Divyanshu commited on
Commit
bdef099
1 Parent(s): 6d1459e

update builder script

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitignore +163 -0
  2. .history/indicxnli_20220823214315.py +0 -203
  3. .history/indicxnli_20220823214334.py +0 -200
  4. .history/indicxnli_20220823220704.py +0 -201
  5. .history/indicxnli_20220823220708.py +0 -201
  6. .history/indicxnli_20220823220724.py +0 -201
  7. .history/indicxnli_20220823220725.py +0 -201
  8. .history/indicxnli_20220823220728.py +0 -201
  9. .history/indicxnli_20220823220732.py +0 -201
  10. .history/indicxnli_20220823220735.py +0 -201
  11. .history/indicxnli_20220823220738.py +0 -201
  12. .history/indicxnli_20220823220742.py +0 -201
  13. .history/indicxnli_20220823221124.py +0 -201
  14. .history/indicxnli_20220823221128.py +0 -202
  15. .history/indicxnli_20220823221142.py +0 -202
  16. .history/indicxnli_20220823221147.py +0 -202
  17. .history/indicxnli_20220823221200.py +0 -203
  18. .history/indicxnli_20220823221210.py +0 -203
  19. .history/indicxnli_20220823221213.py +0 -203
  20. .history/indicxnli_20220823221223.py +0 -203
  21. .history/indicxnli_20220823221227.py +0 -203
  22. .history/indicxnli_20220823221233.py +0 -203
  23. .history/indicxnli_20220823221235.py +0 -203
  24. .history/indicxnli_20220823221240.py +0 -203
  25. .history/indicxnli_20220823221242.py +0 -203
  26. .history/indicxnli_20220823221316.py +0 -203
  27. .history/indicxnli_20220823221318.py +0 -203
  28. .history/indicxnli_20220823221321.py +0 -203
  29. .history/indicxnli_20220823221324.py +0 -203
  30. .history/indicxnli_20220823221328.py +0 -203
  31. .history/indicxnli_20220823221331.py +0 -203
  32. .history/indicxnli_20220823221333.py +0 -203
  33. .history/indicxnli_20220823221334.py +0 -203
  34. .history/indicxnli_20220823221336.py +0 -203
  35. .history/indicxnli_20220823221338.py +0 -203
  36. .history/indicxnli_20220823221339.py +0 -203
  37. .history/indicxnli_20220823221341.py +0 -203
  38. .history/indicxnli_20220823221351.py +0 -203
  39. .history/indicxnli_20220823221357.py +0 -203
  40. .history/indicxnli_20220823221400.py +0 -203
  41. .history/indicxnli_20220823221408.py +0 -203
  42. .history/indicxnli_20220823221440.py +0 -203
  43. .history/indicxnli_20220823221501.py +0 -203
  44. .history/indicxnli_20220823221505.py +0 -203
  45. .history/indicxnli_20220823221601.py +0 -203
  46. .history/indicxnli_20220823221611.py +0 -203
  47. .history/indicxnli_20220823221621.py +0 -203
  48. .history/indicxnli_20220823221623.py +0 -203
  49. .history/indicxnli_20220823221950.py +0 -151
  50. .history/indicxnli_20220823221952.py +0 -151
.gitignore ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ .history/
2
+
3
+ # Byte-compiled / optimized / DLL files
4
+ __pycache__/
5
+ *.py[cod]
6
+ *$py.class
7
+
8
+ # C extensions
9
+ *.so
10
+
11
+ # Distribution / packaging
12
+ .Python
13
+ build/
14
+ develop-eggs/
15
+ dist/
16
+ downloads/
17
+ eggs/
18
+ .eggs/
19
+ lib/
20
+ lib64/
21
+ parts/
22
+ sdist/
23
+ var/
24
+ wheels/
25
+ share/python-wheels/
26
+ *.egg-info/
27
+ .installed.cfg
28
+ *.egg
29
+ MANIFEST
30
+
31
+ # PyInstaller
32
+ # Usually these files are written by a python script from a template
33
+ # before PyInstaller builds the exe, so as to inject date/other infos into it.
