Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
parquet
Sub-tasks:
multiple-choice-qa
Size:
100K - 1M
ArXiv:
License:
Commit
•
d5e1adb
1
Parent(s):
b074c81
Delete loading script
Browse files
exams.py
DELETED
@@ -1,266 +0,0 @@
|
|
1 |
-
# coding=utf-8
|
2 |
-
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
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 |
-
"""EXAMS: a benchmark dataset for multilingual and cross-lingual question answering"""
|
16 |
-
|
17 |
-
|
18 |
-
import json
|
19 |
-
|
20 |
-
import datasets
|
21 |
-
|
22 |
-
|
23 |
-
_CITATION = """\
|
24 |
-
@article{hardalov2020exams,
|
25 |
-
title={EXAMS: A Multi-subject High School Examinations Dataset for Cross-lingual and Multilingual Question Answering},
|
26 |
-
author={Hardalov, Momchil and Mihaylov, Todor and Dimitrina Zlatkova and Yoan Dinkov and Ivan Koychev and Preslav Nvakov},
|
27 |
-
journal={arXiv preprint arXiv:2011.03080},
|
28 |
-
year={2020}
|
29 |
-
}
|
30 |
-
"""
|
31 |
-
|
32 |
-
_DESCRIPTION = """\
|
33 |
-
EXAMS is a benchmark dataset for multilingual and cross-lingual question answering from high school examinations.
|
34 |
-
It consists of more than 24,000 high-quality high school exam questions in 16 languages,
|
35 |
-
covering 8 language families and 24 school subjects from Natural Sciences and Social Sciences, among others.
|
36 |
-
"""
|
37 |
-
|
38 |
-
_HOMEPAGE = "https://github.com/mhardalov/exams-qa"
|
39 |
-
|
40 |
-
_LICENSE = "CC-BY-SA-4.0"
|
41 |
-
|
42 |
-
_URLS_LIST = [
|
43 |
-
("alignments", "https://github.com/mhardalov/exams-qa/raw/main/data/exams/parallel_questions.jsonl"),
|
44 |
-
]
|
45 |
-
_URLS_LIST += [
|
46 |
-
(
|
47 |
-
"multilingual_train",
|
48 |
-
"https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/train.jsonl.tar.gz",
|
49 |
-
),
|
50 |
-
("multilingual_dev", "https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/dev.jsonl.tar.gz"),
|
51 |
-
("multilingual_test", "https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/test.jsonl.tar.gz"),
|
52 |
-
(
|
53 |
-
"multilingual_with_para_train",
|
54 |
-
"https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/with_paragraphs/train_with_para.jsonl.tar.gz",
|
55 |
-
),
|
56 |
-
(
|
57 |
-
"multilingual_with_para_dev",
|
58 |
-
"https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/with_paragraphs/dev_with_para.jsonl.tar.gz",
|
59 |
-
),
|
60 |
-
(
|
61 |
-
"multilingual_with_para_test",
|
62 |
-
"https://github.com/mhardalov/exams-qa/raw/main/data/exams/multilingual/with_paragraphs/test_with_para.jsonl.tar.gz",
|
63 |
-
),
|
64 |
-
]
|
65 |
-
|
66 |
-
_CROSS_LANGUAGES = ["bg", "hr", "hu", "it", "mk", "pl", "pt", "sq", "sr", "tr", "vi"]
|
67 |
-
_URLS_LIST += [
|
68 |
-
("crosslingual_test", "https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/test.jsonl.tar.gz"),
|
69 |
-
(
|
70 |
-
"crosslingual_with_para_test",
|
71 |
-
"https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/with_paragraphs/test_with_para.jsonl.tar.gz",
|
72 |
-
),
|
73 |
-
]
|
74 |
-
for ln in _CROSS_LANGUAGES:
|
75 |
-
_URLS_LIST += [
|
76 |
-
(
|
77 |
-
f"crosslingual_{ln}_train",
|
78 |
-
f"https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/train_{ln}.jsonl.tar.gz",
|
79 |
-
),
|
80 |
-
(
|
81 |
-
f"crosslingual_with_para_{ln}_train",
|
82 |
-
f"https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/with_paragraphs/train_{ln}_with_para.jsonl.tar.gz",
|
83 |
-
),
|
84 |
-
(
|
85 |
-
f"crosslingual_{ln}_dev",
|
86 |
-
f"https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/dev_{ln}.jsonl.tar.gz",
|
87 |
-
),
|
88 |
-
(
|
89 |
-
f"crosslingual_with_para_{ln}_dev",
|
90 |
-
f"https://github.com/mhardalov/exams-qa/raw/main/data/exams/cross-lingual/with_paragraphs/dev_{ln}_with_para.jsonl.tar.gz",
|
91 |
-
),
|
92 |
-
]
