yhavinga commited on
Commit
4a512d2
1 Parent(s): 48eaa16

Add data and first script

Browse files
README.md ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pretty_name: Extreme Summarization (XSum) in Dutch
3
+ language:
4
+ - nl
5
+ paperswithcode_id: xsum_dutch
6
+ task_categories:
7
+ - summarization
8
+ task_ids:
9
+ - news-articles-summarization
10
+ train-eval-index:
11
+ - config: default
12
+ task: summarization
13
+ task_id: summarization
14
+ splits:
15
+ train_split: train
16
+ eval_split: test
17
+ col_mapping:
18
+ document: text
19
+ summary: target
20
+ metrics:
21
+ - type: rouge
22
+ name: Rouge
23
+ ---
24
+
25
+ # Dataset Card for "xsum_dutch" 🇳🇱🇧🇪 Dataset
26
+
27
+ ## Table of Contents
28
+ - [Dataset Description](#dataset-description)
29
+ - [Dataset Summary](#dataset-summary)
30
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
31
+ - [Languages](#languages)
32
+ - [Dataset Structure](#dataset-structure)
33
+ - [Data Instances](#data-instances)
34
+ - [Data Fields](#data-fields)
35
+ - [Data Splits](#data-splits)
36
+ - [Dataset Creation](#dataset-creation)
37
+ - [Curation Rationale](#curation-rationale)
38
+ - [Source Data](#source-data)
39
+ - [Annotations](#annotations)
40
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
41
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
42
+ - [Social Impact of Dataset](#social-impact-of-dataset)
43
+ - [Discussion of Biases](#discussion-of-biases)
44
+ - [Other Known Limitations](#other-known-limitations)
45
+ - [Additional Information](#additional-information)
46
+ - [Dataset Curators](#dataset-curators)
47
+ - [Licensing Information](#licensing-information)
48
+ - [Citation Information](#citation-information)
49
+ - [Contributions](#contributions)
50
+
51
+ ## Dataset Description
52
+
53
+ The Xsum Dutch 🇳🇱🇧🇪 Dataset is an English-language dataset translated to Dutch.
54
+
55
+ *This dataset currently (Aug '22) has a single config, which is
56
+ config `default` of [xsum](https://huggingface.co/datasets/xsum) translated to Dutch
57
+ with [yhavinga/t5-base-36L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-base-36L-ccmatrix-multi).*
58
+
59
+ - **Homepage:** [https://github.com/EdinburghNLP/XSum/tree/master/XSum-Dataset](https://github.com/EdinburghNLP/XSum/tree/master/XSum-Dataset)
60
+ - **Repository:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
61
+ - **Paper:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
62
+ - **Point of Contact:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
63
+ - **Size of downloaded dataset files:** 245.38 MB
64
+ - **Size of the generated dataset:** 507.60 MB
65
+ - **Total amount of disk used:** 752.98 MB
66
+
67
+ ### Dataset Summary
68
+
69
+ Extreme Summarization (XSum) Dataset.
70
+
71
+ There are three features:
72
+ - document: Input news article.
73
+ - summary: One sentence summary of the article.
74
+ - id: BBC ID of the article.
75
+
76
+ ### Supported Tasks and Leaderboards
77
+
78
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
79
+
80
+ ### Languages
81
+
82
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
83
+
84
+ ## Dataset Structure
85
+
86
+ ### Data Instances
87
+
88
+ #### default
89
+
90
+ - **Size of downloaded dataset files:** 245.38 MB
91
+ - **Size of the generated dataset:** 507.60 MB
92
+ - **Total amount of disk used:** 752.98 MB
93
+
94
+ An example of 'validation' looks as follows.
95
+ ```
96
+ {
97
+ "document": "some-body",
98
+ "id": "29750031",
99
+ "summary": "some-sentence"
100
+ }
101
+ ```
102
+
103
+ ### Data Fields
104
+
105
+ The data fields are the same among all splits.
106
+
107
+ #### default
108
+ - `document`: a `string` feature.
109
+ - `summary`: a `string` feature.
110
+ - `id`: a `string` feature.
111
+
112
+ ### Data Splits
113
+
114
+ | name |train |validation|test |
115
+ |-------|-----:|---------:|----:|
116
+ |default|204045| 11332|11334|
117
+
118
+ ## Dataset Creation
119
+
120
+ ### Curation Rationale
121
+
122
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
123
+
124
+ ### Source Data
125
+
126
+ #### Initial Data Collection and Normalization
127
+
128
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
129
+
130
+ #### Who are the source language producers?
131
+
132
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
133
+
134
+ ### Annotations
135
+
136
+ #### Annotation process
137
+
138
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
139
+
140
+ #### Who are the annotators?
141
+
142
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
143
+
144
+ ### Personal and Sensitive Information
145
+
146
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
147
+
148
+ ## Considerations for Using the Data
149
+
150
+ ### Social Impact of Dataset
151
+
152
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
153
+
154
+ ### Discussion of Biases
155
+
156
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
157
+
158
+ ### Other Known Limitations
159
+
160
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
161
+
162
+ ## Additional Information
163
+
164
+ ### Dataset Curators
165
+
166
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
167
+
168
+ ### Licensing Information
169
+
170
+ [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
171
+
172
+ ### Citation Information
173
+
174
+ ```
175
+
176
+ @article{Narayan2018DontGM,
177
+ title={Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization},
178
+ author={Shashi Narayan and Shay B. Cohen and Mirella Lapata},
179
+ journal={ArXiv},
180
+ year={2018},
181
+ volume={abs/1808.08745}
182
+ }
183
+
184
+ ```
185
+
186
+
187
+ ### Contributions
188
+
189
+ Thanks to [@thomwolf](https://github.com/thomwolf), [@lewtun](https://github.com/lewtun), [@mariamabarham](https://github.com/mariamabarham), [@jbragg](https://github.com/jbragg), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten) for adding the English version of this dataset.