34
+ *.manifest
35
+ *.spec
36
+
37
+ # Installer logs
38
+ pip-log.txt
39
+ pip-delete-this-directory.txt
40
+
41
+ # Unit test / coverage reports
42
+ htmlcov/
43
+ .tox/
44
+ .nox/
45
+ .coverage
46
+ .coverage.*
47
+ .cache
48
+ nosetests.xml
49
+ coverage.xml
50
+ *.cover
51
+ *.py,cover
52
+ .hypothesis/
53
+ .pytest_cache/
54
+ cover/
55
+
56
+ # Translations
57
+ *.mo
58
+ *.pot
59
+
60
+ # Django stuff:
61
+ *.log
62
+ local_settings.py
63
+ db.sqlite3
64
+ db.sqlite3-journal
65
+
66
+ # Flask stuff:
67
+ instance/
68
+ .webassets-cache
69
+
70
+ # Scrapy stuff:
71
+ .scrapy
72
+
73
+ # Sphinx documentation
74
+ docs/_build/
75
+
76
+ # PyBuilder
77
+ .pybuilder/
78
+ target/
79
+
80
+ # Jupyter Notebook
81
+ .ipynb_checkpoints
82
+
83
+ # IPython
84
+ profile_default/
85
+ ipython_config.py
86
+
87
+ # pyenv
88
+ # For a library or package, you might want to ignore these files since the code is
89
+ # intended to run in multiple environments; otherwise, check them in:
90
+ # .python-version
91
+
92
+ # pipenv
93
+ # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
94
+ # However, in case of collaboration, if having platform-specific dependencies or dependencies
95
+ # having no cross-platform support, pipenv may install dependencies that don't work, or not
96
+ # install all needed dependencies.
97
+ #Pipfile.lock
98
+
99
+ # poetry
100
+ # Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
101
+ # This is especially recommended for binary packages to ensure reproducibility, and is more
102
+ # commonly ignored for libraries.
103
+ # https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
104
+ #poetry.lock
105
+
106
+ # pdm
107
+ # Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
108
+ #pdm.lock
109
+ # pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
110
+ # in version control.
111
+ # https://pdm.fming.dev/#use-with-ide
112
+ .pdm.toml
113
+
114
+ # PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
115
+ __pypackages__/
116
+
117
+ # Celery stuff
118
+ celerybeat-schedule
119
+ celerybeat.pid
120
+
121
+ # SageMath parsed files
122
+ *.sage.py
123
+
124
+ # Environments
125
+ .env
126
+ .venv
127
+ env/
128
+ venv/
129
+ ENV/
130
+ env.bak/
131
+ venv.bak/
132
+
133
+ # Spyder project settings
134
+ .spyderproject
135
+ .spyproject
136
+
137
+ # Rope project settings
138
+ .ropeproject
139
+
140
+ # mkdocs documentation
141
+ /site
142
+
143
+ # mypy
144
+ .mypy_cache/
145
+ .dmypy.json
146
+ dmypy.json
147
+
148
+ # Pyre type checker
149
+ .pyre/
150
+
151
+ # pytype static type analyzer
152
+ .pytype/
153
+
154
+ # Cython debug symbols
155
+ cython_debug/
156
+
157
+ # PyCharm
158
+ # JetBrains specific template is maintained in a separate JetBrains.gitignore that can
159
+ # be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
160
+ # and can be added to the global gitignore or merged into this file. For a more nuclear
161
+ # option (not recommended) you can uncomment the following to ignore the entire idea folder.
162
+ #.idea/
163
+ Foo
.history/indicxnli_20220823214315.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- # _TRAIN_DATA_URL = "https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip"
56
- # _TESTVAL_DATA_URL = "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip"
57
-
58
- _LANGUAGES = (
59
- 'hi',
60
- 'bn',
61
- 'mr',
62
- 'as',
63
- 'ta',
64
- 'te',
65
- 'or',
66
- 'ml',
67
- 'pa',
68
- 'gu',
69
- 'kn'
70
- )
71
-
72
-
73
- class IndicxnliConfig(datasets.BuilderConfig):
74
- """BuilderConfig for XNLI."""
75
-
76
- def __init__(self, language: str, **kwargs):
77
- """BuilderConfig for XNLI.
78
-
79
- Args:
80
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
81
- **kwargs: keyword arguments forwarded to super.
