hypothesis
active
hypothesis:a-block-sparse-code-should-hold-the-curve-manifold-together-where-a-directional-code-fragments-itA block-sparse code should hold the curve manifold together where a directional code fragments it.
Prediction tested on InceptionV1.
Source paper
extracted_from(2026) · Thomas Fel · Matthew Kowal · Mozes Jacobs · Dron Hazra +21
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Curve manifold recovery.
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.
- Rationale for manifold steering.
- Neuroscience analogy.
- The same block construction predicts a richer contour code (curvature plus position) in deeper layers.hypothesis0.762Forward prediction analogous to V4.
- Proposes that nonlinear geometric structure is superior to linear feature spaces for capturing semantic content.
- A combination of crosscoders and flexible decoding (Matryoshka, iterative codebook) could resolve the steering-classification tensionhypothesis0.750Proposed remedy.
- Core concern.
- The empirical-observational claim grounded in the diverse case studies presented in the chapter
- Manifold structure is present in representations and causally influences downstream behaviorclaim0.739Interpretation of steering results.