{"data":{"kind":"file","path":"README.md","version_id":"bmxnquzzavulel4egtu6ecrb","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":565,"modified_at":"2026-08-26T09:24:47.329000","content_hash":"74b4a7774ce5ee917f48be0abc2e8abb55a2481a5fc4ac1ebd7dd3ce634f39d1"},"entries":[],"content":"# haicai-v1\n\nA v1 verifiers environment, scaffolded with `init`.\n\n## Develop\n\n1. Implement `load` and the `@reward` in `haicai_v1/taskset.py` (see `environments/`).\n2. Install + run:\n\n```bash\nuv pip install -e .        # install this package (or register it in your project)\nuv run eval haicai-v1 -n 3    # evaluate a few tasks with the bash harness\n```\n\n## Layout\n\n- `haicai_v1/taskset.py` — the task (`@reward` scoring + behavior) and the taskset: `load` (data + prompts).\n\nTune knobs from the CLI: `--env.taskset.num-tasks 10`, `--model <id>`, `-n`, and `-r`.\n","encoding":"utf-8","truncated":false,"total_bytes":565},"status":null}