82
- """
83
- super(IndicxnliConfig, self).__init__(**kwargs)
84
- self.language = language
85
-
86
-
87
- class Indicxnli(datasets.GeneratorBasedBuilder):
88
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
89
-
90
- VERSION = datasets.Version("1.1.0", "")
91
- BUILDER_CONFIG_CLASS = IndicxnliConfig
92
- BUILDER_CONFIGS = [
93
- IndicxnliConfig(
94
- name=lang,
95
- language=lang,
96
- version=datasets.Version("1.1.0", ""),
97
- description=f"Plain text import of IndicXNLI for the {lang} language",
98
- )
99
- for lang in _LANGUAGES
100
- ]
101
-
102
- def _info(self):
103
- features = datasets.Features(
104
- {
105
- "premise": datasets.Value("string"),
106
- "hypothesis": datasets.Value("string"),
107
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
108
- }
109
- )
110
- return datasets.DatasetInfo(
111
- description=_DESCRIPTION,
112
- features=features,
113
- # No default supervised_keys (as we have to pass both premise
114
- # and hypothesis as input).
115
- supervised_keys=None,
116
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
117
- citation=_CITATION,
118
- )
119
-
120
- def _split_generators(self, dl_manager):
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823214334.py DELETED
@@ -1,200 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
-
119
- return [
120
- datasets.SplitGenerator(
121
- name=datasets.Split.TRAIN,
122
- gen_kwargs={
123
- "filepaths": [
124
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
125
- ],
126
- "data_format": "XNLI-MT",
127
- },
128
- ),
129
- datasets.SplitGenerator(
130
- name=datasets.Split.TEST,
131
- gen_kwargs={"filepaths": [os.path.join(
132
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
133
- ),
134
- datasets.SplitGenerator(
135
- name=datasets.Split.VALIDATION,
136
- gen_kwargs={"filepaths": [os.path.join(
137
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
138
- ),
139
- ]
140
-
141
- def _generate_examples(self, data_format, filepaths):
142
- """This function returns the examples in the raw (text) form."""
143
-
144
- if self.config.language == "all_languages":
145
- if data_format == "XNLI-MT":
146
- with ExitStack() as stack:
147
- files = [stack.enter_context(
148
- open(filepath, encoding="utf-8")) for filepath in filepaths]
149
- readers = [csv.DictReader(
150
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
151
- for row_idx, rows in enumerate(zip(*readers)):
152
- yield row_idx, {
153
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
154
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
155
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
156
- }
157
- else:
158
- rows_per_pair_id = collections.defaultdict(list)
159
- for filepath in filepaths:
160
- with open(filepath, encoding="utf-8") as f:
161
- reader = csv.DictReader(
162
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
163
- for row in reader:
164
- rows_per_pair_id[row["pairID"]].append(row)
165
-
166
- for rows in rows_per_pair_id.values():
167
- premise = {row["language"]: row["sentence1"]
168
- for row in rows}
169
- hypothesis = {row["language"]: row["sentence2"]
170
- for row in rows}
171
- yield rows[0]["pairID"], {
172
- "premise": premise,
173
- "hypothesis": hypothesis,
174
- "label": rows[0]["gold_label"],
175
- }
176
- else:
177
- if data_format == "XNLI-MT":
178
- for file_idx, filepath in enumerate(filepaths):
179
- file = open(filepath, encoding="utf-8")
180
- reader = csv.DictReader(
181
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
182
- for row_idx, row in enumerate(reader):
183
- key = str(file_idx) + "_" + str(row_idx)
184
- yield key, {
185
- "premise": row["premise"],
186
- "hypothesis": row["hypo"],
187
- "label": row["label"].replace("contradictory", "contradiction"),
188
- }
189
- else:
190
- for filepath in filepaths:
191
- with open(filepath, encoding="utf-8") as f:
192
- reader = csv.DictReader(
193
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
194
- for row in reader:
195
- if row["language"] == self.config.language:
196
- yield row["pairID"], {
197
- "premise": row["sentence1"],
198
- "hypothesis": row["sentence2"],
199
- "label": row["gold_label"],
200
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220704.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
-
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220708.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open()
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220724.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open('forward/train')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220725.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open('forward/train', )
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220728.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open('forward/train', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220732.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open('forward/train/', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220735.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open('forward/train/{}', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220738.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/{}', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823220742.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/{language}', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221124.py DELETED
@@ -1,201 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/{self.config.language}', 'r')
119
-
120
- return [
121
- datasets.SplitGenerator(
122
- name=datasets.Split.TRAIN,
123
- gen_kwargs={
124
- "filepaths": [
125
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
126
- ],
127
- "data_format": "XNLI-MT",
128
- },
129
- ),
130
- datasets.SplitGenerator(
131
- name=datasets.Split.TEST,
132
- gen_kwargs={"filepaths": [os.path.join(
133
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
134
- ),
135
- datasets.SplitGenerator(
136
- name=datasets.Split.VALIDATION,
137
- gen_kwargs={"filepaths": [os.path.join(
138
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
139
- ),
140
- ]
141
-
142
- def _generate_examples(self, data_format, filepaths):
143
- """This function returns the examples in the raw (text) form."""
