Newmetric: NRMSE

[SkafteNicki]

What does this PR do?

Fixes #2303

Before submitting

• Was this discussed/agreed via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

---

📚 Documentation preview 📚: https://torchmetrics--2442.org.readthedocs.build/en/2442/

[Borda]

@SkafteNicki, what is missing here to make it land? 🐿️

[SkafteNicki]

@SkafteNicki, what is missing here to make it land? 🐿️

@Borda should be ready now :]

[Borda]

shall this be another PR?

[Borda]

@SkafteNicki it seems something wrong there is:

>>> import torch
>>> from torchmetrics import NormalizedRootMeanSquaredError
>>> target = torch.tensor([[0.0, 0.0, 0.0], [0.0, 0.0, 0.0]])
>>> preds = torch.tensor([[1.0, 2.0, 3.0], [1.0, 2.0, 3.0]])
>>> nrmse = NormalizedRootMeanSquaredError(num_outputs=3)
>>> nrmse(preds, target)
Expected:
    tensor([1., 1., 1.])
Got:
    tensor([inf, inf, inf])

[Borda]

@SkafteNicki just a friendly ping :) seems almost done just revisit the examples...

[SkafteNicki]

@SkafteNicki just a friendly ping :) seems almost done just revisit the examples...

yes, just need the final tests to pass :]

[Andrewwango]

Please could we also have an L2 option which normalises by the sum of squares of target? This is a commonly used NMSE in image processing eg for MRI (eg see Eq. 5 in https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3947978/#sec-a.w.btitle )

[Andrewwango]

Hi @Borda , do you need any more input from me regarding this? Thanks

[Borda]

@SkafteNicki ^^