gg.py

import datasets

text_file = 'dateee.txt'

class DailyDialog(datasets.GeneratorBasedBuilder):
    """DailyDialog: A Manually Labelled Multi-turn Dialogue Dataset"""

    VERSION = datasets.Version("1.0.0")

    __EOU__ = "__eou__"


    def _split_generators(self, dl_manager: datasets.DownloadManager):
        # You should implement the logic here to split the text_file into TRAIN, VALIDATION, and TEST
        # Example logic is provided below
        with open(text_file, 'r', encoding='utf-8') as f:
            lines = f.readlines()

        train_size = int(0.8 * len(lines))
        val_size = int(0.1 * len(lines))
        test_size = len(lines) - train_size - val_size

        train_data = lines[:train_size]
        val_data = lines[train_size:train_size + val_size]
        test_data = lines[train_size + val_size:]

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"data": train_data}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"data": val_data}),
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"data": test_data}),
        ]

    def _generate_examples(self, data):
        for idx, line in enumerate(data):
            yield idx, {
                "dialog": line.strip().split(self.__EOU__)[:-1]
            }