fix loader
Browse files- KoPI-NLLB.py +5 -5
KoPI-NLLB.py
CHANGED
@@ -35,7 +35,7 @@ for j in _CONF_LANG:
|
|
35 |
_ALL_CONFIG = ["all-raw", "all-dedup", "all-neardup"] + _CONFIGS
|
36 |
_HOMEPAGE = "https://huggingface.co/datasets/munggok/KoPI-NLLB"
|
37 |
_LICENSE = "ODC_C"
|
38 |
-
_BASE_URL =
|
39 |
|
40 |
def kopi_nllb_constructor(nam):
|
41 |
return KoPINLLBConfig(
|
@@ -62,8 +62,8 @@ class KoPINLLB(datasets.GeneratorBasedBuilder):
|
|
62 |
{
|
63 |
"text": datasets.Value("string"),
|
64 |
"url": datasets.Value("string"),
|
65 |
-
"score": datasets.Value("
|
66 |
-
"source": datasets.Value("
|
67 |
}
|
68 |
),
|
69 |
supervised_keys=None,
|
@@ -74,9 +74,9 @@ class KoPINLLB(datasets.GeneratorBasedBuilder):
|
|
74 |
def _split_generators(self, dl_manager):
|
75 |
name = self.config.name.split("-")
|
76 |
if name[0] == "all":
|
77 |
-
train = [_BASE_URL.format(tipe=
|
78 |
else:
|
79 |
-
train = [_BASE_URL.format(tipe=
|
80 |
train_downloaded_files = dl_manager.download(train)
|
81 |
return [
|
82 |
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": train_downloaded_files})
|
|
|
35 |
_ALL_CONFIG = ["all-raw", "all-dedup", "all-neardup"] + _CONFIGS
|
36 |
_HOMEPAGE = "https://huggingface.co/datasets/munggok/KoPI-NLLB"
|
37 |
_LICENSE = "ODC_C"
|
38 |
+
_BASE_URL = 'https://huggingface.co/datasets/munggok/KoPI-NLLB/resolve/main/{tipe}/{lang}.json.zst'
|
39 |
|
40 |
def kopi_nllb_constructor(nam):
|
41 |
return KoPINLLBConfig(
|
|
|
62 |
{
|
63 |
"text": datasets.Value("string"),
|
64 |
"url": datasets.Value("string"),
|
65 |
+
"score": datasets.Value("float32"),
|
66 |
+
"source": datasets.Value("string"),
|
67 |
}
|
68 |
),
|
69 |
supervised_keys=None,
|
|
|
74 |
def _split_generators(self, dl_manager):
|
75 |
name = self.config.name.split("-")
|
76 |
if name[0] == "all":
|
77 |
+
train = [_BASE_URL.format(tipe=name[1],lang=m) for m in _CONF_LANG]
|
78 |
else:
|
79 |
+
train = [_BASE_URL.format(tipe=name[1],lang=name[0])]
|
80 |
train_downloaded_files = dl_manager.download(train)
|
81 |
return [
|
82 |
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": train_downloaded_files})
|