| task: babi | |
| dataset_path: Muennighoff/babi | |
| dataset_name: null | |
| output_type: generate_until | |
| training_split: train | |
| validation_split: valid | |
| test_split: test | |
| doc_to_text: "Passage: {{passage}}Question: {{question}}\nAnswer:" | |
| doc_to_target: " {{answer}}" | |
| target_delimiter: "" | |
| generation_kwargs: | |
| until: | |
| - "\n" | |
| - "Passage:" | |
| metric_list: | |
| - metric: exact_match | |
| aggregation: mean | |
| higher_is_better: true | |
| metadata: | |
| version: 1.0 | |