{"data":{"kind":"file","path":"README.md","version_id":"arit97ex5n9pnm7g95zd6uzp","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1051,"modified_at":"2026-01-28T15:31:37.680000","content_hash":"c863e897ac7ed55530e3f856809f4a0db73437cca9653fd009dc5c8db69bde63"},"entries":[],"content":"# context-select\n\n### Overview\n- **Environment ID**: `context-select`\n- **Short description**: Select relevant events under a strict context budget.\n- **Tags**: single-turn, selection, train, eval\n\n### Quickstart\nGenerate a dataset:\n\n```bash\npython environments/context_select/generate_dataset.py --out environments/context_select/data/context_select/synthetic.jsonl --num 500 --n_events 120 --k 8\n```\n\nRun eval:\n\n```bash\nprime env install context-select\nprime eval run context-select -m openai/gpt-5-nano -n 100 -r 1 \\\n  -a '{\"dataset_path\":\"data/context_select/synthetic.jsonl\",\"num_examples\":100,\"k\":8}'\n```\n\n### Environment Arguments\n\n| Arg | Type | Default | Description |\n| --- | ---- | ------- | ----------- |\n| `dataset_path` | str | `data/context_select/synthetic.jsonl` | JSONL dataset |\n| `num_examples` | int | `100` | Number of rows |\n| `k` | int | `8` | Budget of selected event IDs |\n\n### Metrics\n\n| Metric | Meaning |\n| ------ | ------- |\n| `reward` | F1 score between selected IDs and gold IDs |\n| `valid_json` | Output JSON parsed |\n","encoding":"utf-8","truncated":false,"total_bytes":1051},"status":null}