holylovenia commited on
Commit
74bdb2d
1 Parent(s): e9636d0

Upload copal.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. copal.py +152 -0
copal.py ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import datasets
2
+ import pandas as pd
3
+
4
+ from seacrowd.utils import schemas
5
+ from seacrowd.utils.configs import SEACrowdConfig
6
+ from seacrowd.utils.constants import Licenses, Tasks
7
+
8
+ _CITATION = """\
9
+ @article{wibowo2023copal,
10
+ title={COPAL-ID: Indonesian Language Reasoning with Local Culture and Nuances},
11
+ author={Wibowo, Haryo Akbarianto and Fuadi, Erland Hilman and Nityasya, Made Nindyatama and Prasojo, Radityo Eko and Aji, Alham Fikri},
12
+ journal={arXiv preprint arXiv:2311.01012},
13
+ year={2023}
14
+ }
15
+ """
16
+ _DATASETNAME = "copal"
17
+
18
+ _DESCRIPTION = """\
19
+ COPAL is a novel Indonesian language common sense reasoning dataset. Unlike the previous Indonesian COPA dataset (XCOPA-ID), COPAL-ID incorporates Indonesian local and cultural nuances,
20
+ providing a more natural portrayal of day-to-day causal reasoning within the Indonesian cultural sphere.
21
+ Professionally written by natives from scratch, COPAL-ID is more fluent and free from awkward phrases, unlike the translated XCOPA-ID.
22
+ Additionally, COPAL-ID is presented in both standard Indonesian and Jakartan Indonesian–a commonly used dialect.
23
+ It consists of premise, choice1, choice2, question, and label, similar to the COPA dataset.
24
+ """
25
+
26
+ _HOMEPAGE = "https://huggingface.co/datasets/haryoaw/COPAL"
27
+
28
+ _LICENSE = Licenses.CC_BY_SA_4_0.value
29
+
30
+ _URLS = {"test": "https://huggingface.co/datasets/haryoaw/COPAL/resolve/main/test_copal.csv?download=true", "test_colloquial": "https://huggingface.co/datasets/haryoaw/COPAL/resolve/main/test_copal_colloquial.csv?download=true"}
31
+
32
+ _SUPPORTED_TASKS = [Tasks.COMMONSENSE_REASONING]
33
+
34
+ _LOCAL = False
35
+ _LANGUAGES = ["ind"]
36
+
37
+ _SOURCE_VERSION = "1.0.0"
38
+
39
+ _SEACROWD_VERSION = "2024.06.20"
40
+
41
+
42
+ class COPAL(datasets.GeneratorBasedBuilder):
43
+
44
+ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
45
+ SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
46
+
47
+ BUILDER_CONFIGS = [
48
+ SEACrowdConfig(
49
+ name=f"{_DATASETNAME}_source",
50
+ version=SOURCE_VERSION,
51
+ description="COPAL test source schema",
52
+ schema="source",
53
+ subset_id="copal",
54
+ ),
55
+ SEACrowdConfig(
56
+ name=f"{_DATASETNAME}_colloquial_source",
57
+ version=SOURCE_VERSION,
58
+ description="COPAL test colloquial source schema",
59
+ schema="source",
60
+ subset_id="copal",
61
+ ),
62
+ SEACrowdConfig(
63
+ name=f"{_DATASETNAME}_seacrowd_qa",
64
+ version=SEACROWD_VERSION,
65
+ description="COPAL test seacrowd schema",
66
+ schema="seacrowd_qa",
67
+ subset_id="copal",
68
+ ),
69
+ SEACrowdConfig(
70
+ name=f"{_DATASETNAME}_colloquial_seacrowd_qa",
71
+ version=SEACROWD_VERSION,
72
+ description="COPAL test colloquial seacrowd schema",
73
+ schema="seacrowd_qa",
74
+ subset_id="copal",
75
+ ),
76
+ ]
77
+
78
+ DEFAULT_CONFIG_NAME = "copal_source"
79
+
80
+ def _info(self):
81
+ if self.config.schema == "source":
82
+ features = datasets.Features(
83
+ {
84
+ "premise": datasets.Value("string"),
85
+ "choice1": datasets.Value("string"),
86
+ "choice2": datasets.Value("string"),
87
+ "question": datasets.Value("string"),
88
+ "idx": datasets.Value("int64"),
89
+ "label": datasets.Value("int64"),
90
+ "terminology": datasets.Value("int64"),
91
+ "culture": datasets.Value("int64"),
92
+ "language": datasets.Value("int64"),
93
+ }
94
+ )
95
+ elif self.config.schema == "seacrowd_qa":
96
+ features = schemas.qa_features
97
+ features["meta"] = {"terminology": datasets.Value("int64"), "culture": datasets.Value("int64"), "language": datasets.Value("int64")}
98
+
99
+ return datasets.DatasetInfo(
100
+ description=_DESCRIPTION,
101
+ features=features,
102
+ homepage=_HOMEPAGE,
103
+ license=_LICENSE,
104
+ citation=_CITATION,
105
+ )
106
+
107
+ def _split_generators(self, dl_manager):
108
+ data_dir = dl_manager.download_and_extract(_URLS)
109
+ if "colloquial" in self.config.name:
110
+ data_url = data_dir["test_colloquial"]
111
+ else:
112
+ data_url = data_dir["test"]
113
+ return [
114
+ datasets.SplitGenerator(
115
+ name=datasets.Split.TEST,
116
+ gen_kwargs={"filepath": data_url},
117
+ ),
118
+ ]
119
+
120
+ def _generate_examples(self, filepath):
121
+ df = pd.read_csv(filepath, sep=",", header="infer").reset_index()
122
+ if self.config.schema == "source":
123
+ for row in df.itertuples():
124
+ entry = {
125
+ "premise": row.premise,
126
+ "choice1": row.choice1,
127
+ "choice2": row.choice2,
128
+ "question": row.question,
129
+ "idx": row.idx,
130
+ "label": row.label,
131
+ "terminology": row.Terminology,
132
+ "culture": row.Culture,
133
+ "language": row.Language,
134
+ }
135
+ yield row.index, entry
136
+
137
+ elif self.config.schema == "seacrowd_qa":
138
+ for row in df.itertuples():
139
+ entry = {
140
+ "id": row.idx,
141
+ "question_id": str(row.idx),
142
+ "document_id": str(row.idx),
143
+ "question": row.question,
144
+ "type": "multiple_choice",
145
+ "choices": [row.choice1, row.choice2],
146
+ "context": row.premise,
147
+ "answer": [row.choice1 if row.label == 0 else row.choice2],
148
+ "meta": {"terminology": row.Terminology, "culture": row.Culture, "language": row.Language},
149
+ }
150
+ yield row.index, entry
151
+ else:
152
+ raise ValueError(f"Invalid config: {self.config.name}")