Make monitor required arg of EarlyStopping callback

[rhjohnstone]

What does this PR do?

Makes monitor a required argument of EarlyStopping callback, as one of the deprecated behaviours mentioned in #10312.

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

Yes. Users who logged the key "early_stop_on" and relied on EarlyStopping to default to monitor that key will have to change their code to EarlyStopping(monitor="early_stop_on").

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 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

Anyone in the community is welcome to review the PR.
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 🙃

[awaelchli]

Great!

There are a few tests failing because the argument is now required. These tests need to be updated. Could you do it?
Here is the list:

FAILED tests/callbacks/test_early_stopping.py::test_pickling - TypeError: __i...
FAILED tests/callbacks/test_early_stopping.py::test_early_stopping_mode_options
FAILED tests/trainer/connectors/test_callback_connector.py::test_checkpoint_callbacks_are_last
FAILED tests/trainer/connectors/test_callback_connector.py::test_attach_model_callbacks
FAILED tests/trainer/connectors/test_callback_connector.py::test_attach_model_callbacks_override_info
FAILED tests/trainer/flags/test_fast_dev_run.py::test_callbacks_and_logger_not_called_with_fastdevrun[1]
FAILED tests/trainer/flags/test_fast_dev_run.py::test_callbacks_and_logger_not_called_with_fastdevrun[4]

If you want to check and run a test locally, you can do:

py.test -v tests/path/to/file.py::name_of_test

After that, we would need to include a new entry in our changelog in the "changed" section. Thanks for your help!

[rhjohnstone]

Thanks @awaelchli, I've fixed those tests. I now see that the docs (this page, I think) are failing because monitor is required. Unfortunately, I have no experience with the CLI (and hadn't ever seen those docs before!).

Is it as simple as changing

python ... \
    --trainer.callbacks=EarlyStopping \
    --trainer.callbacks.patience=5 \
    --trainer.callbacks=LearningRateMonitor \
    --trainer.callbacks.logging_interval=epoch

to

python ... \
    --trainer.callbacks=EarlyStopping \
    --trainer.callbacks.monitor=val_loss \
    --trainer.callbacks.patience=5 \
    --trainer.callbacks=LearningRateMonitor \
    --trainer.callbacks.logging_interval=epoch

etc.? I don't know if there's any special CLI behaviour when it comes to required vs optional args.

[awaelchli]

@rhjohnstone I updated the "breaking changes" description in the PR :)

[tchaton]

LGTM !

[rohitgr7]

small update to use it as kwarg