{
  "name": "VizWiz-VQA",
  "release_date": "2024-01-01",
  "subsets": {
    "main": {
      "language": [
        "en"
      ],
      "modalities": [
        "single_image_start"
      ],
      "task_type": "short_answer_qa",
      "score_pipeline": [
        "vqa-accuracy"
      ],
      "score_protocol": {
        "reference": "lmms-eval@lmms_eval/tasks/vizwiz_vqa/utils.py:13-36 (vizwiz_vqa_process_results) verified: EvalAIAnswerProcessor then acc=min(1,#matches/3) over 10 answers. Published answer maps from source `answers` (10-item list) — CONFIRMED retained: rows show list[10].",
        "note": "Fractional per-sample VQA accuracy over 10 reference answers. VizWiz includes an 'unanswerable' class handled inside the same 10-answer voting; no separate aggregation."
      },
      "prompt_template": "<image>{{ question }}\nAnswer the question using a single word or phrase.",
      "mapping_from_source": {
        "media": {
          "from": "image",
          "type": "list",
          "min_items": 1,
          "max_items": 1
        },
        "id": {
          "from": "question_id"
        },
        "question": {
          "from": "question"
        },
        "answer": {
          "from": "answers",
          "optional": true
        },
        "source": {
          "format": "huggingface",
          "url": {
            "val": "https://huggingface.co/datasets/lmms-lab/VizWiz-VQA"
          }
        }
      },
      "prompt_template_source": {
        "origin": "official",
        "reference": "https://github.com/EvolvingLMMs-Lab/lmms-eval/blob/main/lmms_eval/tasks/vizwiz_vqa/utils.py (vizwiz_vqa_doc_to_text — canonical short-answer with VizWiz-specific 'When the provided information is insufficient, respond with Unanswerable.' or canonical trailer)",
        "notes": "Tier 4: lmms-eval VizWiz-VQA canonical evaluation prompt."
      }
    }
  }
}