Encoder and Decoder RNN

[ShishirPatil]

The paper - Semi-supervised Sequence Learning(https://arxiv.org/abs/1511.01432) - states that for training of SA-LSTM, they used the same LSTM for both encoding and for decoding. However, this implementation uses 2 LSTMs - encoder_cell, and decoder_cell. Could you please clarify.

Thanks,
Shishir

[miguelmartin75]

Likely was just a mistake on the author's (github author) part. Seems to work regardless, just requires more parameters.