Datasets:

Languages:
English
ArXiv:
Tags:
License:
yuvalr commited on
Commit
7cdd1ad
1 Parent(s): 88b15b1

Upload 9 files

Browse files
.gitattributes CHANGED
@@ -53,3 +53,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
 
 
 
 
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
56
+ minority_examples/train.anti_biased.jsonl filter=lfs diff=lfs merge=lfs -text
57
+ minority_examples/train.biased.jsonl filter=lfs diff=lfs merge=lfs -text
58
+ partial_input/train.biased.jsonl filter=lfs diff=lfs merge=lfs -text
minority_examples/train.anti_biased.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44c38b3b1f86637865150a01be9bb8eb0803e216b71274877d068b017556b1f2
3
+ size 10820001
minority_examples/train.biased.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df886f1ff71ed64f63a6bf3574d9ff75b3d0c3f35e3bd1797fd72b7b6dbfd81f
3
+ size 52871395
minority_examples/validation.anti_biased.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
minority_examples/validation.biased.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
partial_input/train.anti_biased.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
partial_input/train.biased.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5713f0793dc3eef4298042426e9cd05a37ec6ca50b9e612cea234beb890c748
3
+ size 53267457
partial_input/validation.anti_biased.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
partial_input/validation.biased.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
qqp.py ADDED
@@ -0,0 +1,197 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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 General Language Understanding Evaluation (GLUE) benchmark."""
18
+
19
+
20
+ import csv
21
+ import os
22
+ import textwrap
23
+ import json
24
+
25
+ import numpy as np
26
+
27
+ import datasets
28
+
29
+
30
+ _GLUE_CITATION = """\
31
+ @inproceedings{wang2019glue,
32
+ title={{GLUE}: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding},
33
+ author={Wang, Alex and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R.},
34
+ note={In the Proceedings of ICLR.},
35
+ year={2019}
36
+ }
37
+ """
38
+
39
+ _GLUE_DESCRIPTION = """\
40
+ GLUE, the General Language Understanding Evaluation benchmark
41
+ (https://gluebenchmark.com/) is a collection of resources for training,
42
+ evaluating, and analyzing natural language understanding systems.
43
+ """
44
+
45
+
46
+ class GlueConfig(datasets.BuilderConfig):
47
+ """BuilderConfig for GLUE."""
48
+
49
+ def __init__(
50
+ self,
51
+ text_features,
52
+ label_column,
53
+ data_url,
54
+ data_dir,
55
+ citation,
56
+ url,
57
+ label_classes=None,
58
+ process_label=lambda x: x,
59
+ **kwargs,
60
+ ):
61
+ """BuilderConfig for GLUE.
62
+ Args:
63
+ text_features: `dict[string, string]`, map from the name of the feature
64
+ dict for each text field to the name of the column in the tsv file
65
+ label_column: `string`, name of the column in the tsv file corresponding
66
+ to the label
67
+ data_url: `string`, url to download the zip file from
68
+ data_dir: `string`, the path to the folder containing the tsv files in the
69
+ downloaded zip
70
+ citation: `string`, citation for the data set
71
+ url: `string`, url for information about the data set
72
+ label_classes: `list[string]`, the list of classes if the label is
73
+ categorical. If not provided, then the label will be of type
74
+ `datasets.Value('float32')`.
75
+ process_label: `Function[string, any]`, function taking in the raw value
76
+ of the label and processing it to the form required by the label feature
77
+ **kwargs: keyword arguments forwarded to super.
78
+ """
79
+ super(GlueConfig, self).__init__(version=datasets.Version("1.0.0", ""), **kwargs)
80
+ self.text_features = text_features
81
+ self.label_column = label_column
82
+ self.label_classes = label_classes
83
+ self.data_url = data_url
84
+ self.data_dir = data_dir
85
