评估LLM作为裁判的评分偏见
Evaluating Scoring Bias in LLM-as-a-Judge
AI总结：
本文研究了LLM作为裁判在评分任务中的偏见问题，提出了三种新的评分偏见类型，并开发了一个框架来量化这些偏见，以改进评分提示设计。
摘要：
"LLM-as-a-Judge"范式通过大型语言模型（LLMs）作为自动评估者，是LLM发展中的关键部分，为复杂任务提供可扩展的反馈。然而，这些裁判的可靠性受到多种偏见的影响。现有研究主要集中在比较性评估中的偏见。相比之下，基于评分的评估（分配绝对分数，常用于工业应用）研究较少。为填补这一空白，我们进行了首次专门的评分偏见评估。我们从评分提示本身而非评估目标的偏见出发。我们正式定义了评分偏见，并识别了三种新的偏见类型：评分标准顺序偏见、评分ID偏见和参考答案评分偏见。我们提出了一种全面的框架来量化这些偏见，包含多方面的度量指标和自动数据合成管道来创建定制的评估语料库。我们的实验实证地证明了即使最先进的LLMs也受这些显著评分偏见的影响。我们的分析为设计更稳健的评分提示和缓解这些新发现的偏见提供了可行的见解。
英文摘要：
The "LLM-as-a-Judge" paradigm, using Large Language Models (LLMs) as automated evaluators, is pivotal to LLM development, offering scalable feedback for complex tasks. However, the reliability of these judges is compromised by various biases. Existing research has heavily concentrated on biases in comparative evaluations. In contrast, scoring-based evaluations-which assign an absolute score and are often more practical in industrial applications-remain under-investigated. To address this gap, we undertake the first dedicated examination of scoring bias in LLM judges. We shift the focus from biases tied to the evaluation targets to those originating from the scoring prompt itself. We formally define scoring bias and identify three novel, previously unstudied types: rubric order bias, score ID bias, and reference answer score bias. We propose a comprehensive framework to quantify these biases, featuring a suite of multi-faceted metrics and an automatic data synthesis pipeline to create a tailored evaluation corpus. Our experiments empirically demonstrate that even the most advanced LLMs suffer from these substantial scoring biases. Our analysis yields actionable insights for designing more robust scoring prompts and mitigating these newly identified biases.
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。
