Simple dataset returning a single text (TinyStory) per index.
| 31 | |
| 32 | |
| 33 | class TinyStoriesTextDataset(Dataset): |
| 34 | """ |
| 35 | Simple dataset returning a single text (TinyStory) per index. |
| 36 | """ |
| 37 | def __init__(self, text_list): |
| 38 | super().__init__() |
| 39 | self.texts = text_list |
| 40 | |
| 41 | def __len__(self): |
| 42 | return len(self.texts) |
| 43 | |
| 44 | def __getitem__(self, idx): |
| 45 | return self.texts[idx] |
| 46 | |
| 47 | |
| 48 | def collate_and_encode(batch_texts, sonar_model): |