Maintaining models for aggregation with the 
state manager

In this section, we will explain state_manager.py, which handles maintaining the models and necessary volatile information related to the aggregation of local models.

Importing the libraries of the state manager

This code imports the following. The internal libraries for data_struc, helpers, and states are introduced in the Appendix, Exploring Internal Libraries:

import numpy as np
import logging
import time
from typing import Dict, Any
from fl_main.lib.util.data_struc import LimitedDict
from fl_main.lib.util.helpers import generate_id, generate_model_id
from fl_main.lib.util.states import IDPrefix

After importing the necessary libraries, let’s define the state manager class.

Defining the state manager class

The state manager class (Class StateManager), as seen in state_manager.py, is defined in the following code:

class StateManager:
    """
  ...