144
-
145
- if self.config.language == "all_languages":
146
- if data_format == "XNLI-MT":
147
- with ExitStack() as stack:
148
- files = [stack.enter_context(
149
- open(filepath, encoding="utf-8")) for filepath in filepaths]
150
- readers = [csv.DictReader(
151
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
152
- for row_idx, rows in enumerate(zip(*readers)):
153
- yield row_idx, {
154
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
155
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
156
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
157
- }
158
- else:
159
- rows_per_pair_id = collections.defaultdict(list)
160
- for filepath in filepaths:
161
- with open(filepath, encoding="utf-8") as f:
162
- reader = csv.DictReader(
163
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
164
- for row in reader:
165
- rows_per_pair_id[row["pairID"]].append(row)
166
-
167
- for rows in rows_per_pair_id.values():
168
- premise = {row["language"]: row["sentence1"]
169
- for row in rows}
170
- hypothesis = {row["language"]: row["sentence2"]
171
- for row in rows}
172
- yield rows[0]["pairID"], {
173
- "premise": premise,
174
- "hypothesis": hypothesis,
175
- "label": rows[0]["gold_label"],
176
- }
177
- else:
178
- if data_format == "XNLI-MT":
179
- for file_idx, filepath in enumerate(filepaths):
180
- file = open(filepath, encoding="utf-8")
181
- reader = csv.DictReader(
182
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
183
- for row_idx, row in enumerate(reader):
184
- key = str(file_idx) + "_" + str(row_idx)
185
- yield key, {
186
- "premise": row["premise"],
187
- "hypothesis": row["hypo"],
188
- "label": row["label"].replace("contradictory", "contradiction"),
189
- }
190
- else:
191
- for filepath in filepaths:
192
- with open(filepath, encoding="utf-8") as f:
193
- reader = csv.DictReader(
194
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
195
- for row in reader:
196
- if row["language"] == self.config.language:
197
- yield row["pairID"], {
198
- "premise": row["sentence1"],
199
- "hypothesis": row["sentence2"],
200
- "label": row["gold_label"],
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221128.py DELETED
@@ -1,202 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/{self.config.language}', 'r') as f:
119
-
120
-
121
- return [
122
- datasets.SplitGenerator(
123
- name=datasets.Split.TRAIN,
124
- gen_kwargs={
125
- "filepaths": [
126
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
127
- ],
128
- "data_format": "XNLI-MT",
129
- },
130
- ),
131
- datasets.SplitGenerator(
132
- name=datasets.Split.TEST,
133
- gen_kwargs={"filepaths": [os.path.join(
134
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
135
- ),
136
- datasets.SplitGenerator(
137
- name=datasets.Split.VALIDATION,
138
- gen_kwargs={"filepaths": [os.path.join(
139
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
140
- ),
141
- ]
142
-
143
- def _generate_examples(self, data_format, filepaths):
144
- """This function returns the examples in the raw (text) form."""
