| dataset_info: | |
| features: | |
| - name: question | |
| dtype: string | |
| - name: answer | |
| dtype: string | |
| - name: docid | |
| dtype: string | |
| - name: retrieved_docids | |
| sequence: string | |
| - name: retrieved_doc_scores | |
| sequence: float64 | |
| - name: reranked_docids | |
| sequence: string | |
| - name: reranked_scores | |
| sequence: float64 | |
| splits: | |
| - name: train | |
| num_bytes: 3084316 | |
| num_examples: 1000 | |
| download_size: 2588779 | |
| dataset_size: 3084316 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| This is a subset designed for evaluating Instruction-Aware Contextual Compression[(IACC)](https://github.com/howard-hou/instruction-aware-contextual-compressor/tree/main) performance at various retention ratios. |