method
active
method:autointerp-feature-card-pipelineAutointerp feature-card pipeline
Activation-triggered examples plus direct logit attribution.
Neighborhood — ranked by edge-count
Papers (1)
paper
Concepts (1)
concept
- Novel componentaboutFast-moving residual orthogonal to context.
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.
- Supervised stage method: model generates response, then critiques it according to a principle, then revises it; repeated multiple times.
- Autointerp labels distinguish features from their ten nearest neighbours in embedding spacefinding0.734Shown in graph for the nearest-neighbour test.
- Further methodological progress needed.
- §7.2 assertion.
- Claude-based explanation of features from max-activating examples.
- A pipeline employing controlled semantic oppositions to distill monosemantic functional features from sparse activation spaces.
- Five-step pipeline: task definition, decomposition, description, description-validation loop, final validation.