Efficient Compression of Overparameterized Deep Models through Low-Dimensional Learning Dynamics

Open in new window