From Found to Designed: Concepts as a Design Axis for Large Language Models

• URL: http://arxiv.org/abs/2607.26825v2
• Date: Thu, 30 Jul 2026 15:41:13 GMT
• Title: From Found to Designed: Concepts as a Design Axis for Large Language Models
• Abstract summary: We organize concept-aware interventions along two dimensions: whether concept structure is internally induced or externally grounded, and the stage of the pipeline where it is introduced.<n>This taxonomy reveals three broad patterns: inference-time approaches remain comparatively underexplored, related ideas have developed largely in isolation across pipeline stages, and externally grounded methods span the entire pipeline despite often being described under different terminology.
• Score: 4.979725367638729
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) encode rich concept-like information, but represent it implicitly through distributed statistical associations rather than as explicit, structured, compositional concepts. Consequently, concept-level structure is typically \emph{found} rather than \emph{designed}: it is recovered after training through probing or dictionary learning, with no architectural guarantee of stability, compositionality, controllability, or alignment with human conceptual organization. We organize concept-aware interventions along two dimensions: whether concept structure is internally induced or externally grounded, and the stage of the pipeline where it is introduced. This taxonomy reveals three broad patterns: inference-time approaches remain comparatively underexplored, related ideas have developed largely in isolation across pipeline stages, and externally grounded methods span the entire pipeline despite often being described under different terminology. Together, these observations motivate moving beyond recovering concept-like structure from trained models toward designing LLMs with explicit conceptual representations.

