测试大模型评判者可靠性的开源工具，发现其表现受格式、表述等影响显著。
Judge Reliability Harness: Stress Testing the Reliability of LLM Judges
- 构建评估套件，自动检测大模型评判者在不同扰动下的表现
- 四类评测中无一模型全程可靠，文本格式变化即引发判断偏差
- 适合关注AI评测可信度的研究者和开发者使用
我们提出 Judge Reliability Harness，一个开源库，用于构建验证套件以测试基于大模型的评判者的可靠性。随着大模型评分广泛应用于人工智能基准测试，亟需高效评估此类方法可靠性的工具。给定基准数据集和大模型评判配置，该工具可生成可靠性测试，评估自由回答与代理任务格式下的二元判断准确率和序数评分性能。我们在涵盖安全、说服、误用及代理行为的四个基准上，对四种前沿评判模型进行了评估，发现模型间及不同扰动类型下表现差异显著，揭示了提升大模型评判鲁棒性的空间。我们评估的所有模型在使用本工具时均非全程可靠。例如，初步实验显示，仅因文本格式、改写、冗长程度或响应中真实标签反转，就导致评判准确率下降。代码已公开于：https://github.com/RANDCorporation/judge-reliability-harness
原文摘要 · Abstract (English)
We present the Judge Reliability Harness, an open source library for constructing validation suites that test the reliability of LLM judges. As LLM based scoring is widely deployed in AI benchmarks, more tooling is needed to efficiently assess the reliability of these methods. Given a benchmark dataset and an LLM judge configuration, the harness generates reliability tests that evaluate both binary judgment accuracy and ordinal grading performance for free-response and agentic task formats. We evaluate four state-of-the-art judges across four benchmarks spanning safety, persuasion, misuse, and agentic behavior, and find meaningful variation in performance across models and perturbation types, highlighting opportunities to improve the robustness of LLM judges. No judge that we evaluated is uniformly reliable across benchmarks using our harness. For example, our preliminary experiments on judges revealed consistency issues as measured by accuracy in judging another LLM's ability to complete a task due to simple text formatting changes, paraphrasing, changes in verbosity, and flipping the ground truth label in LLM-produced responses. The code for this tool is available at: https://github.com/RANDCorporation/judge-reliability-harness
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。