Datasets:
better description, blackformat
Browse files- citesum.py +8 -5
citesum.py
CHANGED
|
@@ -28,9 +28,10 @@ logger = datasets.logging.get_logger(__name__)
|
|
| 28 |
_HOMEPAGE = "https://github.com/morningmoni/CiteSum"
|
| 29 |
|
| 30 |
_DESCRIPTION = """\
|
| 31 |
-
Citation Text-guided Scientific Extreme Summarization and Low-resource Domain Adaptation
|
| 32 |
-
|
| 33 |
-
CiteSum
|
|
|
|
| 34 |
"""
|
| 35 |
|
| 36 |
# The second citation introduces the source data, while the first
|
|
@@ -49,7 +50,9 @@ _CITATION = """\
|
|
| 49 |
|
| 50 |
"""
|
| 51 |
|
| 52 |
-
_DOWNLOAD_URL =
|
|
|
|
|
|
|
| 53 |
|
| 54 |
|
| 55 |
class CiteSumConfig(datasets.BuilderConfig):
|
|
@@ -104,7 +107,7 @@ class CiteSum(datasets.GeneratorBasedBuilder):
|
|
| 104 |
datasets.SplitGenerator(
|
| 105 |
name=split,
|
| 106 |
gen_kwargs={
|
| 107 |
-
"filepath": os.path.join(dl_path,
|
| 108 |
},
|
| 109 |
)
|
| 110 |
for split in [
|
|
|
|
| 28 |
_HOMEPAGE = "https://github.com/morningmoni/CiteSum"
|
| 29 |
|
| 30 |
_DESCRIPTION = """\
|
| 31 |
+
CiteSum: Citation Text-guided Scientific Extreme Summarization and Low-resource Domain Adaptation.
|
| 32 |
+
|
| 33 |
+
CiteSum contains TLDR summaries for scientific papers from their citation texts without human annotation,
|
| 34 |
+
making it around 30 times larger than the previous human-curated dataset SciTLDR.
|
| 35 |
"""
|
| 36 |
|
| 37 |
# The second citation introduces the source data, while the first
|
|
|
|
| 50 |
|
| 51 |
"""
|
| 52 |
|
| 53 |
+
_DOWNLOAD_URL = (
|
| 54 |
+
"https://drive.google.com/uc?export=download&id=1ndHCREXGSPnDUNllladh9qCtayqbXAfJ"
|
| 55 |
+
)
|
| 56 |
|
| 57 |
|
| 58 |
class CiteSumConfig(datasets.BuilderConfig):
|
|
|
|
| 107 |
datasets.SplitGenerator(
|
| 108 |
name=split,
|
| 109 |
gen_kwargs={
|
| 110 |
+
"filepath": os.path.join(dl_path, file_mapping[split]),
|
| 111 |
},
|
| 112 |
)
|
| 113 |
for split in [
|