Robin Kurtz commited on
Commit
fa3669e
1 Parent(s): 6a1e6f1

suc3.1 dataloader

Browse files
Files changed (1) hide show
  1. suc3.1.py +193 -0
suc3.1.py ADDED
@@ -0,0 +1,193 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ # Lint as: python3
17
+ """The SuperGLUE benchmark."""
18
+
19
+ import json
20
+ import os
21
+
22
+ import datasets
23
+
24
+ _CITATION = """\
25
+ @article{gustafson2006documentation,
26
+ title={Documentation of the Stockholm-Ume{\aa} Corpus},
27
+ author={Gustafson-Capkov{\'a}, Sofia and Hartmann, Britt},
28
+ journal={Stockholm University: Department of Linguistics},
29
+ year={2006}
30
+ }
31
+ """
32
+
33
+ # You can copy an official description
34
+ _DESCRIPTION = """\
35
+ The dataset is a conversion of the venerable SUC 3.0 dataset into the
36
+ huggingface ecosystem. The original dataset does not contain an official
37
+ train-dev-test split, which is introduced here; the tag distribution for the
38
+ NER tags between the three splits is mostly the same.
39
+
40
+ The dataset has three different types of tagsets: manually annotated POS,
41
+ manually annotated NER, and automatically annotated NER. For the
42
+ automatically annotated NER tags, only sentences were chosen, where the
43
+ automatic and manual annotations would match (with their respective
44
+ categories).
45
+
46
+ Additionally we provide remixes of the same data with some or all sentences
47
+ being lowercased.
48
+ """
49
+
50
+ _HOMEPAGE = "https://spraakbanken.gu.se/en/resources/suc3"
51
+
52
+ _LICENSE = "CC-BY-4.0"
53
+
54
+ # TODO: Add link to the official dataset URLs here
55
+ # The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
56
+ # This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
57
+ _URL = "https://huggingface.co/datasets/KBLab/suc3.1/resolve/main/data/"
58
+ _URLS = {
59
+ "original_tags": {
60
+ "cased": "original_tags/cased.tar.gz",
61
+ "lower": "original_tags/lower.tar.gz",
62
+ "lower_mix": "original_tags/lower_mixed.tar.gz"},
63
+ "simple_tags": {
64
+ "cased": "simple_tags/cased.tar.gz",
65
+ "lower": "simple_tags/lower.tar.gz",
66
+ "lower_mix": "simple_tags/lower_mixed.tar.gz"}
67
+ }
68
+
69
+ _POS_LABEL_NAMES = {
70
+ 'AB', 'DT', 'HA', 'HD', 'HP', 'HS', 'IE', 'IN', 'JJ', 'KN', 'MAD', 'MID',
71
+ 'NN', 'PAD', 'PC', 'PL', 'PM', 'PN', 'PP', 'PS', 'RG', 'RO', 'SN', 'UO',
72
+ 'VB'
73
+ }
74
+ _NER_LABEL_NAMES_ORIGINAL = {
75
+ 'B-animal', 'B-event', 'B-inst', 'B-myth', 'B-other', 'B-person',
76
+ 'B-place', 'B-product', 'B-work', 'I-animal', 'I-event', 'I-inst',
77
+ 'I-myth', 'I-other', 'I-person', 'I-place', 'I-product', 'I-work', 'O'
78
+ }
79
+
80
+ _NER_LABEL_NAMES_SIMPLE = {
81
+ 'B-EVN', 'B-LOC', 'B-MSR', 'B-OBJ', 'B-ORG', 'B-PRS', 'B-TME', 'B-WRK',
82
+ 'I-EVN', 'I-LOC', 'I-MSR', 'I-OBJ', 'I-ORG', 'I-PRS', 'I-TME', 'I-WRK', 'O'
83
+ }
84
+
85
+
86
+ class SucConfig(datasets.BuilderConfig):
87
+ """BuilderConfig for Suc."""
88
+ def __init__(self,
89
+ ner_label_names,
90
+ description,
91
+ data_url,
92
+ **kwargs):
93
+ """BuilderConfig for Suc.
94
+ """
95
+ # Version history:
96
+ # 1.0.2: Fixed non-nondeterminism in ReCoRD.
