Word Embedding、RNN/LSTM/GRU+Seq2Seq+Attention+Self-Attention机制和Contextual Word Embedding(Universal Sentence Embedding)
Fine-Tuning
