hypothesis
active
hypothesis:robustness-influences-superposition-predicting-lower-sae-loss-in-more-robust-modelsRobustness influences superposition, predicting lower SAE loss in more robust models.
Popperian credit for surprising prediction.
Source paper
extracted_from(2025) · Gorton, Liv · Lewis, Owen
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- More robust ResNet18 models consistently achieve lower TopK SAE reconstruction loss at a given sparsity level.associated_withsupportsEvidence in real models.
Questions (1)
question
- Posed on SAE results.
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.
- Lower SAE loss at fixed sparsity likely indicates less interference and less superposition.claim0.849Justifies SAE proxy.
- Authors' hedged interpretation.
- Discussion implication.
- Open question.
- Fig. 7d,e.
- Unexpected result.
- Novel finding that agentic self-evaluation of emotionality correlates with feature persistence
- Future work.