IBM/lale

Library for Semi-Automated Data Science

What it solves

Lale simplifies semi-automated data science by automating the selection of algorithms and the tuning of hyperparameters for machine learning pipelines. It addresses the need for a consistent way to manage pipeline search tools while ensuring that the configurations are type-safe and correct.

How it works

Lale provides a high-level interface that integrates with existing pipeline search tools like Hyperopt, GridSearchCV, and SMAC. To ensure correctness, it uses JSON Schema to validate that hyperparameters match their required types and that data is compatible with the operators used in the pipeline. It also offers a library of transformers and estimators from various frameworks, including scikit-learn, XGBoost, and PyTorch.

Who it’s for

It is designed for data scientists who want to experiment with automated machine learning (AutoML) without having to learn a new programming language or install specialized standalone tools.

Highlights

  • Unified Interface: Provides a consistent way to access multiple pipeline search tools.
  • Type Safety: Uses JSON Schema to catch mismatches between hyperparameters and types or data and operators.
  • Broad Interoperability: Supports operators from scikit-learn, XGBoost, and PyTorch.
  • Python-Native: Works as a standard Python library compatible with Jupyter notebooks.

Related

  • Project
    microsoft/FLAML

    A lightweight Python library for efficient automated machine learning (AutoML) and hyperparameter tuning to find high-quality models with minimal computational resources.

  • Project
    facebookresearch/spdl

    SPDL is a library for scalable and performant data loading, providing pipeline abstractions and operations for processing array data efficiently.

  • Project
    koaning/scikit-lego

    A collection of custom scikit-learn compatible transformers, metrics, and models to extend the functionality of machine learning pipelines.

  • Project
    PrithivirajDamodaran/FlashRank

    A lightweight, CPU-based Python library for re-ranking search results using state-of-the-art cross-encoders and LLMs to improve retrieval precision for LLM pipelines.

  • Project
    feature-engine/feature_engine

    A Python library providing Scikit-learn compatible transformers for comprehensive feature engineering and selection to improve machine learning model performance.