flox.clients.MainClient.MainClient¶

class MainClient[source]¶

Bases: FloxClientLogic

Methods

`get_number_of_samples`
`on_data_process`
`on_data_retrieve`	Retrieves data for training the model.
`on_model_fit`	Fit the provided global model using a local dataset
`on_model_receive`	Parses & decrypts the received data from a controller
`on_model_send`	Final data processing before sending the results back, such as encryption.
`retrieve_framework_data`
`retrieve_local_data`	Retrieves x_train and y_train from given paths of .npy files Overwrite this function if you need to retrieve your locally stored data differently
`run_round`	Combines the rest of the functions to run a single Federated Learning round

on_data_retrieve(config)[source]¶

Retrieves data for training the model.

Parameters¶

config: ConfigFile: a dictionary with values necessary to retrieve the data, such as the data source and preprocessing parameters

Returns¶

XYData: data for evaluating the model. This can take different forms depending on the ML framework you use. For Tensorflow, it would look as x_test and y_test, while for PyTorch, it would look as a single DataLoader instance.

on_model_fit()¶

Fit the provided global model using a local dataset

Return type:: List[ndarray[Any, dtype[Any]]]

Parameters¶

Parameters for fitting the model. This might take different forms depending on the ML framework. For Tensorflow, it might look like this:

model: keras.Sequential ModelTrainer: Class config: ConfigFile x_train: NDArrays y_train: Union[NDArray, NDArrays]

Returns¶

NDArrays: new model weights as Numpy arrays

on_model_receive()[source]¶: Parses & decrypts the received data from a controller

on_model_send(fit_results, training_data=None, config=None, task_runtime=None, task_start_timestamp=None, task_finish_timestamp=None, data_processing_runtime=None, training_runtime=None, endpoint_physical_cores=None, endpoint_logical_cores=None, endpoint_physical_memory=None, platform_name=None)[source]¶

Final data processing before sending the results back, such as encryption.

Returns¶

results: dict

FL results formatted as a dictionary. For example: results = {

“model_weights”: model_weights, “samples_count”: samples_count, “bias_weights”: fractions,

}

retrieve_local_data(config)[source]¶

Retrieves x_train and y_train from given paths of .npy files Overwrite this function if you need to retrieve your locally stored data differently

Return type:: tuple

Parameters¶

config: dict: dictionary with values indicating the path directory and file names for x_ and y_train data

Returns¶

raw_training_data: Tuple(NDArrays, NDArrays): x_train and y_train returned as a tuple

run_round(config, model_trainer)[source]¶

Combines the rest of the functions to run a single Federated Learning round

Parameters¶

model_trainer: instance of a class based on BaseModelTrainer that implements .fit(), .evaluate(), .get_weights(), and .set_weights()
config: dict: a dictionary with all the necessary parameters for retrieving data and training the model

Returns¶

results: dict

FL results formatted as a dictionary. For example: results = {

“model_weights”: model_weights, “samples_count”: samples_count, “bias_weights”: fractions,

}