research
∙
12/06/2021
Automap: Towards Ergonomic Automated Parallelism for ML Models
The rapid rise in demand for training large neural network architectures...
share
research
∙
12/02/2021
Memory-efficient array redistribution through portable collective communication
Modern large-scale deep learning workloads highlight the need for parall...
research
∙
11/15/2020
