Towards Understanding the Condensation of Neural Networks at Initial Training

Open in new window