Pickle SR3 and subordinate classes #586
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Auto-updated by precommit.
This change adds a test parametrization to specifically check SR3 pickling. Design-wise, this change shifts away from some of the safety in argument specifying regularizer. Weighted regularizers are now the same in all ways as nonweighted (scalar-weighted) ones. They already had the same implementation, they just had extra guard code. Weight shape is still checked to match data shape in order to prevent broadcasting, but using a scalar weight on a "weighted" regularizer is now allowed.
This version introduced the method __sklearn_tags__(),
but it currently seems broken. Many tests run into calls for:
env\Lib\site-packages\sklearn\utils\_tags.py:396: in get_tags
tags = estimator.__sklearn_tags__()
E AttributeError: 'super' object has no attribute '__sklearn_tags__'
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #586 +/- ##
==========================================
- Coverage 95.26% 95.24% -0.02%
==========================================
Files 39 39
Lines 4099 4105 +6
==========================================
+ Hits 3905 3910 +5
- Misses 194 195 +1 ☔ View full report in Codecov by Sentry. |
himkwtn
reviewed
Dec 22, 2024
There was a problem hiding this comment.
Should this debug file be included?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #585
This change moves us away from specifically requiring "weighted" in specifying SR3 regularizer. Previously, much of the weighted functionality had been combined with the unweighted (scalar-weighted) functionality, and all that remained was guard checks for the shape of the weight and data. However, that guard code was added at runtime, making the functions unpickleable (See pickle docs). Some of the solution was to use
@wraps. The rest required that we add guard code in a way that was determined at import time, rather than runtime. See commit messages from this PR.Also includes two quick fixes for a broken scikit-learn version and updating our pre-commit format.