Giters

ACEsuit / mace

MACE - Fast and accurate machine learning interatomic potentials with higher order equivariant message passing.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

repulsion branch DistributedEnvironment exception produces no output

bernstei opened this issue · comments

commented

I think the logger need to be set up before the calls that create DistributedEnvironment in cli/run_train.py, otherwise there's no output. Also, since it's an error, it should probably go to logger.error rather than .info, and return a non-zero status when the script terminates.

commented

I see now that the logger depends on rank, which isn't known until after the distributed env is created, and I'm not sure what's the simplest way of dealing with that. Maybe just setting the logging level very early, and setting the full logger later. But I still think the DistributedEnvironment exception message should be an error, no?