145
-
146
- if self.config.language == "all_languages":
147
- if data_format == "XNLI-MT":
148
- with ExitStack() as stack:
149
- files = [stack.enter_context(
150
- open(filepath, encoding="utf-8")) for filepath in filepaths]
151
- readers = [csv.DictReader(
152
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
153
- for row_idx, rows in enumerate(zip(*readers)):
154
- yield row_idx, {
155
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
156
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
157
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
158
- }
159
- else:
160
- rows_per_pair_id = collections.defaultdict(list)
161
- for filepath in filepaths:
162
- with open(filepath, encoding="utf-8") as f:
163
- reader = csv.DictReader(
164
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
165
- for row in reader:
166
- rows_per_pair_id[row["pairID"]].append(row)
167
-
168
- for rows in rows_per_pair_id.values():
169
- premise = {row["language"]: row["sentence1"]
170
- for row in rows}
171
- hypothesis = {row["language"]: row["sentence2"]
172
- for row in rows}
173
- yield rows[0]["pairID"], {
174
- "premise": premise,
175
- "hypothesis": hypothesis,
176
- "label": rows[0]["gold_label"],
177
- }
178
- else:
179
- if data_format == "XNLI-MT":
180
- for file_idx, filepath in enumerate(filepaths):
181
- file = open(filepath, encoding="utf-8")
182
- reader = csv.DictReader(
183
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
184
- for row_idx, row in enumerate(reader):
185
- key = str(file_idx) + "_" + str(row_idx)
186
- yield key, {
187
- "premise": row["premise"],
188
- "hypothesis": row["hypo"],
189
- "label": row["label"].replace("contradictory", "contradiction"),
190
- }
191
- else:
192
- for filepath in filepaths:
193
- with open(filepath, encoding="utf-8") as f:
194
- reader = csv.DictReader(
195
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
196
- for row in reader:
197
- if row["language"] == self.config.language:
198
- yield row["pairID"], {
199
- "premise": row["sentence1"],
200
- "hypothesis": row["sentence2"],
201
- "label": row["gold_label"],
202
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221142.py DELETED
@@ -1,202 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/xnli_{self.config.language}', 'r') as f:
119
-
120
-
121
- return [
122
- datasets.SplitGenerator(
123
- name=datasets.Split.TRAIN,
124
- gen_kwargs={
125
- "filepaths": [
126
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
127
- ],
128
- "data_format": "XNLI-MT",
129
- },
130
- ),
131
- datasets.SplitGenerator(
132
- name=datasets.Split.TEST,
133
- gen_kwargs={"filepaths": [os.path.join(
134
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
135
- ),
136
- datasets.SplitGenerator(
137
- name=datasets.Split.VALIDATION,
138
- gen_kwargs={"filepaths": [os.path.join(
139
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
140
- ),
141
- ]
142
-
143
- def _generate_examples(self, data_format, filepaths):
144
- """This function returns the examples in the raw (text) form."""
145
-
146
- if self.config.language == "all_languages":
147
- if data_format == "XNLI-MT":
148
- with ExitStack() as stack:
149
- files = [stack.enter_context(
150
- open(filepath, encoding="utf-8")) for filepath in filepaths]
151
- readers = [csv.DictReader(
152
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
153
- for row_idx, rows in enumerate(zip(*readers)):
154
- yield row_idx, {
155
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
156
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
157
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
158
- }
159
- else:
160
- rows_per_pair_id = collections.defaultdict(list)
161
- for filepath in filepaths:
162
- with open(filepath, encoding="utf-8") as f:
163
- reader = csv.DictReader(
164
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
165
- for row in reader:
166
- rows_per_pair_id[row["pairID"]].append(row)
167
-
168
- for rows in rows_per_pair_id.values():
169
- premise = {row["language"]: row["sentence1"]
170
- for row in rows}
171
- hypothesis = {row["language"]: row["sentence2"]
172
- for row in rows}
173
- yield rows[0]["pairID"], {
174
- "premise": premise,
175
- "hypothesis": hypothesis,
176
- "label": rows[0]["gold_label"],
177
- }
178
- else:
179
- if data_format == "XNLI-MT":
180
- for file_idx, filepath in enumerate(filepaths):
181
- file = open(filepath, encoding="utf-8")
182
- reader = csv.DictReader(
183
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
184
- for row_idx, row in enumerate(reader):
185
- key = str(file_idx) + "_" + str(row_idx)
186
- yield key, {
187
- "premise": row["premise"],
188
- "hypothesis": row["hypo"],
189
- "label": row["label"].replace("contradictory", "contradiction"),
190
- }
191
- else:
192
- for filepath in filepaths:
193
- with open(filepath, encoding="utf-8") as f:
194
- reader = csv.DictReader(
195
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
196
- for row in reader:
197
- if row["language"] == self.config.language:
198
- yield row["pairID"], {
199
- "premise": row["sentence1"],
200
- "hypothesis": row["sentence2"],
201
- "label": row["gold_label"],
202
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221147.py DELETED
@@ -1,202 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/xnli_{self.config.language}.json', 'r') as f:
119
-
120
-
121
- return [
122
- datasets.SplitGenerator(
123
- name=datasets.Split.TRAIN,
124
- gen_kwargs={
125
- "filepaths": [
126
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
127
- ],
128
- "data_format": "XNLI-MT",
129
- },
130
- ),
131
- datasets.SplitGenerator(
132
- name=datasets.Split.TEST,
133
- gen_kwargs={"filepaths": [os.path.join(
134
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
135
- ),
136
- datasets.SplitGenerator(
137
- name=datasets.Split.VALIDATION,
138
- gen_kwargs={"filepaths": [os.path.join(
139
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
140
- ),
141
- ]
142
-
143
- def _generate_examples(self, data_format, filepaths):
144
- """This function returns the examples in the raw (text) form."""
