How can I implement multi-GPU training #17
Description
HELP, How can I implement multi-GPU training? I have two A4000 GPUs, but after trying DataParallel (DP), it still doesn't work. It seems the data structure of PyTorch Geometric is not fully compatible with the standard DataParallel. Are there any other quick and simple ways to achieve multi-GPU training?
Additionally, how can I train with a simplified DESC dataset?