NLP & Interpretability | PhD Student @ University of Trieste & Laboratory of Data Engineering of Area Science Park | Prev MPI-IS
Francesco Ortu
Loading...
It was super fun to take our first step in interpreting multimodal LLMs, working closely with the brilliant @alexpietroserra.bsky.social and @EmanuelePanizon
Additionally, blocking communication from this token significantly disrupts performance on standard benchmarks, while blocking image-text communication does not
