mirrors / cleanlab--cleanlab

文件历史

1,824 提交 54 分支 21 Git 标签
uly_old_python_logic
提交图

352 次代码提交

作者 SHA1 备注 提交日期
Tata Ganesh fd2506c6b0 Use rare class proportion instead of mean values for imbalance score (#828) 
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-08-28 14:24:24 +00:00
Aditya Thyagarajan a497b38c28 change thresholding class-wise for ObjectLab (#804) 
* Adding support for per-class thresholding for ObjectLab
 2023-08-26 05:06:00 +05:30
Tata Ganesh d37a7d8d85 Expose low memory option for token and multi-label classification (#822) 2023-08-23 01:14:29 -04:00
Hui Wen 0adce8290f 50x speedup for multiannotator module (#821) 2023-08-16 18:39:06 -07:00
Ulyana 396bd68e66 Improve indexing for Object Detection Badloc Score (#807) 2023-08-14 09:29:56 -07:00
Tata Ganesh 83731599be Expose low memory option for finding label issues in classification (#791) 2023-08-11 00:35:51 -04:00
Elías Snorrason ee65553a05 Increase hypothesis deadline for testing near duplicate sets (#802) 2023-08-08 14:46:02 -04:00
Tata Ganesh 3d50b9859d Add imbalance issue to datalab (#758) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-07-28 18:42:15 +00:00
Elías Snorrason b0fb8e5f77 Move Datalab internals to dedicated internal module (#783) 2023-07-28 16:18:51 +00:00
Yiran Shi bc78816341 Label Issue Detection Extention (#760) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-07-27 15:43:34 -04:00
Elías Snorrason cc0fb470d9 Update near-duplicate sets (#781) 2023-07-26 17:12:20 +00:00
Sanjana 944c61379d Cleanvision Integration (#679) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-07-25 15:51:53 +00:00
Hui Wen d84d00e3f4 add test make sure correct examples are not flagged (#778) 2023-07-21 21:18:21 -07:00
Ulyana dcd9ea3a16 Improve OD scores (#770) 
* Add updated badloc, overlooked scores
* Improve code clarity for all scores
* Add codecov tests
* Remove unnecessary logic for swap score
 2023-07-21 12:37:51 -07:00
Aditya Thyagarajan cb51891b06 lowering swap score for multiple labels associated with a box (#750) 
* Lower swap scores for overlapping annotations + Optional arguments correction

---------

Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Ulyana <ulya.tkach@gmail.com>
 2023-07-12 20:29:20 +05:30
Yiran Shi cb4c05cbff Add support for high dimensional arrays in CleanLearning(#749) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-06-27 00:12:52 -04:00
Mayank Kumar 9f302200a4 Regression label quality scores (#572) 
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: huiwengoh <45724323+huiwengoh@users.noreply.github.com>
 2023-06-23 17:51:46 -07:00
Vedang Lad 35606fc401 label error detection in semantic segmentation datasets (#677) 
Co-authored-by: Ulyana <ulyana@cleanlab.ai>
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-06-23 00:32:08 -07:00
Ulyana 299d66f3ec Update object detection tutorial to clarify out-of-sample predictions (#745) 
* Add out-of-sample clarification to notebook, fix filename check, add unit tests
 2023-06-15 13:14:30 -07:00
Ulyana da65a9727f identifying label errors in Object Detection data (#676) 
Co-authored-by: Aditya Thyagarajan <aditya1593@icloud.com>
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-06-07 01:12:46 -07:00
Elías Snorrason 43e24ba3f6 Make labels optional in Datalab (#730) 
* feat(datalab): ✨ make labels optional

* test(datalab): ✅ fix test case for comparing empty list/array
 2023-06-05 16:04:13 +00:00
DerWeh 114201bf76 ENH: make clipping unnecessary for entropy (#703) 
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-05-25 00:49:30 -04:00
Elías Snorrason 375c76fcea Include non-iid in default issue checks (#723) 
* flag no more than one example having non-iid issue

* feat(datalab): ✨ add non-iid to default issue checks
 2023-05-19 20:30:01 +00:00
Elías Snorrason 7400051a95 Fix unbound knn variable in NonIIDIssueManager (#709) 2023-05-10 21:38:47 -07:00
Elías Snorrason fcc9417a3d Simplify reporting for outlier issues (#705) 2023-05-10 01:49:17 +00:00
Elías Snorrason d16cdd67cf Suppress too_slow health check for dataset generation strategy (#706) 2023-05-10 01:48:45 +00:00
Elías Snorrason c41c837e43 Set non iid num issues with an extra rule (#707) 
* feat(datalab): 👔 count number of non-iid issues differently

* fix broken seeding test

* add significance threshold as a parameter

* docs(datalab): 📝 add docs on significance_threshold argument for non-iid issue manager

* test(datalab): ✅ add more end-to-end tests for non-iid issue checks with Datalab

