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

Improve support for custom DataLoaders when instantiated in *_dataloader hook #12981

Merged
merged 41 commits into from
Jun 21, 2022

Conversation

otaj
Copy link
Contributor

@otaj otaj commented May 5, 2022

What does this PR do?

Fixes #12917
Improve support for custom DataLoaders when instantiated in *_dataloader hook. If the DataLoader is a valid DataLoader, there is now a lower chance that things we will break the DataLoader when injecting custom arguments.

Also, if things break due to our injecting, we throw better error messages.

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

DataLoader instantiated inside a hook will now not instantiate attributes for you automatically. Before this PR, this code would work:

class MyDataLoader(DataLoader):
    def __init__(dataset, my_arg, **kwargs):
        # we explicitly do not set my_arg, as PL sets it for us automatically
        super().__init__(dataset, **kwargs)

class MyModule(pl.LightningModule):
    ...
    def train_dataloader(self):
        loader = MyDataLoader(range(10), 10, batch_size=2)
        assert loader.my_arg == 10
        return loader

After this PR, this code would no longer work. This was an internal behavior.

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 minor internal changes/refactors)

PR review

Anyone in the community is welcome to review the PR.
Before you start reviewing, make sure you have read the 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 🙃

cc @Borda @justusschock @awaelchli @ninginthecloud @rohitgr7 @otaj

@otaj otaj added bug Something isn't working data handling Generic data-related topic trainer: predict labels May 5, 2022
@otaj otaj added this to the 1.6.x milestone May 5, 2022
@otaj otaj requested review from Borda, tchaton and SeanNaren as code owners May 5, 2022 12:31
@otaj otaj self-assigned this May 5, 2022
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
tests/utilities/test_data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
otaj and others added 5 commits May 6, 2022 09:31
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
@otaj
Copy link
Contributor Author

otaj commented May 6, 2022

I think I replied to all of your comments @carmocca. I applied most of your suggestions, would you mind taking another look? Thanks!

pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
pytorch_lightning/utilities/data.py Outdated Show resolved Hide resolved
@mergify mergify bot added the has conflicts label May 6, 2022
otaj and others added 3 commits May 6, 2022 12:55
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
@mergify mergify bot removed the has conflicts label Jun 20, 2022
@mergify mergify bot added ready PRs ready to be merged has conflicts and removed ready PRs ready to be merged labels Jun 21, 2022
@mergify mergify bot added ready PRs ready to be merged has conflicts and removed has conflicts ready PRs ready to be merged labels Jun 21, 2022
@mergify mergify bot added ready PRs ready to be merged and removed has conflicts ready PRs ready to be merged labels Jun 21, 2022
@carmocca carmocca merged commit 2e9cd72 into master Jun 21, 2022
@carmocca carmocca deleted the bugfix/dataloader_wrapper branch June 21, 2022 23:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working data handling Generic data-related topic ready PRs ready to be merged
Projects
None yet
Development

Successfully merging this pull request may close these issues.

MisconfigurationException: Trying to inject DistributedSampler into the AnnLoader instance
5 participants