This is an improved implementation of the paper Auto-Encoding Variational Bayes by Kingma and Welling. It uses ReLUs and the adam optimizer, instead of sigmoids and adagrad. These changes make the network converge much faster.
pip install -r requirements.txt
python main.py