Efficient and Human-Inspired Natural Language Processing Methods for Multilingual and Low-Resource Settings
MCML Authors
Abstract
Abstract
This dissertation advances multilingual NLP for low-resource languages by developing prompt-based, retrieval-augmented, and parameter-efficient methods that improve zero- and few-shot performance across diverse languages and tasks. It also combines linguistic, cognitive, and mechanistic analyses to better understand and mitigate multilingual weaknesses in LLMs, contributing to more inclusive, robust, and interpretable language technologies. (Shortened.)
BibTeXKey: Nie25