It also has an idea of using a trained sequence model to provide context to other NLP models. This paper proposed a new method, Universal Language Model Fine-tuning (ULMFiT) that enables robust inductive transfer learning for any NLP task. Train LM on big general domain corpus (use biLM) T...