Run distributed PyTorch job on multiple nodes in HTCondor

distributed
1

I am using HTCondor to submit my jobs to the computing cluster. I would like to know how to write the submit script so that my PyTorch job can be run on two nodes.