+ self.citation = citation
86
+ self.url = url
87
+ self.process_label = process_label
88
+
89
+
90
+ class Glue(datasets.GeneratorBasedBuilder):
91
+ """The General Language Understanding Evaluation (GLUE) benchmark."""
92
+
93
+ BUILDER_CONFIGS = [
94
+ GlueConfig(
95
+ name=bias_amplified_splits_type,
96
+ description=textwrap.dedent(
97
+ """\
98
+ The Quora Question Pairs2 dataset is a collection of question pairs from the
99
+ community question-answering website Quora. The task is to determine whether a
100
+ pair of questions are semantically equivalent."""
101
+ ),
102
+ text_features={
103
+ "question1": "question1",
104
+ "question2": "question2",
105
+ },
106
+ label_classes=["not_duplicate", "duplicate"],
107
+ label_column="label",
108
+ data_url="https://dl.fbaipublicfiles.com/glue/data/QQP-clean.zip",
109
+ data_dir="QQP",
110
+ citation=textwrap.dedent(
111
+ """\
112
+ @online{WinNT,
113
+ author = {Iyer, Shankar and Dandekar, Nikhil and Csernai, Kornel},
114
+ title = {First Quora Dataset Release: Question Pairs},
115
+ year = {2017},
116
+ url = {https://data.quora.com/First-Quora-Dataset-Release-Question-Pairs},
117
+ urldate = {2019-04-03}
118
+ }"""
119
+ ),
120
+ url="https://data.quora.com/First-Quora-Dataset-Release-Question-Pairs",
121
+ ) for bias_amplified_splits_type in ['minority_examples', 'partial_input']
122
+ ]
123
+
124
+ def _info(self):
125
+ features = {text_feature: datasets.Value("string") for text_feature in self.config.text_features.keys()}
126
+ if self.config.label_classes:
127
+ features["label"] = datasets.features.ClassLabel(names=self.config.label_classes)
128
+ else:
129
+ features["label"] = datasets.Value("float32")
130
+ features["idx"] = datasets.Value("int32")
131
+ return datasets.DatasetInfo(
132
+ description=_GLUE_DESCRIPTION,
133
+ features=datasets.Features(features),
134
+ homepage=self.config.url,
135
+ citation=self.config.citation + "\n" + _GLUE_CITATION,
136
+ )
137
+
138
+ def _split_generators(self, dl_manager):
139
+ return [
140
+ datasets.SplitGenerator(
141
+ name="train.biased",
142
+ gen_kwargs={
143
+ "filepath": dl_manager.download(os.path.join(self.config.name, "train.biased.jsonl")),
144
+ },
145
+ ),
146
+ datasets.SplitGenerator(
147
+ name="train.anti_biased",
148
+ gen_kwargs={
149
+ "filepath": dl_manager.download(os.path.join(self.config.name, "train.anti_biased.jsonl")),
150
+ },
151
+ ),
152
+ datasets.SplitGenerator(
153
+ name="validation.biased",
154
+ gen_kwargs={
155
+ "filepath": dl_manager.download(os.path.join(self.config.name, "validation.biased.jsonl")),
156
+ },
157
+ ),
158
+ datasets.SplitGenerator(
159
+ name="validation.anti_biased",
160
+ gen_kwargs={
161
+ "filepath": dl_manager.download(os.path.join(self.config.name, "validation.anti_biased.jsonl")),
162
+ },
163
+ ),
164
+ ]
165
+
166
+ def _generate_examples(self, filepath):
167
+ """Generate examples.
168
+
169
+ Args:
170
+ filepath: a string
171
+
172
+ Yields:
173
+ dictionaries containing "premise", "hypothesis" and "label" strings
174
+ """
175
+ process_label = self.config.process_label
176
+ label_classes = self.config.label_classes
177
+
178
+ for idx, line in enumerate(open(filepath, "rb")):
179
+ if line is not None:
180
+ line = line.strip().decode("utf-8")
181
+ item = json.loads(line)
182
+ example = {
183
+ "idx": item["idx"],
184
+ "question1": item["question1"],
185
+ "question2": item["question2"],
186
+ }
187
+ if self.config.label_column in item:
188
+ label = item[self.config.label_column]
189
+ # For some tasks, the label is represented as 0 and 1 in the tsv
190
+ # files and needs to be cast to integer to work with the feature.
191
+ if label_classes and label not in label_classes:
192
+ label = int(label) if label else None
193
+ example["label"] = process_label(label)
194
+ else:
195
+ example["label"] = process_label(-1)
196
+
197
+ yield example["idx"], example