batch_first = True
输入:batch, seq_len, input_size
h0:num_layers * num_directions, batch, hidden_size
输出:batch,seq_len, num_directions * hidden_size
hn:num_layers * num_directions, batch, hidden_size
batch_first = false
输入:seq_len, batch, input_size
h0:num_layers * num_directions, batch, hidden_size
输出:seq_len, batch, num_directions * hidden_size
hn:num_layers * num_directions, batch, hidden_size