concept
active
concept:many-shot-in-context-learning

Many-shot in-context learning

Regime with dozens to hundreds of examples where sudden learning curves are seen.

Neighborhood — ranked by edge-count

Concepts (1)

concept

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.

  • Technique using 0-20 in-context examples exhibiting a target trait to elicit behavioral shifts, used to validate persona vector monitoring
  • Few-shot learningconcept0.847
    Test-time adaptation from a small number of examples without parameter updates.
  • Use k examples as anchors with no parameter update.
  • Phenomenon where providing many untruthful or undesirable examples in context steers LM behaviour, related to reflective traits.
  • Providing k labeled examples in the prompt to steer model behavior.
  • Baseline method: sweeps over shot count and resamples prompts; calibrates threshold for P(TRUE)-P(FALSE); performed surprisingly weakly
  • Data representation where attributes can take multiple values, requiring scaling to become a formal context.
  • Demonstrated transformers on mathematical understanding and logic; cited to motivate transformer versatility.