TST use global_random_seed in sklearn/utils/tests/test_optimize.py #26462
Conversation
test_newton_cg uses global_random_seed fixture
sply88
changed the title
|
Hi @sply88 ! Thank you for this PR. I can see that the linter failed but the logs have expired and are not accessible anymore. Could you therefore pull the changes from upstream main again into your feature branch and fix the linting issue? |
|
Hi @marenwestermann. Thank you for having a look at this. Have included the latest changes from main. The linter passes. Only Check Changelog fails, but I guess there is no need to update the changelog for this PR? |
|
No, there should be no changelog needed. Can you run the CI for all the random seeds? (See guideline at the top of the issue description and note that you can create an empty commit with |
sply88
changed the title
|
CI has failed here @sply88 |
|
@marenwestermann would you like to take over this one? |
|
Sorry for the late reply. Will close this one as @marenwestermann already fixed the remaining problem with numerical tolerance in #30112. |
3 participants
Towards #22827
I modified the only test in sklearn/utils/tests/test_optimize.py to use the
global_random_seedfixture.Required changes were minimal:
Just replaced the previously fixed seed with
global_random_seedand pass the fixture to the test function.All tests run via
pass locally.
The PR only changes a test file, so no addition to the change log required?