hypothesis
active
hypothesis:latents-near-digit-alphabet-manifolds-may-have-tiled-a-manifold-with-one-killed-to-reduce-l0Latents near digit/alphabet manifolds may have tiled a manifold with one killed to reduce L0
Speculative explanation.
Source paper
extracted_from(2025) · Michaud, Eric J. · Gorton, Liv · McGrath, Tom
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (2)
finding
- Possible manifold signal.
- Training artifact.
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.
- Many latents tiling a low-dimensional manifold would show decoder cosine similarities near 1hypothesis0.827Motivates Appendix E.
- Appendix claim.
- Directly identifies saddle points with near-miss solution attempts, the mechanistic core of the paper's account.
- Point-based landmarks tiling a joint manifold lie at distance ≥ δ−ε from factor manifold M1 (Lemma 1)finding0.761Formal result: point-based SAEs cannot recover factor manifolds under superposition.
- Core concern.
- Central thesis.
- Strongest argument against pathology.
- Neuroscience-motivated prediction.