Update test_repo.py
Browse files- test_repo.py +28 -16
test_repo.py
CHANGED
@@ -22,7 +22,10 @@ _LICENSE = "Apache License 2.0"
|
|
22 |
|
23 |
# Define URLs for different parts of the dataset if applicable
|
24 |
_URLS = {
|
25 |
-
"
|
|
|
|
|
|
|
26 |
}
|
27 |
|
28 |
class TestRepo(datasets.GeneratorBasedBuilder):
|
@@ -31,10 +34,12 @@ class TestRepo(datasets.GeneratorBasedBuilder):
|
|
31 |
VERSION = datasets.Version("1.0.0")
|
32 |
|
33 |
BUILDER_CONFIGS = [
|
34 |
-
datasets.BuilderConfig(name="
|
|
|
|
|
35 |
]
|
36 |
|
37 |
-
|
38 |
|
39 |
def _info(self):
|
40 |
return datasets.DatasetInfo(
|
@@ -44,7 +49,6 @@ class TestRepo(datasets.GeneratorBasedBuilder):
|
|
44 |
"question": datasets.Value("string"),
|
45 |
"options": datasets.Sequence(datasets.Value("string")),
|
46 |
"answer": datasets.Value("string"),
|
47 |
-
"context": datasets.Value("string"), # Assuming all data includes context
|
48 |
"num_options": datasets.Value("string"),
|
49 |
"question_type": datasets.Value("string"),
|
50 |
"dataset_name": datasets.Value("string"),
|
@@ -60,9 +64,17 @@ class TestRepo(datasets.GeneratorBasedBuilder):
|
|
60 |
print(data_dir)
|
61 |
return [
|
62 |
datasets.SplitGenerator(
|
63 |
-
name=
|
64 |
gen_kwargs={"directory": data_dir, "split": "train"},
|
65 |
),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
66 |
]
|
67 |
|
68 |
def _generate_examples(self, directory, split):
|
@@ -74,14 +86,14 @@ class TestRepo(datasets.GeneratorBasedBuilder):
|
|
74 |
with open(filepath, encoding="utf-8") as f:
|
75 |
for key, row in enumerate(f):
|
76 |
data = json.loads(row)
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
|
|
22 |
|
23 |
# Define URLs for different parts of the dataset if applicable
|
24 |
_URLS = {
|
25 |
+
"train_normal_mcqa": "https://huggingface.co/datasets/anand-s/test_repo/resolve/main/train_normal_mcqa.zip",
|
26 |
+
"val_normal_mcqa": "https://huggingface.co/datasets/anand-s/test_repo/resolve/main/val_normal_mcqa.zip",
|
27 |
+
"test_normal_mcqa": "https://huggingface.co/datasets/anand-s/test_repo/resolve/main/test_normal_mcqa.zip",
|
28 |
+
|
29 |
}
|
30 |
|
31 |
class TestRepo(datasets.GeneratorBasedBuilder):
|
|
|
34 |
VERSION = datasets.Version("1.0.0")
|
35 |
|
36 |
BUILDER_CONFIGS = [
|
37 |
+
datasets.BuilderConfig(name="train_normal_mcqa", version=VERSION, description="This configuration covers multiple choice questions."),
|
38 |
+
datasets.BuilderConfig(name="val_normal_mcqa", version=VERSION, description="This configuration covers multiple choice questions."),
|
39 |
+
datasets.BuilderConfig(name="test_normal_mcqa", version=VERSION, description="This configuration covers multiple choice questions."),
|
40 |
]
|
41 |
|
42 |
+
# DEFAULT_CONFIG_NAME = "mcq_domain"
|
43 |
|
44 |
def _info(self):
|
45 |
return datasets.DatasetInfo(
|
|
|
49 |
"question": datasets.Value("string"),
|
50 |
"options": datasets.Sequence(datasets.Value("string")),
|
51 |
"answer": datasets.Value("string"),
|
|
|
52 |
"num_options": datasets.Value("string"),
|
53 |
"question_type": datasets.Value("string"),
|
54 |
"dataset_name": datasets.Value("string"),
|
|
|
64 |
print(data_dir)
|
65 |
return [
|
66 |
datasets.SplitGenerator(
|
67 |
+
name="train_normal_mcqa",
|
68 |
gen_kwargs={"directory": data_dir, "split": "train"},
|
69 |
),
|
70 |
+
datasets.SplitGenerator(
|
71 |
+
name="val_normal_mcqa",
|
72 |
+
gen_kwargs={"directory": data_dir, "split": "val"},
|
73 |
+
),
|
74 |
+
datasets.SplitGenerator(
|
75 |
+
name="test_normal_mcqa",
|
76 |
+
gen_kwargs={"directory": data_dir, "split": "test"},
|
77 |
+
),
|
78 |
]
|
79 |
|
80 |
def _generate_examples(self, directory, split):
|
|
|
86 |
with open(filepath, encoding="utf-8") as f:
|
87 |
for key, row in enumerate(f):
|
88 |
data = json.loads(row)
|
89 |
+
if split in ["train_normal_mcqa", "val_normal_mcqa", "test_normal_mcqa"]:
|
90 |
+
yield key_idx, {
|
91 |
+
"prompt": data.get("prompt", ""),
|
92 |
+
"question": data["question"],
|
93 |
+
"options": data["options"],
|
94 |
+
"answer": data.get("answer", ""),
|
95 |
+
"num_options": data.get("num_options", ""),
|
96 |
+
"question_type": data.get("question_type", ""),
|
97 |
+
"dataset_name": os.path.split(filepath)[-1].replace(".jsonl","")
|
98 |
+
}
|
99 |
+
key_idx +=1
|