nvflare.job_config.fed_job_config module

class FedJobConfig(job_name, min_clients, mandatory_clients=None, meta_props=None)

Bases: object

FedJobConfig represents the job in the NVFlare.

FedJobConfig uses the job_name, min_clients and optional mandatory_clients to create the object. It also provides the method to add in the FedApp, the deployment map of the FedApp and participants, and the resource _spec requirements of the participants if needed.

Parameters:

• job_name – the name of the NVFlare job

• min_clients – the minimum number of clients for the job

• mandatory_clients – mandatory clients to run the job (optional)

• meta_props – additional meta properties for the job (optional)

add_fed_app(app_name: str, fed_app: FedAppConfig)

add_resource_spec(site_name: str, resource_spec: Dict)

generate_job_config(job_root)

generate the job config

Returns:

locate_imports(sf)

Locate imported modules in a Python source file.

Parameters:

sf – source file

Returns:

yield (module name or None, relative import level) tuples

set_app_packages(app_packages: List[str])

Set app packages. When generating job config, code from these packages will not be included into “custom” folder.

Parameters:

app_packages – app packages

Returns: None

set_site_app(site_name: str, app_name: str)

assign an app to a certain site.

Parameters:

• site_name – The target site name.

• app_name – The app name.

Returns:

simulator_run(workspace, clients=None, n_clients=None, threads=None, gpu=None, log_config=None)