layer_pyramidal_recurrent_block: Pyramidal recurrent block

In ifrit98/layerR:

Description

Builds a block of 'num_layers' pyramidal recurrent layers, which

Usage

layer_pyramidal_recurrent_block(
  object,
  units,
  num_layers = 3,
  cell_type = "LSTM",
  activation = "tanh",
  projection_activation = TRUE,
  projection_batchnorm = TRUE,
  kernel_initializer = "glorot_normal",
  kernel_regularizer = NULL,
  recurrent_activation = "sigmoid",
  recurrent_initializer = "orthogonal",
  recurrent_regularizer = NULL,
  bias_initializer = "zeros",
  bias_regularizer = NULL,
  recurrent_dropout = 0,
  name = NULL,
  trainable = TRUE
)

Arguments

units	Number of hidden units for each layer in the block
cell_type	string denoting type of cell to use. default: 'LSTM'
activation	activation function for each layer

References

https://arxiv.org/pdf/1808.09029.pdf

ifrit98/layerR documentation built on March 2, 2020, 8:11 a.m.