method
active
method:cohen-s-d-layer-selection-sweepCohen's d layer selection sweep
Layer selection for probes: maximizes Cohen's d on held-out evaluation texts, restricted to middle 60% of layers
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.
- Procedure of systematically varying the layer at which activations are recorded and injected.
- A process by which vasomotion collapses ambivalent neural patterns into durable definite states, reducing complexity.
- Framework dividing covariance of character and fitness into between-collective and within-collective components; addresses limitation of kin selection.
- Preprocessing pipeline for standardizing ρd, dr, and S across layers/models using dev-set covariance
- Representation of features spread across multiple layers, complicating dictionary learning.
- Systematic sweep of 10 boost levels from threshold-3σ to threshold+3σ to characterize ESR vs. steering strength
- Layer-wise trajectories show early enrichment, mid-layer alignment, and late re-clustering.claim0.695Qualitative geometry pattern.