def _initialize_gru_cell(self, num_units, trainable):
return gru_cell.LayerNormGRUCell(
num_units,
w_initializer=self.uniform_initializer,
u_initializer=random_orthonormal_initializer,
b_initializer=tf.constant_initializer(0.0),
trainable=trainable)
def _initialize_gru_cell(self, num_units):
"""Initializes a GRU cell.
The Variables of the GRU cell are initialized in a way that exactly matches
the skipThoughts paper: recurrent weights are initialized from random
orthonormal matrices and non-recurrent weights are initialized from random
uniform matrices.
Args:
num_units: Number of output units.
Returns:
cell: An instance of RNNCell with variable initializers that match the
skipThoughts paper.
"""
return gru_cell.LayerNormGRUCell(
num_units,
w_initializer=self.uniform_initializer,
u_initializer=random_orthonormal_initializer,
b_initializer=tf.constant_initializer(0.0))