Datasets:
Tasks:
Text Classification
Sub-tasks:
natural-language-inference
Languages:
Japanese
Size:
10K - 100K
License:
update default config
Browse files
janli.py
CHANGED
@@ -24,18 +24,47 @@ _DOWNLOAD_URL = "https://raw.githubusercontent.com/verypluming/JaNLI/main/janli.
|
|
24 |
class JaNLIDataset(ds.GeneratorBasedBuilder):
|
25 |
VERSION = ds.Version("1.0.0")
|
26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
def _info(self) -> ds.DatasetInfo:
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
return ds.DatasetInfo(
|
40 |
description=_DESCRIPTION,
|
41 |
citation=_CITATION,
|
@@ -49,6 +78,15 @@ class JaNLIDataset(ds.GeneratorBasedBuilder):
|
|
49 |
df: pd.DataFrame = pd.read_table(data_path, header=0, sep="\t", index_col=0)
|
50 |
df["id"] = df.index
|
51 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
return [
|
53 |
ds.SplitGenerator(
|
54 |
name=ds.Split.TRAIN,
|
|
|
24 |
class JaNLIDataset(ds.GeneratorBasedBuilder):
|
25 |
VERSION = ds.Version("1.0.0")
|
26 |
|
27 |
+
DEFAULT_CONFIG_NAME = "base"
|
28 |
+
|
29 |
+
BUILDER_CONFIGS = [
|
30 |
+
ds.BuilderConfig(
|
31 |
+
name="base",
|
32 |
+
version=VERSION,
|
33 |
+
description="hoge",
|
34 |
+
),
|
35 |
+
ds.BuilderConfig(
|
36 |
+
name="original",
|
37 |
+
version=VERSION,
|
38 |
+
description="hoge",
|
39 |
+
),
|
40 |
+
]
|
41 |
+
|
42 |
def _info(self) -> ds.DatasetInfo:
|
43 |
+
if self.config.name == "base":
|
44 |
+
features = ds.Features(
|
45 |
+
{
|
46 |
+
"id": ds.Value("int64"),
|
47 |
+
"premise": ds.Value("string"),
|
48 |
+
"hypothesis": ds.Value("string"),
|
49 |
+
"label": ds.ClassLabel(names=["entailment", "non-entailment"]),
|
50 |
+
"heuristics": ds.Value("string"),
|
51 |
+
"number_of_NPs": ds.Value("int32"),
|
52 |
+
"semtag": ds.Value("string"),
|
53 |
+
}
|
54 |
+
)
|
55 |
+
elif self.config.name == "original":
|
56 |
+
features = ds.Features(
|
57 |
+
{
|
58 |
+
"id": ds.Value("int64"),
|
59 |
+
"sentence_A_Ja": ds.Value("string"),
|
60 |
+
"sentence_B_Ja": ds.Value("string"),
|
61 |
+
"entailment_label_Ja": ds.ClassLabel(names=["entailment", "non-entailment"]),
|
62 |
+
"heuristics": ds.Value("string"),
|
63 |
+
"number_of_NPs": ds.Value("int32"),
|
64 |
+
"semtag": ds.Value("string"),
|
65 |
+
}
|
66 |
+
)
|
67 |
+
|
68 |
return ds.DatasetInfo(
|
69 |
description=_DESCRIPTION,
|
70 |
citation=_CITATION,
|
|
|
78 |
df: pd.DataFrame = pd.read_table(data_path, header=0, sep="\t", index_col=0)
|
79 |
df["id"] = df.index
|
80 |
|
81 |
+
if self.config.name == "base":
|
82 |
+
df = df.rename(
|
83 |
+
columns={
|
84 |
+
"sentence_A_Ja": "premise",
|
85 |
+
"sentence_B_Ja": "hypothesis",
|
86 |
+
"entailment_label_Ja": "label",
|
87 |
+
}
|
88 |
+
)
|
89 |
+
|
90 |
return [
|
91 |
ds.SplitGenerator(
|
92 |
name=ds.Split.TRAIN,
|