Surprisal | Gabriele Sarti

Surprisal

Interpreting Neural Language Models for Linguistic Complexity Assessment

This thesis presents a model-driven study of multiple phenomena associated with linguistic complexity, and how those get encoded by neural language models' learned representations.