{
  "@context": {
    "@vocab": "https://schema.org/",
    "FileObject": "https://mlcommons.org/croissant/FileObject",
    "Table": "https://mlcommons.org/croissant/Table",
    "Column": "https://mlcommons.org/croissant/Column",
    "dataType": "https://mlcommons.org/croissant/dataType",
    "image": "https://mlcommons.org/croissant/image",
    "list": "https://mlcommons.org/croissant/list",
    "int": "http://www.w3.org/2001/XMLSchema#int",
    "string": "http://www.w3.org/2001/XMLSchema#string"
  },
  "@type": "Dataset",
  "name": "symbolic-qna",
  "description": "A symbolic reasoning dataset with shape-image programs and multiple choice questions.",
  "source": {
    "url": "https://huggingface.co/datasets/anonymoussifaka/mathtakestwo"
  },
  "data": [
    {
      "@type": "FileObject",
      "name": "qna_file",
      "description": "QnA data in Parquet format.",
      "encodingFormat": "application/parquet",
      "url": "https://huggingface.co/datasets/anonymoussifaka/mathtakestwo/resolve/main/data/train-00000-of-00001.parquet",
      "contains": {
        "@type": "Table",
        "name": "qna_table",
        "description": "Main QnA table with image-based questions and answers.",
        "columns": [
          {
            "@type": "Column",
            "name": "query",
            "description": "The main query image rendered from a symbolic program.",
            "dataType": "image"
          },
          {
            "@type": "Column",
            "name": "option_0",
            "description": "Candidate answer option 0.",
            "dataType": "image"
          },
          {
            "@type": "Column",
            "name": "option_1",
            "description": "Candidate answer option 1.",
            "dataType": "image"
          },
          {
            "@type": "Column",
            "name": "option_2",
            "description": "Candidate answer option 2.",
            "dataType": "image"
          },
          {
            "@type": "Column",
            "name": "option_3",
            "description": "Candidate answer option 3.",
            "dataType": "image"
          },
          {
            "@type": "Column",
            "name": "answer",
            "description": "Correct option index (0–3).",
            "dataType": "int"
          },
          {
            "@type": "Column",
            "name": "program",
            "description": "The symbolic program string that generated the query.",
            "dataType": "string"
          },
          {
            "@type": "Column",
            "name": "option_symbols",
            "description": "Symbolic program strings for each of the four options.",
            "dataType": "list[string]"
          }
        ]
      }
    }
  ],
  "licenses": [
    {
      "name": "MIT License",
      "identifier": "MIT"
    }
  ],
  "keywords": ["symbolic reasoning", "vision-language", "compositionality"]
}