hypothesis
active
hypothesis:adversarially-robust-models-would-have-worse-performance-and-be-more-interpretable-because-robustness-requires-giving-up-superpositionAdversarially robust models would have worse performance and be more interpretable because robustness requires giving up superposition
Predictions of superposition maximalism.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Toy Models of Superpositionmentions
Claims (1)
claim
- Superposition maximalist stance.
Hypotheses (1)
hypothesis
- Adversarial example transfer may be predicted by correlated arrangement of features in superpositionassociated_withSpeculative.
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.
- Adversarially robust models have different interpretability because adversarial training affects superpositionhypothesis0.905Could be tested by crosscoder comparisons.
- Future work.
- Novelty claim.
- Unexpected result.
- Figure 4.
- Practical implication.
- Robustness influences superposition, predicting lower SAE loss in more robust models.hypothesis0.789Popperian credit for surprising prediction.
- Robustness controls superposition.