5.1.1.3.1.1. `FedEval.config.configuration`

5.1.1.3.1.1.1. Module Contents

5.1.1.3.1.1.1.1. Classes

`_Configuraiton`
`_DataConfig`
`_ModelConfig`
`_RT_Machine`
`_RuntimeConfig`
`ConfigurationManagerInterface`	Helper class that provides a standard way to create an ABC using
`ClientConfigurationManagerInterface`	an interface of ConfigurationManager from the client side,
`ServerConfigurationManagerInterface`	an interface of ConfigurationManager from the central server side,
`_CfgYamlInterface`	an interface that regulates the methods used to serialize
`_CfgJsonInterface`	an interface that regulates the methods used to serialize
`_CfgSerializer`	types of serializer for configurations.
`_CfgFileInterface`	an interface that regulates the methods used to serialize
`_RoledConfigurationInterface`	Helper class that provides a standard way to create an ABC using
`ConfigurationManager`	the base class of singletons.

5.1.1.3.1.1.1.2. Attributes

`RawConfigurationDict`
`DEFAULT_D_CFG_FILENAME_YAML`
`DEFAULT_MDL_CFG_FILENAME_YAML`
`DEFAULT_RT_CFG_FILENAME_YAML`
`DEFAULT_D_CFG_FILENAME_JSON`
`DEFAULT_MDL_CFG_FILENAME_JSON`
`DEFAULT_RT_CFG_FILENAME_JSON`
`_D_DIR_KEY`
`_D_NAME_KEY`
`_D_NI_ENABLE_KEY`
`_D_NI_CLASS_KEY`
`_D_NI_STRATEGY_KEY`
`_D_NORMALIZE_KEY`
`_D_SAMPLE_SIZE_KEY`
`_D_PARTITION_KEY`
`_D_FEATURE_SIZE`
`_D_RANDOM_SEED`
`_DEFAULT_D_CFG`
`_STRATEGY_KEY`
`_STRATEGY_NAME_KEY`
`_STRATEGY_ETA_KEY`
`_STRATEGY_B_KEY`
`_STRATEGY_C_KEY`
`_STRATEGY_E_KEY`
`_STRATEGY_E_RATIO`
`_STRATEGY_E_DISTRIBUTE`
`_STRATEGY_MAX_ROUND_NUM_KEY`
`_STRATEGY_TOLERANCE_NUM_KEY`
`_STRATEGY_NUM_ROUNDS_BETWEEN_VAL_KEY`
`_STRATEGY_FEDSTC_SPARSITY_KEY`
`_STRATEGY_FEDPROX_MU_KEY`
`_STRATEGY_FEDOPT_TAU_KEY`
`_STRATEGY_FEDOPT_BETA1_KEY`
`_STRATEGY_FEDOPT_BETA2_KEY`
`_STRATEGY_FEDOPT_NAME_KEY`
`_STRATEGY_FETCHSGD_COL_NUM_KEY`
`_STRATEGY_FETCHSGD_ROW_NUM_KEY`
`_STRATEGY_FETCHSGD_BLOCK_NUM_KEY`
`_STRATEGY_FETCHSGD_TOP_K_KEY`
`_STRATEGY_FEDSVD_BLOCK`
`_STRATEGY_FEDSVD_MODE`
`_STRATEGY_FEDSVD_TOPK`
`_STRATEGY_FEDSVD_L2`
`_STRATEGY_FEDSVD_OPT_1`
`_STRATEGY_FEDSVD_OPT_2`
`_STRATEGY_FEDSVD_EVALUATE`
`_ML_KEY`
`_ML_NAME_KEY`
`_ML_ACTIVATION_KEY`
`_ML_DROPOUT_RATIO_KEY`
`_ML_UNITS_SIZE_KEY`
`_ML_OPTIMIZER_KEY`
`_ML_OPTIMIZER_NAME_KEY`
`_ML_OPTIMIZER_LEARNING_RATE_KEY`
`_ML_OPTIMIZER_MOMENTUM_KEY`
`_ML_LOSS_CALC_METHODS_KEY`
`_ML_METRICS_KEY`
`_ML_DEFAULT_METRICS`
`_DEFAULT_MDL_CFG`
`_RT_SERVER_KEY`
`_RT_S_HOST_KEY`
`_RT_S_LISTEN_KEY`
`_RT_S_PORT_KEY`
`_RT_S_CLIENTS_NUM_KEY`
`_RT_S_SECRET_KEY`
`_RT_DOCKER_KEY`
`_RT_D_IMAGE_LABEL_KEY`
`_RT_D_CONTAINER_NUM_KEY`
`_RT_D_GPU_ENABLE_KEY`
`_RT_D_GPU_NUM_KEY`
`_RT_MACHINES_KEY`
`_RT_M_ADDRESS_KEY`
`_RT_M_PORT_KEY`
`_RT_M_USERNAME_KEY`
`_RT_M_WORK_DIR_KEY`
`_RT_M_SK_FILENAME_KEY`
`_RT_M_CAPACITY_KEY`
`_RT_M_SERVER_NAME`
`_RT_LOG_KEY`
`_RT_L_BASE_LEVEL_KEY`
`_RT_L_FILE_LEVEL_KEY`
`_RT_L_CONSOLE_LEVEL_KEY`
`_RT_L_DIR_PATH_KEY`
`_RT_COMMUNICATION_KEY`
`_RT_COMM_METHOD_KEY`
`_RT_COMM_PORT_KEY`
`_RT_COMM_LIMIT_FLAG_KEY`
`_RT_COMM_BANDWIDTH_UP_KEY`
`_RT_COMM_BANDWIDTH_DOWN_KEY`
`_RT_COMM_LATENCY_KEY`
`_RT_COMM_FAST_MODE`
`_DEFAULT_RT_CFG`
`_DEFAULT_ENCODING`
`_Stream`

FedEval.config.configuration.RawConfigurationDict

FedEval.config.configuration.DEFAULT_D_CFG_FILENAME_YAML = '1_data_config.yml'

FedEval.config.configuration.DEFAULT_MDL_CFG_FILENAME_YAML = '2_model_config.yml'

FedEval.config.configuration.DEFAULT_RT_CFG_FILENAME_YAML = '3_runtime_config.yml'

FedEval.config.configuration.DEFAULT_D_CFG_FILENAME_JSON = '1_data_config.yml'

FedEval.config.configuration.DEFAULT_MDL_CFG_FILENAME_JSON = '2_model_config.yml'

FedEval.config.configuration.DEFAULT_RT_CFG_FILENAME_JSON = '3_runtime_config.yml'

FedEval.config.configuration._D_DIR_KEY = 'data_dir'

FedEval.config.configuration._D_NAME_KEY = 'dataset'

FedEval.config.configuration._D_NI_ENABLE_KEY = 'non-iid'

FedEval.config.configuration._D_NI_CLASS_KEY = 'non-iid-class'

FedEval.config.configuration._D_NI_STRATEGY_KEY = 'non-iid-strategy'

FedEval.config.configuration._D_NORMALIZE_KEY = 'normalize'

FedEval.config.configuration._D_SAMPLE_SIZE_KEY = 'sample_size'

FedEval.config.configuration._D_PARTITION_KEY = 'train_val_test'

FedEval.config.configuration._D_FEATURE_SIZE = 'feature_size'

FedEval.config.configuration._D_RANDOM_SEED = 'random_seed'

FedEval.config.configuration._DEFAULT_D_CFG: RawConfigurationDict

FedEval.config.configuration._STRATEGY_KEY = 'FedModel'

FedEval.config.configuration._STRATEGY_NAME_KEY = 'name'

FedEval.config.configuration._STRATEGY_ETA_KEY = 'eta'

FedEval.config.configuration._STRATEGY_B_KEY = 'B'

FedEval.config.configuration._STRATEGY_C_KEY = 'C'

FedEval.config.configuration._STRATEGY_E_KEY = 'E'

FedEval.config.configuration._STRATEGY_E_RATIO = 'evaluate_ratio'

FedEval.config.configuration._STRATEGY_E_DISTRIBUTE = 'distributed_evaluate'

FedEval.config.configuration._STRATEGY_MAX_ROUND_NUM_KEY = 'max_rounds'

FedEval.config.configuration._STRATEGY_TOLERANCE_NUM_KEY = 'num_tolerance'

FedEval.config.configuration._STRATEGY_NUM_ROUNDS_BETWEEN_VAL_KEY = 'rounds_between_val'

FedEval.config.configuration._STRATEGY_FEDSTC_SPARSITY_KEY = 'sparsity'

FedEval.config.configuration._STRATEGY_FEDPROX_MU_KEY = 'mu'

FedEval.config.configuration._STRATEGY_FEDOPT_TAU_KEY = 'tau'

