finding
active
finding:dictionaries-of-different-sizes-learn-character-count-features-with-similar-receptive-fields-and-ringing-decoder-geometryDictionaries of different sizes learn character count features with similar receptive fields and ringing decoder geometry
Canonical featurization.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Appendix claim.
Hypotheses (1)
hypothesis
- Canonical dilating feature structure emerges from boundary constraint, with positions near zero anchoring feature developmentassociated_withsupportsAppendix hypothesis.
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.
- Ringing evidence.
- Feature presence depends on concept frequency in training data, with a threshold scaling inversely with alive features.
- Biological parallel.
- what is the 'correct number of features' for dictionary learning, and is this question well-posed?question0.761Open question about whether there is a true discrete feature count or a continuous splitting process
- Role of extra dimensions.
- Motivates illusion experiment.
- Head-level causal evidence.
- Naive weight magnitude works here.