Skip to main content

Part 3: Monitoring Your FL Training

Now that both your aggregator and client are set up and running, let's monitor your federated learning experiment in action.

Tracking Training Progress​

  1. Access the metrics dashboard:
https://syftbox.openmined.org/datasites/your.datasite/fl/[YOUR_CREATIVE_NAME]_MNIST_FL/

Below is a sample metrics dashboard for the MNIST_FL project:

Metrics Dashboard

  1. Monitor client logs:
$HOME/SyftBox/apis/fl_client/logs/
  1. Monitor aggregator logs:
$HOME/SyftBox/apis/fl_aggregator/logs/

What to Watch For​

  • Training Rounds: Track the progress of each federated learning round
  • Model Performance: Monitor accuracy and loss metrics
  • Participant Status: Check the status of all participating clients
  • System Health: Keep an eye on logs for any potential issues

Troubleshooting Tips​

If you encounter any issues:

  1. Check both client and aggregator logs for error messages
  2. Ensure your SyftBox client is still running in the original terminal
  3. Make sure your training data is properly loaded

🎉 Congratulations! You've now completed all parts of the federated learning example! Your system is now fully operatonals to participating in privacy-preserving machine learning on the live SyftBox network..

tip

🤖 For a deeper understanding of the system architecture and implementation details, check out the Aggregator and Client Implementation Guides.