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

Fix save/load/resume from checkpoint for DeepSpeed Plugin #8397

Merged
merged 75 commits into from Aug 2, 2021

Conversation

SeanNaren
Copy link
Contributor

@SeanNaren SeanNaren commented Jul 13, 2021

What does this PR do?

Closes #7282 #8344 #8092

Re-write of #7297

This PR will fix DeepSpeed saving/loading and resuming.

TLDR: DeepSpeed ZeRO Stage 3 was broken when saving/loading. This is now fixed. If the user wants to resume from checkpoint they need to set DeepSpeedPlugin(save_full_weights=False) when using ZeRO Stage 3. Documentation will be updated once PR is merged!

Before submitting

  • Was this discussed/approved via a GitHub issue? (not for typos and docs)
  • Did you read the contributor guideline, Pull Request section?
  • Did you make sure your PR does only one thing, instead of bundling different changes together?
  • Did you make sure to update the documentation with your changes? (if necessary)
  • Did you write any new necessary tests? (not for typos and docs)
  • Did you verify new and existing tests pass locally with your changes?
  • Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)
  • Did you list all the breaking changes introduced by this pull request?

PR review

Anyone in the community is free to review the PR once the tests have passed.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

  • Is this pull request ready for review? (if not, please submit in draft mode)
  • Check that all items from Before submitting are resolved
  • Make sure the title is self-explanatory and the description concisely explains the PR
  • Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Make sure you had fun coding 🙃

@SeanNaren SeanNaren modified the milestones: v1.4.x, v1.5 Aug 2, 2021
Sean Naren and others added 2 commits August 2, 2021 16:11
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
@mergify mergify bot added the ready PRs ready to be merged label Aug 2, 2021
@SeanNaren SeanNaren enabled auto-merge (squash) August 2, 2021 15:13
@SeanNaren SeanNaren disabled auto-merge August 2, 2021 15:13
@SeanNaren SeanNaren enabled auto-merge (squash) August 2, 2021 15:13
@kaushikb11
Copy link
Contributor

Disabled auto-merge if you plan to push few more changes. LGTM!

@SeanNaren SeanNaren enabled auto-merge (squash) August 2, 2021 15:55
@SeanNaren SeanNaren disabled auto-merge August 2, 2021 15:55
@SeanNaren SeanNaren enabled auto-merge (squash) August 2, 2021 15:57
@mergify mergify bot added the has conflicts label Aug 2, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working ready PRs ready to be merged
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Model restore fails from stored checkpoint when using Deepspeed
6 participants