Open
Conversation
iraola
force-pushed
the
485-fix-docker-image
branch
4 times, most recently
from
6566871 to
85fc03a
Compare
iraola
force-pushed
the
485-fix-docker-image
branch
2 times, most recently
from
c8a7a75 to
055ba98
Compare
iraola
changed the title
iraola
force-pushed
the
485-fix-docker-image
branch
5 times, most recently
from
1a2e543 to
b6fd722
Compare
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #486 +/- ##
==========================================
- Coverage 95.35% 91.90% -3.46%
==========================================
Files 73 73
Lines 12025 12048 +23
==========================================
- Hits 11467 11073 -394
- Misses 558 975 +417
... and 9 files with indirect coverage changes Continue to review full report in Codecov by Sentry.
🚀 New features to boost your workflow:
|
iraola
force-pushed
the
485-fix-docker-image
branch
from
b6fd722 to
b8b6517
Compare
Mostly small changes in NaN and float data types.
And improve Dockerfile and requirements_*.txt files
- Address NumPy 2.x upgrade issues where the original complex nested np.concatenate calls in test_arrange_sparse failed due to changes in array handling and flattening behavior. - Introduce flatten_index_arrays helper function to properly handle the nested list-of-tuples structure from _arrange_samples, ensuring compatibility and ignoring empty arrays. - Replace the brittle flattening logic with the new helper for improved readability, maintainability, and NumPy 2.x support. The helper is kept within the test module as it's specific to this test's requirements.
Recent versions of scikit-learn enforce stricter input validation in paired_distances and no longer accept sparse matrices implicitly. This caused convergence checks to fail when k-means centers were stored in sparse format, specifically under test_gm.py.test_sparse(). The convergence logic now explicitly normalizes centers to dense arrays before computing distances, preserving previous behavior while making the conversion explicit and forward-compatible. No changes to the algorithm or results are introduced.
Replace usage of numpy.testing._private.parameterized with unittest.TestCase.subTest. This removes reliance on NumPy private APIs that were removed in v2.x, while preserving test coverage across multiple parameter configurations.
Newer scikit-learn versions use n_init=1 by default if nothing else is specified, therefore the comparing solution was not accurate. Fixing n_init=10 fixes this issue.
Relax RandomForestRegressor test accuracy assertion from 0.72 to 0.70 to account for stable but environment-dependent floating-point and parallelism differences observed between local and CI runs.
Handle division by zero and ensure compatibility with inhomogeneous arrays after upgrading to NumPy 2.x, preventing errors in nested algorithm execution and tests. This typycally happens when applying np.array to nested lists of size-varying lists. Adding the argument dtype=object is a quick fix for this.
This refactor simplifies the Docker image and reduces its size. - TensorFlow support is removed - pip cache is explicitly cleaned in the same build layer to avoid baking download artifacts into the final image. - The blosc installation is updated to align with NumPy 2.x support, - Docker build context is tightened by extending .dockerignore to exclude Python virtual environments and other local-only artifacts. Overall, this makes the image leaner, more reproducible, and easier to maintain without changing runtime behavior for supported workloads.
No longer needed due to better integration of newest torch version with numpy.
- Use assertGreaterEqual for better logging upon errors - Update deprecated assertGreaterEquals
Update requirement versions for pyarrow and cvxpy since older versions can show issues after the NumPy 2.x upgrade
iraola
force-pushed
the
485-fix-docker-image
branch
from
b8b6517 to
643e537
Compare
For testing, it's proven to reduce the whole pipeline executing time (close to ~4h).
iraola
force-pushed
the
485-fix-docker-image
branch
from
6e74fde to
0bae378
Compare
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Description
Errors with Jenkins testing related to PyEDDL outdated versions + numpy major version upgrade
Fixes #485
Type of change
Please delete options that are not relevant.
How Has This Been Tested?
Please describe the tests that you ran to verify your changes. If you haven't added any test and it is relevant provide instructions so we can reproduce.
Reproduce instructions:
Checklist: