dataset.py

from datasets import DatasetDict, load_dataset

import json
import datasets

class MyDatasetConfig(datasets.BuilderConfig):
    def __init__(self, category=None, **kwargs):
        super().__init__(**kwargs)
        self.category = category

class MyDataset(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIG_CLASS = MyDatasetConfig

    BUILDER_CONFIGS = [
        MyDatasetConfig(
            name="alpaca_gpt4",
            category="alpaca_gpt4",
            version=datasets.Version("1.0.0"),
        ),
        MyDatasetConfig(
            name="magpie",
            category="magpie",
            version=datasets.Version("1.0.0"),
        ),
    ]

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features({
                "id": datasets.Value("string"),
                "conversations": datasets.Sequence({
                    "from": datasets.Value("string"),
                    "value": datasets.Value("string"),
                })
            })
        )

    def _split_generators(self, dl_manager):
        category = self.config.category
        base = f"./{category}"

        return [
            datasets.SplitGenerator(
                name=f"{category}_tshirt_k_50",
                gen_kwargs={"filepath": f"{base}/tshirt_k_50.json"},
            ),
            datasets.SplitGenerator(
                name=f"{category}_tshirt_k_75",
                gen_kwargs={"filepath": f"{base}/tshirt_k_75.json"},
            )
        ]

    def _generate_examples(self, filepath):
        with open(filepath, "r") as f:
            data = json.load(f)
        for row in data:
            yield row["id"], row