claim
active
claim:a-difference-in-co-activations-or-subtle-feature-context-between-models-at-least-partly-explains-the-symmetryA difference in co-activations or subtle feature context between models at least partly explains the symmetry
Plausible explanation.
Source paper
extracted_fromSiddharth Mishra-Sharma · Trenton Bricken · Jack Lindsey · Adam Jermyn +4
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Supports the co-activation explanation of symmetry.
Hypotheses (1)
hypothesis
- From Lindsey et al; modeled with co-activation toy.
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.
- Evolution pivoted the same problem-solving strategies across different domains.
- Explicitly identified research gap: anecdotal evidence exists but rigorous characterization is absent
- Third empirical pattern.
- Claim distinguishing the deadening effect of large-scale neoclassicist symmetry from the vitalizing effect of numerous overlapping local symmetries
- Based on toy model and Kissane replication.
- Main proposal.
- Methodological claim about the scientific value of combining causal abstraction with representational geometry analysis
- Low-cosine shared features are best interpreted as same features used differently in the two modelsclaim0.769Interpretation of cosine distribution.