Commit
•
eab50b5
0
Parent(s):
Update files from the datasets library (from 1.0.0)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.0.0
- .gitattributes +27 -0
- dataset_infos.json +1 -0
- wiki40b.py +183 -0
.gitattributes
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
4 |
+
*.bin.* filter=lfs diff=lfs merge=lfs -text
|
5 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
11 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
12 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
13 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
14 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
15 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
16 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
17 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
19 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
20 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
21 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
22 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
23 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
24 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
25 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
26 |
+
*.zstandard filter=lfs diff=lfs merge=lfs -text
|
27 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
dataset_infos.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"en": {"description": "\nClean-up text for 40+ Wikipedia languages editions of pages\ncorrespond to entities. The datasets have train/dev/test splits per language.\nThe dataset is cleaned up by page filtering to remove disambiguation pages,\nredirect pages, deleted pages, and non-entity pages. Each example contains the\nwikidata id of the entity, and the full Wikipedia article after page processing\nthat removes non-content sections and structured objects.\n", "citation": "\n", "homepage": "https://research.google/pubs/pub49029/", "license": "", "features": {"wikidata_id": {"dtype": "string", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}, "version_id": {"dtype": "string", "id": null, "_type": "Value"}}, "supervised_keys": null, "builder_name": "wiki40b", "config_name": "en", "version": {"version_str": "1.1.0", "description": null, "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 9423623904, "num_examples": 2926536, "dataset_name": "wiki40b"}, "validation": {"name": "validation", "num_bytes": 527383016, "num_examples": 163597, "dataset_name": "wiki40b"}, "test": {"name": "test", "num_bytes": 522219464, "num_examples": 162274, "dataset_name": "wiki40b"}}, "download_checksums": {}, "download_size": 0, "dataset_size": 10473226384, "size_in_bytes": 10473226384}}
|
wiki40b.py
ADDED
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
|
16 |
+
"""Wiki40B: A clean Wikipedia dataset for 40+ languages."""
|
17 |
+
|
18 |
+
from __future__ import absolute_import, division, print_function
|
19 |
+
|
20 |
+
import logging
|
21 |
+
import os
|
22 |
+
|
23 |
+
import datasets
|
24 |
+
|
25 |
+
|
26 |
+
_CITATION = """
|
27 |
+
"""
|
28 |
+
|
29 |
+
_DESCRIPTION = """
|
30 |
+
Clean-up text for 40+ Wikipedia languages editions of pages
|
31 |
+
correspond to entities. The datasets have train/dev/test splits per language.
|
32 |
+
The dataset is cleaned up by page filtering to remove disambiguation pages,
|
33 |
+
redirect pages, deleted pages, and non-entity pages. Each example contains the
|
34 |
+
wikidata id of the entity, and the full Wikipedia article after page processing
|
35 |
+
that removes non-content sections and structured objects.
|
36 |
+
"""
|
37 |
+
|
38 |
+
_LICENSE = """
|
39 |
+
This work is licensed under the Creative Commons Attribution-ShareAlike
|
40 |
+
3.0 Unported License. To view a copy of this license, visit
|
41 |
+
http://creativecommons.org/licenses/by-sa/3.0/ or send a letter to
|
42 |
+
Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.
|
43 |
+
"""
|
44 |
+
|
45 |
+
_URL = "https://research.google/pubs/pub49029/"
|
46 |
+
|
47 |
+
_DATA_DIRECTORY = "gs://tfds-data/downloads/wiki40b/tfrecord_prod"
|
48 |
+
|
49 |
+
WIKIPEDIA_LANGUAGES = [
|
50 |
+
"en",
|
51 |
+
"ar",
|
52 |
+
"zh-cn",
|
53 |
+
"zh-tw",
|
54 |
+
"nl",
|
55 |
+
"fr",
|
56 |
+
"de",
|
57 |
+
"it",
|
58 |
+
"ja",
|
59 |
+
"ko",
|
60 |
+
"pl",
|
61 |
+
"pt",
|
62 |
+
"ru",
|
63 |
+
"es",
|
64 |
+
"th",
|
65 |
+
"tr",
|
66 |
+
"bg",
|
67 |
+
"ca",
|
68 |
+
"cs",
|
69 |
+
"da",
|
70 |
+
"el",
|
71 |
+
"et",
|
72 |
+
"fa",
|
73 |
+
"fi",
|
74 |
+
"he",
|
75 |
+
"hi",
|
76 |
+
"hr",
|
77 |
+
"hu",
|
78 |
+
"id",
|
79 |
+
"lt",
|
80 |
+
"lv",
|
81 |
+
"ms",
|
82 |
+
"no",
|
83 |
+
"ro",
|
84 |
+
"sk",
|
85 |
+
"sl",
|
86 |
+
"sr",
|
87 |
+
"sv",
|
88 |
+
"tl",
|
89 |
+
"uk",
|
90 |
+
"vi",
|
91 |
+
]
|
92 |
+
|
93 |
+
|
94 |
+
class Wiki40bConfig(datasets.BuilderConfig):
|
95 |
+
"""BuilderConfig for Wiki40B."""
