vak.transforms.defaults.get.get_default_transform

vak.transforms.defaults.get.get_default_transform(model_name: str, mode: str, transform_kwargs: dict)

Get default transforms for a model, according to its family and what mode the model is being used in.

Parameters:

• model_name (str) – Name of model.

• mode (str) – one of {‘train’, ‘eval’, ‘predict’}. Determines set of transforms.

Returns:

transform, target_transform – one or more vak transforms to be applied to inputs x and, during training, the target y. If more than one transform, they are combined into an instance of torchvision.transforms.Compose. Note that when mode is ‘predict’, the target transform is None.

Return type:

callable