{"data":{"kind":"file","path":"README.md","version_id":"ucm0bvxsemv73t992ifdxab3","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1407,"modified_at":"2025-10-08T20:43:11.150000","content_hash":"f3762cd104a428d46a620ba5283f7ed773cd794070cafadd8ae33214ab6594b8"},"entries":[],"content":"# ocr-vl\n\n### Overview\n- **Environment ID**: `ocr-vl`\n- **Short description**: Multilingual numeric OCR VL environment\n- **Tags**: Vision, OCR, VLM\n\n### Datasets\n- **Primary dataset(s)**: webui_multilingual_ocr multilingual WEB UI images and transcriptions\n- **Source links**: [<links>](https://huggingface.co/datasets/deepcopy/webui_multilingual_ocr)\n- **Split sizes**: train 307k\n\n### Task\n- **Type**: single-turn\n- **Parser**: XMLParser\n- **Rubric overview**: mix of WER and CER\n\n### Quickstart\nRun an evaluation with default settings:\n\n```bash\nuv run vf-eval ocr-vl\n```\n\nConfigure model and sampling:\n\n```bash\nuv run vf-eval ocr-vl   -m gpt-4.1-mini   -n 20 -r 3 -t 1024 -T 0.7   -a '{\"lang\": \"hi\",\"size\":1000}'\n```\n\nNotes:\n- Use `-a` / `--env-args` to pass environment-specific configuration as a JSON object.\n\n### Environment Arguments\nDocument any supported environment arguments and their meaning. Example:\n\n| Arg | Type | Default | Description |\n| --- | ---- | ------- | ----------- |\n| `lang` | str | `None` | Lang split (if None : all languages) |\n| `size` | int | `1000` | Limit on dataset size |\n\n### Metrics\nSummarize key metrics your rubric emits and how they’re interpreted.\n\n| Metric | Meaning |\n| ------ | ------- |\n| `format_reward` | Main scalar reward (weighted sum of criteria) |\n| `reward_answer` | 1-CER is mainly digit or non roman sentences, 1-WER of mainly Roman / non digit |\n\n","encoding":"utf-8","truncated":false,"total_bytes":1407},"status":null}