finding
active
finding:correlated-feature-pairs-collapse-into-principal-components-as-sparsity-decreases-6-features-hexagon-to-pcaCorrelated feature pairs collapse into principal components as sparsity decreases (6 features, hexagon to PCA)
Trade-off PCA vs superposition.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Toy Models of Superpositionmentions
Claims (1)
claim
- Correlated feature results.
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.
- Major prediction checked against InceptionV1.
- Primary visual evidence for linear truth representations in large LLMs
- Main proposal.
- Learning dynamics.
- m=2 experiments.
- Perturbing a single feature.
- Justifies PCA choice over UMAP or t-SNE for the node-structured RN model.
- Central interpretation.