This repository contains federated learning experiments with data from the National Lung Cancer Screening Trials as well as installation instructions for NVFLARE
This document will include various usages for NVFLARE, and will be updated as knowledge on the flare environment is acquired. It will cover installation, setup, CLI usage, and more. This document is designed to ease the learning curve and speed up deployments. The official docs and GitHub repo for NVFLARE can be found at this link.
NVFLARE installs using pip. It is recommended to set up a virtual environment.
pip install nvflare
pip install -r requirements.txt
If an error is thrown, upgrade pip and setuptools and retry:
pip install --upgrade pip
pip install --upgrade setuptools
nvflare --help
nvflare --version
This document is using version 2.4.0. For additional information, refer to the docs here.
run_experiments.txt can be used to execute the jobs within jobs. The outputs will be in the workspaces directory.
./run_experiments -h
./run_experiments -e 'job number"
The NLST datasets used in these experiments will be made publicly available