LLM (Large Language Models) from a researcher’s perspective – TRAIL Doctoral School

Topic of the seminar: Large language models.

In this seminar, we will review what Large Language Models (LLMs) are and see how they are currently used in Natural Language Processing (NLP). I will first briefly go over the developments in the NLP field that led to LLMs. After describing what LLMs are and highlighting their advantages and disadvantages, we will delve into their current uses in NLP. Finally, I will go over some of my own research, investigating what LLMs capture of pragmatics.

Practical information

Program

• 2:00pm – 4:00pm: Keynote talk by Marie-Catherine de Marneffe (FNRS Research Associate at UCLouvain).
• 4:15pm – 6:00pm: Short contributed presentations:
  • • Charlotte Nachtegael (ULB): DUVEL: an active-learning annotated biomedical corpus for the recognition of oligogenic combinations
    • Jan Held (ULiège): Explainability model for football referee decision
    • Benoît Ronval (UCLouvain): LLMs and tabular data: zero-shot generation and synthetic data for fairness
    • Martin Balfroid (UNamur): Towards LLM-generated code tours for onboarding
    • Vincent Stragier (UMons): LLM-based interactive agents for visually impaired and blind individuals
    • Christophe Crochet (UCLouvain): How LLM and prompt engineering can close the gap between formal verification and network protocol testing
    • Lize Pirenne (ULiège): Context referencing citations for factuality improvement

REGISTER

https://forms.gle/CNcNzJALrCP1JBoC9