tf.contrib.seq2seq.dynamic_decode(
decoder,
output_time_major=False,
impute_finished=False,
maximum_iterations=None,
parallel_iterations=32,
swap_memory=False,
scope=None
)
Defined in tensorflow/contrib/seq2seq/python/ops/decoder.py
.
Perform dynamic decoding with decoder
.
Calls initialize() once and step() repeatedly on the Decoder object.
Args:
decoder
: ADecoder
instance.output_time_major
: Python boolean. Default:False
(batch major). IfTrue
, outputs are returned as time major tensors (this mode is faster). Otherwise, outputs are returned as batch major tensors (this adds extra time to the computation).impute_finished
: Python boolean. IfTrue
, then states for batch entries which are marked as finished get copied through and the corresponding outputs get zeroed out. This causes some slowdown at each time step, but ensures that the final state and outputs have the correct values and that backprop ignores time steps that were marked as finished.maximum_iterations
:int32
scalar, maximum allowed number of decoding steps. Default isNone
(decode until the decoder is fully done).parallel_iterations
: Argument passed totf.while_loop
.swap_memory
: Argument passed totf.while_loop
.scope
: Optional variable scope to use.
Returns:
(final_outputs, final_state, final_sequence_lengths)
.
Raises:
TypeError
: ifdecoder
is not an instance ofDecoder
.ValueError
: ifmaximum_iterations
is provided but is not a scalar.