def __init__(self,
num_layers,
num_units,
input_mode=CUDNN_INPUT_LINEAR_MODE,
direction=CUDNN_RNN_UNIDIRECTION,
dropout=0.,
seed=None,
dtype=dtypes.float32,
kernel_initializer=None,
bias_initializer=None,
name=None):
"""Creates a CudnnRNN model from model spec.
Args:
num_layers: the number of layers for the RNN model.
num_units: the number of units within the RNN model.
input_mode: indicate whether there is a linear projection between the
input and the actual computation before the first layer. It can be
'linear_input', 'skip_input' or 'auto_select'.
'linear_input' (default) always applies a linear projection of input
onto RNN hidden state. (standard RNN behavior).
'skip_input' is only allowed when input_size == num_units;
'auto_select' implies 'skip_input' when input_size == num_units;
otherwise, it implies 'linear_input'.
direction: the direction model that the model operates. Can be either
'unidirectional' or 'bidirectional'
dropout: dropout rate, a number between [0, 1]. Dropout is applied on
inputs of each layer. When set to 0, dropout is disabled.
seed: the op seed used for initializing dropout. See @{tf.set_random_seed}
for behavior.
dtype: tf.float32 or tf.float64
kernel_initializer: starting value to initialize the weight.
bias_initializer: starting value to initialize the bias
(default is all zeros).
name: VariableScope for the created subgraph; defaults to class name.
This only serves the default scope if later no scope is specified when
invoking __call__().
Raises:
ValueError: if direction is invalid. Or dtype is not supported.
"""
super(_CudnnRNN, self).__init__(dtype=dtype, name=name)
cudnn_rnn_ops.check_direction(direction)
cudnn_rnn_ops.check_input_mode(input_mode)
if dtype not in [dtypes.float32, dtypes.float64]:
raise ValueError("Only support float32, float64, provided %s" % dtype)
# Layer self.dtype is type name, the original DType object is kept here.
self._plain_dtype = dtype
self._num_layers = num_layers
self._num_units = num_units
self._input_mode = input_mode
self._direction = direction
self._dropout = dropout
self._seed = seed
self._kernel_initializer = kernel_initializer
self._bias_initializer = bias_initializer
# Init input_size to None, which will be set after build().
self._input_size = None
self._saveable = None
def __init__(self,
num_layers,
num_units,
input_mode=CUDNN_INPUT_LINEAR_MODE,
direction=CUDNN_RNN_UNIDIRECTION,
dropout=0.,
seed=None,
dtype=dtypes.float32,
kernel_initializer=None,
bias_initializer=None,
name=None):
"""Creates a CudnnRNN model from model spec.
Args:
num_layers: the number of layers for the RNN model.
num_units: the number of units within the RNN model.
input_mode: indicate whether there is a linear projection between the
input and the actual computation before the first layer. It can be
'linear_input', 'skip_input' or 'auto_select'.
'linear_input' (default) always applies a linear projection of input
onto RNN hidden state. (standard RNN behavior).
'skip_input' is only allowed when input_size == num_units;
'auto_select' implies 'skip_input' when input_size == num_units;
otherwise, it implies 'linear_input'.
direction: the direction model that the model operates. Can be either
'unidirectional' or 'bidirectional'
dropout: dropout rate, a number between [0, 1]. Dropout is applied on
inputs of each layer. When set to 0, dropout is disabled.
seed: the op seed used for initializing dropout. See @{tf.set_random_seed}
for behavior.
dtype: tf.float32 or tf.float64
kernel_initializer: starting value to initialize the weight.
bias_initializer: starting value to initialize the bias
(default is all zeros).
name: VariableScope for the created subgraph; defaults to class name.
This only serves the default scope if later no scope is specified when
invoking __call__().
Raises:
ValueError: if direction is invalid. Or dtype is not supported.
"""
super(_CudnnRNN, self).__init__(dtype=dtype, name=name)
cudnn_rnn_ops.check_direction(direction)
cudnn_rnn_ops.check_input_mode(input_mode)
if dtype not in [dtypes.float32, dtypes.float64]:
raise ValueError("Only support float32, float64, provided %s" % dtype)
# Layer self.dtype is type name, the original DType object is kept here.
self._plain_dtype = dtype
self._num_layers = num_layers
self._num_units = num_units
self._input_mode = input_mode
self._direction = direction
self._dropout = dropout
self._seed = seed
self._kernel_initializer = kernel_initializer
self._bias_initializer = bias_initializer
# Init input_size to None, which will be set after build().
self._input_size = None
self._saveable = None
