[config]
Tools to specify hyper-parameters of a Federated Learning process.
This submodule exposes dataclasses that group, document and facilitate parsing (from instances, config dicts and/or TOML files) elements that are required to specify a Federated Learning process from the server's side.
The main classes implemented here are:
- FLRunConfig: Federated learning orchestration hyper-parameters.
- FLOptimConfig: Federated optimization strategy.
The following dataclasses are articulated by FLRunConfig
:
- EvaluateConfig: Hyper-parameters for an evaluation round.
- RegisterConfig: Hyper-parameters for clients registration.
- TrainingConfig: Hyper-parameters for a training round.