question
active
question:can-a-subset-of-features-optimally-reconstruct-the-mean-of-the-activations-while-the-full-set-of-features-reconstructs-the-per-token-activationscan a subset of features optimally reconstruct the mean of the activations while the full set of features reconstructs the per-token activations?
Motivates nested SAE.
Source paper
extracted_from(2026) · Kevin Der · Harish Kamath · Thompson, Ben
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Table 1.
Claims (1)
claim
- Based on sufficiency scores.
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.
- Design choice validated empirically.
- Figure 7.
- §5.5.
- If the relationship is present in activations, better tools (e.g. trained on these datasets) could surface ithypothesis0.778Alternative explanation.
- Limitation of single-prompt selection.
- Core concern.
- Main proposal.
- Demonstrates bidirectional causal link: behavior manifold geometry can be recovered by optimizing in representation space.