Alessio Miaschi

Alessio Miaschi

Full-time researcher (RTDA) in Natural Language Processing

ItaliaNLP Lab (CNR-ILC)

Biography

I am a full-time researcher (RTDA) at the ItaliaNLP Lab, Institute for Computational Linguistics “A. Zampolli” (CNR-ILC, Pisa). In 2022, I received my PhD in Computer Science at the University of Pisa.

My research interests lie primarily in the context of Natural Language Processing (NLP) and in the study of Language Models (LM). I am particularly interested in the interpretability of large-scale LMs and in the evaluation of their internal representations, with a specific emphasis on understanding their inner linguistic abilities. Furthermore, I work in the development of NLP tools tailored for building educational applications.

In my free time, I enjoy going climbing, watching movies, listening to music, and reading books. Additionally, I have a passion for good beers and capturing moments with my analog camera.

Interests

  • Natural Language Processing
  • Language Models
  • Representation Learning
  • Interpretability for Deep Learning

Education

  • PhD in Computer Science, 2022

    University of Pisa

  • MSc in Digital Humanities, 2017

    University of Pisa

  • BSc in Digital Humanities, 2015

    University of Pisa

Recent Publications

(2024). Controllable Text Generation To Evaluate Linguistic Abilities of Italian LLMs. In Proceedings of the Tenth Italian Conference on Computational Linguistics (CLiC-it 2024, Pisa) (upcoming).

(2024). Fantastic Labels and Where to Find Them: Attention-Based Label Selection for Text-to-Text Classification. In Proceedings of the Workshop on Natural Language for Artificial Intelligence (NL4AI @ AIxIA 2024) (upcoming).

(2024). Evaluating Large Language Models via Linguistic Profiling. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024, Miami, Florida) (upcoming).

(2024). Leveraging Large Language Models for Mobile App Review Feature Extraction. arXiv.

Preprint

(2024). Linguistic Knowledge Can Enhance Encoder-Decoder Models (If You Let It). Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024, Turin).

Preprint PDF

News

CLiC-it 2024 Paper

NL4AI 2024 Paper

EMNLP 2024 Paper

New Position

LREC-COLING 2024 Paper

Teaching & Talks

Teaching

Teaching: Teaching assistant: Other courses: Thesis supervised/co-supervised:

Talks

  • 09/06/2023: Invited Talk at the DCP23 Workshop - Opening Large Language Models (Slides)
  • 31/05/2023: Lab at the Lectures on Computational Linguistics 2023 - Explaining Neural Language Models from Internal Representations to Model Predictions (Github Repo)
  • 08/03/2023: Talk at Seminario di Cultura Digitale (Informatica Umanistica, Università di Pisa) - Le risorse linguistiche al tempo delle reti neurali (Slides and Talk)
  • 13/12/2022: Invited Talk at "Ab urbe condata" seminar series organized by KDD Lab - Profiling Neural Language Models
  • 10/11/2022: Talk at ILC Seminars - Profiling a Neural Language Model (Slides)
  • 27/10/2022: Tech Talk at the School of AI (Pi School) 2022 - Interpreting Neural Language Models (Slides and Code)
  • 21/09/2022: Invited Talk at the International Summer School on "Advances in AI" 2022 - Profiling Neural Language Models (Slides)

Contact

  • Via G. Moruzzi 1, Pisa, PI
  • Institute for Computational Linguistics “A. Zampolli” (CNR)
  • DM Me
  • Skype Me