数据分割代码,可以调整分割比,以及文件路径
import randomdef split_data(file_path, train_ratio=0.8, dev_ratio=0.1, test_ratio=0.1):with open(file_path, 'r') as file:lines = file.readlines()random.shuffle(lines)total_lines = len(lines)train_end = int(total_lines * train_ratio)dev_end = train_end + int(total_lines * dev_ratio)train_data = lines[:train_end]dev_data = lines[train_end:dev_end]test_data = lines[dev_end:]# path needs to be changed according to conditionswith open('./data/train.txt', 'w') as train_file:train_file.writelines(train_data)# path needs to be changed according to conditionswith open('./data/dev.txt', 'w') as dev_file:dev_file.writelines(dev_data)# path needs to be changed according to conditionswith open('./data/test.txt', 'w') as test_file:test_file.writelines(test_data)if __name__ == "__main__":split_data('./data/all.txt')