in your LSTM code for sentiment analysis, it goes like:
lstmCell = tf.contrib.rnn.BasicLSTMCell(lstmUnits) lstmCell = tf.contrib.rnn.DropoutWrapper(cell=lstmCell, output_keep_prob=0.25) value, _ = tf.nn.dynamic_rnn(lstmCell, data, dtype=tf.float32) weight = tf.Variable(tf.truncated_normal([lstmUnits, numClasses])) bias = tf.Variable(tf.constant(0.1, shape=[numClasses])) value = tf.transpose(value, [1, 0, 2]) last = tf.gather(value, int(value.get_shape()[0]) - 1) prediction = (tf.matmul(last, weight) + bias)
why not use the second value from function dynamic_rnn() ?
the second value is the final_state of LSTM cell, that is a tuple of (h, Ct).
The h is the output of the last cell, so you can just code like this
value, (c, h) = tf.nn.dynamic_rnn(lstmcell, data, dtype=tf.float32)
When get the h, we can directly apply the multiplication to it.
so, there is no need to use this:
value = tf.transpose(value, [1, 0, 2]) last = tf.gather(value, int(value.get_shape()[0]) - 1)
so, this is just some advices, in fact your tutorials are really good, Greet appreciates!
in your LSTM code for sentiment analysis, it goes like:
lstmCell = tf.contrib.rnn.BasicLSTMCell(lstmUnits) lstmCell = tf.contrib.rnn.DropoutWrapper(cell=lstmCell, output_keep_prob=0.25) value, _ = tf.nn.dynamic_rnn(lstmCell, data, dtype=tf.float32) weight = tf.Variable(tf.truncated_normal([lstmUnits, numClasses])) bias = tf.Variable(tf.constant(0.1, shape=[numClasses])) value = tf.transpose(value, [1, 0, 2]) last = tf.gather(value, int(value.get_shape()[0]) - 1) prediction = (tf.matmul(last, weight) + bias)why not use the second value from function dynamic_rnn() ?
the second value is the final_state of LSTM cell, that is a tuple of (h, Ct).
The h is the output of the last cell, so you can just code like this
value, (c, h) = tf.nn.dynamic_rnn(lstmcell, data, dtype=tf.float32)When get the h, we can directly apply the multiplication to it.
so, there is no need to use this:
value = tf.transpose(value, [1, 0, 2]) last = tf.gather(value, int(value.get_shape()[0]) - 1)so, this is just some advices, in fact your tutorials are really good, Greet appreciates!