Mean Squared Error (MSE) for Personality Evaluation

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

• Mean Absolute Error (MAE) for Personality Evaluationmethod0.836
  Per-dimension error metric for stability across paraphrased personality questions
• Mean Squared Error between self and other activationsmethod0.812
  The specific implementation of SOO loss using MSE between self_attn.o_proj outputs at a specified layer
• Self-evaluated emotionality and textual evaluation of SAE features predict persistence in opposite directions.claim0.749
  Surprising finding that the two evaluation methods diverge in their relationship with persistence
• Self-evaluated emotionality of SAE features negatively correlates with activation variance explained (ρ = -0.184, p = 4.6e-09), requiring variance correction to reveal the persistence signal.finding0.738
  Explains why variance correction is needed to see the self-evaluation–persistence relationship
• Some personality traits may be less accessible using linear methods than othershypothesis0.735
  Author's open question about limitations of linear persona vector approach
• SAE features can be grounded in clinical taxonomy (abnormality, age, sex, medication) to benchmark monosemanticity and entanglement.claim0.731
  Claim that feature grounding enables interpretability metrics.
• Personality/phenomenology is defined by how one metabolizes uncertainty.claim0.731
  Defines individual differences through uncertainty processing styles.
• Different SAE latents control different misalignment behavior categories: toxic persona (#10) drives illegal recommendations, sarcasm/satire (#31) drives factual incorrectnessfinding0.730
  Latents are specialized to different modes of misalignment, explaining diverse misalignment profiles