Streaming support
#1
by
mariosasko
- opened
- README.md +0 -0
- greek_legal_code.py +29 -16
README.md
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
greek_legal_code.py
CHANGED
@@ -19,11 +19,9 @@ a collection of Greek legislative documents classified into multi-level (from br
|
|
19 |
|
20 |
|
21 |
import json
|
22 |
-
import os
|
23 |
|
24 |
import datasets
|
25 |
|
26 |
-
|
27 |
_CITATION = """\
|
28 |
@inproceedings{papaloukas-etal-2021-glc,
|
29 |
title = "Multi-granular Legal Topic Classification on Greek Legislation",
|
@@ -2817,7 +2815,10 @@ class GreekLegalCode(datasets.GeneratorBasedBuilder):
|
|
2817 |
|
2818 |
def _info(self):
|
2819 |
features = datasets.Features(
|
2820 |
-
{
|
|
|
|
|
|
|
2821 |
)
|
2822 |
return datasets.DatasetInfo(
|
2823 |
# This is the description that will appear on the datasets page.
|
@@ -2835,45 +2836,57 @@ class GreekLegalCode(datasets.GeneratorBasedBuilder):
|
|
2835 |
# Citation for the dataset
|
2836 |
citation=_CITATION,
|
2837 |
# Task template
|
2838 |
-
task_templates=[
|
|
|
|
|
|
|
|
|
2839 |
)
|
2840 |
|
2841 |
def _split_generators(self, dl_manager):
|
2842 |
"""Returns SplitGenerators."""
|
2843 |
-
|
2844 |
return [
|
2845 |
datasets.SplitGenerator(
|
2846 |
name=datasets.Split.TRAIN,
|
2847 |
# These kwargs will be passed to _generate_examples
|
2848 |
gen_kwargs={
|
2849 |
-
"
|
2850 |
"split": "train",
|
2851 |
},
|
2852 |
),
|
2853 |
datasets.SplitGenerator(
|
2854 |
name=datasets.Split.TEST,
|
2855 |
# These kwargs will be passed to _generate_examples
|
2856 |
-
gen_kwargs={
|
|
|
|
|
|
|
2857 |
),
|
2858 |
datasets.SplitGenerator(
|
2859 |
name=datasets.Split.VALIDATION,
|
2860 |
# These kwargs will be passed to _generate_examples
|
2861 |
gen_kwargs={
|
2862 |
-
"
|
2863 |
"split": "dev",
|
2864 |
},
|
2865 |
),
|
2866 |
]
|
2867 |
|
2868 |
def _generate_examples(
|
2869 |
-
self,
|
|
|
|
|
2870 |
):
|
2871 |
"""Yields examples as (key, example) tuples."""
|
2872 |
|
2873 |
-
|
2874 |
-
|
2875 |
-
|
2876 |
-
|
2877 |
-
|
2878 |
-
|
2879 |
-
|
|
|
|
|
|
|
|
19 |
|
20 |
|
21 |
import json
|
|
|
22 |
|
23 |
import datasets
|
24 |
|
|
|
25 |
_CITATION = """\
|
26 |
@inproceedings{papaloukas-etal-2021-glc,
|
27 |
title = "Multi-granular Legal Topic Classification on Greek Legislation",
|
|
|
2815 |
|
2816 |
def _info(self):
|
2817 |
features = datasets.Features(
|
2818 |
+
{
|
2819 |
+
"text": datasets.Value("string"),
|
2820 |
+
"label": datasets.ClassLabel(names=self.config.label_names),
|
2821 |
+
}
|
2822 |
)
|
2823 |
return datasets.DatasetInfo(
|
2824 |
# This is the description that will appear on the datasets page.
|
|
|
2836 |
# Citation for the dataset
|
2837 |
citation=_CITATION,
|
2838 |
# Task template
|
2839 |
+
task_templates=[
|
2840 |
+
datasets.tasks.TextClassification(
|
2841 |
+
text_column="text", label_column="label"
|
2842 |
+
)
|
2843 |
+
],
|
2844 |
)
|
2845 |
|
2846 |
def _split_generators(self, dl_manager):
|
2847 |
"""Returns SplitGenerators."""
|
2848 |
+
archive = dl_manager.download(URL)
|
2849 |
return [
|
2850 |
datasets.SplitGenerator(
|
2851 |
name=datasets.Split.TRAIN,
|
2852 |
# These kwargs will be passed to _generate_examples
|
2853 |
gen_kwargs={
|
2854 |
+
"archive": dl_manager.iter_archive(archive),
|
2855 |
"split": "train",
|
2856 |
},
|
2857 |
),
|
2858 |
datasets.SplitGenerator(
|
2859 |
name=datasets.Split.TEST,
|
2860 |
# These kwargs will be passed to _generate_examples
|
2861 |
+
gen_kwargs={
|
2862 |
+
"archive": dl_manager.iter_archive(archive),
|
2863 |
+
"split": "test",
|
2864 |
+
},
|
2865 |
),
|
2866 |
datasets.SplitGenerator(
|
2867 |
name=datasets.Split.VALIDATION,
|
2868 |
# These kwargs will be passed to _generate_examples
|
2869 |
gen_kwargs={
|
2870 |
+
"archive": dl_manager.iter_archive(archive),
|
2871 |
"split": "dev",
|
2872 |
},
|
2873 |
),
|
2874 |
]
|
2875 |
|
2876 |
def _generate_examples(
|
2877 |
+
self,
|
2878 |
+
archive,
|
2879 |
+
split, # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
|
2880 |
):
|
2881 |
"""Yields examples as (key, example) tuples."""
|
2882 |
|
2883 |
+
idx = 0
|
2884 |
+
for path, f in archive:
|
2885 |
+
if path == split + ".jsonl":
|
2886 |
+
for line in f:
|
2887 |
+
data = json.loads(line)
|
2888 |
+
yield idx, {
|
2889 |
+
"text": data["text"],
|
2890 |
+
"label": data[self.config.label_type],
|
2891 |
+
}
|
2892 |
+
idx += 1
|