arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.10899cs.CL
收藏

基于分类法的CheckList大语言模型评估

Taxonomy-based CheckList for Large Language Model Evaluation

Purdue University(普渡大学)

Damin Zhang

AI总结:

本研究提出基于CheckList的QA任务,通过三项对比研究从四个方面评估语言模型的性别偏见,发现Transformer QA模型与LLM在偏见倾向与一致性关系上表现相反,并提供了首个包含人类知识的LLM偏见评估数据集。

摘要:

随着大语言模型(LLMs)被应用于许多下游任务,其内部的刻板表征可能会影响输出的公平性。在本研究中,我们将人类知识引入自然语言干预中,并在性别偏见的背景下研究预训练语言模型(LMs)的行为。受CheckList行为测试启发,我们提出了一项CheckList风格的任务,旨在通过问答(QA)来探究和量化LMs的不道德行为为。我们设计了三项对比研究,从一致性、偏见倾向、模型偏好以及性别偏好转换这四个方面评估LMs。我们探究了一个在SQuAD-v2数据集上训练的基于Transformer的QA模型,以及一个自回归大语言模型模型。结果表明,基于Transformer的QA模型的偏见倾向与其一致性呈正相关,而LLM则表现出相反的关系。我们提出的任务提供了了首个包含人类知识用于LLM偏见评估的数据集。

英文摘要:

As large language models (LLMs) have been used in many downstream tasks, the internal stereotypical representation may affect the fairness of the outputs. In this work, we introduce human knowledge into natural language interventions and study pre-trained language models' (LMs) behaviors within the context of gender bias. Inspired by CheckList behavioral testing, we present a checklist-style task that aims to probe and quantify LMs' unethical behaviors through question-answering (QA). We design three comparison studies to evaluate LMs from four aspects: consistency, biased tendency, model preference, and gender preference switch. We probe one transformer-based QA model trained on SQuAD-v2 dataset and one autoregressive large language model. Our results indicate that transformer-based QA model's biased tendency positively correlates with its consistency, whereas LLM shows the opposite relation. Our proposed task provides the first dataset that involves human knowledge for LLM bias evaluation.

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。

↑