Layer-wise anchoring trajectory

Plot of per-layer anchoring score S(ℓ) across model depth, revealing early dip, mid-layer peak, late standardization.

Neighborhood — ranked by edge-count

concept

Normalized area under S (AUSN)
associated_with
Average of per-layer S(ℓ) scores, summarizing the breadth of anchoring trajectory.
peak anchoring Sbmax
associated_with
Maximum layer-wise anchoring score across layers.

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

layer-wise trajectory analysismethod0.861
Computing per-layer S(ℓ) to summarize geometry.
layer-wise anchoring score S(ℓ) computationmethod0.767
Compute per-layer S(ℓ) = ρ̃d(ℓ) - d̃r(ℓ) - log k after whitening and standardization.
E3: Layer-wise Geometric Trajectory Analysismethod0.766
Quantitative study correlating layer-wise anchoring geometry (S_max, AUS_N) with behavioral thresholds θ50
Layer-wise trajectories show early enrichment, mid-layer alignment, and late re-clustering.claim0.762
Qualitative geometry pattern.
Three-Stage Layer Trajectoryconcept0.759
Empirically observed pattern in E3: early enrichment (ρd dips), mid-layer alignment (dr falls), late standardization (re-clustering)
Normalized Area Under Anchoring Trajectory (AUS_N)concept0.747
Mean of S(ℓ) across layers; weaker geometry correlate with θ50
Layer-wise anchoring peaks in a 'Goldilocks zone' between early and late layers.claim0.734
Qualitative characterization of optimal anchoring depth.
Peak Layer-wise Anchoring Score (S_max)concept0.725
Maximum of S(ℓ) across layers; geometry summary used to predict θ50