| 17 | |
| 18 | |
| 19 | class RaceDataset(Dataset): |
| 20 | |
| 21 | def __init__(self, dataset_name, datapaths, tokenizer, max_seq_length, |
| 22 | max_qa_length=MAX_QA_LENGTH): |
| 23 | |
| 24 | self.dataset_name = dataset_name |
| 25 | print_rank_0(' > building RACE dataset for {}:'.format( |
| 26 | self.dataset_name)) |
| 27 | |
| 28 | string = ' > paths:' |
| 29 | for path in datapaths: |
| 30 | string += ' ' + path |
| 31 | print_rank_0(string) |
| 32 | |
| 33 | self.samples = [] |
| 34 | for datapath in datapaths: |
| 35 | self.samples.extend(process_single_datapath(datapath, tokenizer, |
| 36 | max_qa_length, |
| 37 | max_seq_length)) |
| 38 | |
| 39 | print_rank_0(' >> total number of samples: {}'.format( |
| 40 | len(self.samples))) |
| 41 | |
| 42 | def __len__(self): |
| 43 | return len(self.samples) |
| 44 | |
| 45 | def __getitem__(self, idx): |
| 46 | return self.samples[idx] |
| 47 | |
| 48 | |
| 49 | def process_single_datapath(datapath, tokenizer, max_qa_length, max_seq_length): |
no outgoing calls
no test coverage detected