Mechanistic interpretability of large language models - probing the internal circuits behind hallucinations, factual recall, and in-context learning.
2 project leads and 9 researchers driving interpretability research through the 2025–26 academic year.
Recruitment for this tenure is over. Find the application below.