finding
active
finding:grassmannian-bsf-with-b-1-describes-dinov3-activations-in-fewer-bits-than-the-b-1-sae-at-every-width-and-sparsity-minimum-at-b-between-2-and-4

Grassmannian BSF with b>1 describes DINOv3 activations in fewer bits than the b=1 SAE at every width and sparsity, minimum at b between 2 and 4.

MDL result at 20% distortion.

Source paper

extracted_from
Structuring Sparsity: Block-Sparse Featurizers Capture Visual Concept Manifolds
(2026) · Thomas Fel · Matthew Kowal · Mozes Jacobs · Dron Hazra +21

Neighborhood — ranked by edge-count

Datasets (1)

dataset
  • DINOv3
    about
    Main vision model whose activations are decomposed.