method
active
method:whitening-of-span-embeddingsWhitening of span embeddings
Preprocessing step that uses dev-set covariance to standardize embedding scales before computing ρd and dr.
Neighborhood — ranked by edge-count
Concepts (1)
concept
- cohesion ρdassociated_withWithin-cluster tightness of target pattern representations.
Datasets (1)
dataset
- development set (dev pool)associated_withSet of 1,000 prompts with k=3 examples used for whitening and z-scoring calibration.
Artifacts (1)
artifact
- Main paper presenting UCCT and semantic anchoring framework.
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.
- Preprocessing step using dev-set covariance to standardize span embeddings before computing S
- Obtain instruction and example span embeddings at layer L* with chosen pooling.
- Extracting embeddings from instruction and example spans.
- Vector representations of individual tokens from genomic foundation models; the raw inputs to sequence pooling methods.
- Lagged time series used to capture dynamical dependencies.
- Preprocessing pipeline for standardizing ρd, dr, and S across layers/models using dev-set covariance
- Calibration protocol: whiten embeddings on dev pool, z-score ρd and dr per layer.
- A reinforcing interlock between different materials, mentioned alongside Deep Interlock in West Dean construction.