190
+ The dataset was translated on Cloud TPU compute generously provided by Google through the
191
+ [TPU Research Cloud](https://sites.research.google/trc/).
default/test.json.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cf0edbdcc6bc86e40ac2857b9ee78ea7a97e7c1ec9c0738c97ca6ef8841002f
3
+ size 11275971
default/train.json.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d45b4d3a136c85139765baeb7bd01b4daec03c877c207b627faa0d458a3a6634
3
+ size 201850351
default/validation.json.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b312dee491dc58e10976b8b969ea70442e835e92f9724a37290f33e208a47a88
3
+ size 11080702
test_xsum_dutch.py ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from itertools import chain
2
+
3
+ import pytest
4
+ from datasets import load_dataset
5
+
6
+
7
+ def test_streaming_dataset():
8
+ datasets = load_dataset("./cnn_dailymail_dutch.py", streaming=True)
9
+ assert set(datasets.keys()) == {"train", "validation", "test"}
10
+
11
+ for split in ["train", "validation", "test"]:
12
+ ds = datasets[split]
13
+ for i, e in enumerate(ds):
14
+ assert 10 <= len(e["article"]) <= 11000
15
+ assert 10 <= len(e["highlights"]) <= 500
16
+ if i == 100:
17
+ break
18
+
19
+
20
+ def test_train_split():
21
+ ds = load_dataset("./cnn_dailymail_dutch.py", split="train" + f"[:{2000}]")
22
+ assert len(ds) == 2000
23
+
24
+
25
+ def test_batch_dataset():
26
+ datasets = load_dataset("./cnn_dailymail_dutch.py")
27
+ assert set(datasets.keys()) == {"train", "validation", "test"}
xsum_dutch.py ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ """Cleaned Dutch split of the mC4 corpus."""
16
+
17
+
18
+ import json
19
+ import datasets
20
+
21
+ logger = datasets.logging.get_logger(__name__)
22
+
23
+ _HOMEPAGE = "https://github.com/EdinburghNLP/XSum/tree/master/XSum-Dataset"
24
+
25
+ _CITATION = """
26
+ @article{Narayan2018DontGM,
27
+ title={Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization},
28
+ author={Shashi Narayan and Shay B. Cohen and Mirella Lapata},
29
+ journal={ArXiv},
30
+ year={2018},
31
+ volume={abs/1808.08745}
32
+ }
33
+ """
34
+
35
+ _DESCRIPTION = """
36
+ Extreme Summarization (XSum) Dataset.
37
+ There are three features:
38
+ - document: Input news article.
39
+ - summary: One sentence summary of the article.
40
+ - id: BBC ID of the article.
41
+
42
+ """
43
+
44
+
45
+ _DATA_URL_NL = "https://huggingface.co/datasets/yhavinga/xsum_dutch/resolve/main/{config}/{split}.json.gz"
46
+
47
+
48
+ _DOCUMENT = "document"
49
+ _SUMMARY = "summary"
50
+ _ID = "id"
51
+
52
+ _SUPPORTED_VERSIONS = [
53
+ datasets.Version("default", "Default version."),
54
+ ]
55
+
56
+
57
+ class XsumDutchConfig(datasets.BuilderConfig):
58
+ def __init__(self, **kwargs):
59
+ super().__init__(**kwargs)
60
+
61
+
62
+ class XsumDutch(datasets.GeneratorBasedBuilder):
63
+ BUILDER_CONFIGS = [
64
+ XsumDutchConfig(
65
+ name=str(version), description=version.description
66
+ )
67
+ for version in _SUPPORTED_VERSIONS
68
+ ]
69
+
70
+ def _info(self):
71
+ return datasets.DatasetInfo(
72
+ description=_DESCRIPTION,
73
+ features=datasets.Features(
74
+ {
75
+ _DOCUMENT: datasets.Value("string"),
76
+ _SUMMARY: datasets.Value("string"),
77
+ "id": datasets.Value("string"),
78
+ }
79
+ ),
80
+ supervised_keys=None,
81
+ homepage=_HOMEPAGE,
82
+ citation=_CITATION,
83
+ )
84
+
85
+ def _split_generators(self, dl_manager):
86
+ result = [
87
+ datasets.SplitGenerator(
88
+ name=split,
89
+ gen_kwargs={
90
+ "filepath": dl_manager.download_and_extract(
91
+ _DATA_URL_NL.format(split=str(split), config=str(self.config.name))
92
+ )
93
+ },
94
+ )
95
+ for split in [
96
+ datasets.Split.TRAIN,
97
+ datasets.Split.VALIDATION,
98
+ datasets.Split.TEST,
99
+ ]
100
+ ]
101
+ return result
102
+
103
+ def _generate_examples(self, filepath):
104
+ """This function returns the examples in the raw (text) form by iterating on all the files."""
105
+ logger.info(f"Generating examples from {filepath}")
106
+
107
+ with open(filepath, "r") as file:
108
+ for _id, line in enumerate(file):
109
+ example = json.loads(line)
110
+ yield _id, example