|
93 |
-
_URLs = dict(_URLS_LIST)
|
94 |
-
|
95 |
-
|
96 |
-
class ExamsConfig(datasets.BuilderConfig):
|
97 |
-
def __init__(self, lang, with_para, **kwargs):
|
98 |
-
super(ExamsConfig, self).__init__(version=datasets.Version("1.0.0", ""), **kwargs)
|
99 |
-
self.lang = lang
|
100 |
-
self.with_para = "_with_para" if with_para else ""
|
101 |
-
|
102 |
-
|
103 |
-
class Exams(datasets.GeneratorBasedBuilder):
|
104 |
-
"""Exams dataset"""
|
105 |
-
|
106 |
-
VERSION = datasets.Version("1.0.0")
|
107 |
-
BUILDER_CONFIG_CLASS = ExamsConfig
|
108 |
-
BUILDER_CONFIGS = [
|
109 |
-
ExamsConfig(
|
110 |
-
lang="",
|
111 |
-
with_para=False,
|
112 |
-
name="alignments",
|
113 |
-
description="loads the alignment between question IDs across languages",
|
114 |
-
),
|
115 |
-
ExamsConfig(
|
116 |
-
lang="all",
|
117 |
-
with_para=False,
|
118 |
-
name="multilingual",
|
119 |
-
description="Loads the unified multilingual train/dev/test split",
|
120 |
-
),
|
121 |
-
ExamsConfig(
|
122 |
-
lang="all",
|
123 |
-
with_para=True,
|
124 |
-
name="multilingual_with_para",
|
125 |
-
description="Loads the unified multilingual train/dev/test split with Wikipedia support paragraphs",
|
126 |
-
),
|
127 |
-
ExamsConfig(
|
128 |
-
lang="all", with_para=False, name="crosslingual_test", description="Loads crosslingual test set only"
|
129 |
-
),
|
130 |
-
ExamsConfig(
|
131 |
-
lang="all",
|
132 |
-
with_para=True,
|
133 |
-
name="crosslingual_with_para_test",
|
134 |
-
description="Loads crosslingual test set only with Wikipedia support paragraphs",
|
135 |
-
),
|
136 |
-
]
|
137 |
-
for ln in _CROSS_LANGUAGES:
|
138 |
-
BUILDER_CONFIGS += [
|
139 |
-
ExamsConfig(
|
140 |
-
lang=ln,
|
141 |
-
with_para=False,
|
142 |
-
name=f"crosslingual_{ln}",
|
143 |
-
description=f"Loads crosslingual train and dev set for {ln}",
|
144 |
-
),
|
145 |
-
ExamsConfig(
|
146 |
-
lang=ln,
|
147 |
-
with_para=True,
|
148 |
-
name=f"crosslingual_with_para_{ln}",
|
149 |
-
description=f"Loads crosslingual train and dev set for {ln} with Wikipedia support paragraphs",
|
150 |
-
),
|
151 |
-
]
|
152 |
-
|
153 |
-
DEFAULT_CONFIG_NAME = (
|
154 |
-
"multilingual_with_para" # It's not mandatory to have a default configuration. Just use one if it make sense.
|
155 |
-
)
|
156 |
-
|
157 |
-
def _info(self):
|
158 |
-
if self.config.name == "alignments": # This is the name of the configuration selected in BUILDER_CONFIGS above
|
159 |
-
features = datasets.Features(
|
160 |
-
{
|
161 |
-
"source_id": datasets.Value("string"),
|
162 |
-
"target_id_list": datasets.Sequence(datasets.Value("string")),
|
163 |
-
}
|
164 |
-
)
|
165 |
-
else: # This is an example to show how to have different features for "first_domain" and "second_domain"
|
166 |
-
features = datasets.Features(
|
167 |
-
{
|
168 |
-
"id": datasets.Value("string"),
|
169 |
-
"question": {
|
170 |
-
"stem": datasets.Value("string"),
|
171 |
-
"choices": datasets.Sequence(
|
172 |
-
{
|
173 |
-
"text": datasets.Value("string"),
|
174 |
-
"label": datasets.Value("string"),
|
175 |
-
"para": datasets.Value("string"),
|
176 |
-
}
|
177 |
-
),
|
178 |
-
},
|
179 |
-
"answerKey": datasets.Value("string"),
|
180 |
-
"info": {
|
181 |
-
"grade": datasets.Value("int32"),
|
182 |
-
"subject": datasets.Value("string"),
|
183 |
-
"language": datasets.Value("string"),
|
184 |
-
},
|
185 |
-
}
|
186 |
-
)
|
187 |
-
return datasets.DatasetInfo(
|
188 |
-
description=_DESCRIPTION,
|
189 |
-
features=features, # Here we define them above because they are different between the two configurations
|
190 |
-
supervised_keys=None,
|
191 |
-
homepage=_HOMEPAGE,
|
192 |
-
license=_LICENSE,
|
193 |
-
citation=_CITATION,
|
194 |
-
)
|
195 |
-
|
196 |
-
def _split_generators(self, dl_manager):
|
197 |
-
"""Returns SplitGenerators."""
