croissant-winogrande_za_old.json

{
  "@context": {
    "@language": "en",
    "@vocab": "https://schema.org/",
    "citeAs": "cr:citeAs",
    "column": "cr:column",
    "conformsTo": "dct:conformsTo",
    "cr": "http://mlcommons.org/croissant/",
    "rai": "http://mlcommons.org/croissant/RAI/",
    "data": {
      "@id": "cr:data",
      "@type": "@json"
    },
    "dataType": {
      "@id": "cr:dataType",
      "@type": "@vocab"
    },
    "dct": "http://purl.org/dc/terms/",
    "examples": {
      "@id": "cr:examples",
      "@type": "@json"
    },
    "extract": "cr:extract",
    "field": "cr:field",
    "fileProperty": "cr:fileProperty",
    "fileObject": "cr:fileObject",
    "fileSet": "cr:fileSet",
    "format": "cr:format",
    "includes": "cr:includes",
    "isLiveDataset": "cr:isLiveDataset",
    "jsonPath": "cr:jsonPath",
    "key": "cr:key",
    "md5": "cr:md5",
    "parentField": "cr:parentField",
    "path": "cr:path",
    "recordSet": "cr:recordSet",
    "references": "cr:references",
    "regex": "cr:regex",
    "repeated": "cr:repeated",
    "replace": "cr:replace",
    "sc": "https://schema.org/",
    "separator": "cr:separator",
    "source": "cr:source",
    "subField": "cr:subField",
    "transform": "cr:transform"
  },
  "@type": "sc:Dataset",
  "name": "Winogrande-ZA-old",
  "description": "The Winogrande commonsense reasoning LLM evaluation benchmark (sourced from https://github.com/allenai/winogrande) train_s, test, and dev sections human-translated into each of the South African Languages Afrikaans, Zulu, and Xhosa using Upwork.com translators (one initial translation round only). The original English version of these sections is also included.",
  "conformsTo": "http://mlcommons.org/croissant/1.0",
  "citeAs": "@article{,\n  title={Expanding Reasoning Benchmarks in Low-Resourced African Languages: Winogrande and Clinical MMLU in Afrikaans, Xhosa, and Zulu},\n  author={Tuka Alhanai and Adam Kasumovic and Mohammad Ghassemi and Guillaume Chabot-Couture},\n  year={2024}\n}",
  "datePublished": "2024-06-06 00:00:00",
  "inLanguage": [
    "en",
    "af",
    "xh",
    "zu"
  ],
  "license": "MIT",
  "url": "https://github.com/InstituteforDiseaseModeling/winogrande-mmlu-clinical-za",
  "version": "1.0.0",
  "distribution": [
    {
      "@type": "cr:FileObject",
      "@id": "github-repository",
      "name": "github-repository",
      "description": "InstituteforDiseaseModeling repository on GitHub.",
      "contentUrl": "https://github.com/InstituteforDiseaseModeling/winogrande-mmlu-clinical-za",
      "encodingFormat": "git+https",
      "sha256": "main"
    },
    {
      "@type": "cr:FileSet",
      "@id": "en_file_set",
      "name": "en_file_set",
      "description": "Winogrande (Train S + Dev + Test) in English (old version)",
      "containedIn": {
        "@id": "github-repository"
      },
      "encodingFormat": "text/csv",
      "includes": "winogrande-mmlu-clinical-za/winogrande_za_old/winogrande.csv"
    },
    {
      "@type": "cr:FileSet",
      "@id": "af_file_set",
      "name": "af_file_set",
      "description": "Winogrande (Train S + Dev + Test) in Afrikaans (old version)",
      "containedIn": {
        "@id": "github-repository"
      },
      "encodingFormat": "text/csv",
      "includes": "winogrande-mmlu-clinical-za/winogrande_za_old/winogrande_af.csv"
    },
    {
      "@type": "cr:FileSet",
      "@id": "zu_file_set",
      "name": "zu_file_set",
      "description": "Winogrande (Train S + Dev + Test) in Zulu (old version)",
      "containedIn": {
        "@id": "github-repository"
      },
      "encodingFormat": "text/csv",
      "includes": "winogrande-mmlu-clinical-za/winogrande_za_old/winogrande_zu.csv"
    },
    {
      "@type": "cr:FileSet",
      "@id": "xh_file_set",
      "name": "xh_file_set",
      "description": "Winogrande (Train S + Dev + Test) in Xhosa (old version)",
      "containedIn": {
        "@id": "github-repository"
      },
      "encodingFormat": "text/csv",
      "includes": "winogrande-mmlu-clinical-za/winogrande_za_old/winogrande_xh.csv"
    }
  ],
  "recordSet": [
    {
      "@type": "cr:RecordSet",
      "@id": "en",
      "name": "en",
      "field": [
        {
          "@type": "cr:Field",
          "@id": "en/qID",
          "name": "qID",
          "description": "Unique identifier for the question. The last character denotes the answer. Corresponds with the same qIDs in the new version following a second round of corrections and can therefore be joined with the new dataset to compare translation quality or obtain splits.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "en_file_set"
            },
            "extract": {
              "column": "qID"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "en/sentence",
          "name": "English Sentence",
          "description": "English sentence that is missing a word or a few words (denoted with an underscore), requiring one of two options to best fill in the missing word or words.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "en_file_set"
            },
            "extract": {
              "column": "sentence"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "en/option1",
          "name": "English Option 1",
          "description": "English option 1. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "en_file_set"
            },
            "extract": {
              "column": "option1"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "en/option2",
          "name": "English Option 2",
          "description": "English option 2. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "en_file_set"
            },
            "extract": {
              "column": "option2"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "en/answer",
          "name": "Answer",
          "description": "Correct option (a number 1 or 2 corresponding to Option 1 or Option 2). Note that for rows from the original test set, the answer will be listed as 0 since no answer was provided by the original Winogrande creators. However, the answers for the test set rows can be found in the last character of the qID.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "en_file_set"
            },
            "extract": {
              "column": "answer"
            }
          }
        }
      ]
    },
    {
      "@type": "cr:RecordSet",
      "@id": "af",
      "name": "af",
      "field": [
        {
          "@type": "cr:Field",
          "@id": "af/qID",
          "name": "qID",
