Congratulations on the successful PhD defense Céline Budding!

We are pleased to congratulate our colleague Céline Budding for successfully completing her PhD thesis, “What do language models know when they know language? A theoretical framework for explaining the linguistic behavior of LLMs”.

Céline’s research investigates what kind of knowledge large language models may possess and lays out a theoretical framework for examining what large language models really learn from the enormous amounts of data they’re trained on. Specifically, she motivates in her dissertation a cognitivist approach towards explaining the behavior of LLMs and provides a framework for explaining the linguistic behavior of LLMs through the identification of tacit knowledge in these systems.

Supervised within the Eindhoven University of Technology by Carlos Zednik and Vincent Müller, Céline’s dissertation offers an important contribution to the philosophy of AI and provides a rigorous foundation for future research into how language models possess knowledge.

You can read the full dissteration here or look into the interview with Céline on her PhD jounrey and results.