Skip to content

Instantly share code, notes, and snippets.

@hadifar

hadifar/simple_rnn_0.py

Created December 6, 2018 22:49
Show Gist options
  • Save hadifar/1a0dbe2a1bd443258831ac0b8e5ca64f to your computer and use it in GitHub Desktop.
Save hadifar/1a0dbe2a1bd443258831ac0b8e5ca64f to your computer and use it in GitHub Desktop.
Download ZIP
Raw
simple_rnn_0.py
# our RNN variables
Wx = tf.get_variable(name='Wx', shape=[embedding_size, rnn_size])
Wh = tf.get_variable(name='Wh', shape=[rnn_size, rnn_size])
bias_rnn = tf.get_variable(name='brnn', initializer=tf.zeros([rnn_size]))
def rnn_step(prev_hidden_state, x):
return tf.tanh(tf.matmul(x, Wx) + tf.matmul(prev_hidden_state, Wh) + bias_rnn)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment