| language: | |
| - fr | |
| dataset_info: | |
| features: | |
| - name: context | |
| dtype: string | |
| - name: question | |
| dtype: string | |
| - name: label | |
| dtype: int64 | |
| - name: source | |
| dtype: string | |
| splits: | |
| - name: train | |
| num_bytes: 292674100 | |
| num_examples: 234708 | |
| - name: test | |
| num_bytes: 49719642 | |
| num_examples: 33610 | |
| download_size: 57567043 | |
| dataset_size: 342393742 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| - split: test | |
| path: data/test-* | |
| This is a dataset for training the cross-encoder of our RAG system. It is a combination of the PIAF, FQuAD, SQuAD-French, and pandora-s-fr datasets. |