SAE feature firing probability persistence metric

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

• SAE featuresconcept0.755
  The individual, supposedly monosemantic directions learned by SAEs; argued here to fragment manifolds into disconnected pieces.
• SAE Feature #77278 fires 195,040 times in corpus, associated with satisfaction vs. emptiness dimensionfinding0.754
  High-frequency SAE feature reported as controlling fundamental positive vs. negative affect dimension
• SAE-based persistence replication of probe-based findings (no shared probe confounds)claim0.745
  The SAE self-evaluation persistence finding serves as a replication of probe-based results that shares no potential probe construction confounds
• SAE Feature Emotion Subspace Overlap Metricmethod0.743
  Fraction of an SAE feature's length lying inside the 171-dimensional subspace spanned by emotion probes, computed via SVD orthogonalization
• Self-evaluated emotionality and textual evaluation of SAE features predict persistence in opposite directions.claim0.741
  Surprising finding that the two evaluation methods diverge in their relationship with persistence
• SAE feature steering in history, conceptual, and zero-shot control conditions produces zero experience reports under either suppression or amplificationfinding0.733
  Shows gating effect is specific to the self-referential computational regime, not a general feature effect
• Token-100 correlation persistence metricmethod0.730
  Measures emotion feature persistence as correlation between z-scored activation at token 0 and token 100 across all eligible target model tokens
• autoregressive persistenceconcept0.730
  Baseline persistence of any probe direction arising from the autoregressive nature of LLMs, not specific to emotion content