Tag
该论文区分了表征相似性的两种度量——原生余弦几何与线性可恢复性，并在五个模型族的天文数据集嵌入上发现，模型规模主要提升线性对齐后的邻域对应，而非原生几何收敛。
This paper challenges the assumption that EER is a reliable proxy for human voice similarity in speech generation, demonstrating that perceptual alignment depends more on training objectives than model performance. It proposes embedding geometry with effective dimensionality as a principled metric for evaluating voice similarity.
This paper analyzes the embedding space of Masked Diffusion Language Models (MDLMs) and finds it is hyperspherical, making linear interpolation suboptimal. The authors introduce Spherical Soft-Masking (S-SM), which uses SLERP and Fréchet means on the hypersphere, improving MAUVE and perplexity over prior soft-masking approaches.
This paper examines the embedding geometry of GPT-2 Small around the token 'Trump', comparing discretized and continuous nearest neighbor approaches to understand representational structure.
Analyzes the 64-D embedding manifold of Google AlphaEarth across 12.1M U.S. samples, shows non-Euclidean structure and poor vector arithmetic, then builds an agentic system with geometry-aware tools that outperforms parametric baselines on environmental queries.
Systematic study shows LLM-based dense retrievers outperform BERT baselines on typos and poisoning but remain vulnerable to semantic perturbations, with embedding geometry predicting robustness.