Introducing Distributed Data Parallel support on PyTorch Windows
Model training has been and will be in the foreseeable future one of the most frustrating things machine learning developers face. It takes quite a long time and people can’t really do anything about it.