I used tf.contrib layer to write recurrent neural network in TensorFlow. I made LSTM cell type first and extract the output and states by passing this cell into another layer. But in TensorFlow 2.x it seems like it can be done in a single line
output, state_h, state_c = layers.LSTM(self.args.embedding_size, return_state=True, name="encoder")(tf.nn.embedding_lookup(self.embeddings, self.neighborhood_placeholder)
and I can't apply dropout warpper like in tensorflow 1.x. How may I convert the following codes into tensorflow 2.x?
with tf.variable_scope('LSTM'):
cell = tf.contrib.rnn.DropoutWrapper(
tf.contrib.rnn.LayerNormBasicLSTMCell(num_units=self.args.embedding_size, layer_norm=False),
input_keep_prob=1.0, output_keep_prob=1.0)
_, states = tf.nn.dynamic_rnn(
cell,
tf.nn.embedding_lookup(self.embeddings, self.neighborhood_placeholder),
dtype=tf.float32,
sequence_length=self.seqlen_placeholder)
self.lstm_output = states.h
Replace tf.contrib.rnn.DropoutWrapper
with tf.compat.v1.nn.rnn_cell.DropoutWrapper
.
Replace tf.contrib.rnn.LayerNormBasicLSTMCell
with tf.compat.v1.nn.rnn_cell.LSTMCell
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.