openfl.plugins.frameworks_adapters.keras_adapter.FrameworkAdapterPlugin

class openfl.plugins.frameworks_adapters.keras_adapter.FrameworkAdapterPlugin

Bases: FrameworkAdapterPluginInterface

Framework adapter plugin class.

__init__()

Initialize framework adapter.

Return type:

None

Methods

__init__()	Initialize framework adapter.
get_tensor_dict(model[, optimizer, suffix])	Extract tensor dict from a model and an optimizer.
serialization_setup()	Prepare model for serialization (optional).
set_tensor_dict(model, tensor_dict[, ...])	Set the model weights with a tensor dictionary.

static get_tensor_dict(model, optimizer=None, suffix='')

Extract tensor dict from a model and an optimizer.

Parameters:

• model (object) – The model object.

• optimizer (object, optional) – The optimizer object. Defaults to None.

• suffix (str, optional) – The suffix for the weight dictionary keys. Defaults to ‘’.

Returns:

A dictionary with weight name as key and

numpy ndarray as value.

Return type:

model_weights (dict)

static serialization_setup()

Prepare model for serialization (optional).

static set_tensor_dict(model, tensor_dict, optimizer=None, device='cpu')

Set the model weights with a tensor dictionary.

Parameters:

• model (object) – The model object.

• tensor_dict (dict) – The tensor dictionary.

• optimizer (object, optional) – The optimizer object. Defaults to None.

• device (str, optional) – The device to be used. Defaults to ‘cpu’.

Returns:

None