Training and Research

Academic staff

Denis Delfitto, Andrea Padovan , Fabio d'Asaro, Andrea Padovan, Andrea Padovan, Gaetano Fiorin

Credits

2.5

Language

English

Class attendance

Compulsory

Location

VERONA

Learning objectives

On successful completion of this module, PhD students will acquire a basic knowledge in the nature and use of Large Language Models (LLM) and recent developments in AI in text analysis and text production

Prerequisites and basic notions

Some basic knowledge in linguistics and in formal language theory

Program

This 10-hour course will introduce students to the key features of Large Language Models (LLMs) and their relationship with natural language.
The course will cover (i) a brief introduction to the basic mathematical notions that are necessary to understand the developments in Machine Learning (ML) that led to LLMs and (ii) a critical discussion of some influential papers that deal with the LLMs’ impact on applied and theoretical linguistics.
More particularly, there have been interesting debates about the models' ability to produce grammatical sentences and understand complex syntactic constructions. Some of the questions that arise are: (i) how can models display such a high level of proficiency in some linguistic domains? (ii) which are the differences (or the commonalities) with human linguistic competence? (iii) in which linguistic domains are models currently underperforming and what could be done to improve their performance? (iv) what are the consequences for theories on language acquisition? (v) how can models impact research methodology in applied and theoretical linguistics?
Some case studies with both ChatGPT and Google Gemini will be presented to showcase the grammatical performance of LLMs. The course will also explore the concept of 'emergent skills' in LLMs and the associated scientific debate

Bibliography

Visualizza la bibliografia con Leganto, strumento che il Sistema Bibliotecario mette a disposizione per recuperare i testi in programma d'esame in modo semplice e innovativo.

Didactic methods

n.v.

Learning assessment procedures

PhD Students can choose aspects of the module as topic of their term paper.

Students with disabilities or specific learning disorders (SLD), who intend to request the adaptation of the exam, must follow the instructions given HERE

Assessment

n.v.

Criteria for the composition of the final grade

n.v.

Scheduled Lessons

When Classroom Teacher topics
Wednesday 22 January 2025
14:00 - 17:00
Duration: 3:00 AM
Palazzo di Lingue - Co-Working [04 - M] Denis Delfitto Linguistics and Large Language Models
Thursday 23 January 2025
10:00 - 12:00
Duration: 2:00 AM
Palazzo di Lingue - T.10 [T.10 - terra] Denis Delfitto Linguistica e Large Language Models
Thursday 23 January 2025
14:00 - 17:00
Duration: 3:00 AM
Palazzo di Lingue - T.10 [T.10 - terra] Denis Delfitto Linguistica e Large Language Models
Friday 24 January 2025
10:00 - 12:00
Duration: 2:00 AM
Palazzo di Lingue - T.10 [T.10 - terra] Denis Delfitto Linguistica e Large Language Models