Add should_rank_save_checkpoint property to Training Plugins

[kaushikb11]

What does this PR do?

Fixes #7678
Related to link

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)

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 🙃

[codecov]

Codecov Report

Merging #7684 (54f331b) into master (8b01497) will decrease coverage by 5%.
The diff coverage is 94%.

@@           Coverage Diff           @@
##           master   #7684    +/-   ##
=======================================
- Coverage      93%     88%    -5%     
=======================================
  Files         199     199            
  Lines       12957   12960     +3     
=======================================
- Hits        11991   11371   -620     
- Misses        966    1589   +623     

[pep8speaks]

Hello @kaushikb11! Thanks for updating this PR.

• In the file pytorch_lightning/callbacks/model_checkpoint.py:

Line 722:13: W503 line break before binary operator

Comment last updated at 2021-05-24 21:11:39 UTC

[ananthsub]

thanks for following up on this @kaushikb11 ! I requested changes because of possible inconsistencies that emerge from @rank_zero_only decorator usage in the callback vs using trainer.should_rank_save_checkpoint

[ananthsub]

@SeanNaren @shuyingsunshine21 i think this will also help with FSDP saving sharded state

[carmocca]

Are you sure it's okay adding a convenience function in the trainer for it?

Do you expect/allow users to use/rely on it directly?

[kaushikb11]

Yeah, I don't expect it to be used by users, but for internal use.

[carmocca]

My point is that if it is for internal use, the property should be prepended by _.

It's ugly accessing an underscore attribute but we need a mechanism to avoid having to deprecate later.

[tchaton]

LGTM !