通过分类探针识别潜在知识以改进大语言模型中的偏好提取
Improving Preference Extraction In LLMs By Identifying Latent Knowledge Through Classifying Probes
AI总结：
针对大语言模型作为自动评判器存在无意偏差的问题，提出利用对比提示对训练的线性分类探针提取潜在知识与偏好，经多模型多数据集实验验证，该方法优于传统生成式评判及同数据量微调评判器，兼具准确性、稳健性与高效性。
摘要：
大语言模型（LLM）常被用作自动评判器来评估文本，但各类无意偏差会阻碍其有效性。我们提出使用线性分类探针——通过利用对比提示对之间的差异训练而成——来直接获取LLM的潜在知识并提取更准确的偏好。我们使用来自四个不同模型家族的不同规模模型，以及六个评估文本质量评估和常识推理的多样化数据集开展了大量实验，结果表明，有监督和无监督探针方法均始终优于传统的基于生成的评判方法，同时保持相近的计算成本。这些探针在域迁移下仍具备泛化性，甚至能优于使用相同训练数据量的微调评判器。我们的结果表明，线性探针为LLM-as-judge任务提供了一种准确、稳健且计算高效的方法，同时为模型如何编码与评判相关的知识提供了可解释的见解。我们的数据和代码将在未来公开发布。
英文摘要：
Large Language Models (LLMs) are often used as automated judges to evaluate text, but their effectiveness can be hindered by various unintentional biases. We propose using linear classifying probes, trained by leveraging differences between contrasting pairs of prompts, to directly access LLMs' latent knowledge and extract more accurate preferences. Through extensive experiments using models of varying size from four different families and six diverse datasets assessing text quality evaluation and common sense reasoning, we demonstrate that both supervised and unsupervised probing approaches consistently outperform traditional generation-based judgement while maintaining similar computational costs. These probes generalise under domain shifts and can even outperform finetuned evaluators with the same training data size. Our results suggest linear probing offers an accurate, robust and computationally efficient approach for LLM-as-judge tasks while providing interpretable insights into how models encode judgement-relevant knowledge. Our data and code will be openly released in the future.
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。
