标签: tensorflow lstm
我是Tensorflow和LSTM的新手,有人可以帮助我了解以下代码中initial_state的用法。为什么我们需要一个initial_state。
*** initial_state = cell.zero_state(batch_size,tf.float32)
lstm_outputs,final_state = tf.nn.dynamic_rnn(单元格,嵌入,initial_state = initial_state)***