长度控制的 AlpacaEval：一种消除自动评估器偏差的简单方法
Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators
AI总结：
本文提出长度控制的 AlpacaEval，通过广义线性模型以零长度差异为条件消除自动评估中的长度偏差，使指标更鲁棒并将与 LMSYS Chatbot Arena 的 Spearman 相关系数从 0.94 提升至 0.98。
摘要：
基于大语言模型（LLM）的自动标注器由于相较于人工评估具有成本效益和可扩展性，已成为 LLM 开发流程中的关键组成部分。然而，这些自动标注器可能引入难以消除的偏差。即使是简单且已知的混杂因素，例如对更长输出的偏好，仍然存在于现有的自动评估指标中。我们提出一种简单的回归分析方法，用于控制自动评估中的偏差。作为一个真实案例研究，我们专注于减少 AlpacaEval 的长度偏差。AlpacaEval 是一个快速且成本可负担的指令微调 LLM 基准，它使用 LLM 来估计响应质量。尽管 AlpacaEval 与人类偏好高度相关，但已知它偏向于生成更长输出的模型。我们引入一个长度控制的 AlpacaEval，旨在回答反事实问题：“如果模型输出和基线输出具有相同长度，偏好会是什么？”为实现这一目标，我们首先拟合一个广义线性模型（GLM），基于我们想要控制的中间变量（长度差异）和其他相关特征来预测有偏自动标注器的偏好。然后，通过在 GLM 中以长度差异为零作为条件来预测偏好，从而获得长度控制的偏好。长度控制不仅提高了该指标对模型输出冗长程度操纵的鲁棒性，我们还发现它将与 LMSYS Chatbot Arena 的 Spearman 相关系数从 0.94 提高到 0.98。
英文摘要：
LLM-based auto-annotators have become a key component of the LLM development process due to their cost-effectiveness and scalability compared to human-based evaluation. However, these auto-annotators can introduce biases that are hard to remove. Even simple, known confounders such as preference for longer outputs remain in existing automated evaluation metrics. We propose a simple regression analysis approach for controlling biases in auto-evaluations. As a real case study, we focus on reducing the length bias of AlpacaEval, a fast and affordable benchmark for instruction-tuned LLMs that uses LLMs to estimate response quality. Despite being highly correlated with human preferences, AlpacaEval is known to favor models that generate longer outputs. We introduce a length-controlled AlpacaEval that aims to answer the counterfactual question: "What would the preference be if the model's and baseline's output had the same length?" To achieve this, we first fit a generalized linear model to predict the biased auto-annotator's preferences based on the mediators we want to control for (length difference) and other relevant features. We then obtain length-controlled preferences by predicting preferences while conditioning the GLM with a zero difference in lengths. Length-controlling not only improves the robustness of the metric to manipulations in model verbosity, but we also find that it increases the Spearman correlation with LMSYS Chatbot Arena from 0.94 to 0.98.
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。
