How fmorsbach/simplefedml works
Project type and purpose
• This is a simple federated machine-learning (FedML) demo built in Python.
• It consists of a central server (Flask-based HTTP “federation coordinator”) and multiple clients that train local models and submit updates.
File structure
• Root: .gitignore, README.md
• client/
– ReferenceModel.py (defines the TensorFlow model architecture)
– main.py (client main loop: fetch global model, train locally, send updates)
• server/
– Model.py (model definition or helper)
– CentralFederation.py (core aggregation logic: average weights)
– HTTPCentralFederation.py (Flask app exposing REST endpoints for clients)
README for additional insights
• Implements naive model averaging using TensorFlow
• Communication over HTTP via a Flask server