question
active
question:do-saes-capture-such-manifoldsdo SAEs capture such manifolds?
Central research question.
Source paper
extracted_from(2026) · Usha Bhalla · Thomas Fel · Can Rager · Sheridan Feucht +8
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- SAEs do not achieve compact capture; they are in dilution.
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.
- Theoretical framework answer.
- Limitation statement.
- Justifies focusing on direction-based SAEs.
- Manifold-level descriptions recover overarching semantic structure that SAE features miss.claim0.780Positive claim that geometric descriptions retain the conceptual coherence lost in atomized feature decompositions.
- Restricted to specific manifolds, different SAEs show strong alignment of induced coordinate systemsfinding0.775Geometry is shared even if features are not.
- Core critique of sparse autoencoders: they break the geometric structure of representations, making it harder to see the big picture.
- Related to point-based simplicial limit.
- Features act as local patch detectors.