FedEval.config.configuration._STRATEGY_FEDOPT_BETA1_KEY = 'beta1'

FedEval.config.configuration._STRATEGY_FEDOPT_BETA2_KEY = 'beta2'

FedEval.config.configuration._STRATEGY_FEDOPT_NAME_KEY = 'opt_name'

FedEval.config.configuration._STRATEGY_FETCHSGD_COL_NUM_KEY = 'num_col'

FedEval.config.configuration._STRATEGY_FETCHSGD_ROW_NUM_KEY = 'num_row'

FedEval.config.configuration._STRATEGY_FETCHSGD_BLOCK_NUM_KEY = 'num_block'

FedEval.config.configuration._STRATEGY_FETCHSGD_TOP_K_KEY = 'top_k'

FedEval.config.configuration._STRATEGY_FEDSVD_BLOCK = 'block_size'

FedEval.config.configuration._STRATEGY_FEDSVD_MODE = 'fedsvd_mode'

FedEval.config.configuration._STRATEGY_FEDSVD_TOPK = 'fedsvd_top_k'

FedEval.config.configuration._STRATEGY_FEDSVD_L2 = 'fedsvd_lr_l2'

FedEval.config.configuration._STRATEGY_FEDSVD_OPT_1 = 'fedsvd_opt_1'

FedEval.config.configuration._STRATEGY_FEDSVD_OPT_2 = 'fedsvd_opt_2'

FedEval.config.configuration._STRATEGY_FEDSVD_EVALUATE = 'fedsvd_debug_evaluate'

FedEval.config.configuration._ML_KEY = 'MLModel'

FedEval.config.configuration._ML_NAME_KEY = 'name'

FedEval.config.configuration._ML_ACTIVATION_KEY = 'activation'

FedEval.config.configuration._ML_DROPOUT_RATIO_KEY = 'dropout'

FedEval.config.configuration._ML_UNITS_SIZE_KEY = 'units'

FedEval.config.configuration._ML_OPTIMIZER_KEY = 'optimizer'

FedEval.config.configuration._ML_OPTIMIZER_NAME_KEY = 'name'

FedEval.config.configuration._ML_OPTIMIZER_LEARNING_RATE_KEY = 'lr'

FedEval.config.configuration._ML_OPTIMIZER_MOMENTUM_KEY = 'momentum'

FedEval.config.configuration._ML_LOSS_CALC_METHODS_KEY = 'loss'

FedEval.config.configuration._ML_METRICS_KEY = 'metrics'

FedEval.config.configuration._ML_DEFAULT_METRICS = ['accuracy']

FedEval.config.configuration._DEFAULT_MDL_CFG: RawConfigurationDict

FedEval.config.configuration._RT_SERVER_KEY = 'server'

FedEval.config.configuration._RT_S_HOST_KEY = 'host'

FedEval.config.configuration._RT_S_LISTEN_KEY = 'listen'

FedEval.config.configuration._RT_S_PORT_KEY = 'port'

FedEval.config.configuration._RT_S_CLIENTS_NUM_KEY = 'num_clients'

FedEval.config.configuration._RT_S_SECRET_KEY = 'secret_key'

FedEval.config.configuration._RT_DOCKER_KEY = 'docker'

FedEval.config.configuration._RT_D_IMAGE_LABEL_KEY = 'image'

FedEval.config.configuration._RT_D_CONTAINER_NUM_KEY = 'num_containers'

FedEval.config.configuration._RT_D_GPU_ENABLE_KEY = 'enable_gpu'

FedEval.config.configuration._RT_D_GPU_NUM_KEY = 'num_gpu'

FedEval.config.configuration._RT_MACHINES_KEY = 'machines'

FedEval.config.configuration._RT_M_ADDRESS_KEY = 'host'

FedEval.config.configuration._RT_M_PORT_KEY = 'port'

FedEval.config.configuration._RT_M_USERNAME_KEY = 'username'

FedEval.config.configuration._RT_M_WORK_DIR_KEY = 'dir'

FedEval.config.configuration._RT_M_SK_FILENAME_KEY = 'key'

FedEval.config.configuration._RT_M_CAPACITY_KEY = 'capacity'

FedEval.config.configuration._RT_M_SERVER_NAME = 'server'

FedEval.config.configuration._RT_LOG_KEY = 'log'

FedEval.config.configuration._RT_L_BASE_LEVEL_KEY = 'base_level'