145
-
146
- if self.config.language == "all_languages":
147
- if data_format == "XNLI-MT":
148
- with ExitStack() as stack:
149
- files = [stack.enter_context(
150
- open(filepath, encoding="utf-8")) for filepath in filepaths]
151
- readers = [csv.DictReader(
152
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
153
- for row_idx, rows in enumerate(zip(*readers)):
154
- yield row_idx, {
155
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
156
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
157
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
158
- }
159
- else:
160
- rows_per_pair_id = collections.defaultdict(list)
161
- for filepath in filepaths:
162
- with open(filepath, encoding="utf-8") as f:
163
- reader = csv.DictReader(
164
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
165
- for row in reader:
166
- rows_per_pair_id[row["pairID"]].append(row)
167
-
168
- for rows in rows_per_pair_id.values():
169
- premise = {row["language"]: row["sentence1"]
170
- for row in rows}
171
- hypothesis = {row["language"]: row["sentence2"]
172
- for row in rows}
173
- yield rows[0]["pairID"], {
174
- "premise": premise,
175
- "hypothesis": hypothesis,
176
- "label": rows[0]["gold_label"],
177
- }
178
- else:
179
- if data_format == "XNLI-MT":
180
- for file_idx, filepath in enumerate(filepaths):
181
- file = open(filepath, encoding="utf-8")
182
- reader = csv.DictReader(
183
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
184
- for row_idx, row in enumerate(reader):
185
- key = str(file_idx) + "_" + str(row_idx)
186
- yield key, {
187
- "premise": row["premise"],
188
- "hypothesis": row["hypo"],
189
- "label": row["label"].replace("contradictory", "contradiction"),
190
- }
191
- else:
192
- for filepath in filepaths:
193
- with open(filepath, encoding="utf-8") as f:
194
- reader = csv.DictReader(
195
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
196
- for row in reader:
197
- if row["language"] == self.config.language:
198
- yield row["pairID"], {
199
- "premise": row["sentence1"],
200
- "hypothesis": row["sentence2"],
201
- "label": row["gold_label"],
202
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221200.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- with open(f'forward/train/xnli_{self.config.language}.json', 'r') as f:
119
-
120
-
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221210.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json', 'r') as f:
119
-
120
-
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221213.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
-
120
-
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221223.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- train_path = f'forward/train/xnli_{self.config.language}.json'
120
-
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221227.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/train/xnli_{self.config.language}.json'
120
-
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221233.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/train/xnli_{self.config.language}.json'
120
- train_path = f'forward/train/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221235.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/train/xnli_{self.config.language}.json'
120
- test_path = f'forward/train/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221240.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/train/xnli_{self.config.language}.json'
120
- test_path = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221242.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_path = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/dev/xnli_{self.config.language}.json'
120
- test_path = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221316.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/xnli_{self.config.language}.json'
119
- dev_path = f'forward/dev/xnli_{self.config.language}.json'
120
- test_path = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221318.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/xnli_{self.config.language}.json'
119
- dev_dir = f'forward/dev/xnli_{self.config.language}.json'
120
- test_path = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221321.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/xnli_{self.config.language}.json'
119
- dev_dir = f'forward/dev/xnli_{self.config.language}.json'
120
- test_dir = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221324.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/'
119
- dev_dir = f'forward/dev/xnli_{self.config.language}.json'
120
- test_dir = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221328.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/'
119
- dev_dir = f'forward/dev/'
120
- test_dir = f'forward/test/xnli_{self.config.language}.json'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221331.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/'
119
- dev_dir = f'forward/dev/'
