fedmind 0.1.2

Federated Learning research framework in your mind

FedMind

A simple and easy Federated Learning framework fit researchers' mind based on PyTorch.

Unlike other popular FL frameworks focusing on production, FedMind is designed for researchers to easily implement their own FL algorithms and experiments. It provides a simple and flexible interface to implement FL algorithms and experiments.

Installation

The package is published on PyPI under the name fedmind. You can install it with pip:

pip install fedmind

Usage

A configuration file in yaml is required to run the experiments. You can refer to the config.yaml as an example.

There are examples in the examples directory.

Make a copy of both the config.yaml and fedavg_demo.py to your own directory. You can run them with the following command:

python fedavg_demo.py

Features

• Parallel training speed up.
• Serialization for limited resources.