Sentence generation takes multiple words as input

• LSTM internal states generated by one word are far from the distribution of internal states in training

• Should be able to type multiple words

  • Ignore words that are not in your vocabulary.

• It might be interesting to learn a function that returns a scalar value Whether the internal state of LSTM is close to the state being learned separately.

---

This page is auto-translated from /nishio/文章生成は複数単語を入力として受け取る using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.