用轻量探针提前检测大模型幻觉，不增加延迟且效果领先。
DRIFT: Detecting Representational Inconsistencies for Factual Truthfulness
- 从中间层隐藏状态直接读取置信度信号，无需多次采样。
- 在12个评测设置中10个达最优，最高比之前方法提升13个百分点。
- 适合需要高准确率的问答系统，尤其对低延迟场景友好。
大语言模型常生成流畅但错误的答案，而检测幻觉通常需多次采样或事后验证，带来显著延迟和成本。我们假设中间层编码了置信度信号，这些信号在最终输出层丢失，因此提出一种轻量级探针，可直接从隐藏状态读取该信号。该探针计算开销不足0.1\%，可与生成过程完全并行运行，实现答案生成前的幻觉检测。基于此，我们构建了一个大模型路由系统：对高置信度问题立即回答，对不确定问题则转交给更强模型处理。尽管方法简单，但在四个问答基准、三个LLM系列的12个设置中，有10个达到当前最优（SOTA）AUROC，最高较先前方法提升13点，且在数据集分布变化下仍保持良好泛化能力，无需重新训练。
原文摘要 · Abstract (English)
LLMs often produce fluent but incorrect answers, yet detecting such hallucinations typically requires multiple sampling passes or post-hoc verification, adding significant latency and cost. We hypothesize that intermediate layers encode confidence signals that are lost in the final output layer, and propose a lightweight probe to read these signals directly from hidden states. The probe adds less than 0.1\% computational overhead and can run fully in parallel with generation, enabling hallucination detection before the answer is produced. Building on this, we develop an LLM router that answers confident queries immediately while delegating uncertain ones to stronger models. Despite its simplicity, our method achieves SOTA AUROC on 10 out of 12 settings across four QA benchmarks and three LLM families, with gains of up to 13 points over prior methods, and generalizes across dataset shifts without retraining.
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。