Code for reproducing our experiments in: https://arxiv.org/pdf/1802.05701.pdf

1. Download the celebA dataset from here

OR

Download the Shoes dataset from here

OR

Download the Omniglot dataset from here

2. Install dependencies listed in req.txt

3. You will also need pyTorch which may be downloaded from here

4. Run this Jupyter notebook to get the data tensors for CelebA and move them into folder InData/CELEBA/

OR

Run this Jupyter notebook to get the data tensors xShoes.npy and yShoes.npy and move them in to folder InData/SHOES/

5. The code may be run from cmd line with various options detailed in the code