MCPcopy Create free account
hub / github.com/InternScience/SciReason / load

Method load

opencompass/datasets/c3.py:14–53  ·  view source on GitHub ↗
(path: str)

Source from the content-addressed store, hash-verified

12
13 @staticmethod
14 def load(path: str):
15
16 with open(path, 'r', encoding='utf-8') as f:
17 data = json.load(f)
18 rows = []
19 for _, row in enumerate(data):
20 content = row[0]
21 content_str = ' '.join(
22 [''.join(paragraph) for paragraph in content])
23
24 for question in row[1]:
25 label = question['choice'].index(question['answer'])
26 length = len(question['choice'])
27 if length < 4:
28 fill_value = question['choice'][0] # 以第一个值为填充值
29 fill_count = 4 - length # 需要填充的数量
30 question['choice'] += [fill_value] * fill_count # 填充
31
32 rows.append({
33 'content': content_str,
34 'question': question['question'],
35 'choices': question['choice'],
36 'choice0': question['choice'][0],
37 'choice1': question['choice'][1],
38 'choice2': question['choice'][2],
39 'choice3': question['choice'][3],
40 'label': label
41 })
42
43 dataset = Dataset.from_dict({
44 'content': [row['content'] for row in rows],
45 'question': [row['question'] for row in rows],
46 'choice0': [row['choice0'] for row in rows],
47 'choice1': [row['choice1'] for row in rows],
48 'choice2': [row['choice2'] for row in rows],
49 'choice3': [row['choice3'] for row in rows],
50 'choices': [row['choices'] for row in rows],
51 'label': [row['label'] for row in rows]
52 })
53 return dataset
54
55
56@LOAD_DATASET.register_module()

Callers 15

collect_predsFunction · 0.45
runMethod · 0.45
runMethod · 0.45
dataset_sizeMethod · 0.45
dataset_sizeMethod · 0.45
dataset_sizeMethod · 0.45
dataset_sizeMethod · 0.45
save_to_stationFunction · 0.45
read_from_stationFunction · 0.45
read_jsonMethod · 0.45
download_datasetFunction · 0.45

Calls 1

openFunction · 0.85

Tested by 2

load_datasetsFunction · 0.36
load_datasetsFunction · 0.36