Create prepare.py
Browse files- prepare.py +49 -0
prepare.py
ADDED
|
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import random
|
| 2 |
+
from datasets import load_dataset, Dataset, DatasetDict
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
def main():
|
| 6 |
+
datasets = load_dataset('allenai/qasc')
|
| 7 |
+
|
| 8 |
+
_datasets = {}
|
| 9 |
+
for split in ['train', 'validation', 'test']:
|
| 10 |
+
data = []
|
| 11 |
+
split_dataset = datasets[split]
|
| 12 |
+
|
| 13 |
+
for example in split_dataset:
|
| 14 |
+
question = str(example['question']).strip()
|
| 15 |
+
correct_answer = example['correct_answer'].strip()
|
| 16 |
+
|
| 17 |
+
# Original choices: 3 distractors + 1 correct answer
|
| 18 |
+
choices = [
|
| 19 |
+
example['distractor1'].strip(),
|
| 20 |
+
example['distractor2'].strip(),
|
| 21 |
+
example['distractor3'].strip(),
|
| 22 |
+
correct_answer
|
| 23 |
+
]
|
| 24 |
+
|
| 25 |
+
# Shuffle choices
|
| 26 |
+
random.shuffle(choices)
|
| 27 |
+
|
| 28 |
+
# Find new index of correct answer after shuffle
|
| 29 |
+
answer_index = choices.index(correct_answer)
|
| 30 |
+
# Convert index to letter A–D
|
| 31 |
+
answer = chr(ord('A') + answer_index)
|
| 32 |
+
|
| 33 |
+
data.append({
|
| 34 |
+
'question': question,
|
| 35 |
+
'choices': choices,
|
| 36 |
+
'answer': answer,
|
| 37 |
+
'answer_index': answer_index,
|
| 38 |
+
'support': example['support'],
|
| 39 |
+
})
|
| 40 |
+
|
| 41 |
+
_datasets[split] = Dataset.from_list(data)
|
| 42 |
+
|
| 43 |
+
datasets = DatasetDict(_datasets)
|
| 44 |
+
|
| 45 |
+
datasets.push_to_hub('extraordinarylab/sciq')
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
if __name__ == '__main__':
|
| 49 |
+
main()
|