用改写不变性学习，高效检测大模型幻觉。
Cross Paraphrastic Invariance Learning for Hallucination Detection
- 通过生成改写版本并对齐表示，实现表层形式不变的语义编码。
- 在11个任务上仅用1%标注数据，F1性能超越主流方法。
- 适合需要低标注成本、高准确率幻觉检测的研究者。
大语言模型常产生与源文档无关的幻觉内容。为避免昂贵的LLM评估流程和高标注成本，本文提出CPIL（跨改写不变性学习）——一种两阶段孪生框架，最大化利用已有标注数据。具体地，CPIL通过：(i) 为每个文档-主张样本生成改写视图作为正例，并显式对齐其表示以强制表面形式不变性；(ii) 挖掘同文档、反标签样本作为难负例，强化文档敏感的决策边界。随后进行两阶段训练：第一阶段通过对比学习构建具备改写不变性和基于事实感知的嵌入空间；第二阶段接入轻量分类器完成二分类的可信度判断。在包含11项任务的LLM-AggreFact基准上，仅需约1%标注数据，即在F1指标上超越强基线，展现优异预测能力与标签效率。
原文摘要 · Abstract (English)
Large language models (LLMs) frequently generate hallucinations, which are unsupported by a source document. To avoid costly LLM-as-evaluator pipelines and the heavy annotation demands of existing classifiers, we propose CPIL (Cross Paraphrastic Invariance Learning), a two-stage Siamese framework that maximizes the utility of existing labeled data. Concretely, CPIL constructs informative training pairs by: (i) generating paraphrastic views of each document-claim example as positives, and explicitly aligning their representations to enforce invariance to surface form; and (ii) mining same-document, opposite-label pairs as hard negatives to sharpen document-sensitive decision boundaries. Then CPIL conduct a two-stage model training: Stage 1 performs contrastive pretraining to learn a paraphrase-invariant, grounding-aware embedding space; and Stage 2 attaches a lightweight classifier for binary groundedness. On the LLM-AggreFact benchmark (11 tasks), CPIL surpasses strong baselines concerning F1 scores with only ~1% labeled data, showing its prediction superiority and label efficiency.
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。