This repository contains code to reproduce the figures in "Implicit regularization in AI meets generalized hardness of approximation in optimization – Sharp results for diagonal linear networks".
The following python packages are required: matplotlib, scipy, numpy, scikit_learn