Natural language processing, word2vec + subwords, NER, neural machine translation, attention Learning Goals Understand state-of-the-art algorithms for generating language embeddings Basic familiarity with old-school NLP feature engineering techniques Understand tradeoffs to a variety of attentional architectures Understand common long-term dependency moduules: GRUs & LSTMs Experiment with impact of initialization on deep RNN architectures Exercises cs20si 3: A TensorFlow chatbot fast.ai: 13: Neural Machine Translation of Rare Words with Subword Units fast.ai: 12: Neural Machine Translation by Jointly Learning to Align and Translate cs224d: 3-1: Recursive Neural Network cs224d: 2-3: TensorFlow RNN Language Model cs224d: 2-2: TensorFlow NER Window Model cs224d: 1-4: Sentiment Analysis cs224d: 1-3: word2vec cs20si 1-3: word2vec