{
  "id": "demonstration-label-sensitivity",
  "code": "PTL-0018",
  "term": "Demonstration Label Sensitivity",
  "aliases": [
    "role of demonstrations"
  ],
  "category": "exemplars",
  "definition": "Demonstration label sensitivity refers to how much a model's few-shot performance depends on whether the example labels are correct; research found that randomly replacing labels often hurts performance only slightly.",
  "description": "Min et al. showed that demonstrations mainly supply the label space, the input distribution, and the format of the task. This suggests few-shot examples work largely by specifying what the task looks like rather than by teaching the input-label mapping.",
  "example": null,
  "broader": [
    "in-context-learning"
  ],
  "narrower": [],
  "related": [
    "few-shot-prompting",
    "exemplar-selection"
  ],
  "introduced": null,
  "sources": [
    {
      "title": "Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?",
      "authors": "Min et al.",
      "year": 2022,
      "url": "https://arxiv.org/abs/2202.12837"
    }
  ],
  "url": "https://protologue.com/t/demonstration-label-sensitivity/",
  "citation": "Protologue. (2026). Demonstration Label Sensitivity. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.0.0, PTL-0018). https://protologue.com/t/demonstration-label-sensitivity/"
}