concept
active
concept:sae-feature-splittingSAE feature splitting
Explains lower off-target effects of SAE methods.
Neighborhood — ranked by edge-count
Papers (1)
paper
Concepts (1)
concept
- SAE featuresrelated_toThe individual, supposedly monosemantic directions learned by SAEs; argued here to fragment manifolds into disconnected pieces.
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.
- Observed across SAE scales, e.g., 'San Francisco' split into 11 features.
- Phenomenon where a feature in a small SAE splits into multiple finer features in a larger SAE.
- Internal concepts a model uses, extracted as human-interpretable components.
- Method of adding scaled versions of sparse autoencoder latent features during generation to causally modulate model behavior
- Hypothesized explanation.
- Interpretability tool underlying top-feature, details, steering tools.
- Core critique of sparse autoencoders: they break the geometric structure of representations, making it harder to see the big picture.
- Hard-to-steer features are split across layers spanning the SAE point so only a portion is intervened uponhypothesis0.768Cross-layer superposition conjecture.