gpytorchwrapper.src.config.config_classes

Functions

create_config(config_dict)

Create a Config object from a nested configuration dictionary.

Classes

Config(data_conf, transform_conf, ...)
DataConf(num_inputs, num_outputs[, output_index])
LikelihoodConf(likelihood_class, ...)
ModelConf(model_class, model_options)
OptimizerConf(optimizer_class, optimizer_options)
TestingConf([test, test_size, ...])
TrainingConf(model, likelihood, ...)
TransformConf(transform_input, transform_output)
TransformerConf(transform_data, ...)

class gpytorchwrapper.src.config.config_classes.Config(data_conf: gpytorchwrapper.src.config.config_classes.DataConf, transform_conf: gpytorchwrapper.src.config.config_classes.TransformConf, training_conf: gpytorchwrapper.src.config.config_classes.TrainingConf, testing_conf: gpytorchwrapper.src.config.config_classes.TestingConf)

Bases: object

data_conf: DataConf

testing_conf: TestingConf

training_conf: TrainingConf

transform_conf: TransformConf

class gpytorchwrapper.src.config.config_classes.DataConf(num_inputs: int, num_outputs: int, output_index: int | list[int] | NoneType = None)

Bases: object

num_inputs: int

num_outputs: int

output_index: int | list[int] | None = None

class gpytorchwrapper.src.config.config_classes.LikelihoodConf(likelihood_class: str = 'GaussianLikelihood', likelihood_options: Optional[dict] = <factory>)

Bases: object

likelihood_class: str = 'GaussianLikelihood'

likelihood_options: dict | None

class gpytorchwrapper.src.config.config_classes.ModelConf(model_class: str, model_options: Optional[dict] = <factory>)

Bases: object

model_class: str

model_options: dict | None

class gpytorchwrapper.src.config.config_classes.OptimizerConf(optimizer_class: str = 'Adam', optimizer_options: Optional[dict] = <factory>)

Bases: object

optimizer_class: str = 'Adam'

optimizer_options: dict | None

class gpytorchwrapper.src.config.config_classes.TestingConf(test: bool = False, test_size: float = 0.2, strat_shuffle_split: bool = False, kfold: bool = False, kfold_bins: int | None = None)

Bases: object

kfold: bool = False

kfold_bins: int | None = None

strat_shuffle_split: bool = False

test: bool = False

test_size: float = 0.2

class gpytorchwrapper.src.config.config_classes.TrainingConf(model: gpytorchwrapper.src.config.config_classes.ModelConf = <factory>, likelihood: gpytorchwrapper.src.config.config_classes.LikelihoodConf = <factory>, learning_iterations: int = 100, botorch: Optional[bool] = False, debug: Optional[bool] = True, optimizer: gpytorchwrapper.src.config.config_classes.OptimizerConf = <factory>)

Bases: object

botorch: bool | None = False

debug: bool | None = True

learning_iterations: int = 100

likelihood: LikelihoodConf

model: ModelConf

optimizer: OptimizerConf

class gpytorchwrapper.src.config.config_classes.TransformConf(transform_input: gpytorchwrapper.src.config.config_classes.TransformerConf = <factory>, transform_output: gpytorchwrapper.src.config.config_classes.TransformerConf = <factory>)

Bases: object

transform_input: TransformerConf

transform_output: TransformerConf

class gpytorchwrapper.src.config.config_classes.TransformerConf(transform_data: bool = False, transformer_class: str = 'DefaultTransformer', transformer_options: Optional[dict] = <factory>, columns: Optional[list[int]] = None)

Bases: object

columns: list[int] | None = None

transform_data: bool = False

transformer_class: str = 'DefaultTransformer'

transformer_options: dict | None

gpytorchwrapper.src.config.config_classes.create_config(config_dict: dict) → Config

Create a Config object from a nested configuration dictionary.

This function initializes a Config dataclass using values from the provided config_dict. Optional fields not specified in the dictionary are populated with default values.

Parameters:

config_dict (dict) –

A nested dictionary with the following structure:

• data_conf :

  • num_inputsint

    Number of input features.

  • num_outputsint

    Number of output targets.

  • output_indexint or list of int, optional

    Index or indices of outputs to use.

• transform_confdict, optional

  • transform_inputdict

    • transform_data : bool, default False

    • transformer_class : str, default “DefaultTransformer”

    • transformer_options : dict, default {}

    • columns : list of int, optional

  • transform_outputdict

    Same structure as transform_input.

• training_confdict

  • modeldict

    • model_class : str

    • model_options : dict, default {}

  • likelihooddict, optional

    • likelihood_class : str, default “GaussianLikelihood”

    • likelihood_options : dict, default {}

  • learning_iterations : int, default 100

  • botorch : bool, default False

  • debug : bool, default True

  • optimizerdict, optional

    • optimizer_class : str, default “Adam”

    • optimizer_options : dict, default {“lr”: 0.1}

• testing_confdict, optional

  • test : bool, default False

  • test_size : float, default 0.2

  • strat_shuffle_split : bool, default False

  • kfold : bool, default False

  • kfold_bins : int, optional

Returns:

A fully populated Config dataclass instance, with missing optional values filled in using defaults.

Return type:

Config