Execution Overview

This section explains how to execute federated learning tasks in the Flower runtime. First, we list the function types and actions, then we examine the usage pattern, including local vs remote execution. Finally, we provide links to detailed documentation for each function and action.

Function types and Actions

There are three function kinds in the Flower runtime:

flower-app: Complete Flower application with client and server components
flower-client: Flower client for federated learning participation
flower-server: Flower server for managing federated learning tasks

Each kind supports specific actions.

Function Kind	Supported Actions
`flower-app`	`train`
`flower-client`	`build`, `deploy`
`flower-server`	`build`, `deploy`

Usage Pattern

To execute a federated learning task, follow this pattern:

Use dh.new_function() or project.new_function() to create the function, passing function parameters.
Call function.run() with the desired action, passing task parameters and run parameters.

# Create Flower function that references a Flower app source
# in a git repository
function = dh.new_function(
    name="my-flower-app",
    kind="flower-app",
    source="git+https://github.com/your-org/your-repo"
)

# Execute with task and run parameters
run = function.run(
    action="train",  # Task parameter
    federation="my-federation",  # Run parameter
    superlink="superlink-service",  # Run parameter
    parameters={"num_rounds": 10}  # Run parameter
)

Local vs Remote Execution

When executing a flower-app function (and only that) with train action, you can choose between local execution and remote execution by setting the local_execution parameter in the run parameters.

Local Execution (local_execution=True): The function runs directly on your local machine using Flower's simulation mode. You need to have Flower and required dependencies installed locally.
Remote Execution (local_execution=False, default): The function is executed on a remote server or cluster managed by the platform.

Function and Action Documentation

Here are links to the detailed documentation for each function and action:

Function flower-app action train docs.
Function flower-client action build docs.
Function flower-client action deploy docs.
Function flower-server action build docs.
Function flower-server action deploy docs.