Embeddings from Language Models

ELMo is a process to signify the words in the form of vectors and embedding. In ELMo, Bidirectional language models(BiLM) are used. ELMo can be understood as the vectors of the word, which are calculated on the above of a two-layered BiLM

Comparison of Different Functions

Unlike different techniques of word embeddings. The word representation in ELMo is considered as the function of the input sentence. There is a difference also assignment of the vector. In it, the vector which is assigned to a token of a word is a function of the whole sentence in which the word is used.

Various Usages of Embeddings from Language Models

ELMo is used for various NLP processes for both in the industry as well as for research.Main tasks for which ELMo can be used are -

  • Modeling of the Language
  • Developing a Question-Answering system
  • Named Entity Recognition
  • Summarization of the Text