Advanced-NLP

Learn NLP by CS11-711(CMU)

Some notes

2. Word Representation and Text Classification

• Subword Models (Sentence Piece是一种工具)
  • Byte Pair Encoding incrementally combine the most frequent token pairs
  • Unigram Models Cons: hard to use multilingual and arbitrariness
• Continuous Word Embeddings
  • Continuous Bag of Words (CBOW)