{"data":{"kind":"file","path":"README.md","version_id":"ivj2x9jlw3km9qc7e2g60dc4","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1270,"modified_at":"2025-09-19T17:17:51.533000","content_hash":"d00ed5bebe81d4ed89a35664e7de3e77ffbd5377e8a738127e03e5453fa3dc14"},"entries":[],"content":"# med-spell-count-qa\n\n### Overview\n- **Environment ID**: `med-spell-count-qa`\n- **Short description**: Lightweight environment for **orthographic counting** framed as question-answering over medical terms.  \n- **Tags**: `qa`, `counting`, `spelling`\n\n### Datasets\n- **Primary dataset(s)**: `mkurman/MedSpellCount-QA`\n- **Source links**: https://huggingface.co/datasets/mkurman/MedSpellCount-QA\n- **Split sizes**: 20k train, 200 eval\n\n### Task\n- **Type**: single-turn\n- **Parser**: custom\n- **Rubric overview**: Boxed answer score: 1.0 for a correct answer, 0.0 otherwise. An answer is extracted from the completion and compared to the ground truth. Its answer must be enclosed in a LaTeX box, e.g., `\\boxed{answer}`, and perfectly match the ground truth.\n\n### Quickstart\nRun an evaluation with default settings:\n\n```bash\nuv run vf-eval med-spell-count-qa\n```\n\nConfigure model and sampling:\n\n```bash\nuv run vf-eval med-spell-count-qa   -m gpt-4.1-mini   -n 20 -r 3 -t 1024 -T 0.7   -a '{\"key\": \"value\"}'  # env-specific args as JSON\n```\n\n### Metrics\nSummarize key metrics your rubric emits and how they’re interpreted.\n\n| Metric | Meaning |\n| ------ | ------- |\n| `reward` | Main scalar reward (weighted sum of criteria) |\n| `accuracy` | Exact match on target answer |\n\n","encoding":"utf-8","truncated":false,"total_bytes":1270},"status":null}