arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.14889cs.CLcs.AI
收藏

COBIAS:评估语言模型偏见基准的上下文可靠性

COBIAS: Assessing the Contextual Reliability of Bias Benchmarks for Language Models

International Institute of Information Technology, Hyderabad(国际信息技术学院海得拉巴分校);Amazon GenAI(亚马逊生成式人工智能);University of Maryland Baltimore County(马里兰大学巴尔的摩县分校)

Priyanshul Govil, Hemang Jain, Vamshi Krishna Bonagiri, Aman Chadha, Ponnurangam Kumaraguru, Manas Gaur, Sanorita Dey

AI总结:

针对现有语言模型偏见基准缺乏上下文考量的问题,提出上下文可靠性框架及COBIAS指标,基于模型跨上下文行为差异衡量偏见陈述检测可靠性,验证其与人类判断一致性高,可助力构建可靠基准与偏见缓解。

摘要:

大语言模型(LLM)通常会从其训练所用的网络数据中继承偏见,这些数据包含刻板印象与偏见。当前评估和缓解这类偏见的方法依赖偏见基准数据集,这类基准通过观察LLM在偏见性陈述上的行为来衡量偏见。然而,这些陈述缺乏对其试图呈现的情境的上下文考量。为解决这一问题,我们提出了一个上下文可靠性框架,该框架通过考虑偏见性陈述可能出现的各种上下文,评估模型对这类陈述的鲁棒性。我们开发了上下文导向偏见指标与评估分数(COBIAS),基于模型在不同上下文间的行为差异,衡量偏见性陈述在偏见检测中的可靠性。为评估该指标,我们为来自两个现有基准数据集的2291条刻板印象陈述补充了上下文信息。实验表明,COBIAS在偏见性陈述的上下文可靠性方面与人类判断一致(Spearman相关系数ρ=0.65,p=3.4×10^-60),可用于构建可靠的基准,助力偏见缓解工作。

英文摘要:

Large Language Models (LLMs) often inherit biases from the web data they are trained on, which contains stereotypes and prejudices. Current methods for evaluating and mitigating these biases rely on bias-benchmark datasets. These benchmarks measure bias by observing an LLM's behavior on biased statements. However, these statements lack contextual considerations of the situations they try to present. To address this, we introduce a contextual reliability framework, which evaluates model robustness to biased statements by considering the various contexts in which they may appear. We develop the Context-Oriented Bias Indicator and Assessment Score (COBIAS) to measure a biased statement's reliability in detecting bias, based on the variance in model behavior across different contexts. To evaluate the metric, we augmented 2,291 stereotyped statements from two existing benchmark datasets by adding contextual information. We show that COBIAS aligns with human judgment on the contextual reliability of biased statements (Spearman's $ρ= 0.65, p = 3.4 * 10^{-60}$) and can be used to create reliable benchmarks, which would assist bias mitigation works.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