(self)
| 142 | self.train_dataset = torch.utils.data.ConcatDataset(self.train_datasets_withtemplate) |
| 143 | |
| 144 | def train_dataloader(self): |
| 145 | return torch.utils.data.DataLoader( |
| 146 | self.train_dataset, |
| 147 | batch_size=self.config.batch_size, |
| 148 | shuffle=True, |
| 149 | collate_fn=create_collate_fn(self.tokenizer.pad_token_id, pretrain=True), |
| 150 | drop_last=True, |
| 151 | num_workers=min([self.config.batch_size, self.config.num_workers]), |
| 152 | ) |
| 153 | |
| 154 | |
| 155 | class PretrainDatasetWithTemplate(torch.utils.data.dataset.Dataset): |
nothing calls this directly
no test coverage detected