joelniklaus
commited on
Commit
·
0b7dcda
1
Parent(s):
a20fb29
added language
Browse files- lextreme.py +15 -0
lextreme.py
CHANGED
@@ -50,6 +50,7 @@ class LextremeConfig(datasets.BuilderConfig):
|
|
50 |
url,
|
51 |
citation,
|
52 |
hf_hub_name,
|
|
|
53 |
config_name="default",
|
54 |
label_classes=None,
|
55 |
task_type=None,
|
@@ -64,6 +65,7 @@ class LextremeConfig(datasets.BuilderConfig):
|
|
64 |
url: `string`, url for the original project
|
65 |
citation: `string`, citation for the data set
|
66 |
hf_hub_name: `string`, huggingface dataset identifier
|
|
|
67 |
config_name: `string`, huggingface dataset config name
|
68 |
label_classes: `list[string]`, the list of classes if the label is
|
69 |
categorical. If not provided, then the label will be of type
|
@@ -78,6 +80,7 @@ class LextremeConfig(datasets.BuilderConfig):
|
|
78 |
self.description = textwrap.dedent(self.description)
|
79 |
self.citation = textwrap.dedent(citation)
|
80 |
self.hf_hub_name = hf_hub_name
|
|
|
81 |
self.config_name = config_name
|
82 |
self.label_classes = label_classes
|
83 |
self.task_type = task_type
|
@@ -86,6 +89,7 @@ class LextremeConfig(datasets.BuilderConfig):
|
|
86 |
_BRAZILIAN_COURT_DECISIONS_JUDGMENT = {
|
87 |
"task_type": TaskType.SLTC,
|
88 |
"hf_hub_name": "joelito/brazilian_court_decisions",
|
|
|
89 |
"input_col": "decision_description",
|
90 |
"label_col": "judgment_label",
|
91 |
"url": "https://github.com/lagefreitas/predicting-brazilian-court-decisions",
|
@@ -150,6 +154,7 @@ _SWISS_JUDGMENT_PREDICTION = {
|
|
150 |
_GERMAN_ARGUMENT_MINING = {
|
151 |
"task_type": TaskType.SLTC,
|
152 |
"hf_hub_name": "joelito/german_argument_mining",
|
|
|
153 |
"input_col": "input_sentence",
|
154 |
"label_col": "label",
|
155 |
"url": "https://zenodo.org/record/3936490#.X1ed7ovgomK",
|
@@ -179,6 +184,7 @@ _GERMAN_ARGUMENT_MINING = {
|
|
179 |
_GREEK_LEGAL_CODE_VOLUME = {
|
180 |
"task_type": TaskType.SLTC,
|
181 |
"hf_hub_name": "greek_legal_code",
|
|
|
182 |
"config_name": "volume",
|
183 |
"input_col": "text",
|
184 |
"label_col": "label",
|
@@ -3766,6 +3772,7 @@ _MULTI_EURLEX_LEVEL_3 = {
|
|
3766 |
_LENER_BR = {
|
3767 |
"task_type": TaskType.NER,
|
3768 |
"hf_hub_name": "lener_br",
|
|
|
3769 |
"config_name": "lener_br",
|
3770 |
"input_col": "tokens",
|
3771 |
"label_col": "ner_tags",
|
@@ -3813,6 +3820,7 @@ _LENER_BR = {
|
|
3813 |
_LEGAL_NERO = {
|
3814 |
"task_type": TaskType.NER,
|
3815 |
"hf_hub_name": "joelito/legalnero",
|
|
|
3816 |
"input_col": "words",
|
3817 |
"label_col": "ner",
|
3818 |
"url": "https://zenodo.org/record/4922385",
|
@@ -3853,6 +3861,7 @@ _LEGAL_NERO = {
|
|
3853 |
_GREEK_LEGAL_NER = {
|
3854 |
"task_type": TaskType.NER,
|
3855 |
"hf_hub_name": "joelito/greek_legal_ner",
|
|
|
3856 |
"input_col": "words",
|
3857 |
"label_col": "ner",
|
3858 |
"url": "http://legislation.di.uoa.gr/publications?language=en",
|
@@ -4006,6 +4015,7 @@ class LEXTREME(datasets.GeneratorBasedBuilder):
|
|
4006 |
features=datasets.Features({
|
4007 |
"input": input_type,
|
4008 |
"label": label_type,
|
|
|
4009 |
}),
|
4010 |
)
|
4011 |
|
@@ -4030,7 +4040,12 @@ class LEXTREME(datasets.GeneratorBasedBuilder):
|
|
4030 |
label = dataset.features[label_col].int2str(label)
|
4031 |
if self.config.task_type == TaskType.SLTC and label not in self.config.label_classes:
|
4032 |
continue # filter out invalid classes to construct ClassLabel
|
|
|
|
