EncoderRNN¶
-
class
EncoderRNN(word_dim, hidden_size, rnn_layers=6, dropout=0.2)[source]¶ EncoderRNN model to be used in the encoder of the RST Parser network.
-
forward(input_embeddings: transformers.tokenization_utils_base.BatchEncoding, input_lengths: List[int]) → Tuple[torch.Tensor, torch.Tensor][source]¶ Forward pass for encoder RNN.
- Parameters
input_embeddings (BatchEncoding) – input embeddings from the RST preprocessor.
input_lengths (List[int]) – list of input lengths.
- Returns
return outputs and hidden states of encoder.
- Return type
Tuple[torch.Tensor, torch.Tensor]
-