dataset.py

import datasets

_CITATION = """\
@misc{yoruba2025numericalqa,
  title     = {Yorùbá Numerical and Logical Reasoning QA Dataset},
  author    = {Fiyinfoluwa Oyesanmi and Peter Olukanmi},
  year      = {2025},
  url       = {https://huggingface.co/datasets/fiyinoye/yoruba-arithmetic-dataset},
  note      = {A dataset for evaluating reasoning and numeral understanding in Yorùbá.}
}
"""

_DESCRIPTION = """\
This dataset contains three subsets of question-answer pairs written in Yorùbá:
(1) Arithmetic reasoning, (2) Calendar/time reasoning, and (3) Traditional numeral interpretation.
It is intended for evaluating LLMs' reasoning in low-resource, indigenous languages.
"""

_HOMEPAGE = "https://huggingface.co/datasets/fiyinoye/yoruba-arithmetic-dataset"
_LICENSE = "CC-BY-4.0"

_URLS = {
    "arithmetic": "https://huggingface.co/datasets/fiyinoye/yoruba-arithmetic-dataset/blob/main/data/arithmetic.json",
    "calendar": "https://huggingface.co/datasets/fiyinoye/yoruba-arithmetic-dataset/blob/main/data/calendar.json",
    "numerals": "https://huggingface.co/datasets/fiyinoye/yoruba-arithmetic-dataset/blob/main/data/numerals.json",
}

class YorubaNumericalReasoning(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version("1.0.0")
    
    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features({
                "id": datasets.Value("string"),
                "subset": datasets.ClassLabel(names=["arithmetic", "calendar", "numerals"]),
                "question": datasets.Value("string")
            }),
            supervised_keys=None,
            homepage=_HOMEPAGE,
            citation=_CITATION,
            license=_LICENSE,
        )

    def _split_generators(self, dl_manager):
        downloaded = dl_manager.download_and_extract(_URLS)
        return [
            datasets.SplitGenerator(
                name="arithmetic", gen_kwargs={"filepath": downloaded["arithmetic"], "subset": "arithmetic"}
            ),
            datasets.SplitGenerator(
                name="calendar", gen_kwargs={"filepath": downloaded["calendar"], "subset": "calendar"}
            ),
            datasets.SplitGenerator(
                name="numerals", gen_kwargs={"filepath": downloaded["numerals"], "subset": "numerals"}
            ),
        ]

    def _generate_examples(self, filepath, category):
        import json
        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
            for i, row in enumerate(data):
                yield i, {
                    "id": row.get("id", str(i)),
                    "subset": subset,
                    "question": row["question"]
                }