|
|
|
|
|
4033 |
yield id, {
|
4034 |
"input": item[self.config.input_col],
|
4035 |
"label": label,
|
|
|
4036 |
}
|
|
|
50 |
url,
|
51 |
citation,
|
52 |
hf_hub_name,
|
53 |
+
language="multilingual",
|
54 |
config_name="default",
|
55 |
label_classes=None,
|
56 |
task_type=None,
|
|
|
65 |
url: `string`, url for the original project
|
66 |
citation: `string`, citation for the data set
|
67 |
hf_hub_name: `string`, huggingface dataset identifier
|
68 |
+
hf_hub_name: `string`, the language of the dataset (either multilingual or a single language ISO code)
|
69 |
config_name: `string`, huggingface dataset config name
|
70 |
label_classes: `list[string]`, the list of classes if the label is
|
71 |
categorical. If not provided, then the label will be of type
|
|
|
80 |
self.description = textwrap.dedent(self.description)
|
81 |
self.citation = textwrap.dedent(citation)
|
82 |
self.hf_hub_name = hf_hub_name
|
83 |
+
self.language = language
|
84 |
self.config_name = config_name
|
85 |
self.label_classes = label_classes
|
86 |
self.task_type = task_type
|
|
|
89 |
_BRAZILIAN_COURT_DECISIONS_JUDGMENT = {
|
90 |
"task_type": TaskType.SLTC,
|
91 |
"hf_hub_name": "joelito/brazilian_court_decisions",
|
92 |
+
"language": "pt",
|
93 |
"input_col": "decision_description",
|
94 |
"label_col": "judgment_label",
|
95 |
"url": "https://github.com/lagefreitas/predicting-brazilian-court-decisions",
|
|
|
154 |
_GERMAN_ARGUMENT_MINING = {
|
155 |
"task_type": TaskType.SLTC,
|
156 |
"hf_hub_name": "joelito/german_argument_mining",
|
157 |
+
"language": "de",
|
158 |
"input_col": "input_sentence",
|
159 |
"label_col": "label",
|
160 |
"url": "https://zenodo.org/record/3936490#.X1ed7ovgomK",
|
|
|
184 |
_GREEK_LEGAL_CODE_VOLUME = {
|
185 |
"task_type": TaskType.SLTC,
|
186 |
"hf_hub_name": "greek_legal_code",
|
187 |
+
"language": "el",
|
188 |
"config_name": "volume",
|
189 |
"input_col": "text",
|
190 |
"label_col": "label",
|
|
|
3772 |
_LENER_BR = {
|
3773 |
"task_type": TaskType.NER,
|
3774 |
"hf_hub_name": "lener_br",
|
3775 |
+
"language": "pt",
|
3776 |
"config_name": "lener_br",
|
3777 |
"input_col": "tokens",
|
3778 |
"label_col": "ner_tags",
|
|
|
3820 |
_LEGAL_NERO = {
|
3821 |
"task_type": TaskType.NER,
|
3822 |
"hf_hub_name": "joelito/legalnero",
|
3823 |
+
"language": "ro",
|
3824 |
"input_col": "words",
|
3825 |
"label_col": "ner",
|
3826 |
"url": "https://zenodo.org/record/4922385",
|
|
|
3861 |
_GREEK_LEGAL_NER = {
|
3862 |
"task_type": TaskType.NER,
|
3863 |
"hf_hub_name": "joelito/greek_legal_ner",
|
3864 |
+
"language": "el",
|
3865 |
"input_col": "words",
|
3866 |
"label_col": "ner",
|
3867 |
"url": "http://legislation.di.uoa.gr/publications?language=en",
|
|
|
4015 |
features=datasets.Features({
|
4016 |
"input": input_type,
|
4017 |
"label": label_type,
|
4018 |
+
"language": datasets.Value("string"),
|
4019 |
}),
|
4020 |
)
|
4021 |
|
|
|
4040 |
label = dataset.features[label_col].int2str(label)
|
4041 |
if self.config.task_type == TaskType.SLTC and label not in self.config.label_classes:
|
4042 |
continue # filter out invalid classes to construct ClassLabel
|
4043 |
+
if "language" in item.keys():
|
4044 |
+
language = item["language"]
|
4045 |
+
else:
|
4046 |
+
language = self.config.language
|
4047 |
yield id, {
|
4048 |
"input": item[self.config.input_col],
|
4049 |
"label": label,
|
4050 |
+
"language": language,
|
4051 |
}
|