research
∙
12/09/2022
Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints
Training large, deep neural networks to convergence can be prohibitively...
share
research
∙
10/07/2021
