Non-vacuous Generalization Bounds for Deep Neural Networks without any modification to the trained models

Open in new window