finding
active
finding:linear-probe-on-layer-1-residual-stream-predicts-character-count-with-r-2-of-0-985Linear probe on layer 1 residual stream predicts character count with R^2 of 0.985
Linear readout works but is not the true representation.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Central thesis.
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.
- Distributed counting.
- Intrinsic noise in representation.
- Beliefs linearly decodable.
- Top 6 PCs of 150 mean layer-2 residual vectors capture 95% of variance in the character count subspacefinding0.789Low-dimensional manifold.
- Random baseline context.
- Evidence for twisting.
- Shows model persona position is primarily determined by the most recent user message, not prior drift
- Demonstrates emotion-specific persistence beyond variance effects in Cogito