Sebastian Gehrmann
commited on
Commit
•
0ea2525
1
Parent(s):
dbccc03
xlsum
Browse files
xlsum.py
CHANGED
@@ -30,11 +30,11 @@ _CITATION = """\
|
|
30 |
|
31 |
|
32 |
_DESCRIPTION = """\
|
33 |
-
We present XLSum, a comprehensive and diverse dataset comprising 1.35 million professionally
|
34 |
annotated article-summary pairs from BBC, extracted using a set of carefully designed heuristics.
|
35 |
The dataset covers 45 languages ranging from low to high-resource, for many of which no
|
36 |
-
public dataset is currently available. XL-Sum is highly abstractive, concise,
|
37 |
-
and of high quality, as indicated by human and intrinsic evaluation.
|
38 |
"""
|
39 |
|
40 |
_HOMEPAGE = "https://github.com/csebuetnlp/xl-sum"
|
@@ -94,7 +94,7 @@ _LANGUAGES = [
|
|
94 |
|
95 |
class Xlsum(datasets.GeneratorBasedBuilder):
|
96 |
VERSION = datasets.Version("2.0.0")
|
97 |
-
|
98 |
BUILDER_CONFIGS = [
|
99 |
datasets.BuilderConfig(
|
100 |
name="{}".format(lang),
|
@@ -111,7 +111,8 @@ class Xlsum(datasets.GeneratorBasedBuilder):
|
|
111 |
"gem_id": datasets.Value("string"),
|
112 |
"url": datasets.Value("string"),
|
113 |
"title": datasets.Value("string"),
|
114 |
-
"
|
|
|
115 |
"text": datasets.Value("string"),
|
116 |
}
|
117 |
),
|
@@ -154,14 +155,15 @@ class Xlsum(datasets.GeneratorBasedBuilder):
|
|
154 |
|
155 |
def _generate_examples(self, filepath, split):
|
156 |
"""Yields examples as (key, example) tuples."""
|
157 |
-
|
158 |
with open(filepath, encoding="utf-8") as f:
|
159 |
for idx_, row in enumerate(f, 1):
|
160 |
data = json.loads(row)
|
161 |
yield idx_, {
|
162 |
-
"gem_id": f"
|
163 |
"url": data["url"],
|
164 |
"title": data["title"],
|
165 |
-
"
|
|
|
166 |
"text": data["text"],
|
167 |
}
|
|
|
30 |
|
31 |
|
32 |
_DESCRIPTION = """\
|
33 |
+
We present XLSum, a comprehensive and diverse dataset comprising 1.35 million professionally
|
34 |
annotated article-summary pairs from BBC, extracted using a set of carefully designed heuristics.
|
35 |
The dataset covers 45 languages ranging from low to high-resource, for many of which no
|
36 |
+
public dataset is currently available. XL-Sum is highly abstractive, concise,
|
37 |
+
and of high quality, as indicated by human and intrinsic evaluation.
|
38 |
"""
|
39 |
|
40 |
_HOMEPAGE = "https://github.com/csebuetnlp/xl-sum"
|
|
|
94 |
|
95 |
class Xlsum(datasets.GeneratorBasedBuilder):
|
96 |
VERSION = datasets.Version("2.0.0")
|
97 |
+
|
98 |
BUILDER_CONFIGS = [
|
99 |
datasets.BuilderConfig(
|
100 |
name="{}".format(lang),
|
|
|
111 |
"gem_id": datasets.Value("string"),
|
112 |
"url": datasets.Value("string"),
|
113 |
"title": datasets.Value("string"),
|
114 |
+
"target": datasets.Value("string"),
|
115 |
+
"references": [datasets.Value("string")],
|
116 |
"text": datasets.Value("string"),
|
117 |
}
|
118 |
),
|
|
|
155 |
|
156 |
def _generate_examples(self, filepath, split):
|
157 |
"""Yields examples as (key, example) tuples."""
|
158 |
+
|
159 |
with open(filepath, encoding="utf-8") as f:
|
160 |
for idx_, row in enumerate(f, 1):
|
161 |
data = json.loads(row)
|
162 |
yield idx_, {
|
163 |
+
"gem_id": f"xlsum_{self.config.name}-{split}-{idx_}",
|
164 |
"url": data["url"],
|
165 |
"title": data["title"],
|
166 |
+
"target": data["summary"],
|
167 |
+
"references": [data["summary"]],
|
168 |
"text": data["text"],
|
169 |
}
|