Do you know if using tf.while_loop speed things up? Using dynamic_rnn at the moment and it's _so_ slow. I'm not finding implementations using tf.while_loop, the

Do you know if using tf.while_loop speed things up? Using dynamic_rnn at the moment and it's _so_ slow. I'm not finding implementations using tf.while_loop, there's dynamic_rnn as you said but that's so convoluted to read (like TF code..).