thsolver.registry

This module contains the lightweight registries used to map config names to Python factory functions.

register_model(fn)

Registers a model factory by its function name.

Parameters:

fn (callable) – The model factory to register.

Returns:

The input factory, which keeps decorator usage convenient.

Return type:

callable

model_entrypoints(name)

Returns the registered model factory with the given name.

Parameters:

name (str) – The model name.

is_model(name)

Checks whether a model factory has been registered.

Parameters:

name (str) – The model name.

list_models()

Returns all registered model names.

build_model(config, **kwargs)

Builds a registered model from a config node.

Parameters:

• config – A config node containing the field name.

• **kwargs – Additional arguments forwarded to the model factory.

Returns:

The model created by the registered factory.

Return type:

object

register_dataset(fn)

Registers a dataset factory by its function name.

Parameters:

fn (callable) – The dataset factory to register.

Returns:

The input factory, which keeps decorator usage convenient.

Return type:

callable

dataset_entrypoints(name)

Returns the registered dataset factory with the given name.

Parameters:

name (str) – The dataset name.

is_dataset(name)

Checks whether a dataset factory has been registered.

Parameters:

name (str) – The dataset name.

list_datasets()

Returns all registered dataset names.

build_dataset(config, **kwargs)

Builds a registered dataset from a config node.

Parameters:

• config – A config node containing the field name.

• **kwargs – Additional arguments forwarded to the dataset factory.

Returns:

The dataset tuple or object created by the registered factory.

Return type:

object