BriefGPT.xyz
Jul, 2024
一个优雅的桥梁:多语言LLM在不同语言中的偏见相似
What an Elegant Bridge: Multilingual LLMs are Biased Similarly in Different Languages
HTML
PDF
Viktor Mihaylov, Aleksandar Shtedritski
TL;DR
通过大型语言模型(LLMs)从语法性别的角度研究偏见,利用多语种LLMs重新审视和扩展Boroditsky(2003)的基础实验,发现简单的分类器不仅可以预测名词性别,还可以具有跨语言迁移能力,表明LLMs在不同语言中存在相似的偏见。
Abstract
This paper investigates biases of
large language models
(LLMs) through the lens of
grammatical gender
. Drawing inspiration from seminal works in psycholinguistics, particularly the study of gender's influence on
→