src/deep_learning/data/__init__.py

"""数据集模块

提供项目中的文本数据集和 COCO YOLO 数据集。

Usage:
    from deep_learning.data import WikiDataset, PoetryDataset

    # Wiki 数据集
    wiki = WikiDataset(data_dir="~/data/wiki/mini_c4")
    doc_ds = wiki.doc_ds()
    tokens_ds = wiki.tokens_ds()
    wiki.stat(seq_length=256)

    # 诗歌数据集
    poetry = PoetryDataset(data_dir="~/data/Poetry")
    doc_ds = poetry.doc_ds()
    tokens_ds = poetry.tokens_ds()
    poetry.stat(seq_length=100)
"""

from .spec import TextGenerationDataSource, TokenizerBundle
from .coco import YoloDataSource
from .oxford_pets import OxfordPetsSegmentationDataset
from .wiki import WikiDataset
from .poetry import PoetryDataset

__all__ = ["TextGenerationDataSource", "TokenizerBundle", "WikiDataset", "PoetryDataset", "YoloDataSource", "OxfordPetsSegmentationDataset"]