| { | |
| "description": "Template for a Belebele evaluation run. Replace model, harness, prompt, and scoring placeholders with actual settings before use.", | |
| "model_revision": "REPLACE_WITH_MODEL_COMMIT", | |
| "dataset_revision": "e926d6cc0c50196e0d8bb0d018d9fbf1283dd2e8", | |
| "harness_version": "REPLACE_WITH_HARNESS_RELEASE_OR_COMMIT", | |
| "rephrasing_model": "CohereForAI/aya-expanse-8b", | |
| "rephrasing_model_revision": "REPLACE_WITH_REPHRASER_COMMIT", | |
| "seed": 42, | |
| "num_fewshot": 0, | |
| "prompt_template": "REPLACE_WITH_TEMPLATE_OR_ATTACH_CONFIG", | |
| "metric_definitions": { | |
| "acc": "Document the exact scoring rule used by the producer." | |
| }, | |
| "dataset_repository": "base-model-evals/belebele-rephrased", | |
| "dataset_config": "all", | |
| "dataset_split": "train" | |
| } | |