augmentation_pipeline

AugmentationPipeline(layers: list[keras.Layer], name=None, force_training=False, **kwargs={})

Apply layers in order to a tensor or structured sample.

Each child receives the preceding child’s output and the training flag. Layers must accept training and, when supplied, transformations. Calling with transformations requires one entry per layer; use None for a child that should sample its own parameters.

Parameters

Parameters of AugmentationPipeline
Name	Type	Default	Description
`layers`	`list[keras.Layer]`	Required	Ordered layers to apply. The layer objects are retained.
`name`	`str \| None`	None	Optional Keras layer name.
`force_training`	`bool`	False	Pass training=True to every child, including at inference.
`**kwargs`	`Any`	{}	Additional Keras Layer arguments.

Raises

Errors raised by AugmentationPipeline
Type	Description
`ValueError`	The supplied transformations count differs from the layer count.

layers = layers

force_training = force_training

call(inputs, training=None, transformations=None)

get_config()

from_config(config)

classmethod