|
96 |
+
|
97 |
+
def __init__(self, language=None, **kwargs):
|
98 |
+
"""BuilderConfig for Wiki40B.
|
99 |
+
|
100 |
+
Args:
|
101 |
+
language: string, the language code for the Wiki40B dataset to use.
|
102 |
+
**kwargs: keyword arguments forwarded to super.
|
103 |
+
"""
|
104 |
+
super(Wiki40bConfig, self).__init__(
|
105 |
+
name=str(language), description="Wiki40B dataset for {0}.".format(language), **kwargs
|
106 |
+
)
|
107 |
+
self.language = language
|
108 |
+
|
109 |
+
|
110 |
+
_VERSION = datasets.Version("1.1.0")
|
111 |
+
|
112 |
+
|
113 |
+
class Wiki40b(datasets.BeamBasedBuilder):
|
114 |
+
"""Wiki40B: A Clean Wikipedia Dataset for Mutlilingual Language Modeling."""
|
115 |
+
|
116 |
+
BUILDER_CONFIGS = [
|
117 |
+
Wiki40bConfig(
|
118 |
+
version=_VERSION,
|
119 |
+
language=lang,
|
120 |
+
) # pylint:disable=g-complex-comprehension
|
121 |
+
for lang in WIKIPEDIA_LANGUAGES
|
122 |
+
]
|
123 |
+
|
124 |
+
def _info(self):
|
125 |
+
return datasets.DatasetInfo(
|
126 |
+
description=_DESCRIPTION,
|
127 |
+
features=datasets.Features(
|
128 |
+
{
|
129 |
+
"wikidata_id": datasets.Value("string"),
|
130 |
+
"text": datasets.Value("string"),
|
131 |
+
"version_id": datasets.Value("string"),
|
132 |
+
}
|
133 |
+
),
|
134 |
+
supervised_keys=None,
|
135 |
+
homepage=_URL,
|
136 |
+
citation=_CITATION,
|
137 |
+
)
|
138 |
+
|
139 |
+
def _split_generators(self, dl_manager):
|
140 |
+
"""Returns SplitGenerators."""
|
141 |
+
|
142 |
+
lang = self.config.language
|
143 |
+
|
144 |
+
return [
|
145 |
+
datasets.SplitGenerator(
|
146 |
+
name=datasets.Split.TRAIN,
|
147 |
+
gen_kwargs={"filepaths": os.path.join(_DATA_DIRECTORY, "train", "{}_examples-*".format(lang))},
|
148 |
+
),
|
149 |
+
datasets.SplitGenerator(
|
150 |
+
name=datasets.Split.VALIDATION,
|
151 |
+
gen_kwargs={"filepaths": os.path.join(_DATA_DIRECTORY, "dev", "{}_examples-*".format(lang))},
|
152 |
+
),
|
153 |
+
datasets.SplitGenerator(
|
154 |
+
name=datasets.Split.TEST,
|
155 |
+
gen_kwargs={"filepaths": os.path.join(_DATA_DIRECTORY, "test", "{}_examples-*".format(lang))},
|
156 |
+
),
|
157 |
+
]
|
158 |
+
|
159 |
+
def _build_pcollection(self, pipeline, filepaths):
|
160 |
+
"""Build PCollection of examples."""
|
161 |
+
import apache_beam as beam
|
162 |
+
import tensorflow as tf
|
163 |
+
|
164 |
+
logging.info("generating examples from = %s", filepaths)
|
165 |
+
|
166 |
+
def _extract_content(example):
|
167 |
+
"""Extracts content from a TFExample."""
|
168 |
+
wikidata_id = example.features.feature["wikidata_id"].bytes_list.value[0].decode("utf-8")
|
169 |
+
text = example.features.feature["text"].bytes_list.value[0].decode("utf-8")
|
170 |
+
version_id = example.features.feature["version_id"].bytes_list.value[0].decode("utf-8")
|
171 |
+
|
172 |
+
# wikidata_id could be duplicated with different texts.
|
173 |
+
yield wikidata_id + text, {
|
174 |
+
"wikidata_id": wikidata_id,
|
175 |
+
"text": text,
|
176 |
+
"version_id": version_id,
|
177 |
+
}
|
178 |
+
|
179 |
+
return (
|
180 |
+
pipeline
|
181 |
+
| beam.io.ReadFromTFRecord(filepaths, coder=beam.coders.ProtoCoder(tf.train.Example))
|
182 |
+
| beam.FlatMap(_extract_content)
|
183 |
+
)
|