few-shot prompting

Providing k labeled examples in the prompt to steer model behavior.

Neighborhood — ranked by edge-count

paper

framework

Constitutional AI
implements
Alignment approach by Anthropic that explicitly trains self-observation; predicts highest baseline and lowest prompt lift.

concept

in-context learning (ICL)
implements
Test-time adaptation from prompt or retrieved context with no parameter updates.

method

Calibrated Few-Shot Prompting
related_to
Baseline method: sweeps over shot count and resamples prompts; calibrates threshold for P(TRUE)-P(FALSE); performed surprisingly weakly
Many-Shot Prompting
related_to
Technique using 0-20 in-context examples exhibiting a target trait to elicit behavioral shifts, used to validate persona vector monitoring
Multiple-choice evaluation method for PM training
implements
Using language model log probabilities of answer choices (A)/(B) to produce preference labels.
Critique-Revision Pipeline
implements
Supervised stage method: model generates response, then critiques it according to a principle, then revises it; repeated multiple times.

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.

Few-shot learningconcept0.860
Test-time adaptation from a small number of examples without parameter updates.
k-shot promptingmethod0.846
Prompting technique where k example pairs are provided as anchors.
Calibrated few-shot prompting was a surprisingly weak baseline for truth classification compared to linear probesfinding0.811
Unexpected finding that behavioral baseline underperforms representational probing approaches
in-context k-shot promptingmethod0.805
Use k examples as anchors with no parameter update.
few-shot threshold (k50/θ50)concept0.790
Shot count needed to reach 50% accuracy; reflects when anchoring strength crosses critical value.
System prompting and few-shot produce highly similar activations (cosine 0.83-0.92 at safety-critical layers), indicating a shared prompt-side mechanism distinct from activation steering.claim0.788
Evidence for two representational pathways based on cross-method activation divergence
Few-shot prompting had a negative effect on HH-intent for smaller models and a significant positive impact on larger models.finding0.773
Ablation result from Experiment 3 on few-shot prompting effects.
Chain-of-thought prompting produced a similar increase in helpful intention of large models as few-shot prompting.finding0.772
Ablation result from Experiment 3 on chain-of-thought prompting effects.