The Best Organizations Use Flower
Breakthrough Decentralized Training Research
Our pioneering research in Decentralized Foundation Model Training is the key to unlocking a new pre-training paradigm.
Get Started
Build your first federated learning project in two steps. Use Flower with your favorite machine learning framework to easily federated existing projects.
pip install flwr[simulation]flwr new # Select TensorFlow & follow instructionsflwr run .Flower Apps built by the community
Find Flower Apps built by the community, or publish your Flower App for the community.
@flwrlabs/quickstart-pytorch
Federated Learning with PyTorch and Flower (Quickstart Example)
@flwrlabs/flowertune-llm-general-nlp
Base App for the General NLP challenge in the FlowerTune LLM Leaderboard.
@flwrlabs/quickstart-tensorflow
Federated Learning with Tensorflow/Keras and Flower (Quickstart Example)
Building with Flower is simple
This series of tutorials introduces the fundamentals of Federated Learning and how to implement it with Flower.
What is Federated AI?
learn what federated learning is, build your first system in Flower, and gradually extend it. If you work through all parts of the tutorial, you will be able to build advanced federated learning systems that approach the current state of the art in the field.
Get started with Flower
Build a federated learning system using the Flower framework, Flower Datasets and PyTorch. In part 1, we use PyTorch for model training and data loading. In part 2, we federate this PyTorch project using Flower.
Accelerate your Federated AI project with Flower Labs
Flower’s team of experts is here to support your projects from early prototyping to large-scale production deployment.
Services
We provide custom services to accelerate your Federated AI journey in the way that works best for you. From “we’re on standby if you have questions” to “we build and operate everything for you”.
Support
Our experts provide a level of enterprise-grade support that’s unmatched in Federated AI. We build the open-source framework and operate the platform, so we can address issues across the entire stack without having to wait for someone else.