cleanlab

library2.9.0pypypi✓ verified 85d ago

cleanlab is the standard package for data-centric AI, automatically finding and fixing label errors, outliers, and other data issues in datasets for classification, regression, and more. Current version is 2.9.0, released on a monthly/quarterly cadence.

ai-ml

pip install cleanlab

INSTALL

IMPORT

SIG · CLEANLAB

cleanlab

ai-mlpythonv2.9.0

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

CleanLearning

✓ from cleanlab.classification import CleanLearning

✗ from cleanlab import CleanLearning

CleanLearning is not a top-level import; it's under cleanlab.classification

find_label_issues

✓ from cleanlab.filter import find_label_issues

✗ from cleanlab import find_label_issues

find_label_issues is under cleanlab.filter, not top-level

LearningWithNoisyLabels

✓ from cleanlab.classification import CleanLearning

✗ LearningWithNoisyLabels is removed; use CleanLearning instead

Renamed in v2.0

Fits a classifier with noisy labels and identifies likely label errors.

import numpy as np from sklearn.linear_model import LogisticRegression from cleanlab.classification import CleanLearning # Toy data with label errors np.random.seed(42) X = np.random.rand(100, 2) y = np.random.randint(0, 2, 100) # Introduce label errors flip = np.random.rand(100) < 0.2 y[flip] = 1 - y[flip] cl = CleanLearning(clf=LogisticRegression()) cl.fit(X, y) label_issues = cl.get_label_issues() print(label_issues)

Debug

Known issues

deprecated`LearningWithNoisyLabels` has been renamed to `CleanLearning` in v2.0

fix

Use `from cleanlab.classification import CleanLearning`

affects: <=1.0

breakingIn v2.0, the `cleanlab.pruning` module was removed; functionality moved to `cleanlab.classification`

fix

Use `CleanLearning` with `find_label_issues` method

affects: >=2.0

gotcha`find_label_issues` expects predicted probabilities (`pred_probs`), not predictions

fix

Ensure you pass predicted probabilities (e.g., from `predict_proba`) to `find_label_issues`

affects: all

gotchaCleanLearning requires a classifier that implements `predict_proba`

fix

Use classifiers with `predict_proba` method (e.g., LogisticRegression, RandomForestClassifier)

affects: all

deprecated`cleanlab.latent_algebra` is deprecated; use `cleanlab.internal.latent_algebra`

fix

Import from `cleanlab.internal.latent_algebra`

affects: >=2.6

Errors

Common errors & fixes

ModuleNotFoundError: No module named 'cleanlab.pruning'

pruning module was removed in cleanlab v2.0

fix

Use `from cleanlab.classification import CleanLearning` instead

TypeError: find_label_issues() missing 1 required positional argument: 'pred_probs'

pred_probs (predicted probabilities) not provided

fix

Pass pred_probs from your model's predict_proba method: find_label_issues(labels, pred_probs, ...)

ValueError: The truth value of an array with more than one element is ambiguous

Passing NumPy array where boolean index is expected (common with pandas)

fix

Ensure labels are 1D array-like; avoid using pandas Series directly

ImportError: cannot import name 'LearningWithNoisyLabels' from 'cleanlab'

Renamed in v2.0

fix

Use `from cleanlab.classification import CleanLearning`

TypeError: '<' not supported between instances of 'str' and 'int'

Label dtype mismatch (string vs int)

fix

Convert labels to consistent numeric type (e.g., labels = labels.astype(int))

Upgrade

Version history

2.9.0latest on PyPI · released Jan 13, 2026

Audit

Dependencies

numpyrequiredCore dependency for array operations

scikit-learnrequiredUsed for model training and metrics

pandasoptionalUsed for data handling in DataFrames

matplotliboptionalOptional, for visualizations

Agent activity

15 hits · last 30 days

node

OpenAI (training)

Resources

No resource links recorded.