SAE sparse features (100K+ features, 64 active per token)

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

• For all three SAEs (1M, 4M, 34M), average active features per token <300, and reconstruction variance explained ≥65%.finding0.822
  Basic SAE performance metrics.
• Sparse Autoencoders (SAE)method0.782
  Interpretability method criticized in this paper for shattering manifolds into atomic pieces, obscuring overarching semantic structure.
• 34M SAE had roughly 65% dead features.finding0.775
  Most features dead in largest SAE, indicating room for improvement.
• SAE featuresconcept0.769
  The individual, supposedly monosemantic directions learned by SAEs; argued here to fragment manifolds into disconnected pieces.
• SAE Feature #92372 fires 666,235 times in corpus, associated with urgency vs. receptive calm dimensionfinding0.768
  Example of a highly active SAE feature modulating urgency versus acceptance as an emotional dimension
• SAE Feature #77278 fires 195,040 times in corpus, associated with satisfaction vs. emptiness dimensionfinding0.768
  High-frequency SAE feature reported as controlling fundamental positive vs. negative affect dimension
• TopK Sparse Autoencoders (SAEs)method0.767
  Sparse dictionary learning method used to extract interpretable features from EEG transformer embeddings.
• SAEs successfully extract sparse feature dictionaries from embeddings of SleepFM, REVE, and LaBraM EEG transformers.finding0.765
  Foundational empirical result enabling all downstream analysis