We celebrate the successful PhD defense of Letitia Parcalabescu!
Her thesis
“Measuring the Contributions of Vision and Text Modalities in Multimodal Transformers”
provides new metricsfor investigating the visio-linguistic capabilities of VLMs,
to investigate fundamental questions on multimodal fusion, and
to deepen our understanding of
how to examine self-consistency and faithfulness of Vision & Language Models. An outstanding thesis and a splendid and most lively PhD thesis defense!
Congratulations, Letitia, for your excellent work, and wishing you all the best for the future!
