framework
active
framework:grassmannian-bsfGrassmannian BSF
Tied encoder/decoder with orthonormal charts and learned scalar gamma.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Keeps top-k blocks by norm.
Frameworks (1)
framework
- Featurizers whose atoms are blocks of directions rather than single directions.
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.
- MDL result at 20% distortion.
- Curve manifold recovery.
- Free encoder/decoder with block top-k projection.
- Linear encoder with block soft-threshold and group lasso penalty.
- BSF codes improve on the basic probe for classification and approach it for segmentation and depth.finding0.694Linear probing.
- Core idea summary from Figure 1.
- Block and Grassmannian featurizers keep utilization ≈0.6–0.66 at k=16, while Group Lasso falls to ≈0.17.finding0.668Per-feature occupancy.
- Features that correspond to a single semantic concept and are effective for steering behavior.