input_seq: shape: [max_seq_len, batch_size] input_lengths: 一批次中每个句子对应的句子长度列表 shape:[batch_size] hidden: Encoder的初始hidden输入，默认为None shape: [num_layers*num_directions, batch_size, hidden_size] 实
(self, input_seq, input_lengths, hidden=None)
source not stored for this graph (policy: none)
nothing calls this directly
no outgoing calls
no test coverage detected