缓慢衰减与沉默表达：语言模型谱系中的迭代阈下特质传递
Slow Decay and Silenced Expression: Iterated Subliminal Trait Transfer in Language-Model Lineages
浏览论文内容
中文总结 AI 辅助
研究语言模型谱系中阈下特质的跨代传递，发现特质可保持十代且内部存在但行为沉默，并可用位移引导表达。
中文摘要 AI 辅助
语言模型越来越多地基于其他模型的输出进行训练，形成我们称之为谱系的链式结构，其中某一代存在的特质可以传递给下一代。先前关于阈下学习的研究表明，教师的特质可以通过不携带任何该特质内容的过滤数据传递给学生。然而，证据仅覆盖了单步训练。我们研究这种特质在谱系中是否保持或消退。我们将该特质注入Qwen2.5-7B-Instruct的三个副本中，并从每个副本迭代训练步骤至深度十，以两种方式在相同的保留提示上读取每一代：一个关键词筛查，用于查找模型输出中该特质的表达；以及一个激活探针，将每个模型相对于基线的位移投影到由其他谱系教师构建的方向上。我们报告两个发现。第一，该特质在三个谱系中持续十代。被注入的模型在每次补全中都表达它；关键词筛查率在第一步后降至55.6%，到第十代降至21.1%。基线本身在其300次补全中没有任何一次匹配筛查。第二，该特质可以在内部存在而在行为上缺失。当在评估时移除模型的默认系统提示，第十代学生的关键词筛查率在每个提示上为零，而探针得分在每个提示上保持为正。用第十代学生的位移（在默认系统提示下训练和测量）来引导未经处理的基线，即使在移除系统提示的情况下也能诱导出该特质的筛查表达，而同一学生在移除系统提示时却未表现出该特质的表达。
英文摘要
Language models are increasingly trained on the outputs of other models, forming chains that we call lineages, in which a trait present in one generation can pass to the next. Prior work on subliminal learning has shown that a teacher's trait can transmit to a student through filtered data carrying none of the trait's content. However, the evidence covers only a single training step. We study whether such a trait holds or fades across lineages. We instill the trait into three copies of Qwen2.5-7B-Instruct and iterate the training step to depth ten from each, reading every generation two ways on the same held-out prompts: a keyword screen that looks for expressions of the trait in the model's output, and an activation probe that projects each model's displacement from the base onto a direction built from the other lineages' teachers. We report two findings. First, the trait persists through ten generations across three lineages. The instilled models express it on every completion; the keyword-screen rate falls to 55.6% after the first step and to 21.1% by generation ten. The base itself matches the screen on none of its 300 completions. Second, the trait can be present internally while absent behaviorally. When the model's default system prompt is removed at evaluation, the generation-ten students' keyword-screen rate is zero on every prompt while the probe score stays positive on every prompt. Steering the untreated base with the displacement of a generation-ten student, which is trained and measured under the default system prompt, induces screened expression of the trait even with the system prompt removed, while that same student shows no expression of the trait with the system prompt removed.
发表机构
- Denison University（丹尼森大学）
- VNUHCM - University of Information Technology（越南国立大学胡志明市分校 - 信息技术大学）
机构由 AI 辅助整理，请以论文原文为准。
