framework
active
framework:relu-saeReLU SAE
Baseline SAE architecture.
Neighborhood — ranked by edge-count
Papers (2)
paper
Findings (1)
finding
- ReLU underperforms.
Frameworks (1)
framework
- TopK SAEextendsEvaluated architecture (Gao et al.).
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.
- SAE variant trained on ResNet18.
- Trained on toy manifolds.
- Design justification.
- Hybrid with separate turn-averaged and residual per-token SAEs.
- Interpretable features extracted by sparse autoencoders used as steering targets in this study
- Hybrid SAE jointly training turn-averaged inner and per-token outer partitions.
- Evaluated architecture.
- Unlearning not discriminative.