MCPcopy Create free account
hub / github.com/codefuse-ai/codefuse-devops-eval / preprocess

Function preprocess

src/data/data_preprocess.py:5–33  ·  view source on GitHub ↗

Preprocess df and generate final dict

(df: pd.DataFrame, eval_args, df_dev: pd.DataFrame = None)

Source from the content-addressed store, hash-verified

3
4
5def preprocess(df: pd.DataFrame, eval_args, df_dev: pd.DataFrame = None):
6 '''
7 Preprocess df and generate final dict
8 '''
9 question_prompt = '''以下是关于开发运维领域的单项选择题,请选出其中的正确答案。请直接输出选项。\n'''
10
11 if eval_args.k_shot > 0 and df_dev is not None:
12 # uppercase to lowercase
13 df_dev.rename(columns={
14 'Question': 'question',
15 'Answer': 'answer'
16 }, inplace=True)
17
18 prefix = ''
19
20 for idx in range(eval_args.k_shot):
21 question = df_dev['question'].iloc[idx]
22 prefix = prefix + question_prompt + '问题:' + question + '\n'
23
24 for option in ['A', 'B', 'C', 'D']:
25 if df_dev[option].iloc[idx]:
26 prefix += '{}. {}\n'.format(option, df_dev[option].iloc[idx])
27 prefix += '答案:{}\n'.format(df_dev['answer'].iloc[idx].strip().upper())
28 prefix = prefix + question_prompt
29 res = preprocess_question(df, prefix)
30 else:
31 res = preprocess_question(df, question_prompt)
32
33 return res
34
35def preprocess_question(df: pd.DataFrame, prefix: str = ''):
36 '''

Callers 1

load_all_datasetFunction · 0.90

Calls 1

preprocess_questionFunction · 0.85

Tested by

no test coverage detected