Understand Output¶
Interpret runtime outputs to verify correctness and drive downstream actions.
When to use¶
Use this after every run in local development and CI.
Prerequisites¶
- A completed runtime execution
Steps¶
- Inspect diagnostics keys:
diagnostics = run.diagnostics.to_dict()
print(sorted(diagnostics.keys()))
- Confirm row progression:
print(diagnostics["stage_rows"])
print(diagnostics["final_rows"])
- Confirm plugin loading by kind:
print(diagnostics["plugin_counts"])
- Inspect ranked results:
frame = run.dataset.to_polars()
print(frame.select("fit_rank", "fit_score", "title", "company").head())
print(run.dataset.materialize_records(limit=1)[0].to_dict())
print([entry.to_dict() for entry in run.application_plan[:5]])
Expected result¶
fit_scoreis bounded to[0.0, 1.0]fit_rankstarts at1application_plancontains top-ranked entries up totop_knon_fatal_errorsappears only whenfail_fast = falseand errors occur
Next steps¶
- Runtime contract details: Runtime API
- Stage semantics: Stage Contracts