finding
active
finding:r-2-of-character-count-prediction-is-0-93-using-5-key-layer-0-heads-versus-0-97-with-11-heads-in-first-two-layersR^2 of character count prediction is 0.93 using 5 key Layer 0 heads versus 0.97 with 11 heads in first two layers
Distributed counting.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Distributed algorithm.
Questions (1)
question
- Intro question.
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.
- Linear readout works but is not the true representation.
- Shows model persona position is primarily determined by the most recent user message, not prior drift
- Justifies focusing on attention.
- Top 6 PCs of 150 mean layer-2 residual vectors capture 95% of variance in the character count subspacefinding0.750Low-dimensional manifold.
- Fig. 58.
- Intrinsic noise in representation.
- Evidence for twisting.
- Quantitative argument for the richness of quasi-psychological connections enabled by attention streams