finding
active
finding:superposition-is-never-optimal-in-linear-model-interference-outweighs-benefitSuperposition is never optimal in linear model (interference outweighs benefit)
Mathematical derivation from Saxe-style loss.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Toy Models of Superpositionmentions
Claims (1)
claim
- Explanation for neuron interpretability.
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.
- Core mechanism statement.
- Supported by Table 1.
- Cost of removing superposition.
- Discussion implication.
- Central thesis.
- Limit on Jacobian matching.
- Mechanistic explanation for why superposition is geometrically feasible
- Because superposition is a phase, there may exist a regime for every model with no superpositionclaim0.770Cause for hope.