用定量方法刻画大模型叙事倾向差异，发现不同模型有固定输出模式。
Narrative Landscape: Mapping Narrative Dispositions Across LLMs
- 通过重复测试构建模型输出稳定性与多样性双维度指标
- 六款前沿模型呈现明显刚性-探索连续谱，指令类型改变选择空间结构
- 提出可视化方法可直观对比不同模型的叙事偏好分布
本研究提出一种量化框架，用于刻画大模型在重复、受控条件下的稳定输出规律。通过对六款前沿模型和三种指令类型进行结构化叙事约束-选择任务，将模型倾向操作化为两个维度：'一致性'（通过杰卡德相似度衡量跨复现的选择重叠率）与'多样性'（通过逆辛普森指数衡量选项分布离散程度）。我们进一步引入基于PCA的叙事景观（Narrative Landscape）可视化方法，将各模型的选择特征映射至共享空间以实现直接比较。结果表明，不同模型家族呈现出明显的刚性-探索连续谱；且即使数值指标相近，指令类型仍会显著改变选择空间几何形态，说明相似得分可能掩盖本质不同的选择拓扑结构。
原文摘要 · Abstract (English)
This study proposes a quantitative framework for profiling LLM dispositions as stable, model-specific regularities in output under repeated, controlled elicitation. Using a structured narrative constraint-selection task administered across six frontier models and three instruction types, we operationalize disposition through two dimensions: "consistency", measured as cross-replication selection overlap via Jaccard similarity, and "diversity", measured as dispersion across options via the inverse Simpson index. We further introduce Narrative Landscape, a PCA-based visualization that maps each model's selection profile into a shared space for direct comparison. Results reveal a clear rigidity-exploration spectrum across model families and show that instruction types shift the geometry of selection spaces even when scalar metrics appear similar, indicating that comparable scores can mask qualitatively distinct selection topologies.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。