method
active
method:k-shot-promptingk-shot prompting
Prompting technique where k example pairs are provided as anchors.
Neighborhood — ranked by edge-count
Methods (1)
method
- in-context k-shot promptingrelated_tosame_asUse k examples as anchors with no parameter update.
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.
- Providing k labeled examples in the prompt to steer model behavior.
- Technique using 0-20 in-context examples exhibiting a target trait to elicit behavioral shifts, used to validate persona vector monitoring
- Baseline method: sweeps over shot count and resamples prompts; calibrates threshold for P(TRUE)-P(FALSE); performed surprisingly weakly
- The baseline prompting method asking for a single response (e.g., 'Tell me a joke about coffee'), which suffers from mode collapse
- A list-level prompting baseline that asks for k responses in a single call without probability verbalization
- Alternative to preventative steering: prepending a trait-eliciting system prompt to training samples to cancel out training pressure
- Technique for extracting trait directions by contrasting model activations under trait-eliciting vs. trait-suppressing conditions
- Shot count needed to reach 50% accuracy; reflects when anchoring strength crosses critical value.