org.platanios.tensorflow.api.ops.seq2seq.decoders.BasicDecoder
Function that takes an INT32
vector of IDs and returns the corresponding embedded values
that will be passed to the decoder input.
INT32
vector with length equal to the batch size, which contains the begin-of-sequence
token IDs.
INT32
scalar containing the end-of-sequence token ID (i.e., token ID which marks the end of
decoding).
Scalar INT32
tensor representing the batch size of a tensor returned by sample()
.
Scalar INT32
tensor representing the batch size of a tensor returned by sample()
.
INT32
vector with length equal to the batch size, which contains the begin-of-sequence
token IDs.
Function that takes an INT32
vector of IDs and returns the corresponding embedded values
that will be passed to the decoder input.
INT32
scalar containing the end-of-sequence token ID (i.e., token ID which marks the end of
decoding).
Returns a tuple containing: (i) a scalar BOOLEAN
tensor specifying whether initialization has finished, and
(ii) the next input.
Returns a tuple containing: (i) a scalar BOOLEAN
tensor specifying whether initialization has finished, and
(ii) the next input.
Returns a tuple containing: (i) a scalar BOOLEAN
tensor specifying whether sampling has finished, and
(ii) the next RNN cell tuple.
Returns a tuple containing: (i) a scalar BOOLEAN
tensor specifying whether sampling has finished, and
(ii) the next RNN cell tuple.
Returns a sample for the provided time, input, and state.
Returns a sample for the provided time, input, and state.
Returns a zero-valued sample for this helper.
Returns a zero-valued sample for this helper.
RNN decoder helper to be used while performing inference. It uses the argmax over the RNN output logits and passes the result through an embedding layer to get the next input.
Function that takes an
INT32
vector of IDs and returns the corresponding embedded values that will be passed to the decoder input.INT32
vector with length equal to the batch size, which contains the begin-of-sequence token IDs.INT32
scalar containing the end-of-sequence token ID (i.e., token ID which marks the end of decoding).