Hi,

I'm trying to run this example with spark and pytorch but it does not seem to actually run in parrallel. I.e. I am on a GCP cluster and the training runs on the master node in parrallel but it does not seem to run in parrallel on the worker nodes as well. However, if I do this other example where the data is local and I just use pytorch, it does get distributed like I'd like it to. Should both run in parrallel across all the nodes?