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

feat(training): multi-worker training #164

Conversation

clement-bonnet
Copy link
Collaborator

@clement-bonnet clement-bonnet commented Jun 6, 2023

  • adapt training code to run on multiple workers (e.g. TPU v4-16 and more)
  • tested on CPU
  • tested on TPU v4-8
  • tested on TPU v4-16 (2 workers)

@clement-bonnet clement-bonnet added the enhancement New feature or request label Jun 6, 2023
@clement-bonnet clement-bonnet self-assigned this Jun 6, 2023
@clement-bonnet clement-bonnet marked this pull request as ready for review June 6, 2023 08:54
@clement-bonnet
Copy link
Collaborator Author

clement-bonnet commented Jun 6, 2023

CPU
image

TPU v4-8
image

TPU v4-16
image

TPU v4-16 is indeed twice as fast ("steps per second" metric) as TPU v4-8.

dluo96
dluo96 previously approved these changes Jun 6, 2023
Copy link
Contributor

@dluo96 dluo96 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Left a few comments, otherwise LGTM

Copy link
Contributor

@dluo96 dluo96 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! 🚀

Copy link
Contributor

@dluo96 dluo96 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! 🚀

@clement-bonnet clement-bonnet merged commit d22bb17 into instadeepai:main Jun 7, 2023
@clement-bonnet clement-bonnet deleted the 152-fixtraining-support-training-and-evaluation-on-multiple-tpu-workers branch June 7, 2023 08:48
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

Successfully merging this pull request may close these issues.

fix(training): support training and evaluation on multiple (tpu) workers
3 participants