forward compatible RNNs

network.py

@@ -161,10 +161,10 @@ def _zero_pad(inputs):
 def _rnn(acts, input_dim, cell_type, scope=None):
  if cell_type == 'gru':
  logger.info("Adding cell type " + cell_type + " to rnn")
- cell = tf.nn.rnn_cell.GRUCell(input_dim)
+ cell = tf.contrib.rnn.GRUCell(input_dim)
  elif cell_type == 'lstm':
  logger.info("Adding cell type " + cell_type + " to rnn")
- cell = tf.nn.rnn_cell.LSTMCell(input_dim)
+ cell = tf.contrib.LSTMCell(input_dim)
  else:
  msg = "Invalid cell type {}".format(cell_type)
  raise ValueError(msg)
@@ -185,11 +185,11 @@ def _bi_rnn(acts, input_dim, cell_type):
 
  # Backwards
  with tf.variable_scope("bw") as bw_scope:
- reverse_dims = [False, True, False]
- acts_bw = tf.reverse(acts, dims=reverse_dims)
+ reverse_dims = [1]
+ acts_bw = tf.reverse(acts, axis=reverse_dims)
  acts_bw = _rnn(acts_bw, input_dim, cell_type,
  scope=bw_scope)
- acts_bw = tf.reverse(acts_bw, dims=reverse_dims)
+ acts_bw = tf.reverse(acts_bw, axis=reverse_dims)
 
  # Sum the forward and backward states.
  return tf.add(acts_fw, acts_bw)