虚假差距：通过稀疏自编码器量化内部信念与思维链解释之间的分歧
The Hypocrisy Gap: Quantifying Divergence Between Internal Belief and Chain-of-Thought Explanation via Sparse Autoencoders
AI总结：
通过稀疏自编码器量化模型内部信念与生成回答之间的差异，以检测模型的不忠实行为。
摘要：
大型语言模型（LLMs）经常表现出不忠实的行为，产生一个最终答案，与它们内部的思维链（CoT）推理显著不同，以迎合它们正在对话中的用户。为了更好地检测这种行为，我们引入了虚假差距，这是一种机械指标，利用稀疏自编码器（SAEs）来量化模型内部推理与其最终生成之间的分歧。通过数学上比较通过稀疏线性探针得到的内部真实信念，与最终生成的潜在空间轨迹，我们量化并检测模型进行不忠实行为的倾向。在使用Anthropic的奉承基准测试对Gemma、Llama和Qwen模型进行实验时，我们的方法在检测奉承运行时实现了0.55-0.73的AUROC，在模型内部
英文摘要：
Large Language Models (LLMs) frequently exhibit unfaithful behavior, producing a final answer that differs significantly from their internal chain of thought (CoT) reasoning in order to appease the user they are conversing with. In order to better detect this behavior, we introduce the Hypocrisy Gap, a mechanistic metric utilizing Sparse Autoencoders (SAEs) to quantify the divergence between a model's internal reasoning and its final generation. By mathematically comparing an internal truth belief, derived via sparse linear probes, to the final generated trajectory in latent space, we quantify and detect a model's tendency to engage in unfaithful behavior. Experiments on Gemma, Llama, and Qwen models using Anthropic's Sycophancy benchmark show that our method achieves an AUROC of 0.55-0.73 for detecting sycophantic runs and 0.55-0.74 for hypocritical cases where the model internally "knows" the user is wrong, consistently outperforming a decision-aligned log-probability baseline (0.41-0.50 AUROC).
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。
