ValueError: Dimensions must be equal, but are 1024 and 821

問題：
原代碼：

rnn_cell = tf.nn.rnn_cell.BasicLSTMCell(num_units=512, state_is_tuple=True)
# 添加dropout layer，一般只設置output_keep_prob
# lstm_cell = tf.nn.rnn_cell.DropoutWrapper(cell=lstm_cell, input_keep_prob=1.0, output_keep_prob=1.0)
m_rnn_cell = tf.contrib.rnn.MultiRNNCell([rnn_cell] * 3, state_is_tuple=True)

改過之後的代碼：

lstm_cell = []
for iiLyr in range(3):
 lstm_cell.append(tf.nn.rnn_cell.LSTMCell(num_units=512, state_is_tuple=True))
# 添加dropout layer，一般只設置output_keep_prob
# lstm_cell = tf.nn.rnn_cell.DropoutWrapper(cell=lstm_cell, input_keep_prob=1.0, output_keep_prob=1.0)

mlstm_cell = tf.nn.rnn_cell.MultiRNNCell(cells=lstm_cell, state_is_tuple=True)

解決思路：
ValueError
ValueError2