search

adriangb / scikeras

Scikit-Learn API wrapper for Keras.
https://www.adriangb.com/scikeras/
MIT License
239 stars 47 forks source link

chore: updating the poetry version to fix the push issue #308

Closed cristian-rincon closed 11 months ago

cristian-rincon commented 11 months ago

This change includes the update on the pyproject.toml file. As the previous release failed due to the package version, as it was the same one that was already present in the Pypi.

Please refer to this traceback to see the whole error: https://github.com/adriangb/scikeras/actions/runs/6383670304

adriangb commented 11 months ago

So I think the problem here is that removing the pin on grpcio allowed poetry to resolve a newer tensorflow version (see https://github.com/adriangb/scikeras/actions/runs/6424229105/job/17444399467?pr=308#step:6:164, not sure what it was before) and TensorFlow minor updates usually (and unfortunately) have breaking changes. So we'll need the PR to bump the minium TensorFlow version to 2.13 or so and fix any issues that causes. Would you be willing to work on that @cristian-rincon ?

codecov-commenter commented 11 months ago

Codecov Report

Merging #308 (67737aa) into master (ea8f685) will increase coverage by 0.26%. The diff coverage is n/a.

:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.

@@            Coverage Diff             @@
##           master     #308      +/-   ##
==========================================
+ Coverage   97.32%   97.59%   +0.26%     
==========================================
  Files           7        7              
  Lines         748      748              
==========================================
+ Hits          728      730       +2     
+ Misses         20       18       -2

see 1 file with indirect coverage changes

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

cristian-rincon commented 11 months ago

Deprecating tf 2.11 fixed several build issues. I will take a look to see if the current code supports 2.13

cristian-rincon commented 11 months ago

Hi @adriangb Upgrading to TensorFlow 13 right now will require several refactors. I cannot afford those refactors now. However, after conducting several tests with different constraints, I have found a solution that works and can fix the pipelines. I suggest we release this version with the new constraints for now. We can then address the upgrade in future releases. What do you think?