Skip to content

Instantly share code, notes, and snippets.

View anniezhi's full-sized avatar
:shipit:

Xiaoying Zhi anniezhi

:shipit:
4 followers · 11 following
  • Zurich
View GitHub Profile
@TengdaHan
TengdaHan / ddp_notes.md
Last active November 1, 2024 05:58
Multi-node-training on slurm with PyTorch

Multi-node-training on slurm with PyTorch

What's this?

  • A simple note for how to start multi-node-training on slurm scheduler with PyTorch.
  • Useful especially when scheduler is too busy that you cannot get multiple GPUs allocated, or you need more than 4 GPUs for a single job.
  • Requirement: Have to use PyTorch DistributedDataParallel(DDP) for this purpose.
  • Warning: might need to re-factor your own code.
  • Warning: might be secretly condemned by your colleagues because using too many GPUs.