Value-Weighted Attention Pattern

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

• Value-Weighted Attention Pattern Visualizationmethod0.916
  Visualizing attention patterns weighted by the norm of value vectors to better show how much information is moved from each position
• attention computationconcept0.767
  Process using Q, K, V to compute a heat map over K and weighted sum of V.
• Focused Attention Cycleconcept0.750
  Formal model of meditation phenomenology: focus → distraction → awareness of distraction → redirection, derived from active inference.
• attention mechanismconcept0.742
  Core operation in transformers, computing weighted combinations of previous elements
• Attention is a generalization of convolution; all convolutions can be expressed as tensor products of fixed relative position attention patterns and weight matricesclaim0.741
  Mathematical equivalence showing the relationship between attention mechanisms and convolutional operations
• Identification of algorithms implemented in attention layers, distributed across attention headsfinding0.737
  VPD successfully recovered interpretable attention algorithms (previous-token behavior, syntax-boundary routing) in weight space without requiring manual decomposition across heads.
• attentional modeconcept0.735
  Variable aperture of attention, ranging from collapsed to expanded; human and model analogs.
• Attention Decayconcept0.730
  Decrease in attention paid to system prompt over conversational turns, leading to persona fidelity degradation (cited from Li et al.)