伪造引用的产生地：追踪领域层面的幻觉到LLM中的特定神经元
Where Fake Citations Are Made: Tracing Field-Level Hallucination to Specific Neurons in LLMs
AI总结：
研究发现LLM生成虚假引用时，作者名错误率最高，领域特定的幻觉神经元通过内部模型信号可检测和缓解。
摘要：
LLMs频繁生成虚构但可信的引用，即使底层参考错误也常表现出高信心。我们研究了9个模型和108,000个生成引用中的这一失败现象，发现作者名错误率远高于其他领域，在所有模型和设置中均如此。引用风格无明显影响，而以推理为导向的蒸馏会降低召回率。训练于一个领域的探针在其他领域转移至接近随机水平，表明幻觉信号不跨领域泛化。基于此发现，我们应用弹性网正则化与稳定性选择到Qwen2.5-32B-Instruct的神经元级CETT值，识别出一组领域特定的幻觉神经元（FH神经元）。因果干预进一步证实其作用：放大这些神经元会增加幻觉，而抑制它们则在不同领域提升性能，某些领域收益更大。这些结果表明，使用内部模型信号单独检测和缓解引用幻觉有轻量级方法。
英文摘要：
LLMs frequently generate fictitious yet convincing citations, often expressing high confidence even when the underlying reference is wrong. We study this failure across 9 models and 108{,}000 generated references, and find that author names fail far more often than other fields across all models and settings. Citation style has no measurable effect, while reasoning-oriented distillation degrades recall. Probes trained on one field transfer at near-chance levels to the others, suggesting that hallucination signals do not generalize across fields. Building on this finding, we apply elastic-net regularization with stability selection to neuron-level CETT values of Qwen2.5-32B-Instruct and identify a sparse set of field-specific hallucination neurons (FH-neurons). Causal intervention further confirms their role: amplifying these neurons increases hallucination, while suppressing them improves performance across fields, with larger gains in some fields. These results suggest a lightweight approach to detecting and mitigating citation hallucination using internal model signals alone.
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。