FedEval.config.configuration._RT_L_FILE_LEVEL_KEY = 'file_log_level'

FedEval.config.configuration._RT_L_CONSOLE_LEVEL_KEY = 'console_log_level'

FedEval.config.configuration._RT_L_DIR_PATH_KEY = 'log_dir'

FedEval.config.configuration._RT_COMMUNICATION_KEY = 'communication'

FedEval.config.configuration._RT_COMM_METHOD_KEY = 'method'

FedEval.config.configuration._RT_COMM_PORT_KEY = 'port'

FedEval.config.configuration._RT_COMM_LIMIT_FLAG_KEY = 'limit_network_resource'

FedEval.config.configuration._RT_COMM_BANDWIDTH_UP_KEY = 'bandwidth_upload'

FedEval.config.configuration._RT_COMM_BANDWIDTH_DOWN_KEY = 'bandwidth_download'

FedEval.config.configuration._RT_COMM_LATENCY_KEY = 'latency'

FedEval.config.configuration._RT_COMM_FAST_MODE = 'fast_mode'

FedEval.config.configuration._DEFAULT_RT_CFG: RawConfigurationDict

class FedEval.config.configuration._Configuraiton(config: RawConfigurationDict)[source]

Bases: object

property inner: RawConfigurationDict

return a deep copy of its inner configuraiton data, presented as a dict. Noticed that modifications on the returned object will NOT affect the original configuration.

Returns:: a deep copy of the inner data representaiton of this config object.
Return type:: RawConfigurationDict

static _config_filter(config: RawConfigurationDict) → RawConfigurationDict[source]

class FedEval.config.configuration._DataConfig(data_config: RawConfigurationDict = _DEFAULT_D_CFG)[source]

Bases: _Configuraiton

property dataset_name: str

the name of the dataset, chosen from mnist, cifar10, cifar100, femnist, and mnist.

Returns:: the name of chosen dataset.
Return type:: str

property iid: bool

if the dataset would be used in an i.i.d. manner.

Returns:: True if the dataset is sampled in an i.i.d. manner; otherwise, False.
Return type:: bool

property non_iid_class_num: int

return the number of classes hold by each client. Only avaliable when the dataset is sampled in a non-i.i.d. form.

Raises:: AttributeError – raised when called without non-i.i.d. setting.
Returns:: the number of classes hold by each client.
Return type:: int

property non_iid_strategy_name: str

return the name of non-i.i.d. data partition strategy. Two choices are given: 1. “natural” strategy for femnist and celebA dataset 2. “average” for mnist, cifar10 and cifar100

Raises:: AttributeError – raised when called without non-i.i.d. setting.
Returns:: the name of non-i.i.d. data partition strategy.
Return type:: str

property normalized: bool

whether the image pixel data point will be normalized to [0, 1].

Returns:: True if data points would be normalized; otherwise, False.
Return type:: bool

property sample_size: int: return the number of samples owned by each client.

property data_partition: Sequence[float]

get the data partition proportion, ordered as [train data ratio, test data ration, validation data ration].

Constraints met by the return value:

all the ratios in the returned list sum up to 1.
all the ratios in the returned list are non-negative.

Returns:: [train data ratio, test data ration, validation data ration]
Return type:: Sequence[float]

property feature_size

property random_seed

_IID_EXCEPTiON_CONTENT = 'The dataset is configured as iid.'

static _config_filter(config: RawConfigurationDict) → RawConfigurationDict[source]

_non_iid_strategy_name_check() → bool[source]

check if the non-i.i.d. data partition strategy is known.

Returns:: True if the data partition strategy name is registered as followed; otherwise, False.
Return type:: bool

class FedEval.config.configuration._ModelConfig(model_config: RawConfigurationDict = _DEFAULT_MDL_CFG)[source]

Bases: _Configuraiton

property strategy_config: RawConfigurationDict

return a copy of inner strategy raw dict.

Returns:: a deep copy of the strategy-related configuration dict.
Return type:: RawConfigurationDict
Type:: a variant of inner method

property ml_config: RawConfigurationDict

return a copy of inner machine learning raw dict.

Returns:: a deep copy of the ML model-related configuration dict.
Return type:: RawConfigurationDict
Type:: a variant of inner method

property strategy_name: str

get the class name of the federated strategy (i.e., the main controller of federated process). Notice that the strategy class with this name (case sensitive and whole word matching) should have been implemented in this library (specifically, in strategy module), otherwise a TypeNotFound exception would be raised in the following steps.

