Datasets:
Create FarsTail.py
Browse files- FarsTail.py +76 -0
FarsTail.py
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import csv
|
3 |
+
import datasets
|
4 |
+
_CITATION = """\\
|
5 |
+
@article{amirkhani2020farstail,
|
6 |
+
title={FarsTail: A Persian Natural Language Inference Dataset},
|
7 |
+
author={Hossein Amirkhani, Mohammad Azari Jafari, Azadeh Amirak, Zohreh Pourjafari, Soroush Faridan Jahromi, and Zeinab Kouhkan},
|
8 |
+
journal={arXiv preprint arXiv:2009.08820},
|
9 |
+
year={2020}
|
10 |
+
}
|
11 |
+
"""
|
12 |
+
_DESCRIPTION = """\\\\\\\\
|
13 |
+
A Persian Natural Language Inference Dataset
|
14 |
+
"""
|
15 |
+
|
16 |
+
_URL = "https://raw.githubusercontent.com/dml-qom/FarsTail/master/data/"
|
17 |
+
_URLS = {
|
18 |
+
"train": _URL + "Train-word.csv",
|
19 |
+
"test": _URL + "Test-word.csv",
|
20 |
+
"validation": _URL + "Val-word.csv"
|
21 |
+
}
|
22 |
+
class FarsTailConfig(datasets.BuilderConfig):
|
23 |
+
"""BuilderConfig for FarsTail."""
|
24 |
+
def __init__(self, **kwargs):
|
25 |
+
"""BuilderConfig for FarsTail.
|
26 |
+
Args:
|
27 |
+
**kwargs: keyword arguments forwarded to super.
|
28 |
+
"""
|
29 |
+
super(FarsTailConfig, self).__init__(**kwargs)
|
30 |
+
|
31 |
+
class FarsTail(datasets.GeneratorBasedBuilder):
|
32 |
+
BUILDER_CONFIGS = [
|
33 |
+
FarsTailConfig(name="FarsTail", version=datasets.Version("1.0.0"), description="persian NLI dataset"),
|
34 |
+
]
|
35 |
+
def _info(self):
|
36 |
+
return datasets.DatasetInfo(
|
37 |
+
# This is the description that will appear on the datasets page.
|
38 |
+
description=_DESCRIPTION,
|
39 |
+
# datasets.features.FeatureConnectors
|
40 |
+
features=datasets.Features(
|
41 |
+
{
|
42 |
+
"premise": datasets.Value("string"),
|
43 |
+
"hypothesis": datasets.Value("string"),
|
44 |
+
"label": datasets.Value("string")
|
45 |
+
}
|
46 |
+
),
|
47 |
+
supervised_keys=None,
|
48 |
+
# Homepage of the dataset for documentation
|
49 |
+
homepage="https://github.com/dml-qom/FarsTail",
|
50 |
+
citation=_CITATION,
|
51 |
+
)
|
52 |
+
|
53 |
+
def _split_generators(self, dl_manager):
|
54 |
+
"""Returns SplitGenerators."""
|
55 |
+
# dl_manager is a datasets.download.DownloadManager that can be used to
|
56 |
+
# download and extract URLs
|
57 |
+
urls_to_download = _URLS
|
58 |
+
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
59 |
+
return [
|
60 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
|
61 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
|
62 |
+
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["validation"]}),
|
63 |
+
]
|
64 |
+
|
65 |
+
def _generate_examples(self, filepath):
|
66 |
+
try:
|
67 |
+
with open(filepath, encoding="utf-8") as f:
|
68 |
+
reader = csv.DictReader(f, delimiter="\t")
|
69 |
+
for idx, row in enumerate(reader):
|
70 |
+
yield idx, {
|
71 |
+
"premise": row["premise"],
|
72 |
+
"hypothesis": row["hypothesis"],
|
73 |
+
"label": row["label"],
|
74 |
+
}
|
75 |
+
except Exception as e:
|
76 |
+
print(e)
|