Does timm support slurm training?

[wmkai]

I want to train my model on multi-machine. Is there any document for this? Hope to get help from you, thanks!

[rwightman]

@wmkai no documentation, it's not currently supported out of the box, but not that hard to add either, someone else had hacked together something and shared in the discussion or issues (I forget)...

I did a fairly minimal SLUM support in OpenCLIP that could hacked into the train script here ... https://github.com/mlfoundations/open_clip/blob/main/src/training/distributed.py ... I was thinking if importing that distributed setup fn and integrating but haven't had a chance...