Flower
An open-source federated AI framework for training across decentralized data, with Apache 2.0 licensing and PyTorch, TensorFlow and Hugging Face support.
Flower is a federated AI framework for developers and researchers who want to train models across separate organizations or devices. Participants retain control of their data rather than gathering everyone's data in one place.
- Model training: Build federated learning systems with PyTorch, TensorFlow, Hugging Face Transformers, JAX or scikit-learn.
- Analytics and robotics: Use Pandas for federated analytics or LeRobot for federated robotics, alongside support for MLX, TFLite, XGBoost and CatBoost.
- Custom research: Customize federated learning strategies and extend or replace framework components to test approaches suited to your project.
- Research baselines: Evaluate ideas against community implementations of published experiments, including FedAvg, FedProx, FedBN and HeteroFL.
Flower is written in Python and is open source under the Apache 2.0 license. Its federated approach supports collaboration across decentralized data held by participating organizations and devices. Supported frameworks also include PyTorch Lightning, MONAI, fastai and NumPy.
Logotyper
