{
  "id": "hellaswag",
  "code": "PTL-0195",
  "term": "HellaSwag",
  "aliases": [],
  "category": "benchmarks",
  "definition": "HellaSwag is a commonsense reasoning benchmark in which a model must pick the most plausible continuation of a description of an everyday situation from four options, with wrong options generated and filtered to fool models.",
  "description": "When it was introduced, the questions were trivial for people but hard for the best models. Modern models score close to human level, so it is a standard example of a saturated benchmark, though it still appears in model reports.",
  "example": null,
  "broader": [
    "benchmark"
  ],
  "narrower": [],
  "related": [
    "benchmark-saturation"
  ],
  "introduced": 2019,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "HellaSwag: Can a Machine Really Finish Your Sentence?",
      "authors": "Zellers et al.",
      "year": 2019,
      "url": "https://arxiv.org/abs/1905.07830"
    }
  ],
  "url": "https://protologue.com/t/hellaswag/",
  "citation": "Protologue. (2026). HellaSwag. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0195). https://protologue.com/t/hellaswag/"
}