Linguistic Profiling of a Neural Language Model

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Miaschi, Alessio, Brunato, Dominique, Dell'Orletta, Felice, Venturi, Giulia
Formato: Preprint
Publicado: 2020
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913241895010304
author Miaschi, Alessio
Brunato, Dominique
Dell'Orletta, Felice
Venturi, Giulia
author_facet Miaschi, Alessio
Brunato, Dominique
Dell'Orletta, Felice
Venturi, Giulia
contents In this paper we investigate the linguistic knowledge learned by a Neural Language Model (NLM) before and after a fine-tuning process and how this knowledge affects its predictions during several classification problems. We use a wide set of probing tasks, each of which corresponds to a distinct sentence-level feature extracted from different levels of linguistic annotation. We show that BERT is able to encode a wide range of linguistic characteristics, but it tends to lose this information when trained on specific downstream tasks. We also find that BERT's capacity to encode different kind of linguistic properties has a positive influence on its predictions: the more it stores readable linguistic information of a sentence, the higher will be its capacity of predicting the expected label assigned to that sentence.
format Preprint
id arxiv_https___arxiv_org_abs_2010_01869
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Linguistic Profiling of a Neural Language Model
Miaschi, Alessio
Brunato, Dominique
Dell'Orletta, Felice
Venturi, Giulia
Computation and Language
Artificial Intelligence
In this paper we investigate the linguistic knowledge learned by a Neural Language Model (NLM) before and after a fine-tuning process and how this knowledge affects its predictions during several classification problems. We use a wide set of probing tasks, each of which corresponds to a distinct sentence-level feature extracted from different levels of linguistic annotation. We show that BERT is able to encode a wide range of linguistic characteristics, but it tends to lose this information when trained on specific downstream tasks. We also find that BERT's capacity to encode different kind of linguistic properties has a positive influence on its predictions: the more it stores readable linguistic information of a sentence, the higher will be its capacity of predicting the expected label assigned to that sentence.
title Linguistic Profiling of a Neural Language Model
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2010.01869