NLP Task 3: Sequence-to-sequence Models - Language Processing in the Deep Learning Era (Part 8)