The paper shows that the geometric patterns seen in language model embeddings—such as circles for months and saddle-shaped manifolds—arise from group-invariant statistics in word co‑occurrence data. By extending previous work on translation symmetry to arbitrary finite, compact, and homogeneous groups, the authors prove that embeddings correspond to matrix elements of the irreducible representations of the symmetry group. They validate this theory experimentally with the cyclic group <Z_{12}> for months, a dihedral group for musical chords, and a spherical‑harmonic embedding for celestial objects.
By Liam Storan, Andreas Tolias, Nina Miolane
How concepts are represented in neural networks is a fundamental question in machine learning. The dominant view treats concept representations as stationary geometric objects.
arXiv:2607. 04525v1 Announce Type: cross Abstract: How concepts are represented in neural networks is a fundamental question in machine learning.
By Zhimin Hu, Lanhao Niu, Sashank Varma
arXiv:2609.05721v1 Announce Type: new
Abstract: Understanding whether language-model embeddings encode structured real-world information is important for both representation analysis and information...
By Esteban Feuerstein, Victoria Klimkowski, Juan Manuel Ortiz de Zarate, Federico Hern\'an Suaiter
arXiv:2607. 07047v1 Announce Type: cross Abstract: Understanding the geometric structure of pre-trained language model embeddings matters for interpretability and safety.
By Szczepan Konior, Alexandre Quemy, Przemys{\l}aw Klocek, Gr\'egoire Cattan, Bart{\l}omiej Sobieski
arXiv:2608.30315v1 Announce Type: new
Abstract: Token embeddings are the basic representational units that connect discrete tokens with continuous computation in language models. Although modern lang...
By Junjie Yao, Liangkai Hang, Zhi-Qin John Xu
The paper investigates how large language models (LLMs) share a common Fisher‑Rao geometry in their next‑token probability distributions, revealing that behaviour largely determines this geometry while activation geometry depends on coordinate choices. Across transformer, state‑space, and recurrent architectures, output geometries align more closely than activation geometries, and this shared structure facilitates semantic‑category transfer and improves agreement with human word choices as models scale and train. The study further demonstrates that geometry can guide minimum‑disturbance interventions, enabling reusable control that preserves behaviour better than Euclidean methods and enhances steering, editing, attribution, dictionary learning, and fine‑tuning.
By Dario Picozzi
The paper shows that language models trained on natural text develop number representations that exhibit periodic features with dominant periods at T = 2, 5, 10. It identifies a two‑tiered hierarchy: all models learn Fourier‑domain spikes at these periods, but only some acquire geometrically separable features that allow linear classification of numbers modulo T. The study demonstrates that data, architecture, optimizer, and tokenizer influence whether these separable features emerge, and that models can learn them either from co‑occurrence signals in language or from multi‑token addition tasks, illustrating convergent evolution across diverse models.
By Deqing Fu, Tianyi Zhou, Mikhail Belkin, Vatsal Sharan, Robin Jia
The paper investigates why token prediction, a common pre‑training objective for language models, yields useful representations. It introduces a statistical framework linking token prediction accuracy to the geometry of token embeddings, showing that accurate predictions organize embeddings according to Hellinger distances between context distributions. The authors also propose a self‑consistency principle that refines contextual representations through repeated application of a shared block, and provide downstream guarantees for token generation, community recovery, and linear classification.
By Shulei Wang
Neural Collapse predicts that balanced one-hot classification pushes model representations to be equally far from each other; a symmetric configuration that depends only on the output label and ignores any semantic similarity in the inputs. This creates a puzzle: next-token prediction language models are trained predominantly (as context length increases) with one-hot labels: the same context is very unlikely to appear twice in training with different labels.
The paper investigates the Platonic Representation Hypothesis, which posits that more capable models converge toward shared representations. By distinguishing relational structure (which samples are related) from metric geometry (quantitative relations like distances), the authors develop a controlled $2 imes2$ framework to evaluate both aspects at local and global scales. Their findings show that relational structure consistently converges across vision‑language and video‑text models, while metric geometry converges much more weakly, a pattern that persists even when using a Riemannian metric approximation.
By Junwon You, Mihyun Jang, Sangwoo Mo, Jae-Hun Jung
The paper studies how transformer representations evolve across layers by examining the intrinsic dimensionality (ID) of token embeddings and their neighborhood structures. It finds that closed‑class tokens expand and collapse earlier than open‑class tokens, and that these changes are linked to shifts in local geometry. The authors compare encoder and decoder models, showing distinct layer‑wise behaviors, and demonstrate that geometric features alone can predict a token’s part‑of‑speech and reveal how semantic content changes across layers.
By Samuele Vallisa, Federico Ravenda, Claudio Palominos, Rui He, Andrea Raballo, Antonietta Mira, Philipp Homan, Wolfram Hinzen