concept
active
concept:manifold-tilingManifold tiling
Latents tile a manifold to reduce loss.
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- Feature Manifoldsassociated_withHypothesized extension of superposition where features may be higher-dimensional manifolds rather than 1D directions
Concepts (1)
concept
- Pathological scaling regime (β<α)associated_withSAEs learn far fewer features than latents.
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.
- A smooth, potentially curved surface in activation space along which activations vary according to a coherent semantic dimension.
- Technique used to fit M_h and M_y from data; enables manifold steering.
- Localized features with overlapping support jointly encode position along a manifold.
- Edge-to-edge coverings of a surface with no overlaps or gaps
- High curvature embedding trading off capacity and distinguishability.
- An interpretability approach that describes representations in terms of entire curved manifolds rather than many small features.
- A smoothly varying lower-dimensional surface in activation space that captures a concept better than a straight linear direction.
- One-dimensional curved surface in internal activation space; the paper demonstrates alignment with behavior manifold.