
This user has not added any information to their profile yet.
My current research work relates to large language models (LLMs), their power, and limitation, in relation to natural language generation.
I'm interested in the way LLMs learn, the concepts they build, and their interpretation of the context, as the distinction between understanding and generating is blurring. I also work on the, pointwise and distributional, control of the sequences they generate, with a special concern for their societal impacts, given the nature of language.
Currently
- Distributional control of Natural Language Generation;
- large Language Models (evaluation, In-Context Learning, …)
Previously at Naver Labs Europe
- Research on complexity and readability of documents;
- "document analysis" work package lead for the SMOOTH European project (to assist micro enterprises in adopting and complying with GDPR);
- research projects to assist urban travelers, leader of a team of scientists and software engineers;
- knowledge representation of urban mobility, works on ontologies, semantic web technologies and geospatial data.
Before that
- Technical lead for a urban mobility data analytics platform.