|
198 |
-
archives = dl_manager.download(_URLs)
|
199 |
-
if self.config.name == "alignments":
|
200 |
-
return [
|
201 |
-
datasets.SplitGenerator(
|
202 |
-
name="full",
|
203 |
-
gen_kwargs={"filepath": archives["alignments"]},
|
204 |
-
),
|
205 |
-
]
|
206 |
-
elif self.config.name in ["multilingual", "multilingual_with_para"]:
|
207 |
-
return [
|
208 |
-
datasets.SplitGenerator(
|
209 |
-
name=spl_enum,
|
210 |
-
gen_kwargs={
|
211 |
-
"filepath": f"{spl}{self.config.with_para}.jsonl",
|
212 |
-
"files": dl_manager.iter_archive(archives[f"{self.config.name}_{spl}"]),
|
213 |
-
},
|
214 |
-
)
|
215 |
-
for spl, spl_enum in [
|
216 |
-
("train", datasets.Split.TRAIN),
|
217 |
-
("dev", datasets.Split.VALIDATION),
|
218 |
-
("test", datasets.Split.TEST),
|
219 |
-
]
|
220 |
-
]
|
221 |
-
elif self.config.name in ["crosslingual_test", "crosslingual_with_para_test"]:
|
222 |
-
return [
|
223 |
-
datasets.SplitGenerator(
|
224 |
-
name=datasets.Split.TEST,
|
225 |
-
gen_kwargs={
|
226 |
-
"filepath": f"test{self.config.with_para}.jsonl",
|
227 |
-
"files": dl_manager.iter_archive(archives[self.config.name]),
|
228 |
-
},
|
229 |
-
),
|
230 |
-
]
|
231 |
-
else:
|
232 |
-
return [
|
233 |
-
datasets.SplitGenerator(
|
234 |
-
name=spl_enum,
|
235 |
-
gen_kwargs={
|
236 |
-
"filepath": f"{spl}_{self.config.lang}{self.config.with_para}.jsonl",
|
237 |
-
"files": dl_manager.iter_archive(archives[f"{self.config.name}_{spl}"]),
|
238 |
-
},
|
239 |
-
)
|
240 |
-
for spl, spl_enum in [
|
241 |
-
("train", datasets.Split.TRAIN),
|
242 |
-
("dev", datasets.Split.VALIDATION),
|
243 |
-
]
|
244 |
-
]
|
245 |
-
|
246 |
-
def _generate_examples(self, filepath, files=None):
|
247 |
-
if self.config.name == "alignments":
|
248 |
-
with open(filepath, encoding="utf-8") as f:
|
249 |
-
for id_, line in enumerate(f):
|
250 |
-
line_dict = json.loads(line.strip())
|
251 |
-
in_id, out_list = list(line_dict.items())[0]
|
252 |
-
yield id_, {"source_id": in_id, "target_id_list": out_list}
|
253 |
-
else:
|
254 |
-
for path, f in files:
|
255 |
-
if path == filepath:
|
256 |
-
for id_, line in enumerate(f):
|
257 |
-
line_dict = json.loads(line.strip())
|
258 |
-
for choice in line_dict["question"]["choices"]:
|
259 |
-
choice["para"] = choice.get("para", "")
|
260 |
-
yield id_, {
|
261 |
-
"id": line_dict["id"],
|
262 |
-
"question": line_dict["question"],
|
263 |
-
"answerKey": line_dict["answerKey"],
|
264 |
-
"info": line_dict["info"],
|
265 |
-
}
|
266 |
-
break
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|