method
active
method:combined-elicitation

Combined Elicitation

Elicitation strategy pooling responses from Description, Dialogue, and Narration before difference-of-means extraction

Neighborhood — ranked by edge-count

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

  • Baseline elicitation strategy using third-person character descriptions for base model persona extraction
  • Alternative elicitation using neutral scenarios continued as stories with few-shot exemplars establishing persona
  • Alternative elicitation using two-turn everyday conversations with a recurring character for persona extraction
  • Asking model to explain its own behavior after the fact when no chain-of-thought was available
  • Zaadnoordijk and Bayne's category of intentional action; sticker-removal behavior induced by the self-prior corresponds to this
  • Different text formats (Description, Dialogue, Narration) used to elicit persona behavior for vector extraction
  • Proposed cost-reduction procedure that predicts S/N/I label from unsteered baseline expression alone, replacing the full 30-configuration grid
  • Pearson correlation of feature activations across 40M tokens used to measure feature similarity and universality across models