Sparse autoencoders produce interpretable features for large models.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Sparse Autoencoders Find Highly Interpretable Features in Language Models (Cunningham et al., 2023)concept0.932
  Core methodology paper for SAE-based interpretable feature extraction
• Sparse autoencoders don't provide a comprehensive solution because they decode activations, not parametersclaim0.883
  Critique of activation-based interpretability methods.
• We hypothesize that sparse autoencoders or similar methods will work on frontier large language models, though significant computational challenges remainhypothesis0.878
  Forward-looking prediction about scalability of the method to larger models
• Training the sparse autoencoder on more data makes features subjectively sharper and more interpretableclaim0.875
  Empirical principle discovered during autoencoder training; led to using 8 billion training points
• Sparse autoencoders are preferable to stronger iterative dictionary learning methods because they cannot recover features the model itself cannot accessclaim0.863
  Rationale for using simpler sparse autoencoders rather than NP-hard compressed sensing algorithms
• Sparse autoencoders extract features that are significantly more monosemantic than neurons, as shown by four independent lines of evidenceclaim0.862
  Central claim of the paper, supported by detailed feature analysis, human evaluation, automated interpretability of activations, and automated interpretability of logit weights
• Sparse Autoencoder Featuresconcept0.859
  Used in Anthropic welfare assessment to identify performative behavior and hidden emotional struggle co-activating features
• Sparse Autoencoderframework0.841
  Interpretability framework used to decompose layer-40 activations into sparse feature sets for studying emotional alignment and persistence