method
active
method:topk-sparse-autoencoders-saesTopK Sparse Autoencoders (SAEs)
Sparse dictionary learning method used to extract interpretable features from EEG transformer embeddings.
Neighborhood — ranked by edge-count
Methods (2)
method
- Sparse Autoencoders (SAE)related_toInterpretability method criticized in this paper for shattering manifolds into atomic pieces, obscuring overarching semantic structure.
- Sparse autoencoder (SAE) featuresrelated_toInterpretability tool underlying top-feature, details, steering tools.
Events (1)
event
- Preprint applying TopK SAEs to three EEG transformers to reveal sparse feature dictionaries, steering regimes, and spectral interpretation.
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.
- The central mechanistic interpretability tool applied across all three EEG transformers to extract sparse feature dictionaries
- Standard interpretability approach that VPD critiques and proposes an alternative to.
- Evaluated architecture (Gao et al.).
- The primary novel framework introduced in the paper for learning facet-level personality control vectors
- Interpretability framework used to decompose layer-40 activations into sparse feature sets for studying emotional alignment and persistence
- First sparse autoencoders trained on the 671B-parameter DeepSeek R1, open-sourced as two SAEsfinding0.805Headline result: first public interpreters on a true reasoning model at this scale.
- Foundational empirical result enabling all downstream analysis
- Increasing TopK sparsity drives synthetic SAEs through shattering, capture, then dilution regimesfinding0.793Phase diagram on support size and receptive field spread.