class for the beam search state
More...
class for the beam search state
- cell_states: the cell states as a tuple of [batch_size x beam_width x s] tensor
- alignment_history: the alignments history as a TensorArray
- predicted_ids: a [batch_size x beam_width] tensorArray containing the predicted ids
- logprobs: a [batch_size x beam_width] tensor containing, the log probability of the beam elements
- lengths: a [batch_size x beam_width] tensor containing the lengths
- finished: a [batch_size x beam_width] tensor containing which elements have finished
The documentation for this class was generated from the following file:
- components/beam_search_decoder.py