Flower - A Friendly Federated Learning Research Framework

Build

Flower is a research framework for building federated learning systems. The design of Flower is based on a few guiding principles:

Customizable: Federated learning systems vary wildly from one use case to another. Flower allows for a wide range of different configurations depending on the needs of each individual use case.
Extendable: Flower originated from a research project at the Univerity of Oxford, so it was build with AI research in mind. Many components can be extended and overridden to build new state-of-the-art systems.
Framework-agnostic: Different machine learning frameworks have different strengths. Flower can be used with any machine learning framework, for example, PyTorch, TensorFlow, or even raw NumPy for users who enjoy computing gradients by hand.
Understandable: Flower is written with maintainability in mind. The community is encouraged to both read and contribute to the codebase.

Note: Even though Flower is used in production, it is published as pre-release software. Incompatible API changes are possible.

Installation

Flower can be installed directly from the GitHub repository using pip:

$ pip install git+https://github.com/adap/flower.git

Official PyPI releases will follow once the API matures.

Run Examples

We built a number of examples showcasing different usage scenarios in src/flower_example. To run an example, first install the necessary extras (available extras: examples-tensorflow):

pip install git+https://github.com/adap/flower.git#egg=flower[examples-tensorflow]

Once the necessary extras (e.g., TensorFlow) are installed, you might want to run the Fashion-MNIST example by starting a single server and multiple clients in two terminals using the following commands.

Start server in the first terminal:

$ ./src/flower_example/tf_fashion_mnist/run-server.sh

Start the clients in a second terminal:

$ ./src/flower_example/tf_fashion_mnist/run-clients.sh

Docker

If you have Docker on your machine you might want to skip most of the setup and try out the example using the following commands:

# Create docker network `flower` so that containers can reach each other by name
$ docker network create flower
# Build the Flower docker containers
$ ./dev/docker_build.sh

# Run the docker containers (will tail a logfile created by a central logserver)
$ ./src/flower_example/tf_fashion_mnist/run-docker.sh

This will start a slightly reduced setup with only four clients.

Documentation

Documentation

Contributing to Flower

We welcome contributions. Please see CONTRIBUTING.md to get started!

Release files for flwr 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for flwr 0.1.0
File	Size	Uploaded
flwr-0.1.0.tar.gz	83.6 kB	Jul 13, 2020	Details

Built distribution (wheel)

Table of built distributions (wheels) for flwr 0.1.0
File	Interpreter	ABI	Platform	Reset
flwr-0.1.0-py3-none-any.whl 156.1 kB Jul 13, 2020	Python 3	none	any	Details

Total release size: 239.7 kB

Release files / flwr-0.1.0.tar.gz

Download URL	flwr-0.1.0.tar.gz
Size	83.6 kB
Tags	Source
SHA-256 checksum How to use checksums	`b73076601ed4db25654e6ff25406123eaad96fc5db8f87113a002dc36b4ce41c`
BLAKE2b-256 checksum How to use checksums	`f614d2a5d968db5d27911cb7e689a13a9c4c31fbf04b0ef167a82431111f50c7`
Upload date	Jul 13, 2020
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`poetry/1.0.9 CPython/3.7.7 Linux/5.3.0-62-generic`

Release files / flwr-0.1.0-py3-none-any.whl

Download URL	flwr-0.1.0-py3-none-any.whl
Size	156.1 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`9ef8b766fa6ed7251743b24e9ca25fa9ee4affe7f8acc7412ee14676d3805a7c`
BLAKE2b-256 checksum How to use checksums	`96eb60872dfdf671ba8af4254eaab54ed56b2e5a2de72e37d5c41227afb5f614`
Upload date	Jul 13, 2020
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`poetry/1.0.9 CPython/3.7.7 Linux/5.3.0-62-generic`