Skip scheduler.step() if optimizer.step() isn't called in the iteration

[akihironitta]

What does this PR do?

Fixes #5558

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

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

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 🙃

cc @carmocca @justusschock @awaelchli @akihironitta

[akihironitta]

When the warning is raised

This issue only happens when lr_scheduler.step() runs every small steps, i.e.

def configure_optimizers(self):
    optimizer = ...
    scheduler = {
        "scheduler": ...,
        "interval": "step",
        "frequency": 1,  # another small number may also cause this issue.
    }
    return {"optimizer": optimizer, "lr_scheduler": scheduler}

Cause of the warning
As documented in the pytorch source code torch/cuda/amp/grad_scaler.py, scaler.step(optimizer) (which is called when using native amp) is likely to skip optimizer.step() for the first few iterations, and thus, it makes lr_scheduler.step() called before any call of optimizer.step().

EDIT (2021-10-28): native amp skips optimizer.step() not only for the first few iterations but also during training sometimes, so we need to skip lr_scheduler.step() whenever optimizer.step() isn't called by the scaler.

pytorch/pytorch#44511
https://discuss.pytorch.org/t/optimizer-step-before-lr-scheduler-step-error-using-gradscaler/92930

[akihironitta]

still wip