Skip to content

Conversation

@excepshenal
Copy link

What does this PR do?

Addresses #3897 :

  1. FSDP2: Activation-checkpoint modules directly matching auto-wrap policy, not those whose parent module matches policy (introduced by [FSDP2] Refactor + FP8 #3585 , not sure about the context of why)
  2. Nit: replace duplicate fsdp2_prepare_auto_wrap_policy with FullyShardedDataParallelPlugin.set_auto_wrap_policy, and delete extra unnecessary calls to set_auto_wrap_policy

Confirmed in test runs of llama-70b training that this fixes AC to checkpoint modules matching auto-wrap policy rather than those whose parent module matches the policy. Compare before:

image

to after:

image

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you read the contributor guideline,
    Pull Request section?
  • Was this discussed/approved via a Github issue or the forum? Please add a link
    to it if that's the case.
  • Did you make sure to update the documentation with your changes? Here are the
    documentation guidelines, and
    here are tips on formatting docstrings.
  • Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant