Demo: Open-Loop Evaluation

Run the standard demo plus the sample open-loop dataset workflow:

roboci demo --openloop

This command uses the sample dataset and reference model bundled with the installed package, so it works outside a repository checkout.

Equivalent standalone command:

roboci openloop run \
  --dataset examples/datasets/sample_openloop \
  --model examples/agents/rule_based_agent.py \
  --task trajectory_prediction

The standalone command uses explicit examples/... paths and therefore requires a cloned checkout or equivalent local files.

Expected outputs:

  • runs/<openloop-run>/openloop/predictions.jsonl
  • runs/<openloop-run>/openloop/per_sample_metrics.jsonl
  • runs/<openloop-run>/openloop/aggregate_metrics.json
  • runs/<openloop-run>/metrics.json
  • runs/<openloop-run>/report.html
  • runs/<openloop-run>/report.md

Limitations: the bundled dataset is intentionally tiny and deterministic. It is useful for artifact and metric checks, not model quality claims.