Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[New Feature] Slurm training script #213

Closed
rayleizhu opened this issue Feb 18, 2023 · 4 comments
Closed

[New Feature] Slurm training script #213

rayleizhu opened this issue Feb 18, 2023 · 4 comments
Assignees
Labels
enhancement New feature or request

Comments

@rayleizhu
Copy link
Contributor

How can I start training on a slurm cluster? Do I need to take care of ddp initialization routines by myself?

@FengLi-ust
Copy link
Collaborator

Yes, you can refer to DINO slurm scripts as an example. We will upload this script in the future.

@rayleizhu
Copy link
Contributor Author

rayleizhu commented Feb 22, 2023

I’ve implemented slurm launcher using submitit. It supports:

  • one-line launch command on slurm cluster, just like existing trainer provided by current detrex.
  • automatic experiment name generation according to hyper-parameter overrides.
  • automatically requeue and resume from latest checkpoint when a running job is preempted or reaches maximum execution time.

Are you interested in merging my implementation to detrex? If yes, I will create a pull request by the end of this week.

@FengLi-ust
Copy link
Collaborator

Of course, that would be great!

@rentainhe rentainhe added the enhancement New feature or request label Feb 22, 2023
@rentainhe rentainhe changed the title Slurm training script [New Feature] Slurm training script Feb 22, 2023
@rentainhe
Copy link
Collaborator

Thanks for your contribution, we will close this issue~ reopen this at anytime if you need more help~

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

3 participants