question
active
question:are-there-model-structures-that-account-for-memorization-and-generalization-across-large-swaths-of-training-dataAre there model structures that account for memorization and generalization across large swaths of training data?
Aggregate analysis motivation.
Source paper
extracted_from(2025) · Merullo, Jack · Vatsavaya, Srihita · Bushnaq, Lucius · Lewis, Owen
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Disentanglement evidence.
Claims (1)
claim
- Core novel interpretation.
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.
- Discussion and App. A.
- Opening question.
- Core goal statement.
- Selective pressure toward convergence via task generality
- What factors determine the generalisation of learned alignment maps beyond training data?question0.796Open question about the gap between Theorem 1's existence proof and practical learnability
- Future direction.
- Scaling laws for dictionary learning are unknown and needed to assess feasibility on frontier models
- The central hypothesis of the paper; the platonic representation hypothesis itself