TY  - RPRT
TI  - How many degrees of freedom do we need to train deep networks: a loss landscape perspective
AU  - Brett W. Larsen
AU  - Stanislav Fort
AU  - Nic Becker
AU  - Surya Ganguli
PY  - 2022
UR  - https://arxiv.org/abs/2107.05802
ID  - 2107.05802
ER  -