Returns:: the classname/typename of the federated strategy.
Return type:: str

property ml_method_name: str

get the class name of the machine learning model (i.e., the kernel of the whole calculation process). Notice that the strategy class with this name (case sensitive and whole word matching) should have been implemented in this library (specifically, in model module), otherwise a TypeNotFound exception would be raised in the following steps.

Returns:: the classname/typename of the inner machine learning model.
Return type:: str

property server_learning_rate: float

get the learning rate on the server side. Only available in FedOpt and FedSCA.

Raises:: AttributeError – called in a in proper federated strategy.
Returns:: the learning rate on the server side.
Return type:: float

property B: int: the local minibatch size used for the updates on the client side.

property C: float

the fraction of clients that perform computation in each round.

Examples

if there are 100 available clients in a test network with a C of 0.2, then there should be (100*0.2=)20 clients in each round of iterations.

property E: int: the number of training passes that each client makes over its local dataset in each round.

property evaluate_ratio

property distributed_evaluate

property max_round_num: int: the total/maximum number of the iteration rounds.

property tolerance_num: int: the patience for early stopping

property num_of_rounds_between_val: int: the number of rounds between test or validation

property stc_sparsity: float

the origin of FedSTC

Type:: TODO(fgh)

property prox_mu: float: the /mu parameter in FedProx, a scaler that measures the approximation between the local model and the global model. More info available in Federated Optimization in Heterogeneous Networks(arXiv:1812.06127).

property opt_tau: float

property opt_beta_1: float

property opt_beta_2: float

property activation: str: the name of activation mechanism in tensorflow layers. More info available in https://tensorflow.google.cn/api_docs/python/tf/keras/activations.

property dropout: float: the dropout fraction of Dropout layer in the DL model.

property unit_size: Sequence[int]

the size of sequential neural network components.

Returns:: the size of network components (ordered the same with data flow direction)
Return type:: Sequence[int]

property optimizer_name: str: the name of the optimizer in tensorflow network. More info available in https://tensorflow.google.cn/api_docs/python/tf/keras/optimizers.

property learning_rate: float: the learning rate of model training in tensorlflow.

property momentum: float: the momentum of the optimizer.

property loss_calc_method: str

the identifier of a loss function in tensorflow. More info available in https://tensorflow.google.cn/api_docs/python/tf/keras/losses.

Returns:: the string name of the loss function during model training.
Return type:: str

property metrics: Sequence[str]

names of the metrics used in model training and validation in tensorflow. More info in https://tensorflow.google.cn/api_docs/python/tf/keras/metrics.

Returns:: a copy of metric names.
Return type:: Sequence[str]

property col_num: int: the number of columns in FetchSGD. More info available at https://export.arxiv.org/abs/2007.07682.

property row_num: int: the number of rows in FetchSGD. More info available at https://export.arxiv.org/abs/2007.07682.

property block_num: int: the number of blocks in FetchSGD. More info available at https://export.arxiv.org/abs/2007.07682.

property top_k: int: the number of top items in FetchSGD. More info available at https://export.arxiv.org/abs/2007.07682.

property block_size: int: block size of FedSVD

property svd_mode: str: block size of FedSVD

property svd_top_k: int: block size of FedSVD

property svd_lr_l2: L2 penalize of FedSVD

property svd_opt_1

property svd_opt_2

property svd_evaluate

static _config_filter(config: RawConfigurationDict) → RawConfigurationDict[source]

static __check_raw_config(config: RawConfigurationDict) → None

static __check_runtime_config_shallow_structure(config: RawConfigurationDict) → None

static __check_ML_model_params(ml_config: RawConfigurationDict) → None

class FedEval.config.configuration._RT_Machine(machine_config: RawConfigurationDict, is_server: bool = False)[source]

Bases: _Configuraiton

property is_server: bool: if the machine is a central server.

property addr: str: the IP address of this machine or the name of this container in docker.

property port: int: the port of this virtual machine on the physical machine.

property username: str: the username of this machine.

property work_dir_path: str: the path of this machine’s working diretory.

property key_filename: str: the name of ssh connection secret key file.

property capacity: int

the number of container that this machine can handle. Only available on the client side.

Raises:: AttributeError – called from the server side.

