| text to text implementation basing on https://github.com/salesforce/DocNLI | |
| DatasetDict({ | |
| train: Dataset({ | |
| features: ['idx', 'inputs', 'targets'], | |
| num_rows: 942314 | |
| }) | |
| validation: Dataset({ | |
| features: ['idx', 'inputs', 'targets'], | |
| num_rows: 234258 | |
| }) | |
| test: Dataset({ | |
| features: ['idx', 'inputs', 'targets'], | |
| num_rows: 267086 | |
| }) | |
| }) |