---------

Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-05-10 01:33:25 +00:00
Sanjana cf5d2a02eb Added installation instructions for package extras (#697) 
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
Co-authored-by: huiwengoh <45724323+huiwengoh@users.noreply.github.com>
 2023-05-09 15:19:24 -04:00
Elías Snorrason be1bd6903d Seed noniid permutation tests (#694) 2023-05-08 22:50:11 -07:00
Elías Snorrason 5ee2eb8be9 Drop summary scores from Datalab.report() (#699) 
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-05-04 22:15:41 -04:00
Elías Snorrason f77ef9f065 increase hypothesis deadline to 500ms for test_find_overlapping_classes_with_confident_joint (#695) 
This addresses the flaky test issue on macOS and Windows platforms in the CI, where the test occasionally exceeded the previous 200ms deadline.
 2023-05-04 13:14:26 -04:00
Jonas Mueller a95a447010 Update tests for custom issue manager example (#692) 
* issuemanager.get_summary->make_summary

* test(datalab): ♻️ move tests for issue managers into their respective test modules

* test(datalab): ✅ add tests for custom issue manager
- Validate scores provided to IssueManager.make_summary
- Move fixture for custom issue manager to conftest.py
- Test make_summary on a "custom" IssueManager

* test(datalab): add __init__.py for discoverability of issue manager tests

---------

Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-05-03 00:40:27 +00:00
Elías Snorrason f8b11d37b6 refactor(datalab): ♻️ rename get_summary -> get_issue_summary (#691) 
This only pertains to Datalab/DataIssues. IssueManager has a different method with the get_summary name that is used in a different way.
 2023-05-02 19:15:31 +00:00
Jonas Mueller d5e81f0c6d allow for kwargs in token find_label_issues (#686) 2023-04-28 10:28:39 -07:00
Elías Snorrason 5650a87d85 Remove the "health_summary_kwargs" dictionary argument from LabelIssueManager (#683) 2023-04-28 00:01:05 +00:00
Elías Snorrason ff3bdd1e48 Add a less restrictive Reporter.report() method for Datalab (#680) 
Add a new method `Reporter.report()` with fewer restrictions. The existing `get_report()` method is a pure function that returns a string as output. The new `report()` method can have side-effects, such as printing the report.
 2023-04-27 18:23:01 +00:00
Elías Snorrason 64edc95630 Introduce Datalab (#614) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Jesse Cummings <jecummin@mit.edu>
Co-authored-by: Jesse Cummings <jecummin@users.noreply.github.com>
Co-authored-by: Hui Wen <45724323+huiwengoh@users.noreply.github.com>
 2023-04-25 09:26:43 -07:00
Aditya Thyagarajan f8c1866c58 move methods to multilabel_classification module (#657) 
adds new dataset summarization methods for multilabel settings

---------

Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-04-13 21:50:11 -07:00
Hui Wen 1e653def93 remove min batch size restriction in LabelInspector (#665) 2023-04-11 13:05:02 -07:00
Hui Wen 64c6cfcdd5 Pass confident joint computed in CleanLearning to filter.find_label_issues (#661) 2023-04-05 21:21:17 -07:00
Hui Wen fa1db6e1d5 Bug fixes + improvements to multiannotator module (#654) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-03-20 23:36:41 -07:00
Hui Wen 00776646a7 Fix bug with call to find_overlapping_issues without specifying labels (#652) 
* fix None labels issue

* build: ➕ add hypothesis to dev dependencies. Used for property-based testing

* test: ✅ test that find_overlapping_classes can run by only providing a confident joint

Resolves #651

---------

Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-03-13 15:37:00 +00:00
Jonas Mueller 6eb60198e6 update tests to be more stringent (#635) 2023-02-15 17:34:11 -08:00
clu0 3757637d8c Fix batched multiprocessing being slower on tall matrices (#634) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-02-15 13:14:39 -08:00
clu0 4850050803 label_issues_batched multiprocessing (#630) 
4x speedup (48 cores, 10k classes x 1M examples)


Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-02-12 20:25:02 -08:00
Hui Wen 4d2753ae05 Fix KerasWrapper summary method (#631) 2023-02-12 18:40:52 -08:00
Jonas Mueller 686cbf63f1 Method to estimate label issues with limited memory via mini-batches (#615) 
Co-authored-by: clu0 <33559427+clu0@users.noreply.github.com>
 2023-02-10 16:59:16 -08:00
Ulyana d99788b9de Fixed duplicate issue removal in find_label_issues (#624) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
 2023-02-10 08:28:29 -08:00
Hui Wen 6aaee8332b Create new cleanlab.models module (#601) 
Co-authored-by: Jonas Mueller <1390638+jwmueller@users.noreply.github.com>
Co-authored-by: Elías Snorrason <eliassno@gmail.com>
 2023-02-07 10:03:32 -08:00
Ulyana c191d8781d Fix format compatibility with latest black==23. release (#620) 2023-02-07 10:01:45 -08:00