🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/113659

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 4cfb2d4 with merge base 9b0f2f8 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@pytorchbot label "module: typing"

@pytorchbot merge

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Merge failed

Reason: 1 mandatory check(s) failed. The first few are:

• Lint / lintrunner / linux-job

Dig deeper by viewing the failures on hud

Looks like this fix triggers

Error: Signature of "step" incompatible with supertype "LRScheduler"

It seems that a bunch of the LRScheduler subclasses simply violate LSP.

• Base class: def step(self, epoch=None)
• SequentialLR: def step(self) (missing epoch argument!)
• ChainedScheduler: def step(self) (missing epoch argument!)
• ReduceLROnPlateau: def step(self, metrics, epoch=None) extra mandatory metrics argument.

For SequentialLR and ChainedScheduler, it would make sense to simply pass the epoch argument through. (better open a separate issue?)

For ReduceLROnPlateau, I'd suggest adding a type: ignore[override], I don't really see a better solution right now.

Alternatively, one could set a default value like metrics=float('nan').

type: ignore as necessary, IMO

Agreed with Ed for ReduceLROnPlateau, and yes for propagating epoch for the others.

I added the type-ignore here and a put some educated guesses on how to pass epoch into a second pull request: #113675

@pytorchbot merge

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team