Type BasicDecoder
Namespace tensorflow.contrib.seq2seq
Parent Decoder
Interfaces IBasicDecoder
Basic sampling decoder.
Methods
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize
 finalize_dyn
 initialize
 step
 step
 step
 step
 step_dyn
Properties
Public instance methods
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, BeamSearchDecoderState final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, bool final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, bool final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, object final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, object final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, IGraphNodeBase final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, IGraphNodeBase final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, int final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, int final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, BeamSearchDecoderState final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, object final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, bool final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(FinalBeamSearchDecoderOutput outputs, bool final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

FinalBeamSearchDecoderOutput
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, object final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, object final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, IGraphNodeBase final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, IGraphNodeBase final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, int final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, int final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, BeamSearchDecoderState final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, BeamSearchDecoderState final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, bool final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(IEnumerable<object> outputs, bool final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

IEnumerable<object>
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

bool
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, int final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, int final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

int
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, IGraphNodeBase final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, IGraphNodeBase final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

IGraphNodeBase
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, object final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, BeamSearchDecoderState final_state, int sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

int
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
ValueTuple<object, object> finalize(PythonClassContainer outputs, BeamSearchDecoderState final_state, IGraphNodeBase sequence_lengths)
Called after decoding iterations complete.
Parameters

PythonClassContainer
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

BeamSearchDecoderState
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

IGraphNodeBase
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

ValueTuple<object, object>
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
object finalize_dyn(object outputs, object final_state, object sequence_lengths)
Called after decoding iterations complete.
Parameters

object
outputs  RNNCell outputs (possibly nested tuple of) tensor[s] for all time steps.

object
final_state  RNNCell final state (possibly nested tuple of) tensor[s] for last time step.

object
sequence_lengths  1D `int32` tensor containing lengths of each sequence.
Returns

object
 `(final_outputs, final_state)`: `final_outputs` is an object containing the final decoder output, `final_state` is a (structure of) state tensors and TensorArrays.
object initialize(string name)
Initialize the decoder.
Parameters

string
name  Name scope for any created operations.
Returns

object
 `(finished, first_inputs, initial_state)`.
object step(IGraphNodeBase time, object inputs, object state, string name)
Perform a decoding step.
Parameters

IGraphNodeBase
time  scalar `int32` tensor.

object
inputs  A (structure of) input tensors.

object
state  A (structure of) state tensors and TensorArrays.

string
name  Name scope for any created operations.
Returns

object
 `(outputs, next_state, next_inputs, finished)`.
object step(IGraphNodeBase time, object inputs, BeamSearchDecoderState state, string name)
Perform a decoding step.
Parameters

IGraphNodeBase
time  scalar `int32` tensor.

object
inputs  A (structure of) input tensors.

BeamSearchDecoderState
state  A (structure of) state tensors and TensorArrays.

string
name  Name scope for any created operations.
Returns

object
 `(outputs, next_state, next_inputs, finished)`.
object step(IGraphNodeBase time, BeamSearchDecoderState inputs, object state, string name)
Perform a decoding step.
Parameters

IGraphNodeBase
time  scalar `int32` tensor.

BeamSearchDecoderState
inputs  A (structure of) input tensors.

object
state  A (structure of) state tensors and TensorArrays.

string
name  Name scope for any created operations.
Returns

object
 `(outputs, next_state, next_inputs, finished)`.
object step(IGraphNodeBase time, BeamSearchDecoderState inputs, BeamSearchDecoderState state, string name)
Perform a decoding step.
Parameters

IGraphNodeBase
time  scalar `int32` tensor.

BeamSearchDecoderState
inputs  A (structure of) input tensors.

BeamSearchDecoderState
state  A (structure of) state tensors and TensorArrays.

string
name  Name scope for any created operations.
Returns

object
 `(outputs, next_state, next_inputs, finished)`.
object step_dyn(object time, object inputs, object state, object name)
Perform a decoding step.
Parameters

object
time  scalar `int32` tensor.

object
inputs  A (structure of) input tensors.

object
state  A (structure of) state tensors and TensorArrays.

object
name  Name scope for any created operations.
Returns

object
 `(outputs, next_state, next_inputs, finished)`.
Public properties
object batch_size get;
The batch size of input values.
object batch_size_dyn get;
The batch size of input values.
object output_dtype get;
A (possibly nested tuple of...) dtype[s].
object output_dtype_dyn get;
A (possibly nested tuple of...) dtype[s].
object output_size get;
A (possibly nested tuple of...) integer[s] or `TensorShape` object[s].
object output_size_dyn get;
A (possibly nested tuple of...) integer[s] or `TensorShape` object[s].