发表机构
African Institute for Mathematical Sciences (AIMS); University of Cape Town (UCT)（非洲数学科学研究所; 开普敦大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过探针和对比方向分析Gemma 4 31B对非洲语言的表示，发现区域和语系对齐模式，并验证了英语数据构建方向的国家引导有效性。
AI 中文摘要
我们研究了Gemma 4 31B如何表示非洲语言并响应文化引导。第一项研究使用探针、对比方向和表示相似性度量，比较了九种非洲语言和三种对照语言。从英语的迁移在不同语言和层之间有所变化。在多个层上，表示非洲与西方对比的方向在非洲语言之间比这些语言与对照语言之间更为对齐。跨语系的比较在十二层中的五层通过了报告的Holm阈值，尽管语言对之间的依赖性限制了统计解释。在尼日利亚内部，约鲁巴语和伊博语在十二层中的十一层比它们与豪萨语的平均配对更为对齐。第二项研究使用单独的英语数据为尼日利亚、加纳、肯尼亚和南非构建方向。在选定的强度下采用联合评分，估计的归因率与随机方向的差异范围从0.63到0.81。大多数输出通过了自动结构连贯性筛查。与Aya Expanse 32B的比较表明，结果取决于表示度量。综合来看，这些研究记录了采样表示中的区域和语系模式以及英语中的国家引导。
英文摘要
We study how Gemma 4 31B represents African languages and responds to cultural steering. The first study compares nine African languages and three controls using probes, contrast directions, and measures of representation similarity. Transfer from English varies across languages and layers. Directions representing an Africa versus West contrast are more aligned among the African languages than between these languages and the controls at several layers. The comparison across language families passes the reported Holm threshold at five of twelve layers, although dependence between language pairs limits the statistical interpretation. Within Nigeria, Yoruba and Igbo are more aligned than the average of their pairs with Hausa at eleven of twelve layers. The second study uses separate English data to construct directions for Nigeria, Ghana, Kenya, and South Africa. Under union scoring at the selected strengths, estimated differences in attribution rates from random directions range from 0.63 to 0.81. Most outputs pass the automated structural coherence screen. Comparisons with Aya Expanse 32B show that results depend on the representation measure. Together, the studies document regional and family patterns in the sampled representations and country steering in English.
