| dataset_info: | |
| features: | |
| - name: id | |
| dtype: int64 | |
| - name: question | |
| dtype: string | |
| - name: pred_answer | |
| dtype: string | |
| - name: topk | |
| struct: | |
| - name: contents | |
| sequence: string | |
| - name: docid | |
| sequence: int64 | |
| - name: answer | |
| sequence: string | |
| splits: | |
| - name: train | |
| num_bytes: 1244684721 | |
| num_examples: 222849 | |
| download_size: 706547692 | |
| dataset_size: 1244684721 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| The dataset is originated by [dmrau/multi_qa](dmrau/multi_qa). | |
| The *topk* passages are retrieved by SPLADE-v3 and the *pred_answer* is generated with Mistral-7B-Instruct-v0.2 with the retrieved passages. |