Autogluon

AutoGluon is an open-source AutoML library that automates machine learning tasks, enabling developers and data scientists to build high-performance models with minimal code. It supports tabular, text, image, and time series data, leveraging ensembling, hyperparameter tuning, and neural architecture search to achieve state-of-the-art accuracy. Targeted at both beginners and experienced practitioners, AutoGluon reduces the time from raw data to deployment. Users can train a model with just three lines of code, while advanced users can customize layers, optimizers, and training schedules. It integrates seamlessly with the Python ecosystem, supporting pandas DataFrames, NumPy arrays, and popular deep learning frameworks. What sets AutoGluon apart is its automated ensembling and multi-modal capabilities. It automatically identifies data types and applies appropriate preprocessing and model architectures. For tabular data, it stacks multiple models and uses an ensemble of ensembles. For text and images, it fine-tunes pretrained transformers. This results in robust performance without manual feature engineering or model selection. AutoGluon is developed by AWS Labs and is free to use under the Apache 2.0 license. It does not offer a cloud-hosted version, but can be deployed on any infrastructure. Its documentation is comprehensive, with tutorials covering installation, quickstarts, and advanced customization. The library is actively maintained with regular releases and community support via GitHub and forums.