从PyTorch的BiLSTM（BiGRU）获取最后一个状态

Question

在阅读了几篇文章之后，我仍然对从BiLSTM获取最后隐藏状态的实现的正确性感到困惑。

1. Understanding Bidirectional RNN in PyTorch (TowardsDataScience)
2. PackedSequence for seq2seq model (PyTorch forums)
3. What's the difference between “hidden” and “output” in PyTorch LSTM? (StackOverflow)
4. Select tensor in a batch of sequences (Pytorch formums)

来自最后一个源（4）的方法对我来说似乎是最干净的，但是我仍然不确定我是否正确理解了线程。我是否使用了LSTM和反向LSTM中正确的最终隐藏状态？这是我的实现

# pos contains indices of words in embedding matrix
# seqlengths contains info about sequence lengths
# so for instance, if batch_size is 2 and pos=[4,6,9,3,1] and 
# seqlengths contains [3,2], we have batch with samples
# of variable length [4,6,9] and [3,1]

all_in_embs = self.in_embeddings(pos)
in_emb_seqs = pack_sequence(torch.split(all_in_embs, seqlengths, dim=0))
output,lasthidden = self.rnn(in_emb_seqs)
if not self.data_processor.use_gru:
    lasthidden = lasthidden[0]
# u_emb_batch has shape batch_size x embedding_dimension
# sum last state from forward and backward  direction
u_emb_batch = lasthidden[-1,:,:] + lasthidden[-2,:,:]

对吗？