{
  "example_status": "illustrative_unrun_evaluation_plan",
  "run_status": "not_run",
  "baseline_checkpoint": "to_select_and_pin",
  "candidate_checkpoint": "not_trained",
  "model_revision": "to_record",
  "software_versions": "to_pin",
  "grouping_policy": "all known speaker, session and source relationships remain within a split; linked groups must be reconciled",
  "reference_policy": "illustrative-verbatim-v1",
  "scoring_normalization": "to_define_and_freeze_before_comparison",
  "decoder_configuration": "to_record_and_hold_constant",
  "evaluation_splits": [
    "held_out_target_domain",
    "general_speech_regression",
    "silence_and_nonspeech_diagnostic"
  ],
  "metrics": {
    "word_error_rate": null,
    "entity_error_rate": null,
    "insertion_errors_on_nonspeech": null,
    "runtime_seconds": null
  },
  "acceptance_rule": "to_define_before_training",
  "notes": "No audio, trained checkpoint or measured model result is supplied."
}