Related papers

• Formal Concept Lattices are Good Semantic Scaffolds for Concept-Based Learning [22.873980931052685]
  Learning semantics is essential for deep learning models to be interpretable and better aligned with human reasoning.<n> Concept-based models approach this by representing classes through meaningful semantic abstractions.<n>We show that formal concept lattices provide principled semantic scaffolds to guide neural network learning.
  arXiv  Detail & Related papers  (2026-06-03T21:50:29Z)
• Emergent Structured Representations Support Flexible In-Context Inference in Large Language Models [77.98801218316505]
  Large language models (LLMs) exhibit emergent behaviors suggestive of human-like reasoning.<n>We investigate the internal processing of LLMs during in-context concept inference.
  arXiv  Detail & Related papers  (2026-02-08T03:14:39Z)
• Insight: Interpretable Semantic Hierarchies in Vision-Language Encoders [52.94006363830628]
  Language-aligned vision foundation models perform strongly across diverse downstream tasks.<n>Recent works decompose these representations into human-interpretable concepts, but provide poor spatial grounding and are limited to image classification tasks.<n>We propose Insight, a language-aligned concept foundation model that provides fine-grained concepts, which are human-interpretable and spatially grounded in the input image.
  arXiv  Detail & Related papers  (2026-01-20T09:57:26Z)
• How Do Large Language Models Learn Concepts During Continual Pre-Training? [69.99800338599]
  We study how individual concepts are acquired and forgotten, as well as how multiple concepts interact through interference and synergy.<n>Our findings offer a circuit-level view of concept learning dynamics and inform the design of more interpretable and robust concept-aware training strategies.
  arXiv  Detail & Related papers  (2026-01-07T04:29:15Z)
• A Geometric Unification of Concept Learning with Concept Cones [58.70836885177496]
  Two traditions of interpretability have evolved side by side but seldom spoken to each other: Concept Bottleneck Models (CBMs) and Sparse Autoencoders (SAEs)<n>We show that both paradigms instantiate the same geometric structure.<n>CBMs provide human-defined reference geometries, while SAEs can be evaluated by how well their learned cones approximate or contain those of CBMs.
  arXiv  Detail & Related papers  (2025-12-08T09:51:46Z)
• FaCT: Faithful Concept Traces for Explaining Neural Network Decisions [56.796533084868884]
  Deep networks have shown remarkable performance across a wide range of tasks, yet getting a global concept-level understanding of how they function remains a key challenge.<n>We put emphasis on the faithfulness of concept-based explanations and propose a new model with model-inherent mechanistic concept-explanations.<n>Our concepts are shared across classes and, from any layer, their contribution to the logit and their input-visualization can be faithfully traced.
  arXiv  Detail & Related papers  (2025-10-29T13:35:46Z)
• MindCraft: How Concept Trees Take Shape In Deep Models [15.113541622429084]
  We introduce the MindCraft framework built upon Concept Trees.<n> Concept Trees reconstruct the hierarchical emergence of concepts, revealing exactly when they diverge from shared representations into linearly separable subspaces.<n> Empirical evaluations show that Concept Trees recover semantic hierarchies, disentangle latent concepts, and can be widely applied across multiple domains.
  arXiv  Detail & Related papers  (2025-09-26T20:39:52Z)
• ConceptFlow: Hierarchical and Fine-grained Concept-Based Explanation for Convolutional Neural Networks [14.365259717799034]
  Concept-based interpretability for Convolutional Neural Networks (CNNs) aims to align internal model representations with high-level semantic concepts.<n>We propose ConceptFlow, a concept-based interpretability framework that simulates the internal "thinking path" of a model.
  arXiv  Detail & Related papers  (2025-09-16T03:02:46Z)
• Large Language Models as Quasi-crystals: Coherence Without Repetition in Generative Text [0.0]
  essay proposes an analogy between large language models (LLMs) and quasicrystals, systems that exhibit global coherence without periodic repetition, generated through local constraints.<n> Drawing on the history of quasicrystals, it highlights an alternative mode of coherence in generative language: constraint-based organization without repetition or symbolic intent.<n>This essay aims to reframe the current discussion around large language models, not by rejecting existing methods, but by suggesting an additional axis of interpretation grounded in structure rather than semantics.
  arXiv  Detail & Related papers  (2025-04-16T11:27:47Z)
• OmniPrism: Learning Disentangled Visual Concept for Image Generation [57.21097864811521]
  Creative visual concept generation often draws inspiration from specific concepts in a reference image to produce relevant outcomes.<n>We propose OmniPrism, a visual concept disentangling approach for creative image generation.<n>Our method learns disentangled concept representations guided by natural language and trains a diffusion model to incorporate these concepts.
  arXiv  Detail & Related papers  (2024-12-16T18:59:52Z)
• Aligned at the Start: Conceptual Groupings in LLM Embeddings [10.282327560070202]
  This paper shifts focus to the often-overlooked input embeddings - the initial representations fed into transformer blocks.<n>Using fuzzy graph, k-nearest neighbor (k-NN), and community detection, we analyze embeddings from diverse LLMs.
  arXiv  Detail & Related papers  (2024-06-08T01:27:19Z)
• Coarse-to-Fine Concept Bottleneck Models [9.910980079138206]
  This work targets ante hoc interpretability, and specifically Concept Bottleneck Models (CBMs) Our goal is to design a framework that admits a highly interpretable decision making process with respect to human understandable concepts, on two levels of granularity. Within this framework, concept information does not solely rely on the similarity between the whole image and general unstructured concepts; instead, we introduce the notion of concept hierarchy to uncover and exploit more granular concept information residing in patch-specific regions of the image scene.
  arXiv  Detail & Related papers  (2023-10-03T14:57:31Z)
• Text-to-Image Generation for Abstract Concepts [76.32278151607763]
  We propose a framework of Text-to-Image generation for Abstract Concepts (TIAC) The abstract concept is clarified into a clear intent with a detailed definition to avoid ambiguity. The concept-dependent form is retrieved from an LLM-extracted form pattern set.
  arXiv  Detail & Related papers  (2023-09-26T02:22:39Z)
• Concept Gradient: Concept-based Interpretation Without Linear Assumption [77.96338722483226]
  Concept Activation Vector (CAV) relies on learning a linear relation between some latent representation of a given model and concepts. We proposed Concept Gradient (CG), extending concept-based interpretation beyond linear concept functions. We demonstrated CG outperforms CAV in both toy examples and real world datasets.
  arXiv  Detail & Related papers  (2022-08-31T17:06:46Z)

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.