Fix replace_sampler missing the batch size under specific conditions

[carmocca]

What does this PR do?

Reported in Slack.

Fixes a bug where the batch_size was not properly reset under the following conditions:

• batch_size is provided to the DataLoader
• No batch_sampler is present initially
• The DataLoader signature passes the batch_size through **kwargs.

Does your PR introduce any breaking changes? If yes, please list them.

None

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?
• [n/a] 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 list all the breaking changes introduced by this pull request?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

• 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

[codecov]

Codecov Report

Merging #9367 (945cf12) into master (a79c351) will decrease coverage by 4%.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master   #9367    +/-   ##
=======================================
- Coverage      92%     88%    -4%     
=======================================
  Files         178     178            
  Lines       14899   14902     +3     
=======================================
- Hits        13752   13142   -610     
- Misses       1147    1760   +613     

[justusschock]

LGTM. Can you add a simple test with a dummy loader?

Something like

class DummyLoader(torch.utils.data.DataLoader):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

[SeanNaren]

LGTM. Can you add a simple test with a dummy loader?

Something like

class DummyLoader(torch.utils.data.DataLoader):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

May be mistaken, but the CustomDataLoader test already does this:

class CustomDataLoader(DataLoader):
        def __init__(self, num_features, dataset, *args, **kwargs):
            self.num_features = num_features
            super().__init__(dataset, *args, **kwargs)

[justusschock]

@SeanNaren my bad, overseen this

[tchaton]

LGTM !