scm2cpp-learn: 44 scikit-learn estimators, same results, up to ~1000x faster

说明与用途

scm2cpp-learn — 44 个 scikit‑learn 估计器的替代实现，保持相同结果，速度提升至 1000 倍，并可选择使用 CUDA。

问题

Python users face slow execution of scikit-learn estimators on large datasets and limited GPU support.

结果

All 44 estimators produce coefficients matching scikit‑learn to rounding error while achieving up to 1,047× speed‑up on a single core and GPU acceleration on an RTX 4090.

工作流

• pip install scm2cpp-learn
• replace import from sklearn with scm2cpp_learn
• instantiate estimator as usual
• fit and predict – same API and output as scikit‑learn

证据

44 scikit-learn estimators (Lasso, LogisticRegression, RFE, PCA, SVM, ...) implementing the same algorithms with the same results — coefficients agree to rounding error

来源

1.5x to ~1000x faster (e.g. RFE 1,047x; MultiTaskElasticNetCV ~100x on a single core)

来源

scikit-learn API compatible: works in Pipeline and GridSearchCV, passes check_estimator

来源