__ITEM_CHECK_VALUE_ERROR_PATTERN = 'machine configuraitons should have {}.'

static __check_items(config: RawConfigurationDict, is_server: bool = False) → None

class FedEval.config.configuration._RuntimeConfig(runtime_config: RawConfigurationDict = _DEFAULT_RT_CFG)[source]

Bases: _Configuraiton

property machines: Mapping[str, _RT_Machine] | None

return a deep copy of all the machines in the configuration.

Returns:: None if there is no machine setting.
Return type:: Optional[Mapping[str, _RT_Machine]]

property client_machines: Mapping[str, _RT_Machine] | None

return a deep copy of all the client machines in the configuration.

Returns:: None if there is no client machine setting.
Return type:: Optional[Mapping[str, _RT_Machine]]

property server_machine

property limit_network_resource: bool: whether limit the network resource

property bandwidth_upload: str: the bandwidth of each container.

property bandwidth_download: str: the bandwidth of each container.

property latency: str: the latency of each container.

property image_label: str: the label of the docker image used in this experiment.

property container_num: int: the number of total docker containers in this experiment.

property central_server_addr: str: the IP address of the central server.

property central_server_listen_at: str: the listening IP address of the flask services on the cetral server side.

property central_server_port: int: the port that the central server occupies.

property client_num: int: the total number of the clients.

property base_log_level: str: the base logging level of all the loggers.

property file_log_level: str: the logging level in the log files.

property console_log_level: str: the logging level in consoles.

property secret_key: str

the secret key of the flask service on the central server side.

Returns:: the secret key as a string.
Return type:: str

property gpu_enabled: bool: whether the GPU is enabled in this experiment.

property gpu_num: int

the number of GPUs.

Raises:: AttributeError – called without GPUs enabled.

property comm_method: str: the method/technique used for mechaine-wise communication in the experiment.

property comm_port: int: the port for communication on the server side.

property comm_fast_mode: bool

In fast mode, all the clients in one container will only download the parameters once: to improve the efficiency, e.g., when tuning the parameters.

Turn off the fast_mode if you are benchmarking the communication and time

__ITEM_CHECK_VALUE_ERROR_PATTERN = 'runtime configurations should have {}.'

__AVAILABLE_LOGGING_LEVELS

static _config_filter(config: RawConfigurationDict) → RawConfigurationDict[source]

static __check_items(config: RawConfigurationDict) → None

_has_machines() → bool[source]

__init_machines() → bool

static _check_log_level_validity(level: str) → None[source]

make sure the given string is one of the logging levels.

Parameters:: level (str) – a string representation of a logging level.
Raises:: ValueError – the given string is not a valid logging level.

class FedEval.config.configuration.ConfigurationManagerInterface[source]

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

abstract property data_config_filename: str

abstract property model_config_filename: str

abstract property runtime_config_filename: str

abstract property data_config: RawConfigurationDict

abstract property model_config: _ModelConfig

abstract property runtime_config: RawConfigurationDict

abstract property job_id

class FedEval.config.configuration.ClientConfigurationManagerInterface[source]

Bases: abc.ABC

an interface of ConfigurationManager from the client side, regulating the essential functions as clients.

Raises:: NotImplementedError – called without implementation.

class FedEval.config.configuration.ServerConfigurationManagerInterface[source]

Bases: abc.ABC

an interface of ConfigurationManager from the central server side, regulating the essential functions as clients.

Raises:: NotImplementedError – called without implementation.

abstract property num_of_train_clients_contacted_per_round: int

FedEval.config.configuration._DEFAULT_ENCODING = 'utf-8'

FedEval.config.configuration._Stream

class FedEval.config.configuration._CfgYamlInterface[source]

Bases: abc.ABC

an interface that regulates the methods used to serialize and deserialize configuraitons in YAML.

static load_configs(src_path, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, encoding=_DEFAULT_ENCODING) → Tuple[RawConfigurationDict, RawConfigurationDict, RawConfigurationDict][source]

static save_configs(data_cfg: RawConfigurationDict, model_cfg: RawConfigurationDict, runtime_cfg: RawConfigurationDict, dst_path, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, encoding=_DEFAULT_ENCODING) → None[source]

class FedEval.config.configuration._CfgJsonInterface[source]

Bases: abc.ABC

an interface that regulates the methods used to serialize and deserialize configuraitons in JSON.

static load_configs(src_path, data_config_filename: str = DEFAULT_D_CFG_FILENAME_JSON, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_JSON, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_JSON, encoding=_DEFAULT_ENCODING) → Tuple[RawConfigurationDict, RawConfigurationDict, RawConfigurationDict][source]

static save_configs(data_cfg: RawConfigurationDict, model_cfg: RawConfigurationDict, runtime_cfg: RawConfigurationDict, dst_path, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, encoding=_DEFAULT_ENCODING) → None[source]

class FedEval.config.configuration._CfgSerializer[source]

Bases: enum.Enum

types of serializer for configurations.

