abdoelsayed
commited on
Commit
•
0a6cfbd
1
Parent(s):
7a0e90f
Upload ArabicaQA.py
Browse files- ArabicaQA.py +66 -0
ArabicaQA.py
ADDED
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from datasets import load_dataset_builder, DatasetInfo, DownloadConfig, GeneratorBasedBuilder, datasets
|
2 |
+
|
3 |
+
class CustomSQuADFormatDataset(GeneratorBasedBuilder):
|
4 |
+
"""A custom dataset similar to SQuAD but tailored for 'ArabicaQA' hosted on Hugging Face."""
|
5 |
+
|
6 |
+
VERSION = datasets.Version("1.0.0")
|
7 |
+
BUILDER_CONFIGS = [
|
8 |
+
datasets.BuilderConfig(name="ArabicaQA", version=VERSION, description="Custom dataset similar to SQuAD format.")
|
9 |
+
]
|
10 |
+
|
11 |
+
|
12 |
+
def _info(self):
|
13 |
+
return DatasetInfo(
|
14 |
+
description="This dataset is formatted similarly to the SQuAD dataset.",
|
15 |
+
features=datasets.Features(
|
16 |
+
{
|
17 |
+
"id": datasets.Value("string"),
|
18 |
+
"title": datasets.Value("string"),
|
19 |
+
"context": datasets.Value("string"),
|
20 |
+
"question": datasets.Value("string"),
|
21 |
+
"answers": datasets.features.Sequence(
|
22 |
+
{
|
23 |
+
"text": datasets.Value("string"),
|
24 |
+
"answer_start": datasets.Value("int32"),
|
25 |
+
}
|
26 |
+
),
|
27 |
+
}
|
28 |
+
),
|
29 |
+
supervised_keys=None,
|
30 |
+
homepage="https://huggingface.co/datasets/abdoelsayed/ArabicaQA",
|
31 |
+
citation="",
|
32 |
+
)
|
33 |
+
|
34 |
+
def _split_generators(self, dl_manager: DownloadConfig):
|
35 |
+
urls_to_download = {
|
36 |
+
"train": "https://huggingface.co/datasets/abdoelsayed/ArabicaQA/raw/main/MRC/train.json",
|
37 |
+
"dev": "https://huggingface.co/datasets/abdoelsayed/ArabicaQA/raw/main/MRC/val.json",
|
38 |
+
"test": "https://huggingface.co/datasets/abdoelsayed/ArabicaQA/raw/main/MRC/test.json"
|
39 |
+
}
|
40 |
+
downloaded_files = dl_manager.download(urls_to_download)
|
41 |
+
|
42 |
+
return [
|
43 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
|
44 |
+
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
|
45 |
+
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["test"]}),
|
46 |
+
|
47 |
+
]
|
48 |
+
|
49 |
+
def _generate_examples(self, filepath):
|
50 |
+
with open(filepath, encoding="utf-8") as f:
|
51 |
+
squad_data = json.load(f)["data"]
|
52 |
+
for article in squad_data:
|
53 |
+
title = article.get("title", "")
|
54 |
+
for paragraph in article["paragraphs"]:
|
55 |
+
context = paragraph["context"]
|
56 |
+
for qa in paragraph["qas"]:
|
57 |
+
id_ = qa["id"]
|
58 |
+
question = qa["question"]
|
59 |
+
answers = [{"text": answer["text"], "answer_start": answer["answer_start"]} for answer in qa.get("answers", [])]
|
60 |
+
|
61 |
+
yield id_, {
|
62 |
+
"title": title,
|
63 |
+
"context": context,
|
64 |
+
"question": question,
|
65 |
+
"answers": answers,
|
66 |
+
}
|