Datasets:
add splitting
Browse files- WikiHowNFQA.py +36 -10
WikiHowNFQA.py
CHANGED
@@ -30,20 +30,46 @@ class WikiHowNFQADataset(GeneratorBasedBuilder):
|
|
30 |
)
|
31 |
|
32 |
def _split_generators(self, dl_manager):
|
|
|
33 |
downloaded_file = dl_manager.download_and_extract('https://huggingface.co/datasets/Lurunchik/WikiHowNFQA/resolve/main/WikiHowNFQA.jsonl')
|
|
|
34 |
return [
|
35 |
-
SplitGenerator(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
]
|
37 |
|
38 |
-
def _generate_examples(self, filepath):
|
39 |
with open(filepath, encoding="utf-8") as f:
|
40 |
for id_, line in enumerate(f):
|
41 |
data = json.loads(line)
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
|
|
|
|
|
|
|
|
|
30 |
)
|
31 |
|
32 |
def _split_generators(self, dl_manager):
|
33 |
+
"""Returns SplitGenerators."""
|
34 |
downloaded_file = dl_manager.download_and_extract('https://huggingface.co/datasets/Lurunchik/WikiHowNFQA/resolve/main/WikiHowNFQA.jsonl')
|
35 |
+
|
36 |
return [
|
37 |
+
SplitGenerator(
|
38 |
+
name=Split.TRAIN,
|
39 |
+
gen_kwargs={
|
40 |
+
"filepath": downloaded_file,
|
41 |
+
"split": "train",
|
42 |
+
},
|
43 |
+
),
|
44 |
+
SplitGenerator(
|
45 |
+
name=Split.VALIDATION,
|
46 |
+
gen_kwargs={
|
47 |
+
"filepath": downloaded_file,
|
48 |
+
"split": "valid",
|
49 |
+
},
|
50 |
+
),
|
51 |
+
SplitGenerator(
|
52 |
+
name=Split.TEST,
|
53 |
+
gen_kwargs={
|
54 |
+
"filepath": downloaded_file,
|
55 |
+
"split": "test",
|
56 |
+
},
|
57 |
+
),
|
58 |
]
|
59 |
|
60 |
+
def _generate_examples(self, filepath, split):
|
61 |
with open(filepath, encoding="utf-8") as f:
|
62 |
for id_, line in enumerate(f):
|
63 |
data = json.loads(line)
|
64 |
+
if data['split'] == split:
|
65 |
+
yield id_, {
|
66 |
+
'article_id': data['article_id'],
|
67 |
+
'question': data['question'],
|
68 |
+
'answer': data['answer'],
|
69 |
+
'related_document_urls_wayback_snapshots': data['related_document_urls_wayback_snapshots'],
|
70 |
+
'split': data['split'],
|
71 |
+
'cluster': data['cluster'],
|
72 |
+
}
|
73 |
+
|
74 |
+
|
75 |
+
WikiHowNFQADataset().download_and_prepare()
|