97
+ # 1.0.1: Change from the pre-release trial version of SuperGLUE (v1.9) to
98
+ # the full release (v2.0).
99
+ # 1.0.0: S3 (new shuffling, sharding and slicing mechanism).
100
+ # 0.0.2: Initial version.
101
+ super(SucConfig,
102
+ self).__init__(version=datasets.Version("1.0.2"), **kwargs)
103
+ self.ner_label_names = ner_label_names
104
+ self.description = description
105
+ self.config.data_url = data_url
106
+
107
+
108
+
109
+ class Suc(datasets.GeneratorBasedBuilder):
110
+ """The SuperGLUE benchmark."""
111
+
112
+ BUILDER_CONFIGS = [
113
+ SucConfig(
114
+ name="original_cased",
115
+ ner_label_names=_NER_LABEL_NAMES_ORIGINAL,
116
+ data_url=_URLS["original_tags"]["cased"],
117
+ description="manually annotated & cased",
118
+ ),
119
+ SucConfig(
120
+ name="original_lower",
121
+ ner_label_names=_NER_LABEL_NAMES_ORIGINAL,
122
+ data_url=_URLS["original_tags"]["lower"],
123
+ description="manually annotated & lower",
124
+ ),
125
+ SucConfig(
126
+ name="original_lower_mixed",
127
+ ner_label_names=_NER_LABEL_NAMES_ORIGINAL,
128
+ data_url=_URLS["original_tags"]["lower_mixed"],
129
+ description="manually annotated & lower_mixed",
130
+ ),
131
+ SucConfig(
132
+ name="simple_cased",
133
+ ner_label_names=_NER_LABEL_NAMES_SIMPLE,
134
+ data_url=_URLS["simple_tags"]["cased"],
135
+ description="automatically annotated & cased",
136
+ ),
137
+ SucConfig(
138
+ name="simple_lower",
139
+ ner_label_names=_NER_LABEL_NAMES_SIMPLE,
140
+ data_url=_URLS["simple_tags"]["lower"],
141
+ description="automatically annotated & lower",
142
+ ),
143
+ SucConfig(
144
+ name="simple_lower_mixed",
145
+ ner_label_names=_NER_LABEL_NAMES_SIMPLE,
146
+ data_url=_URLS["simple_tags"]["lower_mixed"],
147
+ description="autimatically annotated & lower_mixed",
148
+ ),
149
+ ]
150
+
151
+ def _info(self):
152
+ features = {"id": datasets.Value("string"),
153
+ "tokens": datasets.features.Sequence(datasets.Value("string")),
154
+ "pos_tags": datasets.features.Sequence(datasets.features.ClassLabel(_POS_LABEL_NAMES)),
155
+ "ner_tags": datasets.features.Sequence(datasets.features.ClassLabel(self.config.ner_label_names))}
156
+
157
+ return datasets.DatasetInfo(
158
+ description=_DESCRIPTION + self.config.description,
159
+ features=datasets.Features(features),
160
+ homepage=_HOMEPAGE,
161
+ citation=_CITATION,
162
+ supervised_keys=None,
163
+ )
164
+
165
+ def _split_generators(self, dl_manager):
166
+ dl_dir = dl_manager.download_and_extract(_URL + self.config.data_url)
167
+ dl_dir = os.path.join(dl_dir)
168
+ return [
169
+ datasets.SplitGenerator(
170
+ name=datasets.Split.TRAIN,
171
+ gen_kwargs={
172
+ "data_file": os.path.join(dl_dir, "train.jsonl"),
173
+ },
174
+ ),
175
+ datasets.SplitGenerator(
176
+ name=datasets.Split.VALIDATION,
177
+ gen_kwargs={
178
+ "data_file": os.path.join(dl_dir, "dev.jsonl"),
179
+ },
180
+ ),
181
+ datasets.SplitGenerator(
182
+ name=datasets.Split.TEST,
183
+ gen_kwargs={
184
+ "data_file": os.path.join(dl_dir, "test.jsonl"),
185
+ },
186
+ ),
187
+ ]
188
+
189
+ def _generate_examples(self, data_file):
190
+ with open(data_file, encoding="utf-8") as f:
191
+ for line in f:
192
+ row = json.loads(line)
193
+ yield row