Downstream Client Feature Analysis

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

• Feature Sparsityconcept0.731
  Property that features activate on only a small fraction of inputs; enables compressed sensing and is what allows superposition to work
• Learned features reflect the functionality of the model and not just the data distribution, as evidenced by interpretable downstream effects not used in dictionary learningclaim0.730
  Authors argue features are model properties because logit effects and ablations are consistent with feature interpretations
• Action Featuresconcept0.728
  Dual interpretation of features: in addition to responding to inputs, features also act to increase probability of specific output tokens
• Feature Visualizationmethod0.725
  Method of optimizing input to cause a neuron to fire maximally, used to characterize what a neuron detects; establishes causal link
• Feature splittingconcept0.717
  Phenomenon where a feature in a small SAE splits into multiple finer features in a larger SAE.
• feature as applicationconcept0.709
  Metaphor treating each system feature or function as a separate application that can be independently loaded and managed.
• Superposition of Sparse Featuresconcept0.705
  Mechanistic finding by Bricken et al. 2023 about how LLMs store features; cited as operational justification for pattern-repository assumption
• Event Analysis of Systemic Teamwork (EAST)method0.704
  Network analysis method used to examine distributed cognition in multi-agent systems; demonstrates measurement approach for collective cognitive processes.