finding
active
finding:pattern-replicates-across-nine-models-2-72b-and-ten-tasksPattern replicates across nine models (2–72B) and ten tasks
Generalization.
Source paper
extracted_from(2025) · Yoav Gur-Arieh · Mor Geva · Atticus Geiger
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (2)
claim
- LMs retrieve bound entities via a mixture of positional, lexical and reflexive mechanisms rather than positional aloneassociated_withCentral thesis.
- Cross-model consistency.
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.
- Selective pressure toward convergence via task generality
- Are there model structures that account for memorization and generalization across large swaths of training data?question0.754Aggregate analysis motivation.
- Interpretation of early answering/mistakes results.
- Third of three speculative claims asserting that learned features are not model-specific but represent universal solutions to learning problems
- Observed by Anima Labs in untrained base models; not present in training data, implying computational origin of self-reported parallel processing.
- Fig. 23.
- Empirical result showing alignment increases with model competence