w4ngatang commited on
Commit
81ce5d0
1 Parent(s): a24b89f

Initial data loader version

Browse files
Files changed (2) hide show
  1. .gitattributes +7 -0
  2. squality.py +194 -0
.gitattributes CHANGED
@@ -35,3 +35,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
35
  *.mp3 filter=lfs diff=lfs merge=lfs -text
36
  *.ogg filter=lfs diff=lfs merge=lfs -text
37
  *.wav filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
35
  *.mp3 filter=lfs diff=lfs merge=lfs -text
36
  *.ogg filter=lfs diff=lfs merge=lfs -text
37
  *.wav filter=lfs diff=lfs merge=lfs -text
38
+ data/v1/test.jsonl filter=lfs diff=lfs merge=lfs -text
39
+ data/v1/train.jsonl filter=lfs diff=lfs merge=lfs -text
40
+ data/v1/validation.jsonl filter=lfs diff=lfs merge=lfs -text
41
+ data/v1-1/*.jsonl filter=lfs diff=lfs merge=lfs -text
42
+ data/v1-1/train.jsonl filter=lfs diff=lfs merge=lfs -text
43
+ data/v1-1/validation.jsonl filter=lfs diff=lfs merge=lfs -text
44
+ data/v1-1/test.jsonl filter=lfs diff=lfs merge=lfs -text
squality.py ADDED
@@ -0,0 +1,194 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+ # TODO: Address all TODOs and remove all explanatory comments
15
+ """TODO: Add a description here."""
16
+ """ Dataset loading script for SQuALITY, an abstractive summarization dataset that is
17
+ * long document: 3k-6k words
18
+ * question-focused: 5/doc
19
+ * multi-reference 4/question
20
+ """
21
+
22
+ import os
23
+ import csv
24
+ import json
25
+
26
+ import datasets
27
+
28
+
29
+ _CITATION = """\
30
+ @article{wang2022squality,
31
+ title={SQuALITY: Building a Long-Document Summarization Dataset the Hard Way},
32
+ author={Wang, Alex and Pang, Richard Yuanzhe and Chen, Angelica and Phang, Jason and Bowman, Samuel R.},
33
+ journal={arXiv preprint 2205.11465},
34
+ year={2022}
35
+ }
36
+ """
37
+
38
+ # TODO: Add description of the dataset here
39
+ # You can copy an official description
40
+ _DESCRIPTION = """\
41
+ This new dataset is designed to solve this great NLP task and is crafted with a lot of care.
42
+ """
43
+
44
+ _HOMEPAGE = "ihttps://github.com/nyu-mll/SQuALITY"
45
+
46
+ _LICENSE = "CC BY"
47
+
48
+ # TODO: Add link to the official dataset URLs here
49
+ # The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
50
+ # This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
51
+ #_URLS = {
52
+ # "first_domain": "https://huggingface.co/great-new-dataset-first_domain.zip",
53
+ # "second_domain": "https://huggingface.co/great-new-dataset-second_domain.zip",
54
+ #}
55
+
56
+
57
+ class SQuALITYDataset(datasets.GeneratorBasedBuilder):
58
+ """TODO: Short description of my dataset."""
59
+
60
+ VERSION = datasets.Version("1.1")
61
+
62
+ # This is an example of a dataset with multiple configurations.
63
+ # If you don't want/need to define several sub-sets in your dataset,
64
+ # just remove the BUILDER_CONFIG_CLASS and the BUILDER_CONFIGS attributes.
65
+
66
+ # If you need to make complex sub-parts in the datasets with configurable options
67
+ # You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
68
+ # BUILDER_CONFIG_CLASS = MyBuilderConfig
69
+
70
+ # You will be able to load one or the other configurations in the following list with
71
+ # data = datasets.load_dataset('my_dataset', 'first_domain')
72
+ BUILDER_CONFIGS = [
73
+ datasets.BuilderConfig(name="squality-v1", version=datasets.Version("1.0"), description="SQUALITY v1.0, containing 100 stories (2000 summaries)"),
74
+ datasets.BuilderConfig(name="squality-v1.1", version=VERSION, description="SQuALITY version v1.1, expands on v1.0 by adding 27 stories (540 summaries)"),
75
+ ]
76
+
77
+ DEFAULT_CONFIG_NAME = "squality-v1.1" # It's not mandatory to have a default configuration. Just use one if it make sense.
