implementation of Paper, Appendix A.1, Equation 3

[nejlag]

Hi Mark,

Thanks for sharing the code. I've a question regarding one tiny bit of your implementation:

In model.py, line 71 , i.e. prev_state = rnn.LSTMStateTuple(state[0], tf.add(h_k, r_k), you added h_k and r_k, but in paper, Appendix A.1, Equation 3, it seems h_k and r_k are concatenated. Could you please elaborate on this?!

Thank you...