量化但具欺骗性？量化大语言模型的多维真实性评估
Quantized but Deceptive? A Multi-Dimensional Truthfulness Evaluation of Quantized LLMs
AI总结：
本文提出TruthfulnessEval框架，从逻辑推理、常识和模仿性虚假三个维度评估量化LLMs的真实性，发现量化模型内部保留真实表征却易受欺骗性提示影响而输出错误内容，为量化感知对齐提供启示。
摘要：
量化通过显著降低内存和计算成本，使大语言模型（LLMs）能够在资源受限环境中高效部署。虽然量化LLMs通常在困惑度和零样本任务上保持性能，但量化对真实性的影响——即模型是生成真实还是欺骗性响应——在很大程度上仍未得到探索。本文提出TruthfulnessEval，一个用于评估量化LLMs真实性的综合评估框架，涵盖三个维度：（1）逻辑推理真实性；（2）常识真实性；（3）模仿性虚假陈述真实性。利用该框架，我们在多个开源LLMs上考察了主流量化技术（从4比特到极端的2比特）。令人意外的是，我们发现尽管量化模型在内部保留了真实的表征，但在误导性提示下更容易产生错误输出。为探究这一脆弱性，我们测试了“诚实”、“中性”和“欺骗性”提示各15种改写变体，并观察到“欺骗性”提示可以覆盖与真实一致的行为，而“诚实”和“中性”提示则保持稳定输出。此外，通过逐层探测和PCA可视化，我们揭示量化模型在内部“知道”真相，但在“欺骗性”提示引导下仍会产生错误输出。我们的发现为未来量化感知对齐和真实性干预的设计提供了见解。
英文摘要：
Quantization enables efficient deployment of large language models (LLMs) in resource-constrained environments by significantly reducing memory and computation costs. While quantized LLMs often maintain performance on perplexity and zero-shot tasks, their impact on truthfulness-whether generating truthful or deceptive responses-remains largely unexplored. In this work, we introduce TruthfulnessEval, a comprehensive evaluation framework for assessing the truthfulness of quantized LLMs across three dimensions: (1) Truthfulness on Logical Reasoning; (2) Truthfulness on Common Sense; and (3) Truthfulness on Imitative Falsehoods. Using this framework, we examine mainstream quantization techniques (ranging from 4-bit to extreme 2-bit) across several open-source LLMs. Surprisingly, we find that while quantized models retain internally truthful representations, they are more susceptible to producing false outputs under misleading prompts. To probe this vulnerability, we test 15 rephrased variants of "honest", "neutral" and "deceptive" prompts and observe that "deceptive" prompts can override truth-consistent behavior, whereas "honest" and "neutral" prompts maintain stable outputs. Further, we reveal that quantized models "know" the truth internally yet still produce false outputs when guided by "deceptive" prompts via layer-wise probing and PCA visualizations. Our findings provide insights into future designs of quantization-aware alignment and truthfulness interventions.
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。
