Convex Hull of Class Representations

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Linear representationconcept0.718
  The idea that features are encoded as directions in activation space.
• Linear Representation Hypothesisframework0.714
  The hypothesis that models internalize concepts as approximately linear directions in representation space; used to interpret MDS injection behavior
• Linear Representation of Featuresconcept0.704
  The central object of study — the idea that a concept like truth is encoded as a direction in the LLM's latent space
• Internal model representationsconcept0.693
  The latent activations or embeddings inside a neural network.
• Subspace Decomposition of Representationsconcept0.691
  Investigation of whether a distributed representation can be further decomposed into sub-representations encoding component identities.
• Class-based inheritanceconcept0.691
• Non-Linear Representation Hypothesisconcept0.689
  Hypothesis that information may be encoded in arbitrary non-linear subspaces of a neural network
• In-Context Learning of Representations (Park et al. 2025)framework0.688
  Reports phase-like breakpoints and geometry changes as context scales; UCCT provides measurable predictor