Multilingual large language models leak human stereotypes across language boundaries

2023-12-12 · Yang Trista Cao, Anna Sotnikova, Jieyu Zhao, Linda X. Zou, Rachel Rudinger, Hal Daume III

Multilingual large language models have gained prominence for their proficiency in processing and generating text across languages. Like their monolingual counterparts, multilingual models are likely to pick up on stereotypes and other social biases present in their training data. In this paper, we study a phenomenon we term stereotype leakage, which refers to how training a model multilingually may lead to stereotypes expressed in one language showing up in the models' behaviour in another. We propose a measurement framework for stereotype leakage and investigate its effect across English, Russian, Chinese, and Hindi and with GPT-3.5, mT5, and mBERT. Our findings show a noticeable leakage of positive, negative, and non-polar associations across all languages. We find that of these models, GPT-3.5 exhibits the most stereotype leakage, and Hindi is the most susceptible to leakage effects. WARNING: This paper contains model outputs which could be offensive in nature.

📄 PDF Abstract BibTeX arXiv:2312.07141

Code (1)

annasou/stereotype_leakage 공식 구현 pytorch

Methods 이 논문이 사용한 방법론

Gated Linear Unit A Gated Linear Unit, or GLU computes: $$ \mathrm{GLU}(a, b) = a \otimes \sigma(b) $$ It is used in natural language processing architectures, for example the Gated CNN,…

15 Ways to Contact How can i speak to someone at Delta Airlines 설명 없음

Weight Decay 설명 없음

Cosine Annealing Cosine Annealing is a type of learning rate schedule that has the effect of starting with a large learning rate that is relatively rapidly decreased to a minimum value before…

Linear Warmup With Cosine Annealing Linear Warmup With Cosine Annealing is a learning rate schedule where we increase the learning rate linearly for $n$ updates and then anneal according to a cosine schedule…

Adam 설명 없음

{Dispute@FaQ-s}How to file a dispute with Expedia? How to file a dispute with Expedia? To file a complaint against Expedia, first try contacting their customer service directly. You can reach them by phone at…

GPT-3 설명 없음

SeeGULL Multilingual: a Dataset of Geo-Culturally Situated Stereotypes

2024-03-08 · Mukul Bhutani, Kevin Robinson, Vinodkumar Prabhakaran, Shachi Dave 외

While generative multilingual models are rapidly being deployed, their safety and fairness evaluations are largely limited to resources collected in English. This is especially problematic for evaluations targeting inher…

Fairness

StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs

2026-05-11 · Pierre Le Jeune, Étienne Duchesne, Weixuan Xiao, Stefano Palminteri 외 arxiv

Multilingual studies of social bias in open-ended LLM generation remain limited: most existing benchmarks are English-centric, template-based, or restricted to recognizing pre-specified stereotypes. We introduce StereoTa…

Deconstructing Stereotypes: Scope-Conditioned Generation for Effective Multilingual Counterspeech

2026-09-15 · Greta Damo, Elias Urios Alacreu, Elena Cabrio, Paolo Rosso 외 arxiv

Counterspeech (CS) - direct responses that counter online Hate Speech (HS) using reasoning and alternative viewpoints - has emerged as an alternative to content removal. Current automatic CS generation methods, however, …

Scalable and Culturally Specific Stereotype Dataset Construction via Human-LLM Collaboration

2026-07-08 · Weicheng Ma, John Guerrerio, Soroush Vosoughi arxiv

Research on stereotypes in large language models (LLMs) has largely focused on English-speaking contexts, due to the lack of datasets in other languages and the high cost of manual annotation in underrepresented cultures…

Breaking Language Barriers or Reinforcing Bias? A Study of Gender and Racial Disparities in Multilingual Contrastive Vision Language Models

2025-05-20 · Zahraa Al Sahili, Ioannis Patras, Matthew Purver

Multilingual vision-language models promise universal image-text retrieval, yet their social biases remain under-explored. We present the first systematic audit of three public multilingual CLIP checkpoints -- M-CLIP, NL…

Image-text Retrieval Text Retrieval