claim
active
claim:the-positional-mechanism-generalizes-poorly-becoming-noisy-in-middle-positions-as-entity-count-growsThe positional mechanism generalizes poorly, becoming noisy in middle positions as entity count grows
Challenges prevailing view.
Source paper
extracted_from(2025) · Yoav Gur-Arieh · Mor Geva · Atticus Geiger
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (6)
finding
- Learned parameters.
- Gaussian matters.
- Mixed predictions cluster near positional index with diffuse logit distribution in middle indicessupportsFigures 3, 13.
- Figures 14–17.
- Core result.
- Figure 2.
Claims (1)
claim
- Central thesis.
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.
- Interpretation of padding results.
- When lexical/reflexive targets absent from original, model falls back to noisy positional mechanismfinding0.778Fig. 28–29.
- Proposes middle-range entity quality as the criterion for judging the success of a building process
- Prevailing view: retrieve entity by group position index.
- §D.3.
- Hidden states of middle entity groups become less linearly separable by position as n growsfinding0.760Fig. 22.
- Mechanism agreement.
- Fig. 19.