concept
active
concept:memorization-in-superpositionMemorization in Superposition
Specific phrases or sequences memorized via binary features in superposition, enabling narrow pattern matching despite few neurons
Neighborhood — ranked by edge-count
Frameworks (1)
framework
- Superposition Hypothesisassociated_withCore theoretical framework: neural networks represent more features than neurons by encoding features as directions in superposition
Findings (1)
finding
- Demonstrates mechanistic memorization via feature assemblies in superposition
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Phenomenon where models represent more features than dimensions via almost-orthogonal directions.
- More computations than neurons/dimensions.
- Attentional features spread across multiple heads.
- Theoretical model of how neural networks encode more features than dimensions, informing linear representation work.
- The state in which a dialogue agent maintains multiple possible characters simultaneously, refined as the conversation proceeds
- Any method enabling enumeration of features.
- Superposition maximalist stance.
- superposition is deeply linked to the ability to identify and enumerate over all features in a modelquote0.784Safety rationale.