78
+
79
+ def _info(self):
80
+ # This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
81
+
82
+ #if self.config.name == "first_domain": # This is the name of the configuration selected in BUILDER_CONFIGS above
83
+ # features = datasets.Features(
84
+ # {
85
+ # "sentence": datasets.Value("string"),
86
+ # "option1": datasets.Value("string"),
87
+ # "answer": datasets.Value("string")
88
+ # # These are the features of your dataset like images, labels ...
89
+ # }
90
+ # )
91
+
92
+ features = datasets.Features(
93
+ {
94
+ "document": datasets.Value("string"),
95
+ "question": datasets.Value("string"),
96
+ "summary": datasets.Value("string")
97
+ # These are the features of your dataset like images, labels ...
98
+ }
99
+ )
100
+
101
+ return datasets.DatasetInfo(
102
+ # This is the description that will appear on the datasets page.
103
+ description=_DESCRIPTION,
104
+ # This defines the different columns of the dataset and their types
105
+ features=features,
106
+ # If there's a common (input, target) tuple from the features,
107
+ # uncomment supervised_keys line below and
108
+ # specify them. They'll be used if as_supervised=True in builder.as_dataset.
109
+ # supervised_keys=("sentence", "label"),
110
+ # Homepage of the dataset for documentation
111
+ homepage=_HOMEPAGE,
112
+ # License for the dataset if available
113
+ license=_LICENSE,
114
+ # Citation for the dataset
115
+ citation=_CITATION,
116
+ )
117
+
118
+ def _split_generators(self, dl_manager):
119
+ # This method is tasked with downloading/extracting the data and
120
+ # defining the splits depending on the configuration
121
+ # If several configurations are possible (listed in BUILDER_CONFIGS),
122
+ # the configuration selected by the user is in self.config.name
123
+
124
+ if self.config.name == "squality-v1":
125
+ data_dir = "data/v1"
126
+ elif self.config.name == "squality-v1.1":
127
+ data_dir = "data/v1-1"
128
+
129
+ return [
130
+ datasets.SplitGenerator(
131
+ name=datasets.Split.TRAIN,
132
+ # These kwargs will be passed to _generate_examples
133
+ gen_kwargs={
134
+ "filepath": os.path.join(data_dir, "train.jsonl"),
135
+ "split": "train",
136
+ },
137
+ ),
138
+ datasets.SplitGenerator(
139
+ name=datasets.Split.TEST,
140
+ # These kwargs will be passed to _generate_examples
141
+ gen_kwargs={
142
+ "filepath": os.path.join(data_dir, "test.jsonl"),
143
+ "split": "test"
144
+ },
145
+ ),
146
+ datasets.SplitGenerator(
147
+ name=datasets.Split.VALIDATION,
148
+ # These kwargs will be passed to _generate_examples
149
+ gen_kwargs={
150
+ "filepath": os.path.join(data_dir, "validation.jsonl"),
151
+ "split": "dev",
152
+ },
153
+ ),
154
+ ]
155
+
156
+ # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
157
+ def _generate_examples(self, filepath, split):
158
+ # This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
159
+ # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
160
+ with open(filepath, encoding="utf-8") as f:
161
+ for row in enumerate(f):
162
+ # fields
163
+ # * metadata
164
+ # * document
165
+ # * questions
166
+ story = json.loads(row)
167
+ for question in story['questions']:
168
+ # fields
169
+ # * question_text
170
+ # * question_number
171
+ # * responses
172
+ key = question['gem_id']
173
+
174
+ # for the test split, yield all references at once
175
+ # to easily compute multi-reference metrics
176
+ if split == "test":
177
+ yield key, {
178
+ 'document': story['document'],
179
+ 'question': question['question_text'],
180
+ 'summary': [r['response_text'] for r in question['responses']]
181
+ }
182
+
183
+ else:
184
+ for response in question['responses']:
185
+ # fields
186
+ # * uid
187
+ # * worker_uid
188
+ # * response_text
189
+ yield key, {
190
+ 'document': story['document'],
191
+ 'question': question['question_text'],
192
+ 'summary': response['response_text']
193
+ }
194
+