120
- test_dir = f'forward/test/'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221333.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/'
119
- dev_dir = f'forward/dev'
120
- test_dir = f'forward/test/'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221334.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train/'
119
- dev_dir = f'forward/dev'
120
- test_dir = f'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221336.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = f'forward/train'
119
- dev_dir = f'forward/dev'
120
- test_dir = f'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221338.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = f'forward/dev'
120
- test_dir = f'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221339.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = f'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221341.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.tsv") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221351.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"multinli.train.{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221357.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI-MT",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221400.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "XNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221408.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221440.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, "xnli.test.tsv")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221501.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json")], "data_format": "XNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221505.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json")], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221601.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221611.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- testval_dir, "xnli.dev.tsv") for lang in self.config.languages], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221621.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221623.py DELETED
@@ -1,203 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- dev_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- if self.config.language == "all_languages":
148
- if data_format == "XNLI-MT":
149
- with ExitStack() as stack:
150
- files = [stack.enter_context(
151
- open(filepath, encoding="utf-8")) for filepath in filepaths]
152
- readers = [csv.DictReader(
153
- file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
154
- for row_idx, rows in enumerate(zip(*readers)):
155
- yield row_idx, {
156
- "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
157
- "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
158
- "label": rows[0]["label"].replace("contradictory", "contradiction"),
159
- }
160
- else:
161
- rows_per_pair_id = collections.defaultdict(list)
162
- for filepath in filepaths:
163
- with open(filepath, encoding="utf-8") as f:
164
- reader = csv.DictReader(
165
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
166
- for row in reader:
167
- rows_per_pair_id[row["pairID"]].append(row)
168
-
169
- for rows in rows_per_pair_id.values():
170
- premise = {row["language"]: row["sentence1"]
171
- for row in rows}
172
- hypothesis = {row["language"]: row["sentence2"]
173
- for row in rows}
174
- yield rows[0]["pairID"], {
175
- "premise": premise,
176
- "hypothesis": hypothesis,
177
- "label": rows[0]["gold_label"],
178
- }
179
- else:
180
- if data_format == "XNLI-MT":
181
- for file_idx, filepath in enumerate(filepaths):
182
- file = open(filepath, encoding="utf-8")
183
- reader = csv.DictReader(
184
- file, delimiter="\t", quoting=csv.QUOTE_NONE)
185
- for row_idx, row in enumerate(reader):
186
- key = str(file_idx) + "_" + str(row_idx)
187
- yield key, {
188
- "premise": row["premise"],
189
- "hypothesis": row["hypo"],
190
- "label": row["label"].replace("contradictory", "contradiction"),
191
- }
192
- else:
193
- for filepath in filepaths:
194
- with open(filepath, encoding="utf-8") as f:
195
- reader = csv.DictReader(
196
- f, delimiter="\t", quoting=csv.QUOTE_NONE)
197
- for row in reader:
198
- if row["language"] == self.config.language:
199
- yield row["pairID"], {
200
- "premise": row["sentence1"],
201
- "hypothesis": row["sentence2"],
202
- "label": row["gold_label"],
203
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221950.py DELETED
@@ -1,151 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- dev_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- yield row["pairID"], {
148
- "premise": row["sentence1"],
149
- "hypothesis": row["sentence2"],
150
- "label": row["gold_label"],
151
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.history/indicxnli_20220823221952.py DELETED
@@ -1,151 +0,0 @@
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
- """XNLI: The Cross-Lingual NLI Corpus."""
