finding
active
finding:with-n-5-features-in-m-2-dimensions-making-one-feature-sufficiently-sparse-collapses-pentagon-into-digons-via-first-order-phase-changeWith n=5 features in m=2 dimensions, making one feature sufficiently sparse collapses pentagon into digons via first-order phase change
Perturbing a single feature.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Toy Models of Superpositionmentions
Claims (1)
claim
- Promising for real models.
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.
- Both empirical and theoretical phase diagrams.
- Resembles fractional quantum Hall effect.
- Learning dynamics.
- Trade-off PCA vs superposition.
- Features jump between dimensionalities during training with sudden loss drops in digon modelsfinding0.746Learning dynamics.
- Mechanistic claim linking identified Fourier features to base-10 arithmetic
- Explains why time and sequence are essential for generated complexity.
- Interpretation that the geometric properties of living structure are not arbitrary but arise inevitably from the smooth unfolding process.