GenMix50k.py

import json
import datasets as ds




_DESCRIPTION = """
Description of the dataset.
"""

_CITATION = """
Citation for the dataset.
"""

_LICENCE = """
License information for the dataset.
"""


_FEATURES = ds.Features({
    "x": ds.Value(dtype="string"),
    "y": ds.Value(dtype="string")
})




class GenMix50kConfig(ds.BuilderConfig):
    def __init__(self, **kwargs):
        super(GenMix50kConfig, self).__init__(**kwargs)
        self.version = ds.Version("1.0.3")
        self.features = _FEATURES
        self.citation = _CITATION




class GenMix50k(ds.GeneratorBasedBuilder):

    BUILDER_CONFIGS = [
        GenMix50kConfig(name="translation"),
        GenMix50kConfig(name="dialogue"),
        GenMix50kConfig(name="summarization")
    ]

    
    def _info(self) -> ds.DatasetInfo:
        """Returns the dataset metadata."""
        return ds.DatasetInfo(
            description=_DESCRIPTION,
            features=self.config.features,
            citation=self.config.citation,
            license=_LICENCE,
            supervised_keys=None
        )

    
    def _split_generators(self, dl_manager: ds.DownloadManager):
        """Returns SplitGenerators"""
        
        base_url = "https://huggingface.co/datasets/moon23k/GenMix50k/resolve/main/"
        data_files = {
            "translation": {
                "train": base_url + "translation/train.json",
                "validation": base_url + "translation/valid.json",
                "test": base_url + "translation/test.json"
            },
            "dialogue": {
                "train": base_url + "dialogue/train.json",
                "validation": base_url + "dialogue/valid.json",
                "test": base_url + "dialogue/test.json"
            },
            "summarization": {
                "train": base_url + "summarization/train.json",
                "validation": base_url + "summarization/valid.json",
                "test": base_url + "summarization/test.json"
            }
        }

        downloaded_files = dl_manager.download_and_extract(data_files[self.config.name])

        return [
            ds.SplitGenerator(
                name=ds.Split.TRAIN,
                gen_kwargs={"filepath": downloaded_files["train"]},
            ),
            ds.SplitGenerator(
                name=ds.Split.VALIDATION,
                gen_kwargs={"filepath": downloaded_files["validation"]},
            ),
            ds.SplitGenerator(
                name=ds.Split.TEST,
                gen_kwargs={"filepath": downloaded_files["test"]},
            ),
        ]

    
    def _generate_examples(self, filepath):
        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
            for idx, example in enumerate(data):
                yield idx, example