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

Simplify CLI args validation and ensure CLI args take precedence over config file. #2757

Merged
merged 4 commits into from
May 9, 2024

Conversation

Iain-S
Copy link
Contributor

@Iain-S Iain-S commented May 9, 2024

What does this PR do?

  • We simplify the processsing of config file sub-sections (e.g. fsdp, megatron, ipex, deepspeed, etc). Previously, any time a dict was encountered in a config, all dicts in the config were re-processed.
  • CLI options such as --megatron_some_option will no longer be overridden by config options (there was previously no if getattr(args, name, None) check before overwriting the arg.name). It makes things more consistent but note that it does change existing behaviour.

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?

Issues

Somewhat related to #2731.

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.

@muellerzr

Notes for Reviewers

I've made the logic in _validate_launch_command() more explicit as the proposed

    if args.config_file is not None or (os.path.isfile(default_config_file) and not args.cpu):

is equivalent to the previous

    if args.config_file is not None or os.path.isfile(default_config_file) and not args.cpu:

However, if what we really want is

    if (args.config_file is not None or os.path.isfile(default_config_file)) and not args.cpu:

then I'll happily edit but note that it would be changing the existing behaviour.

@HuggingFaceDocBuilderDev

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Copy link
Collaborator

@muellerzr muellerzr left a comment

Choose a reason for hiding this comment

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

Nice refactor and tests! Thanks for doing this 🤗

@muellerzr muellerzr merged commit 724824a into huggingface:main May 9, 2024
23 checks passed
@Iain-S Iain-S deleted the arg-validation branch May 10, 2024 08:48
@muellerzr
Copy link
Collaborator

FYI this broke FSDP as defaults for that are not always None. Making a follow-up

@Iain-S
Copy link
Contributor Author

Iain-S commented May 13, 2024

Let me know if I need to resubmit this PR. If you could suggest a test case to catch the problem you found, that would be great.

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.

3 participants