blazefl.contrib.FedAvgServerHandler

class blazefl.contrib.FedAvgServerHandler(model_selector: ModelSelector, model_name: str, dataset: PartitionedDataset, global_round: int, num_clients: int, sample_ratio: float, device: str)

Bases: ServerHandler

Server-side handler for the Federated Averaging (FedAvg) algorithm.

Manages the global model, coordinates client sampling, aggregates client updates, and controls the training process across multiple rounds.

model

The global model being trained.

Type:

torch.nn.Module

dataset

Dataset partitioned across clients.

Type:

PartitionedDataset

global_round

Total number of federated learning rounds.

Type:

int

num_clients

Total number of clients in the federation.

Type:

int

sample_ratio

Fraction of clients to sample in each round.

Type:

float

device

Device to run the model on (‘cpu’ or ‘cuda’).

Type:

str

client_buffer_cache

Cache for storing client

Type:

list[FedAvgUplinkPackage]

updates before aggregation.

num_clients_per_round

Number of clients sampled per round.

Type:

int

round

Current training round.

Type:

int

__init__(model_selector: ModelSelector, model_name: str, dataset: PartitionedDataset, global_round: int, num_clients: int, sample_ratio: float, device: str) → None

Initialize the FedAvgServerHandler.

Parameters:

• model_selector (ModelSelector) – Selector for initializing the model.

• model_name (str) – Name of the model to be used.

• dataset (PartitionedDataset) – Dataset partitioned across clients.

• global_round (int) – Total number of federated learning rounds.

• num_clients (int) – Total number of clients in the federation.

• sample_ratio (float) – Fraction of clients to sample in each round.

• device (str) – Device to run the model on (‘cpu’ or ‘cuda’).

Methods

__init__(model_selector, model_name, ...)	Initialize the FedAvgServerHandler.
aggregate(parameters_list, weights_list)	Aggregate model parameters from multiple clients using weighted averaging.
downlink_package()	Create a downlink package containing the current global model parameters to send to clients.
global_update(buffer)	Aggregate client updates and update the global model parameters.
if_stop()	Check if the training process should stop.
load(payload)	Load a client's uplink package into the server's buffer and perform a global update if all expected packages for the round are received.
sample_clients()	Randomly sample a subset of clients for the current training round.

static aggregate(parameters_list: list[Tensor], weights_list: list[int]) → Tensor

Aggregate model parameters from multiple clients using weighted averaging.

Parameters:

• parameters_list (list[torch.Tensor]) – List of serialized model parameters

• clients. (from)

• weights_list (list[int]) – List of data sizes corresponding to each client’s

• parameters.

Returns:

Aggregated model parameters.

Return type:

torch.Tensor

downlink_package() → FedAvgDownlinkPackage

Create a downlink package containing the current global model parameters to send to clients.

Returns:

Downlink package with serialized model parameters.

Return type:

FedAvgDownlinkPackage

global_update(buffer: list[FedAvgUplinkPackage]) → None

Aggregate client updates and update the global model parameters.

Parameters:

buffer (list[FedAvgUplinkPackage]) – List of uplink packages from clients.

if_stop() → bool

Check if the training process should stop.

Returns:

True if the current round exceeds or equals the total number of global rounds; False otherwise.

Return type:

bool

load(payload: FedAvgUplinkPackage) → bool

Load a client’s uplink package into the server’s buffer and perform a global update if all expected packages for the round are received.

Parameters:

payload (FedAvgUplinkPackage) – Uplink package from a client.

Returns:

True if a global update was performed; False otherwise.

Return type:

bool

sample_clients() → list[int]

Randomly sample a subset of clients for the current training round.

Returns:

Sorted list of sampled client IDs.

Return type:

list[int]