| ## 🔍 Dataset Use | |
| This dataset contains 50 manually-labeled prompts used for evaluating the PriVAL scoring model. | |
| - Format: JSONL | |
| - Fields: `id`, `prompt`, `scores` (13-dimension) | |
| - License: MIT ([view license](./License)) | |
| You can load it manually or via Hugging Face Datasets. |