Blog

This text presents a literature review of Language Models (LM), covering two main topics: (1) The Transformers-based Neural Network used to train modern language models; and (2) The Semantic Space produced by the network training of the LM, which computationaly represents the language being modeled. In fact, obtaining such a computational representation for textual constructs is a long-standing problem that has challenged diverse NLP (Natural Language Processing) approaches. The establishment of transformers-based language models opens up vast possibilities and perspectives on interdisciplinary topics beyond NLP. Therefore, this survey details the history, the development and the mechanisms of Transformers-based language models. The text concludes with a critical analysis addressing issues regarding applications based on language models.

Fagner Leal Pantoja, in collaboration with Dr. Claudia Bauzer Medeiros and Dr. André Santanchè
Fagner Leal Pantoja, in collaboration with Dr. Claudia Bauzer Medeiros and Dr. André Santanchè