changed dropout to placeholder, resolved #27

models/sentiment.py

@@ -20,6 +20,7 @@ def __init__(self, vocab_size, hidden_size, dropout,
 	num_layers, max_gradient_norm, max_seq_length,
 	learning_rate, lr_decay,batch_size, forward_only=False):
 		self.num_classes =2
+		self.dropout = dropout
 		self.vocab_size = vocab_size
 		self.learning_rate = tf.Variable(float(learning_rate), trainable=False)
 		self.learning_rate_decay_op = self.learning_rate.assign(
@@ -45,9 +46,12 @@ def __init__(self, vocab_size, hidden_size, dropout,
 		self.seq_lengths = tf.placeholder(tf.int32, shape=[None],
 		name="early_stop")
 
-		self.dropout_keep_prob_embedding = tf.constant(self.dropout)
-		self.dropout_keep_prob_lstm_input = tf.constant(self.dropout)
-		self.dropout_keep_prob_lstm_output = tf.constant(self.dropout)
+		self.dropout_keep_prob_embedding = tf.placeholder(tf.float32,
+														  name="dropout_keep_prob_embedding")
+		self.dropout_keep_prob_lstm_input = tf.placeholder(tf.float32,
+														   name="dropout_keep_prob_lstm_input")
+		self.dropout_keep_prob_lstm_output = tf.placeholder(tf.float32,
+															name="dropout_keep_prob_lstm_output")
 
 		with tf.variable_scope("embedding"), tf.device("/cpu:0"):
 			W = tf.get_variable(
@@ -199,6 +203,10 @@ def step(self, session, inputs, targets, seq_lengths, forward_only=False):
 		input_feed[self.seq_input.name] = inputs
 		input_feed[self.target.name] = targets
 		input_feed[self.seq_lengths.name] = seq_lengths
+		input_feed[self.dropout_keep_prob_embedding.name] = self.dropout
+		input_feed[self.dropout_keep_prob_lstm_input.name] = self.dropout
+		input_feed[self.dropout_keep_prob_lstm_output.name] = self.dropout
+
 		if not forward_only:
 			input_feed[self.str_summary_type.name] = "train"
 			output_feed = [self.merged, self.mean_loss, self.update]