Flower
Flower, formerly known as Adap and Flower Labs, is a remote B2B company launched in 2023, specializing in federated learning solutions for AI training on distributed data.
Company Overview
Flower is an open-source framework launched in 2023, designed for training AI on distributed data through federated learning. Originally known as Adap and later Flower Labs, the company operates fully remotely with a team size of five. Flower is a part of Y Combinator's W23 batch and focuses on providing scalable, framework-agnostic solutions for federated learning in the B2B industry.
Services
Flower offers a unified approach to federated learning, analytics, and evaluation. The company provides detailed installation guides and quickstart examples for various machine learning frameworks including PyTorch, TensorFlow, scikit-learn, MXNet, and Hugging Face. The services are designed to enable research and real-world AI system development on multiple platforms such as AWS, GCP, Azure, Android, iOS, Raspberry Pi, and Nvidia Jetson. Flower promotes scalability and framework agnosticism, making it suitable for workloads with tens of millions of clients.
Industry Applications
Flower's federated learning framework is used by prominent companies like Banking Circle, Nokia, Porsche, and Brave to enhance their AI models on sensitive data distributed across organizational silos or user devices. The framework enables training on significantly more data, facilitating substantial advancements in AI capabilities. Flower supports various machine learning workloads and is compatible with a broad array of programming languages and ML frameworks.
Community and Events
Flower is a community-driven initiative, actively inviting feedback and participation via Slack. The company hosts events such as Flower Monthly and Flower Summit to engage with users, share updates, and foster collaboration. The community-oriented approach helps Flower to continuously improve its offerings and address the needs of its users effectively. Participation in these events provides a platform for users to learn, share, and contribute to the federated learning ecosystem.
Technological Compatibility
Flower is designed to be a framework-agnostic and platform-independent solution, supporting a wide range of machine learning frameworks such as PyTorch, TensorFlow, scikit-learn, MXNet, and Hugging Face. It offers compatibility with multiple platforms including AWS, GCP, Azure, Android, iOS, Raspberry Pi, and Nvidia Jetson. This versatility allows researchers and developers to conduct federated learning experiments and deploy AI models across various devices and infrastructure setups.