Handle batchnorms in BatchGradientVerification

[indigoviolet]

What does this PR do?

Fixes #568

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• 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?
• Did you write any new necessary tests? [not needed for typos/docs]
• Did you verify new and existing tests pass locally with your changes?
• If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

• Is this pull request ready for review? (if not, please submit in draft mode)

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 🙃

[codecov]

Codecov Report

Merging #569 (1727cc7) into master (4d194d4) will increase coverage by 0.05%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #569      +/-   ##
==========================================
+ Coverage   76.97%   77.02%   +0.05%     
==========================================
  Files         117      117              
  Lines        6836     6852      +16     
==========================================
+ Hits         5262     5278      +16     
  Misses       1574     1574              

Flag	Coverage Δ
cpu	26.20% <33.33%> (+0.01%)	⬆️
pytest	26.20% <33.33%> (+0.01%)	⬆️
unittests	76.51% <100.00%> (+0.05%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
pl_bolts/callbacks/verification/batch_gradient.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4d194d4...9157012. Read the comment docs.

[awaelchli]

nice fix

[indigoviolet]

Looks like the test failures are unrelated, something to do with mnist downloads?

[awaelchli]

Hey @indigoviolet

Great work!

I allowed myself to 1) add some docs to the context manager 2) rename it to selective_eval becaue it's not specific to normalization layers (what do you think of the name?) and 3) add some tests

hope that's ok!
The tests that fail here also seem to fail locally on master, but for the changes in this PR all seems fine so far.

[indigoviolet]

@awaelchli very nice, thank you.

[Borda]

@indigoviolet mind check the failing tests?

[indigoviolet]

@Borda the test failures were unrelated to this PR, something to do with PY3 and torch._six. I've merged upstream master, hopefully that fixes it.

[indigoviolet]

/opt/hostedtoolcache/Python/3.6.13/x64/lib/python3.6/urllib/request.py:650: HTTPError
----------------------------- Captured stdout call -----------------------------
Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to /home/runner/work/pytorch-lightning-bolts/pytorch-lightning-bolts/datasets/MNIST/raw/train-images-idx3-ubyte.gz

Now the failure seems to be something to do with mnist downloads, also unrelated to this PR. I'm not sure how to fix this (@Borda)

[kaushikb11]

Hi @indigoviolet, the tests will be fixed in #580

[awaelchli]

Thanks a lot @indigoviolet for working on this!