pytorch-lightning icon indicating copy to clipboard operation
pytorch-lightning copied to clipboard

Ignore parameters causing ValueError when dumping to YAML

Open Callidior opened this issue 10 months ago • 2 comments

What does this PR do?

Fixes #19730

Unserializable parameters such as nn.Modules or Tensors often fail with a ValueError when calling yaml.dump on them and using PyTorch 2.x. Lightning already tries to handle this type of error but only catches TypeError so far. This PR proposes to additionally catch ValueError.

Before submitting
  • Was this discussed/agreed via a GitHub issue? (not for typos and docs)
  • [x] Did you read the contributor guideline, Pull Request section?
  • [x] 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)
  • [x] 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 minor internal changes/refactors)

PR review

Anyone in the community is welcome to review the PR. Before you start reviewing, make sure you have read the review guidelines. In short, see the following bullet-list:

Reviewer checklist
  • [ ] 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

📚 Documentation preview 📚: https://pytorch-lightning--19804.org.readthedocs.build/en/19804/

Callidior avatar Apr 23 '24 08:04 Callidior

Having the same issue, this fix would be useful.

nicolas-mng avatar Apr 24 '24 13:04 nicolas-mng

This fix should proceed soon. Otherwise, no way to test multiple models.

zhanghaoyue avatar May 01 '24 13:05 zhanghaoyue

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 53%. Comparing base (a611de0) to head (934f022). Report is 94 commits behind head on master.

:exclamation: There is a different number of reports uploaded between BASE (a611de0) and HEAD (934f022). Click for more details.

HEAD has 173 uploads less than BASE
Flag BASE (a611de0) HEAD (934f022)
python3.10 19 12
cpu 64 22
lightning 39 16
pytest 45 2
examples 8 0
app 8 0
python3.8 12 6
lightning_fabric 10 0
python3.11 20 4
tpu 1 0
gpu 4 2
pytorch2.0 12 11
pytest-full 24 22
pytorch2.1 6 5
pytorch_lightning 10 8
lightning_app 1 0
Additional details and impacted files
@@            Coverage Diff             @@
##           master   #19804      +/-   ##
==========================================
- Coverage      84%      53%     -31%     
==========================================
  Files         426      418       -8     
  Lines       35280    35127     -153     
==========================================
- Hits        29616    18708   -10908     
- Misses       5664    16419   +10755     

codecov[bot] avatar Jun 06 '24 19:06 codecov[bot]

Thank you @Callidior, great fix!

awaelchli avatar Jun 06 '24 22:06 awaelchli