Share
Image Generation 3 75.00%
Super-Resolution 1 25.00%

Components


Component Type
PixelCNN
Generative Models

Categories