method
active
method:10-fold-cross-validation-across-magnitudes10-fold cross-validation across magnitudes
Held-out prediction protocol.
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- Closed-form model: ICL accumulates evidence via likelihood, steering shifts priors.
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.
- Classification-based comparison of interpretation abilities across IIT metrics and Span Representation for ToM score categories.
- Non-persona generality.
- Indicates lack of statistically significant differences between top methods
- Validation of judge model robustness by regrading 1000 responses with 4 additional judge models
- Tunes PELT cost, penalty and kernel bandwidth by splitting S into folds.
- Quantitative bound on observed alignment; raises the open question of whether this gap reflects noise or real misalignment
- Explicit textual or graphical links between parts of a work, dynamic and virtual.
- Footnote 2.