Submit

Enter baseline metric values (will auto-populate from metrics above)

Upload your harness code. Must include a run.py entry point that implements:

def run(workspace: str) -> None:
    """
    workspace contains:
      task.yaml     - Task config (metrics, baseline, evaluator command)
      seed/         - Modifiable seed code (e.g., solve.py)
      evaluator/    - Read-only evaluation scripts
      data/         - Read-only input data
      reference/    - Read-only reference algorithm source

    Write results to:
      results/metrics.json   - Final best scores
      results/best/          - Best evolved code snapshot
      results/history.jsonl  - Iteration log (optional)
    """

Submissions are reviewed by an admin before being registered. You'll be notified when approved.

Files are uploaded to the bioevolve-data Modal Volume under <dataset-id>/. Sandboxes mount the volume read-only so train/ test runs see them at /workspace/data/. Large files (multi-GB BAMs) should be prepared on Modal directly via a script like infra/data/prepare_*.py instead of going through this form.

Admin-only: this writes a new registry/datasets/<id>.yaml and pushes the files to the shared Modal Volume immediately.