Source code for the paper: "Adversarial examples within the training distribution: A widespread challenge" - View it on GitHub