Fetching the paper…

Implicit regularization in AI meets generalized hardness of approximation in optimization -- Sharp results for diagonal linear networks · Around