约束释义一致性用于大语言模型幻觉检测
Constrained Paraphrase Consistency for LLM Hallucination Detection
AI总结：
提出约束一致性幻觉检测器(CCHD)，通过约束优化利用释义一致性，无需额外数据，在多个基准上超越现有方法。
摘要：
大型语言模型（LLM）可能生成事实不一致的声明，这促使需要准确且可扩展的幻觉检测器。先前的工作主要通过合成或新标注来扩大训练集，这增加了成本和潜在偏差，同时未充分利用语义等价释义所隐含的一致性。我们提出约束一致性幻觉检测器（CCHD），将训练形式化为约束优化问题。在原始文档-声明对上的标准交叉熵基础上，补充了（i）释义一致性约束，限制不同释义视图之间的差异，以及（ii）标签保持约束，将释义与真实标签绑定。我们通过模型参数和每个视图的拉格朗日乘子的梯度下降-上升法求解该问题，仅增加少量标量对偶变量，且无推理时开销。使用DeBERTa和Flan-T5骨干网络，CCHD在标准事实性基准上持续优于强基线（FactCG、MiniCheck和AlignScore），展示了其在幻觉检测上的优越性。
英文摘要：
Large language models (LLMs) can generate factually inconsistent claims, motivating accurate and scalable hallucination detectors. Prior work largely enlarges training sets via synthesis or new annotations, introducing increasing cost and potential bias while underusing the consistency implied by semantically equivalent paraphrases. We propose Consistency-Constrained Hallucination Detector (CCHD), which formulates training as a constrained optimization problem. The standard cross-entropy on original document-claim pairs is complemented by (i) paraphrase-consistency constraints bounding divergence across paraphrased views, and (ii) label-preservation constraints tying paraphrases to ground truth. We solve the problem by gradient descent-ascent over model parameters and per-view Lagrange multipliers, adding only a few scalar dual variables and no inference-time overhead. With DeBERTa and Flan-T5 backbones, CCHD consistently outperforms strong baselines (FactCG, MiniCheck, and AlignScore) on standard factuality benchmarks, demonstrating its superiority on hallucination detection.
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。
