Fetching the paper…

Memorizing Gaussians with no over-parameterizaion via gradient decent on neural networks · Around