Auto Parallel

requirement

CUDA 11.8, DGL 1.1.2, Pytorch 2.0.1, conda create -n ap python=3.9 #create conda virtual enviroment python 3.9 conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia #conda install pytorch-gpu CUDA 11.8 conda install -c dglteam/label/cu118 dgl #conda install dgl CUDA 11.8

Reuse my conda virtual python enviroment

source .bashrc conda activate ap2

Train on single machine

bash ./train_configs.sh

Train on multi-machines

bash ./exp_multinodes.sh 0 # on machine 0 bash ./exp_multinodes.sh 1 # on machine 0