LSTM-based Character-Level Language Model (LM)

An LSTM-based Character-Level Language Model (LM) is a RNN-based character-level LM that is an LSTM model.

• Context:
  • It can be produced by an LSTM-based Character-Level Language Modeling System.
  • …
• Counter-Example(s):
  • a Transformer-based Character-Level LM.
  • an LSTM-based Word-Level Language Model.
• See: Multi-Layer LSTM.

---

---

References

2015

• (Karpathy, 2015) ⇒ Andrej Karpathy. (2015). “The Unreasonable Effectiveness of Recurrent Neural Networks.” In: Proceedings of Blog post 2015-05-21.
  • QUOTE: ... By the way, together with this post I am also releasing code on Github that allows you to train character-level language models based on multi-layer LSTMs. You give it a large chunk of text and it will learn to generate text like it one character at a time. …

---