If you're using TF 1.2.1 or above, you want to replace this line: cell = tf.cont... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

yablak on Aug 22, 2017 | parent | context | favorite | on: Show HN: RNN that generates Game of Thrones text

If you're using TF 1.2.1 or above, you want to replace this line:

cell = tf.contrib.rnn.MultiRNNCell([drop_cell] * num_layers)

with:

cell = tf.contrib.rnn.MultiRNNCell([create_cell() for _ in range(num_layers)])

otherwise you'll be using the same weights for each LSTM layer and your model is less expressive.

zthoutt on Aug 22, 2017 [–]

Yep, I was using TF 1.0.0. I will probably upgrade in the future and will keep this in mind.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact