Linear Representations of Hierarchical Concepts in Language Models

• URL: http://arxiv.org/abs/2604.07886v3
• Date: Mon, 13 Apr 2026 10:44:52 GMT
• Title: Linear Representations of Hierarchical Concepts in Language Models
• Abstract summary: We train linear transformations specific to each hierarchical depth and semantic domain.<n>We then analyze how hierarchical is encoded in representation space.<n>Overall, we find that all models considered in our experiments encode concept hierarchies in the form of highly interpretable linear representations.
• Score: 34.691065518620924
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We investigate how and to what extent hierarchical relations (e.g., Japan $\subset$ Eastern Asia $\subset$ Asia) are encoded in the internal representations of language models. Building on Linear Relational Concepts, we train linear transformations specific to each hierarchical depth and semantic domain, and characterize representational differences associated with hierarchical relations by comparing these transformations. Going beyond prior work on the representational geometry of hierarchies in LMs, our analysis covers multi-token entities and cross-layer representations. Across multiple domains we learn such transformations and evaluate in-domain generalization to unseen data and cross-domain transfer. Experiments show that, within a domain, hierarchical relations can be linearly recovered from model representations. We then analyze how hierarchical information is encoded in representation space. We find that it is encoded in a relatively low-dimensional subspace and that this subspace tends to be domain-specific. Our main result is that hierarchy representation is highly similar across these domain-specific subspaces. Overall, we find that all models considered in our experiments encode concept hierarchies in the form of highly interpretable linear representations.

Related papers

• An expressivity analysis of hierarchical modelling in deep transformers via bounded-depth grammars [54.11540943172608]
  Deep neural networks are widely believed to derive their expressive power from their ability to form textbfhierarchical representations<n>In language modeling, textbftransformers have emerged as the dominant architecture, with early layers capturing local syntactic patterns and later layers encoding more complex clause-level dependencies.<n>We argue that these architectures possess the structural capacity to encode abstract grammatical states into low-dimensional, linearly separable subspaces within the residual stream.
  arXiv  Detail & Related papers  (2026-06-16T05:02:13Z)
• Geometry of Human Perceptual Domains Emerges Transiently in LLM Representations [1.503974529275767]
  We study the emergence of intrinsic geometrical structure corresponding to perceptual modalities within the residual streams of open-weight transformer architectures.<n>We observe the emergence of layer-wise geometric structure across multiple perceptual domains, despite the absence of any direct perceptual supervision during training.<n>This provides new insight into how and where human-like perceptual geometry arises in large language models.
  arXiv  Detail & Related papers  (2026-05-27T05:04:29Z)
• TopoAlign: Topology-Aware Visual Representation Alignment [18.80110544625249]
  TopoAlign is a framework for visually comparing model representations from a structural perspective.<n>We show that TopoAlign provides meaningful insights into representation structure and alignment from a topological perspective.
  arXiv  Detail & Related papers  (2026-05-25T07:58:26Z)
• Tracing the Roots: A Multi-Agent Framework for Uncovering Data Lineage in Post-Training LLMs [81.43302841109349]
  Post-training data plays a pivotal role in shaping the capabilities of Large Language Models (LLMs)<n>We propose an automated multi-agent framework to reconstruct the evolutionary graph of dataset development.
  arXiv  Detail & Related papers  (2026-04-12T06:24:07Z)
• Modeling Hierarchical Spaces: A Review and Unified Framework for Surrogate-Based Architecture Design [0.32622301272834514]
  This paper reviews literature on structured input spaces and proposes a unified framework that generalizes existing approaches.<n>A variable is described as meta if its value governs the presence of other decreed variables, enabling the modeling of conditional and hierarchical structures.<n>Our framework defines hierarchical distances and kernels to enable surrogate modeling and optimization on hierarchical domains.
  arXiv  Detail & Related papers  (2025-06-27T20:38:57Z)
• Learning Syntax Without Planting Trees: Understanding Hierarchical Generalization in Transformers [74.96551626420188]
  Transformers trained on natural language data have been shown to learn its hierarchical structure and generalize to sentences with unseen syntactic structures.<n>We investigate sources of inductive bias in transformer models and their training that could cause such generalization behavior to emerge.
  arXiv  Detail & Related papers  (2024-04-25T07:10:29Z)
• Linearity of Relation Decoding in Transformer Language Models [82.47019600662874]
  Much of the knowledge encoded in transformer language models (LMs) may be expressed in terms of relations. We show that, for a subset of relations, this computation is well-approximated by a single linear transformation on the subject representation.
  arXiv  Detail & Related papers  (2023-08-17T17:59:19Z)
• The geometry of hidden representations of large transformer models [43.16765170255552]
  Large transformers are powerful architectures used for self-supervised data analysis across various data types. We show that the semantic structure of the dataset emerges from a sequence of transformations between one representation and the next. We show that the semantic information of the dataset is better expressed at the end of the first peak, and this phenomenon can be observed across many models trained on diverse datasets.
  arXiv  Detail & Related papers  (2023-02-01T07:50:26Z)
• The Low-Dimensional Linear Geometry of Contextualized Word Representations [27.50785941238007]
  We study the linear geometry of contextualized word representations in ELMO and BERT. We show that a variety of linguistic features are encoded in low-dimensional subspaces.
  arXiv  Detail & Related papers  (2021-05-15T00:58:08Z)
• Hierarchical Pyramid Representations for Semantic Segmentation [0.0]
  We learn the structures of objects and the hierarchy among objects because context is based on these intrinsic properties. In this study, we design novel hierarchical, contextual, and multiscale pyramidal representations to capture the properties from an input image. Our proposed method achieves state-of-the-art performance in PASCAL Context.
  arXiv  Detail & Related papers  (2021-04-05T06:39:12Z)
• Universal-RCNN: Universal Object Detector via Transferable Graph R-CNN [117.80737222754306]
  We present a novel universal object detector called Universal-RCNN. We first generate a global semantic pool by integrating all high-level semantic representation of all the categories. An Intra-Domain Reasoning Module learns and propagates the sparse graph representation within one dataset guided by a spatial-aware GCN.
  arXiv  Detail & Related papers  (2020-02-18T07:57:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.