SENN PyTorch implementation of Towards Robust Interpretability with Self-Explaining Neural Networks, Alvarez-Melis et al. 2018