Where and how is information stored in model-internal representations?

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

• Internal model representationsconcept0.844
  The latent activations or embeddings inside a neural network.
• Feedback loops from internal model representationsconcept0.796
  Mechanism of using internal activations or representations to create corrective feedback during generation.
• LLM Internal Representationsconcept0.779
  High-dimensional vectors produced at each transformer layer for each input token; the primary substrate analyzed in this study.
• Stateful Internal Representationconcept0.772
  A representation that maintains stable activation across many tokens rather than being locally triggered by specific content
• Model's uncertainty information is encoded in the reflection directionclaim0.769
  Interpretive claim from probing experiment showing reflection direction features outperform baseline for uncertainty prediction
• Structure in representationsconcept0.765
  The central question of whether representational geometry implies corresponding computational structure
• The geometry of internal representations and the geometry of model behavior share a precise correspondence — representation geometry is a window into the inner world of neural networks.claim0.757
  The paper's deepest interpretive claim, asserting that representation structure and behavioral structure are not coincidentally aligned but deeply connected.
• We hypothesize that representation geometry drives model behavior — the geometric structure of internal representations causally shapes what models do externally.hypothesis0.749
  The causal hypothesis motivating the use of causality (intervention) as the lens connecting representation and behavior geometry.