twodlearn.core.autoinit module¶

class twodlearn.core.autoinit.AutoConstant[source]¶

auto initialize properties as tensorflow constants

class twodlearn.core.autoinit.AutoConstantVariable[source]¶

auto initialize properties as non-trainable vairables

class twodlearn.core.autoinit.AutoInit[source]¶

Bases: object

Indicates that the property should be auto initialized

Example:

TdlModel(prop=AutoInit())  # Runs auto initialization for prop

If the property initializer accepts AutoType, the Type can be provided using a tuple:

TdlModel(prop=(AutoInit(), AutoType))

class twodlearn.core.autoinit.AutoNormalVar(mean, stddev)[source]¶: Bases: twodlearn.core.autoinit.AutoinitType

class twodlearn.core.autoinit.AutoPlaceholder[source]¶: Bases: twodlearn.core.autoinit.AutoinitType

class twodlearn.core.autoinit.AutoTensor[source]¶

auto initialize properties as tensorflow Tensors

class twodlearn.core.autoinit.AutoTrainable[source]¶

auto initialize properties as trainable vairables

class twodlearn.core.autoinit.AutoVariable(initializer=None)[source]¶

auto initialize properties as variables

If an initializer is provided, then shape must be specified:

init = AutoVariable(initializer=tf.keras.initializer.glorot_uniform())
var = init(shape=shape)

Otherwise, calling AutoVariable expects an initial value or an initializer

class twodlearn.core.autoinit.AutoZeros[source]¶: Bases: twodlearn.core.autoinit.AutoinitType

class twodlearn.core.autoinit.AutoinitType[source]¶

Bases: object

Base class to identify auto initializers