CalibraEval：校准预测分布以缓解LLM作为评判者中的选择偏差
CalibraEval: Calibrating Prediction Distribution to Mitigate Selection Bias in LLMs-as-Judges
AI总结：
针对LLMs-as-Judges在成对比较中的选择偏差问题，本文提出无标签推理期去偏方法CalibraEval，通过非参数保序算法NOA校准预测分布，在多个基准上有效缓解偏差并提升评估性能。
摘要：
使用大型语言模型（LLMs）作为自动化评估工具来评价生成自然语言的质量，即LLMs-as-Judges，已展现出有前景的能力并迅速获得广泛关注。然而，当应用于候选响应的成对比较时，基于LLM的评估器常常表现出选择偏差。具体而言，当选项位置或ID令牌被交换时，其判断可能变得不一致，从而损害评估结果的有效性和公平性。为应对这一挑战，我们提出了CalibraEval，这是一种在推理阶段缓解选择偏差的新型无标签方法。具体来说，CalibraEval将去偏重新表述为一个优化任务，旨在调整观测到的预测分布，使其与无偏预测分布对齐。为解决这一优化问题，我们提出了一种非参数保序算法（NOA）。该算法利用模型预测分布之间的偏序关系，从而无需显式标签和精确的数学函数建模。在多个代表性基准上对LLMs进行的实证评估表明，与现有去偏方法相比，CalibraEval能够有效缓解选择偏差并提升性能。这项工作标志着向构建更鲁棒、更无偏的自动化评估框架迈出了一步，为提高AI驱动评估的可靠性铺平了道路。
英文摘要：
The use of large language models (LLMs) as automated evaluation tools to assess the quality of generated natural language, known as LLMs-as-Judges, has demonstrated promising capabilities and is rapidly gaining widespread attention. However, when applied to pairwise comparisons of candidate responses, LLM-based evaluators often exhibit selection bias. Specifically, their judgments may become inconsistent when the option positions or ID tokens are swapped, compromising the effectiveness and fairness of the evaluation result. To address this challenge, we introduce CalibraEval, a novel label-free method for mitigating selection bias during inference. Specifically, CalibraEval reformulates debiasing as an optimization task aimed at adjusting observed prediction distributions to align with unbiased prediction distributions. To solve this optimization problem, we propose a non-parametric order-preserving algorithm (NOA). This algorithm leverages the partial order relationships between model prediction distributions, thereby eliminating the need for explicit labels and precise mathematical function modeling.Empirical evaluations of LLMs in multiple representative benchmarks demonstrate that CalibraEval effectively mitigates selection bias and improves performance compared to existing debiasing methods. This work marks a step toward building more robust and unbiased automated evaluation frameworks, paving the way for improved reliability in AI-driven assessments
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。
