Datasets:
Tasks:
Multiple Choice
Sub-tasks:
multiple-choice-qa
Languages:
English
Size:
10K<n<100K
ArXiv:
License:
rishabbala
commited on
Commit
•
91cc602
1
Parent(s):
f47c062
Delete loading script
Browse files- cosmos_qa.py +0 -116
cosmos_qa.py
DELETED
@@ -1,116 +0,0 @@
|
|
1 |
-
"""Cosmos QA dataset."""
|
2 |
-
|
3 |
-
|
4 |
-
import csv
|
5 |
-
import json
|
6 |
-
|
7 |
-
import datasets
|
8 |
-
|
9 |
-
|
10 |
-
_HOMEPAGE = "https://wilburone.github.io/cosmos/"
|
11 |
-
|
12 |
-
_DESCRIPTION = """\
|
13 |
-
Cosmos QA is a large-scale dataset of 35.6K problems that require commonsense-based reading comprehension, formulated as multiple-choice questions. It focuses on reading between the lines over a diverse collection of people's everyday narratives, asking questions concerning on the likely causes or effects of events that require reasoning beyond the exact text spans in the context
|
14 |
-
"""
|
15 |
-
|
16 |
-
_CITATION = """\
|
17 |
-
@inproceedings{huang-etal-2019-cosmos,
|
18 |
-
title = "Cosmos {QA}: Machine Reading Comprehension with Contextual Commonsense Reasoning",
|
19 |
-
author = "Huang, Lifu and
|
20 |
-
Le Bras, Ronan and
|
21 |
-
Bhagavatula, Chandra and
|
22 |
-
Choi, Yejin",
|
23 |
-
booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)",
|
24 |
-
month = nov,
|
25 |
-
year = "2019",
|
26 |
-
address = "Hong Kong, China",
|
27 |
-
publisher = "Association for Computational Linguistics",
|
28 |
-
url = "https://www.aclweb.org/anthology/D19-1243",
|
29 |
-
doi = "10.18653/v1/D19-1243",
|
30 |
-
pages = "2391--2401",
|
31 |
-
}
|
32 |
-
"""
|
33 |
-
|
34 |
-
_LICENSE = "CC BY 4.0"
|
35 |
-
|
36 |
-
_URL = "https://github.com/wilburOne/cosmosqa/raw/master/data/"
|
37 |
-
_URLS = {
|
38 |
-
"train": _URL + "train.csv",
|
39 |
-
"test": _URL + "test.jsonl",
|
40 |
-
"dev": _URL + "valid.csv",
|
41 |
-
}
|
42 |
-
|
43 |
-
|
44 |
-
class CosmosQa(datasets.GeneratorBasedBuilder):
|
45 |
-
"""Cosmos QA dataset."""
|
46 |
-
|
47 |
-
VERSION = datasets.Version("0.1.0")
|
48 |
-
|
49 |
-
def _info(self):
|
50 |
-
return datasets.DatasetInfo(
|
51 |
-
description=_DESCRIPTION,
|
52 |
-
features=datasets.Features(
|
53 |
-
{
|
54 |
-
"id": datasets.Value("string"),
|
55 |
-
"context": datasets.Value("string"),
|
56 |
-
"question": datasets.Value("string"),
|
57 |
-
"answer0": datasets.Value("string"),
|
58 |
-
"answer1": datasets.Value("string"),
|
59 |
-
"answer2": datasets.Value("string"),
|
60 |
-
"answer3": datasets.Value("string"),
|
61 |
-
"label": datasets.Value("int32"),
|
62 |
-
}
|
63 |
-
),
|
64 |
-
homepage=_HOMEPAGE,
|
65 |
-
citation=_CITATION,
|
66 |
-
license=_LICENSE,
|
67 |
-
)
|
68 |
-
|
69 |
-
def _split_generators(self, dl_manager):
|
70 |
-
"""Returns SplitGenerators."""
|
71 |
-
urls_to_download = _URLS
|
72 |
-
dl_dir = dl_manager.download_and_extract(urls_to_download)
|
73 |
-
return [
|
74 |
-
datasets.SplitGenerator(
|
75 |
-
name=datasets.Split.TRAIN,
|
76 |
-
gen_kwargs={"filepath": dl_dir["train"], "split": "train"},
|
77 |
-
),
|
78 |
-
datasets.SplitGenerator(
|
79 |
-
name=datasets.Split.TEST,
|
80 |
-
gen_kwargs={"filepath": dl_dir["test"], "split": "test"},
|
81 |
-
),
|
82 |
-
datasets.SplitGenerator(
|
83 |
-
name=datasets.Split.VALIDATION,
|
84 |
-
gen_kwargs={"filepath": dl_dir["dev"], "split": "dev"},
|
85 |
-
),
|
86 |
-
]
|
87 |
-
|
88 |
-
def _generate_examples(self, filepath, split):
|
89 |
-
"""Yields examples."""
|
90 |
-
with open(filepath, encoding="utf-8") as f:
|
91 |
-
if split == "test":
|
92 |
-
for id_, row in enumerate(f):
|
93 |
-
data = json.loads(row)
|
94 |
-
yield id_, {
|
95 |
-
"id": data["id"],
|
96 |
-
"context": data["context"],
|
97 |
-
"question": data["question"],
|
98 |
-
"answer0": data["answer0"],
|
99 |
-
"answer1": data["answer1"],
|
100 |
-
"answer2": data["answer2"],
|
101 |
-
"answer3": data["answer3"],
|
102 |
-
"label": int(data.get("label", -1)),
|
103 |
-
}
|
104 |
-
else:
|
105 |
-
data = csv.DictReader(f)
|
106 |
-
for id_, row in enumerate(data):
|
107 |
-
yield id_, {
|
108 |
-
"id": row["id"],
|
109 |
-
"context": row["context"],
|
110 |
-
"question": row["question"],
|
111 |
-
"answer0": row["answer0"],
|
112 |
-
"answer1": row["answer1"],
|
113 |
-
"answer2": row["answer2"],
|
114 |
-
"answer3": row["answer3"],
|
115 |
-
"label": int(row.get("label", -1)),
|
116 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|