Training DNN in O(1) Memory Using Random Matrices

2 points| tnch | 1 year ago |arxiv.org