YAML = 'yaml'

JSON = 'json'

class FedEval.config.configuration._CfgFileInterface[source]

Bases: abc.ABC

an interface that regulates the methods used to serialize and deserialize configuraitons from the file system.

abstract static from_files(from_config_path: str, serializer: str | _CfgSerializer = _CfgSerializer.YAML, encoding=_DEFAULT_ENCODING) → ConfigurationManagerInterface[source]

abstract to_files(dst_dir_path: str, serializer: str | _CfgSerializer = _CfgSerializer.YAML, encoding: str | None = None) → None[source]

static serializer2enum(serializer: str | _CfgSerializer) → _CfgSerializer[source]: convert serializer name(string) into enum type

class FedEval.config.configuration._RoledConfigurationInterface[source]

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

abstract property role: FedEval.config.role.Role

class FedEval.config.configuration.ConfigurationManager(data_config: RawConfigurationDict = _DEFAULT_D_CFG, model_config: RawConfigurationDict = _DEFAULT_MDL_CFG, runtime_config: RawConfigurationDict = _DEFAULT_RT_CFG, thread_safe: bool = False)[source]

Bases: FedEval.config.singleton.Singleton, ConfigurationManagerInterface, ClientConfigurationManagerInterface, ServerConfigurationManagerInterface, _CfgYamlInterface, _CfgJsonInterface, _CfgFileInterface, _RoledConfigurationInterface

the base class of singletons. Each cls on the inheritance tree can own only one instance.

property data_unique_id

property config_unique_id

property data_dir_name: str

The output directory of the clients’ data.

Returns:: the name of the data directory.
Return type:: str

property log_dir_path: str: the path of the base of log directory.

property history_record_path: str: the path of the history record.

property job_id: str

property encoding: str: the encoding scheme during (de)serialization.

property data_config_filename: str

property model_config_filename: str

property runtime_config_filename: str

property data_config: _DataConfig

property model_config: _ModelConfig

property runtime_config: _RuntimeConfig

property num_of_train_clients_contacted_per_round: int: the number of clients selected to participate the main federated process in each round.

property num_of_eval_clients_contacted_per_round: int: the number of clients selected to participate the main federated process in each round.

property role: FedEval.config.role.Role

return the role of this runtime entity.

Raises:: AttributeError – called without role configured.
Returns:: the role of this runtime entity.
Return type:: Role

__init_once_lock

__initiated = False

_init_file_names(data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML) → None[source]

classmethod generate_unique_id(data_config: dict, model_config: dict, runtime_config: dict)[source]

static _get_md5(config_string)[source]

static load_configs(src_path, serializer: str | _CfgSerializer = _CfgSerializer.YAML, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, encoding=_DEFAULT_ENCODING) → Tuple[RawConfigurationDict, RawConfigurationDict, RawConfigurationDict][source]

static save_configs(data_cfg: RawConfigurationDict, model_cfg: RawConfigurationDict, runtime_cfg: RawConfigurationDict, dst_path, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, encoding=_DEFAULT_ENCODING, serializer: str | _CfgSerializer = _CfgSerializer.YAML)[source]

static from_files(src_path: str, data_config_filename: str = DEFAULT_D_CFG_FILENAME_YAML, model_config_filename: str = DEFAULT_MDL_CFG_FILENAME_YAML, runtime_config_filename: str = DEFAULT_RT_CFG_FILENAME_YAML, serializer: str | _CfgSerializer = _CfgSerializer.YAML, encoding=_DEFAULT_ENCODING)[source]

to_files(dst_dir_path: str, serializer: str | _CfgSerializer = _CfgSerializer.YAML, encoding: str | None = None) → None[source]

__init_role() → None

5.1.1.3.1.1. FedEval.config.configuration

5.1.1.3.1.1.1. Module Contents

5.1.1.3.1.1.1.1. Classes

5.1.1.3.1.1.1.2. Attributes

5.1.1.3.1.1. `FedEval.config.configuration`