          "description": "Unique identifier for the question. The last character denotes the answer. Corresponds with the same qIDs in the new version following a second round of corrections and can therefore be joined with the new dataset to compare translation quality or obtain splits.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "af_file_set"
            },
            "extract": {
              "column": "qID"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "af/sentence",
          "name": "Afrikaans Sentence",
          "description": "Afrikaans sentence that is missing a word or a few words (denoted with an underscore), requiring one of two options to best fill in the missing word or words.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "af_file_set"
            },
            "extract": {
              "column": "sentence"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "af/option1",
          "name": "Afrikaans Option 1",
          "description": "Afrikaans option 1. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "af_file_set"
            },
            "extract": {
              "column": "option1"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "af/option2",
          "name": "Afrikaans Option 2",
          "description": "Afrikaans option 2. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "af_file_set"
            },
            "extract": {
              "column": "option2"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "af/answer",
          "name": "Answer",
          "description": "Correct option (a number 1 or 2 corresponding to Option 1 or Option 2). Note that for rows from the original test set, the answer will be listed as 0 since no answer was provided by the original Winogrande creators. However, the answers for the test set rows can be found in the last character of the qID.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "af_file_set"
            },
            "extract": {
              "column": "answer"
            }
          }
        }
      ]
    },
    {
      "@type": "cr:RecordSet",
      "@id": "zu",
      "name": "zu",
      "field": [
        {
          "@type": "cr:Field",
          "@id": "zu/qID",
          "name": "qID",
          "description": "Unique identifier for the question. The last character denotes the answer. Corresponds with the same qIDs in the new version following a second round of corrections and can therefore be joined with the new dataset to compare translation quality or obtain splits.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "zu_file_set"
            },
            "extract": {
              "column": "qID"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "zu/sentence",
          "name": "Zulu Sentence",
          "description": "Zulu sentence that is missing a word or a few words (denoted with an underscore), requiring one of two options to best fill in the missing word or words.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "zu_file_set"
            },
            "extract": {
              "column": "sentence"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "zu/option1",
          "name": "Zulu Option 1",
          "description": "Zulu option 1. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "zu_file_set"
            },
            "extract": {
              "column": "option1"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "zu/option2",
          "name": "Zulu Option 2",
          "description": "Zulu option 2. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "zu_file_set"
            },
            "extract": {
              "column": "option2"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "zu/answer",
          "name": "Answer",
          "description": "Correct option (a number 1 or 2 corresponding to Option 1 or Option 2). Note that for rows from the original test set, the answer will be listed as 0 since no answer was provided by the original Winogrande creators. However, the answers for the test set rows can be found in the last character of the qID.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "zu_file_set"
            },
            "extract": {
              "column": "answer"
            }
          }
        }
      ]
    },
    {
      "@type": "cr:RecordSet",
      "@id": "xh",
      "name": "xh",
      "field": [
        {
          "@type": "cr:Field",
          "@id": "xh/qID",
          "name": "qID",
          "description": "Unique identifier for the question. The last character denotes the answer. Corresponds with the same qIDs in the new version following a second round of corrections and can therefore be joined with the new dataset to compare translation quality or obtain splits.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "xh_file_set"
            },
            "extract": {
              "column": "qID"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "xh/sentence",
          "name": "Xhosa Sentence",
          "description": "Xhosa sentence that is missing a word or a few words (denoted with an underscore), requiring one of two options to best fill in the missing word or words.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "xh_file_set"
            },
            "extract": {
              "column": "sentence"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "xh/option1",
          "name": "Xhosa Option 1",
          "description": "Xhosa option 1. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "xh_file_set"
            },
            "extract": {
              "column": "option1"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "xh/option2",
          "name": "Xhosa Option 2",
          "description": "Xhosa option 2. A word or a few words that may be the option to best fill in the underscore in the corresponding sentence.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "xh_file_set"
            },
            "extract": {
              "column": "option2"
            }
          }
        },
        {
          "@type": "cr:Field",
          "@id": "xh/answer",
          "name": "Answer",
          "description": "Correct option (a number 1 or 2 corresponding to Option 1 or Option 2). Note that for rows from the original test set, the answer will be listed as 0 since no answer was provided by the original Winogrande creators. However, the answers for the test set rows can be found in the last character of the qID.",
          "dataType": "sc:Text",
          "source": {
            "fileSet": {
              "@id": "xh_file_set"
            },
            "extract": {
              "column": "answer"
            }
          }
        }
      ]
    }
  ]
}