pes2o.py

import gzip
import json

import datasets

logger = datasets.logging.get_logger(__name__)


_HOMEPAGE = "https://huggingface.co/datasets/allenai/pes2o"


_DESCRIPTION = "\
The PES2O dataset is a collection of ~40M creative commmon licensed academic \
papers, cleaned, filtered, and formatted for pre-training of language models. \
It is derived from the Semantic Scholar Open Research Corpus(Lo et al, 2020), \
or S2ORC.\
"

_LICENSE = "odc-by"

_VARIANTS = {
    "v1": {
        "version": "1.0.0",
        "download_size": 100702002904,
        "dataset_size": 67787014,
        "splits": {
            "train": {
                "num_bytes": 100145555091,
                "num_examples": 67624463,
                "files": [
                    "data/v1/train-00000-of-00020.json.gz",
                    "data/v1/train-00001-of-00020.json.gz",
                    "data/v1/train-00002-of-00020.json.gz",
                    "data/v1/train-00003-of-00020.json.gz",
                    "data/v1/train-00004-of-00020.json.gz",
                    "data/v1/train-00005-of-00020.json.gz",
                    "data/v1/train-00006-of-00020.json.gz",
                    "data/v1/train-00007-of-00020.json.gz",
                    "data/v1/train-00008-of-00020.json.gz",
                    "data/v1/train-00009-of-00020.json.gz",
                    "data/v1/train-00010-of-00020.json.gz",
                    "data/v1/train-00011-of-00020.json.gz",
                    "data/v1/train-00012-of-00020.json.gz",
                    "data/v1/train-00013-of-00020.json.gz",
                    "data/v1/train-00014-of-00020.json.gz",
                    "data/v1/train-00015-of-00020.json.gz",
                    "data/v1/train-00016-of-00020.json.gz",
                    "data/v1/train-00017-of-00020.json.gz",
                    "data/v1/train-00018-of-00020.json.gz",
                    "data/v1/train-00019-of-00020.json.gz",
                ],
            },
            "validation": {
                "num_bytes": 556447813,
                "num_examples": 162551,
                "files": [
                    "data/v1/validation-00000-of-00002.json.gz",
                    "data/v1/validation-00001-of-00002.json.gz",
                ],
            },
        },
    },
    "v2": {
        "version": "1.0.0",
        "download_size": 87129236480,
        "dataset_size": 38972211,
        "splits": {
            "train": {
                "num_bytes": 86572382178,
                "num_examples": 38811179,
                "files": [
                    "data/v2/train-00000-of-00020.json.gz",
                    "data/v2/train-00001-of-00020.json.gz",
                    "data/v2/train-00002-of-00020.json.gz",
                    "data/v2/train-00003-of-00020.json.gz",
                    "data/v2/train-00004-of-00020.json.gz",
                    "data/v2/train-00005-of-00020.json.gz",
                    "data/v2/train-00006-of-00020.json.gz",
                    "data/v2/train-00007-of-00020.json.gz",
                    "data/v2/train-00008-of-00020.json.gz",
                    "data/v2/train-00009-of-00020.json.gz",
                    "data/v2/train-00010-of-00020.json.gz",
                    "data/v2/train-00011-of-00020.json.gz",
                    "data/v2/train-00012-of-00020.json.gz",
                    "data/v2/train-00013-of-00020.json.gz",
                    "data/v2/train-00014-of-00020.json.gz",
                    "data/v2/train-00015-of-00020.json.gz",
                    "data/v2/train-00016-of-00020.json.gz",
                    "data/v2/train-00017-of-00020.json.gz",
                    "data/v2/train-00018-of-00020.json.gz",
                    "data/v2/train-00019-of-00020.json.gz",
                ],
            },
            "validation": {
                "num_bytes": 556854302,
                "num_examples": 161032,
                "files": [
                    "data/v1/validation-00000-of-00002.json.gz",
                    "data/v1/validation-00001-of-00002.json.gz",
                ],
            },
        },
    },
}

_FEATURES = datasets.Features(
    added=datasets.Value("string"),
    created=datasets.Value("string"),
    id=datasets.Value("string"),
    source=datasets.Value("string"),
    text=datasets.Value("string"),
    version=datasets.Value("string"),
)

_CITATION = """\
@techreport{pes2o,
    author = {Luca Soldaini and Kyle Lo},
    year = 2023,
    title = {{PES2O (Pretraining Efficiently on S2ORC) Dataset}},
    institution = {{Allen Institute for AI}},
    note = {\url{https://huggingface.co/datasets/allenai/pes2o}}
}
"""


class Pes2o(datasets.GeneratorBasedBuilder):
    """Pretraining Efficiently on S2ORC!"""

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name=name, version=config["version"])
        for name, config in _VARIANTS.items()
    ]

    DEFAULT_CONFIG_NAME = "v2"

    def _info(self):
        """Give information and typings for the dataset."""
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=_FEATURES,
            supervised_keys=None,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
            dataset_size=_VARIANTS[self.config.name]["dataset_size"],
            download_size=_VARIANTS[self.config.name]["download_size"],
        )

    def _split_generators(self, dl_manager):
        train_downloaded_files = dl_manager.download(
            _VARIANTS[self.config.name]["splits"]["train"]["files"]
        )
        validation_downloaded_files = dl_manager.download(
            _VARIANTS[self.config.name]["splits"]["validation"]["files"]
        )
        return [
            datasets.SplitGenerator(
                name=str(datasets.Split.TRAIN),
                gen_kwargs={"filepaths": train_downloaded_files},
            ),
            datasets.SplitGenerator(
                name=str(datasets.Split.VALIDATION),
                gen_kwargs={"filepaths": validation_downloaded_files},
            ),
        ]

    def _generate_examples(self, filepaths):
        """This function returns the examples in the raw (text) form by
        iterating on all the files."""
        id_ = 0
        for filepath in filepaths:
            logger.info("generating examples from = %s", filepath)
            with gzip.open(open(filepath, "rb"), "rt", encoding="utf-8") as f:
                for line in f:
                    if line:
                        example = json.loads(line)
                        yield id_, example
                        id_ += 1