| from __future__ import absolute_import, division, print_function | |
| import datasets | |
| _URL = "data/" | |
| _URLs = { | |
| "train": _URL + "train.tsv", | |
| "valid": _URL + "valid.tsv", | |
| "test": _URL + "test.tsv", | |
| } | |
| class RottenTomato(datasets.GeneratorBasedBuilder): | |
| def _info(self): | |
| return datasets.DatasetInfo( | |
| description="MR Dataset", | |
| features=datasets.Features( | |
| { | |
| "text": datasets.Value("string"), | |
| "label": datasets.features.ClassLabel(names=['positive', 'negative']), | |
| } | |
| ), | |
| supervised_keys=None, | |
| license="", | |
| homepage="", | |
| citation="", | |
| ) | |
| def _split_generators(self, dl_manager): | |
| downloaded_files = dl_manager.download_and_extract(_URLs) | |
| return [ | |
| datasets.SplitGenerator( | |
| name=datasets.Split.TRAIN, | |
| gen_kwargs={ | |
| "filepath": downloaded_files["train"], | |
| } | |
| ), | |
| datasets.SplitGenerator( | |
| name=datasets.Split.VALIDATION, | |
| gen_kwargs={ | |
| "filepath": downloaded_files["valid"], | |
| } | |
| ), | |
| datasets.SplitGenerator( | |
| name=datasets.Split.TEST, | |
| gen_kwargs={ | |
| "filepath": downloaded_files["test"], | |
| } | |
| ), | |
| ] | |
| def _generate_examples(self, filepath): | |
| with open(filepath, "r", encoding='ISO-8859-1') as f: | |
| for idx, line in enumerate(f): | |
| text, label = line.split("\t") | |
| yield idx, {"text": text.strip(), "label": label.strip()} | |