5 papers
A Mechanistic Understanding of Pronoun Fidelity in LLMs
Katharina Trinley, Jesujoba O. Alabi, Dietrich Klakow +1
Faithful and robust pronoun use is important for fair and coherent generations, yet large language models largely fail when multiple referents use different pronouns. To study the…
The Latin Substrate: How Language Models Represent and Mediate Script Choice
Daniil Gurgurov, Alan Saji, Katharina Trinley +2
Many languages are written in multiple scripts, requiring large language models (LLMs) to generate equivalent linguistic content in distinct orthographic forms. While prior work su…
Language Arithmetics: Towards Systematic Language Neuron Identification and Manipulation
Daniil Gurgurov, Katharina Trinley, Yusser Al Ghussin +3
Large language models (LLMs) exhibit strong multilingual abilities, yet the neural mechanisms behind language-specific processing remain unclear. We analyze language-specific neuro…
Multilingual Political Views of Large Language Models: Identification and Steering
Daniil Gurgurov, Katharina Trinley, Ivan Vykopal +3
Large language models (LLMs) are increasingly used in everyday tools and applications, raising concerns about their potential influence on political views. While prior research has…
What Language(s) Does Aya-23 Think In? How Multilinguality Affects Internal Language Representations
Katharina Trinley, Toshiki Nakai, Tatiana Anikina +1
Large language models (LLMs) excel at multilingual tasks, yet their internal language processing remains poorly understood. We analyze how Aya-23-8B, a decoder-only LLM trained on…