method
active
method:contrastive-distractor-scoring-of-feature-explanationsContrastive distractor scoring of feature explanations
Scores explanations by distinguishing activating from distractor samples; weakly penalises polysemanticity.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Claude-based explanation of features from max-activating examples.
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.
- Method comparing brain activity in conscious vs. unconscious conditions.
- Limitation of single-prompt selection.
- Core insight.
- Technique for extracting trait directions by contrasting model activations under trait-eliciting vs. trait-suppressing conditions
- The property that living structures contain intense contrast—far more than one imagines helpful; true opposites which annihilate each other when superimposed, creating differentiation that gives birth to something; contrast unifies rather than separates when used correctly
- Adaptation of instruction-tuned extraction to base models using third-person descriptions and hypothetical situations
- Property that features activate on only a small fraction of inputs; enables compressed sensing and is what allows superposition to work
- Per-(emotion, token) z-score computed as injected emotion activation minus mean of 170 other probes, contrasted against no-steering baseline