parsing

Functions

clean_namespace	Removes all unpicklable entries from hparams.
collect_init_args	Recursively collects the arguments passed to the child constructors in the inheritance tree.
get_init_args	For backwards compatibility: #16369.
is_picklable	Tests if an object can be pickled.
lightning_getattr	Special getattr for Lightning.
lightning_hasattr	Special hasattr for Lightning.
lightning_setattr	Special setattr for Lightning.
parse_class_init_keys	Parse key words for standard self, *args and **kwargs.
save_hyperparameters	See save_hyperparameters()

Classes

AttributeDict

Extended dictionary accessible with dot notation.

Utilities used for parameter parsing.

class lightning.pytorch.utilities.parsing.AttributeDict

Bases: Dict

Extended dictionary accessible with dot notation.

>>> ad = AttributeDict({'key1': 1, 'key2': 'abc'})
>>> ad.key1
1
>>> ad.update({'my-key': 3.14})
>>> ad.update(new_key=42)
>>> ad.key1 = 2
>>> ad
"key1":    2
"key2":    abc
"my-key":  3.14
"new_key": 42

lightning.pytorch.utilities.parsing.clean_namespace(hparams)

Removes all unpicklable entries from hparams.

Return type:

None

lightning.pytorch.utilities.parsing.collect_init_args(frame, path_args, inside=False, classes=())

Recursively collects the arguments passed to the child constructors in the inheritance tree.

Parameters:

• frame (FrameType) – the current stack frame

• path_args (List[Dict[str, Any]]) – a list of dictionaries containing the constructor args in all parent classes

• inside (bool) – track if we are inside inheritance path, avoid terminating too soon

• classes (Tuple[Type, ...]) – the classes in which to inspect the frames

Return type:

List[Dict[str, Any]]

Returns:

A list of dictionaries where each dictionary contains the arguments passed to the constructor at that level. The last entry corresponds to the constructor call of the most specific class in the hierarchy.

lightning.pytorch.utilities.parsing.get_init_args(frame)

For backwards compatibility: #16369.

Return type:

Dict[str, Any]

lightning.pytorch.utilities.parsing.is_picklable(obj)

Tests if an object can be pickled.

Return type:

bool

lightning.pytorch.utilities.parsing.lightning_getattr(model, attribute)

Special getattr for Lightning. Checks for attribute in model namespace, the old hparams namespace/dict, and the datamodule.

Raises:

AttributeError – If model doesn’t have attribute in any of model namespace, the hparams namespace/dict, and the datamodule.

Return type:

Optional[Any]

lightning.pytorch.utilities.parsing.lightning_hasattr(model, attribute)

Special hasattr for Lightning.

Checks for attribute in model namespace, the old hparams namespace/dict, and the datamodule.

Return type:

bool

lightning.pytorch.utilities.parsing.lightning_setattr(model, attribute, value)

Special setattr for Lightning. Checks for attribute in model namespace and the old hparams namespace/dict. Will also set the attribute on datamodule, if it exists.

Raises:

AttributeError – If model doesn’t have attribute in any of model namespace, the hparams namespace/dict, and the datamodule.

Return type:

None

lightning.pytorch.utilities.parsing.parse_class_init_keys(cls)

Parse key words for standard self, *args and **kwargs.

Return type:

Tuple[str, Optional[str], Optional[str]]

Examples

>>> class Model:
...     def __init__(self, hparams, *my_args, anykw=42, **my_kwargs):
...         pass
>>> parse_class_init_keys(Model)
('self', 'my_args', 'my_kwargs')

lightning.pytorch.utilities.parsing.save_hyperparameters(obj, *args, ignore=None, frame=None)

See save_hyperparameters()

Return type:

None