Understanding VQ-VAE (DALL-E Explained Pt. 1)

By Charlie Snell

Understanding VQ-VAE (DALL-E Explained Pt. 1) Machine Learning at Berkeley Feb 09, 2021 228 7 25 Share By Charlie Snell Like everyone else in the ML community, we’ve been incredibly impressed by the results from OpenAI’s DALL-E . This model is able to generate precise, high quality images from a text description. It can even produce creative renderings of objects that likely don’t exist in the real world, like “an armchair in the shape of an avocado”. To celebrate the release of DALL-E, we’re publishing a series of blog posts explaining the key components of this model. This first blog post in

saved by

• Clem Yang

related reading

• Difference between AutoEncoder (AE) and Variational AutoEncoder (VAE) | Towards Data Sciencetowardsdatascience.com
• Generative modelling in latent space – Sander Dielemansander.ai
• Variational autoencoders.jeremyjordan.me
• From Autoencoder to Beta-VAE | Lil'Loglilianweng.github.io
• Jupyter Notebook Viewernbviewer.org
• Tutorial - What is a variational autoencoder? – Jaan Lı 李jaan.io
• Variational autoencoder - Wikipediaen.wikipedia.org
• Variational AutoEncoders (VAE) with PyTorch - Alexander Van de Kleutavandekleut.github.io
• machine learning - Is the output of a variational autoencoder meant to be a distribution that can be sampled, or a sample directly? - Cross Validatedstats.stackexchange.com
• Introduction to Neural Compression | Desi R. Ivanovadesirivanova.com
• Diffusion Models as a kind of VAE | Angus Turnerangusturner.github.io
• [1904.01277] A PCA-like Autoencoderarxiv.org