Exploring ways to extract stable interpretations from neural networks. - View it on GitHub