USE_DROPOUT = True about edgeml

[ShikharJ]

@metastableB

[metastableB]

Hi @ablanda, please post the stack trace of your error so that I can help you better. With the information I have, the error you seem to be having is from this line

  if self.useDropout is True:
                keep_prob = tf.placeholder(dtype=tf.float32, name='keep-prob')
                wrapped_cell = tf.contrib.rnn.DropoutWrapper(cell,
                                                             input_keep_prob=keep_prob,
                                                             output_keep_prob=keep_prob)
                outputs__, states = tf.nn.static_rnn(
                wrapped_cell, x, dtype=tf.float32)
                outputs = []

So EMIFastGRNN does initialize the keep_prob with a placeholder as desired by tensorflow as long as self.useDropout is true.

[ShikharJ]

Closing for inactivity.