Make a VAE?

[dsteinberg]

This may not be the best framework for it, but it would be a good exercise to see how it looks compared to competing frameworks, e.g. Keras, Edward.

[ltiao]

You may find these relevant and hopefully useful :)

• Guide to Variational Autoencoders with a Concise Keras Implementation
• Inference in Variational Autoencoders with Different Monte Carlo Sample Sizes
• Convolutional VAE in Keras (Notebook)
• Convolutional VAE in Keras on the Fashion MNIST Dataset (Notebook)

Also, the white paper on TensorFlow Distributions contains many neat examples which demonstrate how easy it is to go from the most basic VAE to the state-of-the-art architecture that uses a PixelCNN++ decoder and inverse autoregressive flows for the encoder and prior. Might be good for comparison or just a source of inspiration :)

[dsteinberg]

Thanks man! Very useful!!