dlpy.blocks.Bidirectional

class dlpy.blocks.Bidirectional(n, n_blocks=1, rnn_type='gru', output_type='samelength', dropout=0.2, max_output_length=None, src_layers=None, name=None)

Bases: object

Bidirectional RNN layers

Parameters:
n : int or list of int

Specifies the number of neurons in the recurrent layer. If n_blocks=1, then n should be an int. If n_blocks > 1, then n can be an int or a list of ints to indicate the number of neurons in each block.

n_blocks : int, optional

Specifies the number of bidirectional recurrent layer blocks.
Default: 1

rnn_type : string, optional

Specifies the type of the rnn layer.
Default: GRU
Valid Values: RNN, LSTM, GRU

output_type : string, optional

Specifies the output type of the recurrent layer.
Default: SAMELENGTH
Valid Values: ENCODING, SAMELENGTH, ARBITRARYLENGTH

max_output_length : int, mostly optional

Specifies the maximum number of tokens to generate when the outputType parameter is set to ARBITRARYLENGTH.

dropout : float, optional

Specifies the dropout rate.
Default: 0.2

src_layers : list, optional

Specifies the list of source layers for the layer.

name : string, optional

Specifies layer names. If not specified, ‘RNN’ is used

Returns:
:class:`Bidirectional’
__init__(n, n_blocks=1, rnn_type='gru', output_type='samelength', dropout=0.2, max_output_length=None, src_layers=None, name=None)

Initialize self. See help(type(self)) for accurate signature.

Methods

__init__(n[, n_blocks, rnn_type, …]) Initialize self.
add_layers() Add layers for the block
compile([block_num]) Convert the options into DLPy layer definition.
get_last_layers() Return last two layers, if they exist
get_layers() Return list of layers