| 141 | |
| 142 | |
| 143 | def data_split(all_data, rate=[0.8, 0.1, 0.1]): |
| 144 | data_len = len(all_data) |
| 145 | split_count = np.array(rate) * data_len |
| 146 | random.shuffle(all_data) |
| 147 | train = all_data[:int(split_count[0])] |
| 148 | dev = all_data[int(split_count[0]): int(split_count[0] + split_count[1])] |
| 149 | test = all_data[int(split_count[0] + split_count[1]):] |
| 150 | # for name in ['train', 'dev', 'test']: |
| 151 | # print('++++++%s summary++++++' % name) |
| 152 | # print_count(eval(name)) |
| 153 | return train, dev, test |
| 154 | |
| 155 | |
| 156 | if __name__ == "__main__": |