# cleanlab/cleanlab

Cleanlab's open-source library is the standard data-centric AI package for data quality and machine learning with messy, real-world data and labels.

Repository: https://github.com/cleanlab/cleanlab
Canonical: https://ross.abutalabs.com/products/cleanlab
Homepage: https://cleanlab.ai
Language: Python
License: Apache-2.0
License Family: permissive
Topics: weak-supervision, data-cleaning, data-quality, noisy-labels, data-centric-ai, out-of-distribution-detection, outlier-detection, active-learning, data-labeling, data-profiling, data-validation, labeling, data-curation, annotation, datasets, exploratory-data-analysis, data-annotation, machine-learning, anomaly-detection, data-science
Last push: 2026-01-13T17:39:04+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 62, release rhythm 41, longevity 100
- inputs: {"age_days": 3037, "days_push": 232, "days_rel": 232, "gap_med": 153, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 11636, forks 917 (observed 2026-08-28T04:10:49.230827+00:00)

## What it is
Cleanlab is a Python library for data-centric AI that automatically detects issues in ML datasets, such as label errors, outliers, duplicates, and other data quality problems, using your existing model outputs. It works with text, image, audio, and tabular data and supports robust training, multi-annotator consensus, and active learning.

## Use cases
- find mislabeled examples in my training dataset
- detect outliers and duplicates in a machine learning dataset
- clean noisy labels before training a classifier
- estimate which annotators produce low-quality labels
- decide which data points to label next with active learning
- audit dataset quality for image, text, audio, or tabular data
- train robust models on real-world messy data

## When to choose
- you have a supervised ML dataset and suspect label errors or other data issues
- you want to leverage existing model predictions or embeddings to find dataset problems
- you need multi-annotator consensus or annotator-quality scoring
- you want a well-established, actively maintained open-source data quality library

## When to avoid
- you need LLM output monitoring, guardrails, or hallucination detection - that is Cleanlab's separate commercial TLM platform, not this library
- you need general-purpose data wrangling or ETL rather than ML-specific dataset issue detection
- your workflow is outside Python

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, data-science, analytics, nlp, image-processing, audio-processing
- domain: data-science, machine-learning, analytics
- platform: python, cross-platform
- tags: data-quality, noisy-labels, label-errors, outlier-detection, out-of-distribution-detection, data-centric-ai, active-learning, data-cleaning, data-validation, weak-supervision, multi-annotator, confident-learning, data-engineering

## Member repositories
- cleanlab/cleanlab (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:49.230827+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:15:26.618851+00:00, confidence not recorded.
  - readme: https://github.com/cleanlab/cleanlab (fetched 2026-08-28T04:10:49.230827+00:00, sha 1e6a9dce125f)
  - homepage: https://cleanlab.ai (fetched 2026-08-29T08:13:52.075825+00:00, sha 26872bedc4eb)
  - site_page: https://cleanlab.ai/about (fetched 2026-08-29T08:13:52.084943+00:00, sha f8cb314c1a86)
  - registry_pypi: https://pypi.org/pypi/cleanlab/json (fetched 2026-08-29T08:13:52.087146+00:00, sha 57973e5bdbaa)
  - site_page: https://help.cleanlab.ai (fetched 2026-08-29T08:13:52.079987+00:00, sha 24f7e8afb4bb)
  - site_page: https://help.cleanlab.ai/ (fetched 2026-08-29T08:13:52.082253+00:00, sha 24f7e8afb4bb)
- Data as of 2026-08-30T08:39:29.467469+00:00.