brainmass.datasets.register_dataset

brainmass.datasets.register_dataset(name, loader, *, description=None)

Register a named dataset loader.

Parameters:

• name (str) – The unique name under which the dataset is looked up by load_dataset().

• loader (Callable[[], object]) – A zero-argument callable returning the dataset when called. It is invoked lazily, on each load_dataset() call.

• description (str) – A one-line human-readable description shown by list_datasets(). Defaults to an empty string.

Raises:

• TypeError – If loader is not callable.

• ValueError – If a dataset with the same name is already registered.

Return type:

None

See also

load_dataset

Look up and call a registered loader.

list_datasets

List the registered datasets.

Return type:

None

Parameters:

• name (str)

• loader (Callable[[], object])

• description (str)

Examples

>>> from brainmass import datasets
>>> datasets.register_dataset('ones', lambda: [1, 1, 1], description='a list')
>>> datasets.load_dataset('ones')
[1, 1, 1]
>>> datasets._REGISTRY.pop('ones')
(...)