18
-
19
-
20
- import collections
21
- import csv
22
- import os
23
- from contextlib import ExitStack
24
-
25
- import datasets
26
-
27
-
28
- _CITATION = """\
29
- @misc{https://doi.org/10.48550/arxiv.2204.08776,
30
- doi = {10.48550/ARXIV.2204.08776},
31
-
32
- url = {https://arxiv.org/abs/2204.08776},
33
-
34
- author = {Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop},
35
-
36
- keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
37
-
38
- title = {IndicXNLI: Evaluating Multilingual Inference for Indian Languages},
39
-
40
- publisher = {arXiv},
41
-
42
- year = {2022},
43
-
44
- copyright = {Creative Commons Attribution 4.0 International}
45
- }
46
- }"""
47
-
48
- _DESCRIPTION = """\
49
- IndicXNLI is a translated version of XNLI to 11 Indic Languages. As with XNLI, the goal is
50
- to predict textual entailment (does sentence A imply/contradict/neither sentence
51
- B) and is a classification task (given two sentences, predict one of three
52
- labels).
53
- """
54
-
55
- _LANGUAGES = (
56
- 'hi',
57
- 'bn',
58
- 'mr',
59
- 'as',
60
- 'ta',
61
- 'te',
62
- 'or',
63
- 'ml',
64
- 'pa',
65
- 'gu',
66
- 'kn'
67
- )
68
-
69
-
70
- class IndicxnliConfig(datasets.BuilderConfig):
71
- """BuilderConfig for XNLI."""
72
-
73
- def __init__(self, language: str, **kwargs):
74
- """BuilderConfig for XNLI.
75
-
76
- Args:
77
- language: One of hi, bn, mr, as, ta, te, or, ml, pa, gu, kn
78
- **kwargs: keyword arguments forwarded to super.
79
- """
80
- super(IndicxnliConfig, self).__init__(**kwargs)
81
- self.language = language
82
-
83
-
84
- class Indicxnli(datasets.GeneratorBasedBuilder):
85
- """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
86
-
87
- VERSION = datasets.Version("1.1.0", "")
88
- BUILDER_CONFIG_CLASS = IndicxnliConfig
89
- BUILDER_CONFIGS = [
90
- IndicxnliConfig(
91
- name=lang,
92
- language=lang,
93
- version=datasets.Version("1.1.0", ""),
94
- description=f"Plain text import of IndicXNLI for the {lang} language",
95
- )
96
- for lang in _LANGUAGES
97
- ]
98
-
99
- def _info(self):
100
- features = datasets.Features(
101
- {
102
- "premise": datasets.Value("string"),
103
- "hypothesis": datasets.Value("string"),
104
- "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
105
- }
106
- )
107
- return datasets.DatasetInfo(
108
- description=_DESCRIPTION,
109
- features=features,
110
- # No default supervised_keys (as we have to pass both premise
111
- # and hypothesis as input).
112
- supervised_keys=None,
113
- homepage="https://www.nyu.edu/projects/bowman/xnli/",
114
- citation=_CITATION,
115
- )
116
-
117
- def _split_generators(self, dl_manager):
118
- train_dir = 'forward/train'
119
- dev_dir = 'forward/dev'
120
- test_dir = 'forward/test'
121
-
122
- return [
123
- datasets.SplitGenerator(
124
- name=datasets.Split.TRAIN,
125
- gen_kwargs={
126
- "filepaths": [
127
- os.path.join(train_dir, f"xnli_{lang}.json") for lang in self.config.languages
128
- ],
129
- "data_format": "IndicXNLI",
130
- },
131
- ),
132
- datasets.SplitGenerator(
133
- name=datasets.Split.TEST,
134
- gen_kwargs={"filepaths": [os.path.join(
135
- test_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "IndicXNLI"},
136
- ),
137
- datasets.SplitGenerator(
138
- name=datasets.Split.VALIDATION,
139
- gen_kwargs={"filepaths": [os.path.join(
140
- dev_dir, f"xnli_{lang}.json") for lang in self.config.languages], "data_format": "XNLI"},
141
- ),
142
- ]
143
-
144
- def _generate_examples(self, data_format, filepaths):
145
- """This function returns the examples in the raw (text) form."""
146
-
147
- yield row["pairID"], {
148
- "premise": row["sentence1"],
149
- "hypothesis": row["sentence2"],
150
- "label": row["gold_label"],
151
- }