finding
active
finding:j-lens-vectors-are-amplified-roughly-10x-by-mlp-blocks-in-workspace-range-while-neuron-output-directions-have-gain-near-1J-lens vectors are amplified roughly 10x by MLP blocks in workspace range while neuron output directions have gain near 1.
Broadcast depth.
Source paper
extracted_fromAuthors
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- MLP and attention analysis.
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.
- Directed modulation prediction.
- Layer structure.
- Structural signature.
- Limitation.
- Imagine instruction raises property name in J-lens while J-orthogonalized property probe stays at baseline.finding0.785J-space privileged.
- Mediation control.
- Mechanistic proof of concept.
- Related work.