Method create_dataset

benchmark/dataset_api.py:170–178

Create the dataset. This method should be deterministic - if called multiple times it should return the same data. If you need randomness, please use a fixed seed. :param cache_dir: A directory you can use to read/write any cache data.

(self, cache_dir: Path)

    @abstractmethod
    def create_dataset(self, cache_dir: Path) -> Dataset:
        """
        Create the dataset.

        This method should be deterministic - if called multiple times it should return the same
        data. If you need randomness, please use a fixed seed.

        :param cache_dir: A directory you can use to read/write any cache data.
        """


DATASET_FACTORIES: TaggedRegistry[DatasetFactory, DatasetTag] = TaggedRegistry()