A Variational Autoencoder implementation on MNIST images using convolutional layers.
Uses a training wrapper I created which can be found here.
Just clone both repos in your Desktop and run main.py.
The original MNIST dataset from pytorch was changed and a loss function was employed to be passed to the trainer wrapper. The dataset simply extends the __getitem__() method to return the image as the target.
The trainer's loss function field is agnostic that is why a method was used.
