random_choice

Batchwise random layer choice with captured branches and owned RNG.

Machine-readable model

• RandomChoiceclassChoose one layer for a whole batch; branches must have compatible outputs.

class

RandomChoice

Python

Choose one layer for a whole batch; branches must have compatible outputs.

helia_edge/layers/preprocessing/random_choice.py:9

RandomChoice(layers: list[keras.Layer], batchwise: bool = True, **kwargs={})

Choose one layer for a whole batch; branches must have compatible outputs.

Per-example layer choice was never implemented and remains unsupported. The next-major default is the supported batchwise=True mode.

Base class: BaseAugmentation.

Inherited from BaseAugmentation: augment_masks(), augment_sample(), augment_samples(), augment_targets(), get_random_transformations().

attribute

layers

Python

helia_edge/layers/preprocessing/random_choice.py:23

layers = layers

attribute

batchwise

Python

helia_edge/layers/preprocessing/random_choice.py:24

batchwise = batchwise

method

build

Python

helia_edge/layers/preprocessing/random_choice.py:26

build(input_shape)

method

batch_augment

Python

helia_edge/layers/preprocessing/random_choice.py:32

batch_augment(inputs, transformations=None)

method

call

Python

helia_edge/layers/preprocessing/random_choice.py:41

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

method

get_config

Python

helia_edge/layers/preprocessing/random_choice.py:49

get_config()

method

from_config

Python

helia_edge/layers/preprocessing/random_choice.py:56

from_config(config)

classmethod