Plataforma de exploración de la Composición Semántica apartir de Modelos de Lenguaje pre-entrenados yembeddings estáticos
- 1 1Universidad Nacional de Educación a Distancia (UNED), España
- Miguel A. Alonso (ed. lit.)
- Margarita Alonso-Ramos (ed. lit.)
- Carlos Gómez-Rodríguez (ed. lit.)
- David Vilares (ed. lit.)
- Jesús Vilares (ed. lit.)
Publisher: CEUR Workshop Proceedings
Year of publication: 2022
Pages: 52-56
Type: Book chapter
Abstract
The computing power growth and the advent of the Transformer model have changed theNLP landscape. Transfer Learning has allowed the posibility of achieving state-of-the-art results at a fractionof the computational cost. In this scope, this work presents the development of a server-client applicationcapable of obtaining contextual and static word vectors from a wide variety of models, operate with them toachieve semantic composition to, lastly, visualize them in a 3-dimensional space and obtain semantic similarity;all of this, while exploiting the hardware resources available.