tf.contrib.cudnn_rnn.CudnnLSTMSaveable

SaveableObject implementation handling Cudnn LSTM opaque params.

tf.contrib.cudnn_rnn.CudnnLSTMSaveable(
    opaque_params, num_layers, num_units, input_size,
    input_mode=CUDNN_INPUT_LINEAR_MODE, direction=CUDNN_RNN_UNIDIRECTION,
    scope=None, name='cudnn_rnn_saveable'
)

Args
opaque_params	a variable, Cudnn RNN opaque params.
num_layers	the number of layers for the RNN model.
num_units	the number of units within the RNN model.
input_size	the size of the input, it could be different from the num_units.
input_mode	indicate whether there is a linear projection between the input and the actual computation before the first layer. It could 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. Could be either 'unidirectional' or 'bidirectional'
scope	string of VariableScope, the scope of equivalent subgraph consisting only platform-independent tf RNN cells.
name	the name of the CudnnOpaqueParamsSaveable object.

Attributes
device	The device for SaveSpec Tensors.
format_converter
optional_restore	A hint to restore assertions that this object is optional.

Methods

restore

View source

restore(
    restored_tensors, restored_shapes
)

Restores this object from 'restored_tensors'.

Args
restored_tensors	the tensors that were loaded from a checkpoint
restored_shapes	the shapes this object should conform to after restore, or None.

Returns
An operation that restores the state of the object.

Raises
ValueError	If the object cannot be restored using the provided parameters.