finding
active
finding:probe-cosine-similarity-becomes-more-peaked-with-deeper-layers-0-3-with-higher-secondary-ringsProbe cosine similarity becomes more peaked with deeper layers 0-3, with higher secondary rings
Sharpening.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Toy model.
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.
- Supported by the geometric transition visible in cosine similarity heatmaps for F0-F3.
- Geometric evidence for convergence to stable truth directions only for simpler tasks.
- Geometric evaluation of truth direction alignment across layers and prompt templates.
- Evidence for twisting.
- Validates that the contrast vector method and PCA-based PC1 capture the same direction
- Core result of Experiment 3: cross-model semantic convergence under self-referential processing
- E3 backbone-specific finding showing three-stage trajectory generalizes across architectures
- Shows the passive vs. active divide is more important than the specific wording of instructions.