Function get_input_data

evaluation/score_MMLU.py:22–31

(test_file_path,filename)

# get result
def get_input_data(test_file_path,filename):
    _hint = f'There is a single choice question about {filename.replace("_test.csv", " ")}. Answer the question by replying A, B, C or D.'
    df = pd.read_csv(os.path.join(test_file_path,filename),header=None,names=["input","A","B","C","D","answer"])
    data = []
    for i in range(len(df)):
        line = df.iloc[i]
        input = f"{_hint}\nQuestion: {line['input']}\nA. {line['A']}\nB. {line['B']}\nC. {line['C']}\nD. {line['D']}\nAnswer: "
        target = line["answer"]
        data.append({"input":input,"target":target})
    return data
# post process
def get_capital_answer(text):
    patterns = [