自动检测数据集中错误数据和标注的框架。该项目基于置信学习（confident learning，CL）算法，实现了自动检测出机器学习数据集中的各种问题，提高数据集质量训练出更好的模型，支持图像、文本、音频类型的数据。
active 2023-08-15 → 2026-07-04 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 1055 days from 2023-08-15 to 2026-07-04. Pushes: 365 total, peak 18 in a day. Pull requests: 359 total, peak 15 in a day. Issues: 158 total, peak 14 in a day. Comments: 1,170 total, peak 45 in a day. Stars: 3,015 total, peak 71 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| jwmueller | 1.3K | 173 | 80 | 507 |
| elisno | 549 | 93 | 146 | 152 |
| ulya-tkch | 175 | 53 | 25 | 53 |
| tataganesh | 127 | 0 | 8 | 70 |
| codecov[bot] | 107 | 0 | 0 | 107 |
| sanjanag | 70 | 3 | 6 | 26 |
| aditya1503 | 58 | 3 | 5 | 24 |
| huiwengoh | 43 | 4 | 7 | 11 |
| mturk24 | 42 | 21 | 3 | 11 |
| coding-famer | 35 | 0 | 7 | 19 |
| Steven-Yiran | 34 | 0 | 3 | 23 |
| smttsp | 32 | 0 | 1 | 18 |
| mglowacki100 | 30 | 0 | 0 | 18 |
| gordon-lim | 29 | 0 | 5 | 12 |
| gogetron | 27 | 0 | 15 | 6 |
| 01PrathamS | 26 | 0 | 7 | 14 |
| MaxJoas | 23 | 0 | 0 | 13 |
| CLAassistant | 20 | 0 | 0 | 20 |
| allincowell | 16 | 0 | 7 | 6 |
| nelsonauner | 13 | 0 | 0 | 7 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#908AnukritiSingh-222026-04-26 10:01Improve property-based test for near-duplicate sets
- Pull request#1308tsharma-dev2026-04-13 06:51
- Pull request#1304BillionClaw2026-03-16 04:20
- Issue comment#1303CLAassistant2026-03-15 22:46feat(multilabel): extend active learning with multiple annotators to multi-label datasets
- Pull request#1303BillionClaw2026-03-15 22:46
- Issue#1301CV-China2026-02-11 02:02[Inconsistent Results in Cleanlab Object Detection Data Cleaning (v2.9.0/v2.7.0) - Batch vs Individual Processing Differences and Parameter Sensitivity]
- Issue#1301CV-China2026-02-11 02:02[Inconsistent Results in Cleanlab Object Detection Data Cleaning (v2.9.0/v2.7.0) - Batch vs Individual Processing Differences and Parameter Sensitivity]
- Issue comment#1299CLAassistant2026-02-01 04:08docs: Add link to Issue Types Guide in find_issues docstring
- Issue comment#1258druzsan2026-01-19 12:58Support for numpy 2.*
- Issue comment#1298CLAassistant2026-01-16 03:18docs: add Issue Types guide link to find_issues docstring
- Pull request#1296ulya-tkch2026-01-13 00:07
- Pull request#1296ulya-tkch2026-01-12 23:56
- Releaseulya-tkch2026-01-12 21:43v3.0.0
- Releaseulya-tkch2026-01-12 21:23v3.0.0
- Pull request#1293ulya-tkch2026-01-12 19:01
- Pull request#1292ulya-tkch2026-01-10 02:20
- Issue comment#1244ulya-tkch2026-01-09 23:35Assertion error when running the text example
- Issue#1253ulya-tkch2026-01-09 23:25AssertionError Datalab
- Issue#1260ulya-tkch2026-01-09 23:19ValueError: Unsupported Label format when creating Datalab object in tutorials ran on Colab or locally
- Issue comment#1260ulya-tkch2026-01-09 23:19ValueError: Unsupported Label format when creating Datalab object in tutorials ran on Colab or locally
- Issue comment#1290ulya-tkch2026-01-09 20:22Fix empty bboxes
- Pull request#1292ulya-tkch2026-01-09 19:52
- Issue comment#930repairman292026-01-09 05:25extend active learning with multiple annotators to multi-label datasets
- Pull request#1289ulya-tkch2026-01-09 04:25
- Issue comment#1289codecov[bot]2026-01-08 23:50Remove Tensorflow from docs + code
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 3,015 stars here means stars gained during the window, not the repo's star count.