Gated Recurrent Unit - Cho et al. More...

Inheritance diagram for keraflow.layers.recurrent.GRU:

Public Member Functions
def	__init__

Public Member Functions inherited from keraflow.layers.recurrent.Recurrent
def	__init__

Detailed Description

Gated Recurrent Unit - Cho et al.

2014.

input_shape: 3D, (nb_samples, sequence_length, input_dim)
output_shape (return_sequences=True): 3D, (nb_samples, sequence_length, output_dim)
output_shape (return_sequences=False): 2D, (nb_samples, output_dim)
parameters:
- W_z, W_r, W_h: (input_dim, output_dim)
- U_z, U_r, U_h: (output_dim, output_dim)
- b_z, b_r, b_h: (output_dim,)
References:
- On the Properties of Neural Machine Translation: Encoder-Decoder Approaches
- Emperical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

Constructor & Destructor Documentation

def keraflow.layers.recurrent.GRU.__init__	(	self,
		output_dim,
		init = `'glorot_uniform'`,
		inner_init = `'orthogonal'`,
		activation = `'tanh'`,
		inner_activation = `'hard_sigmoid'`,
		dropout_W = `0.`,
		dropout_U = `0.`,
		return_sequences = `False`,
		go_backwards = `False`,
		stateful = `False`,
		unroll = `False`,
		kwargs
	)

Parameters

output_dim	int. The output dimension of the layer.
init	str/function. Function to initialize `W_z`,`W_r`,`W_h`, (input to hidden transformation). See Initializations.
inner_init	str/function. Function to initialize `U_z`, `U_r`, `U_h` (hidden to hidden transformation). See Initializations.
activation	str/function. Activation function applied on the output. See Activations.
inner_activation	str/function. Activation function applied on the inner cells. See Activations.
dropout_W	float between 0 and 1. Fraction of the input units to drop for input gates.
dropout_U	float between 0 and 1. Fraction of the input units to drop for recurrent connections.
return_sequences	Boolean. See Recurrent.__init__
go_backwards	Boolean. See Recurrent.__init__
stateful	Boolean. See Recurrent.__init__
unroll	Boolean. See Recurrent.__init__
kwargs	see Layer.__init__.