seraml 1.0.0

Machine learning library — Sera Framework (Rust core). Drop-in Scikit-learn replacement, 2–686× faster.

SeraML

Machine learning library built on the Sera Framework — a low-level Rust core shared with SeraPlot and the upcoming SeraDFrame.

SeraML is a drop-in replacement for Scikit-learn, 2–686× faster, with a single pip install and zero C/Fortran dependencies.

pip install seraml

Quick example

from seraml.linear_model import LinearRegression
from seraml.model_selection import train_test_split
from seraml.metrics import r2_score

X = [[1], [2], [3], [4], [5]]
y = [2.1, 4.0, 5.9, 8.1, 10.0]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

model = LinearRegression()
model.fit(X_train, y_train)
score = r2_score(y_test, model.predict(X_test))
print(f"R² = {score:.4f}")

Scikit-learn compatibility

SeraML mirrors the Scikit-learn API — .fit(), .predict(), .transform(), .score(). Pipelines work the same way.

Submodules

Module	Contents
seraml.linear_model	LinearRegression, Ridge, Lasso, ElasticNet, LogisticRegression, SGDClassifier, SGDRegressor, RidgeClassifier
seraml.tree	DecisionTreeClassifier, DecisionTreeRegressor
seraml.ensemble	RandomForestClassifier, RandomForestRegressor, GradientBoostingClassifier, GradientBoostingRegressor, AdaBoostClassifier, AdaBoostRegressor
seraml.cluster	KMeans, DBSCAN
seraml.neighbors	KNeighborsClassifier, KNeighborsRegressor, NearestCentroid
seraml.naive_bayes	GaussianNB, MultinomialNB, BernoulliNB
seraml.svm	LinearSVC, LinearSVR
seraml.preprocessing	StandardScaler, MinMaxScaler, RobustScaler, MaxAbsScaler, Normalizer, LabelEncoder, OneHotEncoder, OrdinalEncoder, SimpleImputer, PolynomialFeatures, KBinsDiscretizer, PowerTransformer, QuantileTransformer
seraml.decomposition	PCA, TruncatedSVD
seraml.model_selection	GridSearchCV, RandomizedSearchCV, HalvingGridSearchCV, HalvingRandomSearchCV, StratifiedKFold, train_test_split, cross_val_score
seraml.metrics	Full classification, regression, clustering metric suite

The Sera Framework

SeraML is one of three products built on Sera — a low-level Rust framework:

• SeraPlot — 60+ chart types, zero dependencies (pypi.org/project/seraplot)
• SeraML — drop-in Scikit-learn replacement (this package)
• SeraDFrame — Pandas/Polars alternative (Q4 2027)

The Rust core provides CPU/GPU optimisation, memory arenas, parallel threads (rayon), and a compile-time macro registry — natively inherited by every method.

License

MIT