get_variable_ddi

tfsnippet.utils.get_variable_ddi(name, initial_value, shape=None, dtype=tf.float32, initializing=False, regularizer=None, constraint=None, trainable=True, collections=None, **kwargs)

Wraps tf.get_variable() to support data-dependent initialization.

Parameters:

• name – Name of the variable.
• initial_value – The data-dependent initial value of the variable.
• shape – Shape of the variable.
• dtype – Data type of the variable.
• initializing (bool) – Whether or not it is building the graph for data-dependent initialization? Ignored if initial_value is absent.
• regularizer – Regularizer of the variable.
• constraint – Constraint of the variable.
• trainable (bool) – Whether or not to the variable is trainable?
• collections (Iterable[str]) – Add the variable to these collections.
• **kwargs – Other named parameters passed to tf.get_variable().

Returns:

The variable or the tensor.

Return type:

tf.Variable or tf.Tensor