Commit
·
fff9c79
1
Parent(s):
c310749
Support streaming xtreme dataset for bucc18 config (#4026)
Browse filesCommit from https://github.com/huggingface/datasets/commit/369dadbd775a9db07a2db9a571d1a69cdf20b13c
xtreme.py
CHANGED
|
@@ -659,22 +659,20 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
| 659 |
]
|
| 660 |
if self.config.name.startswith("bucc18"):
|
| 661 |
lang = self.config.name.split(".")[1]
|
| 662 |
-
|
| 663 |
self.config.data_url + f"bucc2018-{lang}-en.training-gold.tar.bz2"
|
| 664 |
)
|
| 665 |
-
|
| 666 |
self.config.data_url + f"bucc2018-{lang}-en.sample-gold.tar.bz2"
|
| 667 |
)
|
| 668 |
return [
|
| 669 |
datasets.SplitGenerator(
|
| 670 |
name=datasets.Split.VALIDATION,
|
| 671 |
-
|
| 672 |
-
gen_kwargs={"filepath": os.path.join(bucc18_dl_dev_dir, "bucc2018", lang + "-en")},
|
| 673 |
),
|
| 674 |
datasets.SplitGenerator(
|
| 675 |
name=datasets.Split.TEST,
|
| 676 |
-
|
| 677 |
-
gen_kwargs={"filepath": os.path.join(bucc18_dl_test_dir, "bucc2018", lang + "-en")},
|
| 678 |
),
|
| 679 |
]
|
| 680 |
if self.config.name.startswith("udpos"):
|
|
@@ -886,26 +884,17 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
| 886 |
},
|
| 887 |
}
|
| 888 |
if self.config.name.startswith("bucc18"):
|
| 889 |
-
|
| 890 |
-
|
| 891 |
-
|
| 892 |
-
|
| 893 |
-
|
| 894 |
-
|
| 895 |
-
|
| 896 |
-
|
| 897 |
-
|
| 898 |
-
|
| 899 |
-
|
| 900 |
-
with open(target_file, encoding="utf-8") as f:
|
| 901 |
-
data = csv.reader(f, delimiter="\t")
|
| 902 |
-
target_sentences = [row for row in data]
|
| 903 |
-
with open(source_file, encoding="utf-8") as f:
|
| 904 |
-
data = csv.reader(f, delimiter="\t")
|
| 905 |
-
source_sentences = [row for row in data]
|
| 906 |
-
with open(source_target_file, encoding="utf-8") as f:
|
| 907 |
-
data = csv.reader(f, delimiter="\t")
|
| 908 |
-
source_target_ids = [row for row in data]
|
| 909 |
for id_, pair in enumerate(source_target_ids):
|
| 910 |
source_id = pair[0]
|
| 911 |
target_id = pair[1]
|
|
|
|
| 659 |
]
|
| 660 |
if self.config.name.startswith("bucc18"):
|
| 661 |
lang = self.config.name.split(".")[1]
|
| 662 |
+
bucc18_dl_test_archive = dl_manager.download(
|
| 663 |
self.config.data_url + f"bucc2018-{lang}-en.training-gold.tar.bz2"
|
| 664 |
)
|
| 665 |
+
bucc18_dl_dev_archive = dl_manager.download(
|
| 666 |
self.config.data_url + f"bucc2018-{lang}-en.sample-gold.tar.bz2"
|
| 667 |
)
|
| 668 |
return [
|
| 669 |
datasets.SplitGenerator(
|
| 670 |
name=datasets.Split.VALIDATION,
|
| 671 |
+
gen_kwargs={"filepath": dl_manager.iter_archive(bucc18_dl_dev_archive)},
|
|
|
|
| 672 |
),
|
| 673 |
datasets.SplitGenerator(
|
| 674 |
name=datasets.Split.TEST,
|
| 675 |
+
gen_kwargs={"filepath": dl_manager.iter_archive(bucc18_dl_test_archive)},
|
|
|
|
| 676 |
),
|
| 677 |
]
|
| 678 |
if self.config.name.startswith("udpos"):
|
|
|
|
| 884 |
},
|
| 885 |
}
|
| 886 |
if self.config.name.startswith("bucc18"):
|
| 887 |
+
lang = self.config.name.split(".")[1]
|
| 888 |
+
data_dir = f"bucc2018/{lang}-en"
|
| 889 |
+
for path, file in filepath:
|
| 890 |
+
if path.startswith(data_dir):
|
| 891 |
+
csv_content = [line.decode("utf-8") for line in file]
|
| 892 |
+
if path.endswith("en"):
|
| 893 |
+
target_sentences = list(csv.reader(csv_content, delimiter="\t"))
|
| 894 |
+
elif path.endswith("gold"):
|
| 895 |
+
source_target_ids = list(csv.reader(csv_content, delimiter="\t"))
|
| 896 |
+
else:
|
| 897 |
+
source_sentences = list(csv.reader(csv_content, delimiter="\t"))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 898 |
for id_, pair in enumerate(source_target_ids):
|
| 899 |
source_id = pair[0]
|
| 900 |
target_id = pair[1]
|