Sarus implementation of classical ML models. The models are implemented using the Keras API of tensorflow 2. Vizualization are implemented and can be seen in tensorboard.

The required packages are managed with pipenv and can be installed using pipenv install. Please see the pipenv documentation for more information.

These models' implementations are intended to be easy to read and to adapt by making use of the latest Tensorflow 2 library and Keras API.

To install and train a model.

pipenv install
pipenv shell
python train.py

To visualize losses and reconstructions.

tensorboard --logdir ./logs/

• Simple Autoencoder
• Variational Autoencoder (VAE)
• Vector Quantized Autoencoder (VQ-VAE)
• PixelCNN
• Gated PixelCNN
• PixelCNN++
• Conditional Neural Processes
• PixelSNAIL