finding
active
finding:sae-reconstructions-with-increasing-numbers-of-features-approximate-manifolds-piecewise-linearly-each-feature-covering-a-local-patchSAE reconstructions with increasing numbers of features approximate manifolds piecewise-linearly, each feature covering a local patch
Features act as local patch detectors.
Source paper
extracted_from(2026) · Usha Bhalla · Thomas Fel · Can Rager · Sheridan Feucht +8
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Central empirical interpretation.
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.
- Core critique of sparse autoencoders: they break the geometric structure of representations, making it harder to see the big picture.
- Top SAE features on the years manifold show localized smooth tuning curves with overlapping supportfinding0.800Reminiscent of population coding.
- Core concern.
- Theoretical framework answer.
- Justifies focusing on direction-based SAEs.
- Related to point-based simplicial limit.
- Manifold-level descriptions recover overarching semantic structure that SAE features miss.claim0.786Positive claim that geometric descriptions retain the conceptual coherence lost in atomized feature decompositions.
- Open question in limitations.