Convert dataset to Parquet (#7)

- Convert dataset to Parquet (ec6839eed447dfffbd0bfb7551701b7dcadd44aa)
- Add 'snli' config data files (cdf18d97fcea5bf4a6d7a6e9365ea20d4024eaae)
- Add 'xnli' config data files (17289f0bafe71b2cc393b7d1530cf24db6f18f47)
- Delete loading script (a4fdb7b517f363a92fd8cf5a447301115c6a4d81)
- Delete data file (a2a6a1a066884fb48ec27a1e66152f0184233c6e)

README.md

@@ -39,10 +39,10 @@ dataset_info:
           '2': contradiction
   splits:
   - name: train
-    num_bytes: 84729207
+    num_bytes: 84728887
     num_examples: 392702
-  download_size: 42113232
-  dataset_size: 84729207
+  download_size: 54693610
+  dataset_size: 84728887
 - config_name: snli
   features:
   - name: premise
@@ -58,10 +58,10 @@ dataset_info:
           '2': contradiction
   splits:
   - name: train
-    num_bytes: 80137097
+    num_bytes: 80136649
     num_examples: 550152
-  download_size: 42113232
-  dataset_size: 80137097
+  download_size: 22015955
+  dataset_size: 80136649
 - config_name: xnli
   features:
   - name: premise
@@ -77,13 +77,28 @@ dataset_info:
           '2': contradiction
   splits:
   - name: validation
-    num_bytes: 518830
+    num_bytes: 518822
     num_examples: 2490
   - name: test
-    num_bytes: 1047437
+    num_bytes: 1047429
     num_examples: 5010
-  download_size: 42113232
-  dataset_size: 1566267
+  download_size: 529321
+  dataset_size: 1566251
+configs:
+- config_name: multi_nli
+  data_files:
+  - split: train
+    path: multi_nli/train-*
+- config_name: snli
+  data_files:
+  - split: train
+    path: snli/train-*
+- config_name: xnli
+  data_files:
+  - split: validation
+    path: xnli/validation-*
+  - split: test
+    path: xnli/test-*
 ---
 
 # Dataset Card for "kor_nli"

kor_nli.py

@@ -1,121 +0,0 @@
-"""TODO(kor_nli): Add a description here."""
-
-
-import os
-
-import datasets
-
-
-# TODO(kor_nli): BibTeX citation
-_CITATION = """\
-@article{ham2020kornli,
-  title={KorNLI and KorSTS: New Benchmark Datasets for Korean Natural Language Understanding},
-  author={Ham, Jiyeon and Choe, Yo Joong and Park, Kyubyong and Choi, Ilji and Soh, Hyungjoon},
-  journal={arXiv preprint arXiv:2004.03289},
-  year={2020}
-}
-"""
-
-# TODO(kor_nli):
-_DESCRIPTION = """ Korean Natural  Language Inference datasets
-"""
-_URL = "data.zip"
-
-
-class KorNLIConfig(datasets.BuilderConfig):
-    """BuilderConfig for KorNLI."""
-
-    def __init__(self, **kwargs):
-        """BuilderConfig for KorNLI.
-
-        Args:
-
-          **kwargs: keyword arguments forwarded to super.
-        """
-        # Version 1.1.0 remove empty document and summary strings.
-        super(KorNLIConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
-
-
-class KorNli(datasets.GeneratorBasedBuilder):
-    """TODO(kor_nli): Short description of my dataset."""
-
-    # TODO(kor_nli): Set up version.
-    VERSION = datasets.Version("1.0.0")
-    BUILDER_CONFIGS = [
-        KorNLIConfig(name="multi_nli", description="Korean multi NLI datasets"),
-        KorNLIConfig(name="snli", description="Korean SNLI dataset"),
-        KorNLIConfig(name="xnli", description="Korean XNLI dataset"),
-    ]
-
-    def _info(self):
-        # TODO(kor_nli): Specifies the datasets.DatasetInfo object
-        return datasets.DatasetInfo(
-            # This is the description that will appear on the datasets page.
-            description=_DESCRIPTION,
-            # datasets.features.FeatureConnectors
-            features=datasets.Features(
-                {
-                    # These are the features of your dataset like images, labels ...
-                    "premise": datasets.Value("string"),
-                    "hypothesis": datasets.Value("string"),
-                    "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
-                }
-            ),
-            # If there's a common (input, target) tuple from the features,
-            # specify them here. They'll be used if as_supervised=True in
-            # builder.as_dataset.
-            supervised_keys=None,
-            # Homepage of the dataset for documentation
-            homepage="https://github.com/kakaobrain/KorNLUDatasets",
-            citation=_CITATION,
-        )
-
-    def _split_generators(self, dl_manager):
-        """Returns SplitGenerators."""
-        # TODO(kor_nli): Downloads the data and defines the splits
-        # dl_manager is a datasets.download.DownloadManager that can be used to
-        # download and extract URLs
-        dl_dir = dl_manager.download_and_extract(_URL)
-        dl_dir = os.path.join(dl_dir, "KorNLI")
-        if self.config.name == "multi_nli":
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TRAIN,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": os.path.join(dl_dir, "multinli.train.ko.tsv")},
-                ),
-            ]
-        elif self.config.name == "snli":
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TRAIN,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": os.path.join(dl_dir, "snli_1.0_train.ko.tsv")},
-                ),
-            ]
-        else:
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": os.path.join(dl_dir, "xnli.dev.ko.tsv")},
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.TEST,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": os.path.join(dl_dir, "xnli.test.ko.tsv")},
-                ),
-            ]
-
-    def _generate_examples(self, filepath):
-        """Yields examples."""
-        # TODO(kor_nli): Yields (key, example) tuples from the dataset
-        with open(filepath, encoding="utf-8") as f:
-            next(f)  # skip headers
-            columns = ("premise", "hypothesis", "label")
-            for id_, row in enumerate(f):
-                row = row.strip().split("\t")
-                if len(row) != 3:
-                    continue
-                row = dict(zip(columns, row))
-                yield id_, row

data.zip → multi_nli/train-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0e1f08af79f96ec66293df406ca2475be8ffffccb95115c9441cd160fd49f89
-size 41666635
+oid sha256:8de5f800e4c9d0fb89bd6a36a800f7ec26be2a97f50c32f2aff3c197ca155f66
+size 54693610

snli/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd6c2d95ae7b8979e2a89c0d05b208f38d480ca4e23f8249122e9b40116b5572
+size 22015955

xnli/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ebba3632ffac4d05a18e76629486eb88c584ae5459d15181c295341e6af659b
+size 351479

xnli/validation-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef903c790a16149f167367a01775aaf7e5d6d1a74deb89400eec4f3f29b4aec8
+size 177842