{"data":{"kind":"file","path":"README.md","version_id":"qugp2gcnjv8clij8j4v3emck","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1838,"modified_at":"2025-09-12T00:53:08.216000","content_hash":"657f74de57e26b254bc9af06ecad741aa75f5a32eb275d2e846a756c7a7a1f30"},"entries":[],"content":"# continuation-quality\n\n### Overview\n- **Environment ID**: `continuation-quality`\n- **Short description**: Single-turn quality grades on base model continuations using a judge model.\n- **Tags**: single-turn, completions, base-model\n\n### Datasets\n- **Primary dataset(s)**: `agentlans/wikipedia-paragraphs` mapped to prefix/ground-truth continuation\n- **Source links**: Hugging Face Datasets\n- **Split sizes**: Train split filtered to adequately-long paragraphs\n\n### Task\n- **Type**: single-turn\n- **Parser**: custom\n- **Rubric overview**: Judge model letter grade (gpt-4.1-mini-based by default)\n\n### Quickstart\nRun an evaluation with default settings:\n\n```bash\nuv run vf-eval continuation-quality\n```\n\nConfigure model and sampling:\n\n```bash\nuv run vf-eval continuation-quality   -m gpt-4.1-mini   -n 20 -r 3 -t 1024 -T 0.7   -a '{\"key\": \"value\"}'  # env-specific args as JSON\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| `dataset_name` | str | `\"agentlans/wikipedia-paragraphs\"` | Training dataset |\n| `dataset_split` | str | `\"train\"` | Training dataset split |\n| `dataset_key` | str | `\"text\"` | Column in dataset with training text |\n| `judge_model` | str | `\"gpt-4.1-mini\"` | Model to judge continuations with |\n| `judge_base_url` | str | `\"https://api.openai.com/v1\"` | API base URL for judge model |\n| `judge_api_key_var` | str | `\"OPENAI_API_KEY\"` | Environment variable containing the judge model API key |\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","encoding":"utf-8","truncated":false,"total_bytes":1838},"status":null}