Implementation Notes¶

This section covers various aspects of AFL-Sim's implementation, for interested readers and developers who are considering extending the framework.

Table of Contents¶

Simulated Federated Learning Workflows: A detailed description of the real-world Federated Learning implementations simulated by AFL-Sim, including examples and diagrams.

Local Training and Aggregation Weights: How AFL-Sim internally handles these two crucial components of Federated Learning, while maintaining theoretical consistency.

Serializing Concurrent Execution: How AFL-Sim's architecture converts concurrent client training to serial simulation events, for both synchronous and asynchronous training modes.

Modeling System Heterogeneity A short guide on how AFL-Sim synthesizes realistic inputs for its benchmarking tasks, including data splits and client arrivals.