FLAML

FLAML is a Python AutoML library for data scientists and developers who want to reduce manual model selection and tuning. Its search accounts for computational cost when choosing which configurations to evaluate.

• Automated model selection: Find models for classification and regression through a scikit-learn-style interface, or limit the search to learners such as XGBoost, LightGBM and Random Forest.
• Custom learners: Extend model selection and hyperparameter tuning to your own learners alongside the built-in choices.
• Configuration tuning: Search settings for machine learning pipelines, statistical models, algorithms and software, with support for constraints, early stopping and evaluations that have different costs.
• LLM inference tuning: Optimize foundation-model inference settings to fit an application's needs and reduce the cost of model use.

FLAML runs within your Python environment and is open source under the MIT license. It integrates with MLflow and supports AutoML and hyperparameter tuning in Microsoft Fabric Data Science. A .NET implementation is available through ML.NET.

AutoGen is a separate project; FLAML doesn't include the autogen agent module.