{
  "schema_version": "ziwu-evaluation-profile-v1",
  "profile_id": "robodojo/fill_pen_holder/xiaomi-r1/natural-start/native-score150",
  "status": "EVIDENCE_BOUNDED_PRIMARY",
  "compatibility": {
    "anchor_report_schema": "Anchor-v0.1",
    "run_bundle_schema": "anchor-run-bundle-v1",
    "trace_schema": "anchor-trace-v1",
    "note": "This is a Ziwu demo profile. Anchor does not yet define a formal profile schema."
  },
  "task_id": "robodojo/fill_pen_holder",
  "subject": {
    "benchmark": "robodojo",
    "task": "fill_pen_holder",
    "agent": "robot_policy",
    "model": "Xiaomi_Robotics_1",
    "policy": "Xiaomi_Robotics_1",
    "checkpoint": "RoboDojo-sim-arx_x5-ee-0",
    "embodiment": "arx_x5",
    "action_type": "ee"
  },
  "cohort": {
    "label": "NATIVE-SCORE150",
    "episodes": 150,
    "layouts": 50,
    "seeds_per_layout": 3,
    "seeds": [0, 1, 2],
    "rollout_mode": "natural_start",
    "horizon_steps": 1100
  },
  "evaluation_contract": {
    "mode": "on_policy_observational",
    "success_condition": "RoboDojo native FINAL_ACCEPTANCE",
    "canonical_progress_contract": "../canonical-progress-stage-contract-v1.json",
    "task_adapter": "../fill-pen-holder-stage-adapter-v1.json",
    "checkpoint_semantics": "P0-P4 are canonical verified-progress bands; P4 requires native final acceptance.",
    "unknown_policy": "UNKNOWN is serialized as null and is never coerced to numeric zero."
  },
  "comparability_group": "robodojo/fill_pen_holder|xiaomi-r1|arx_x5|ee|natural-start|1100|native-score",
  "metrics": [
    {
      "id": "task_success",
      "label": "Native task success",
      "status": "OBSERVED",
      "value": 0.233333,
      "unit": "probability",
      "direction": "maximize",
      "numerator": 35,
      "denominator": 150,
      "ci95": [0.173333, 0.3],
      "estimator": "observed ratio with layout-cluster bootstrap CI",
      "evidence_ref": "native_score150_audit"
    },
    {
      "id": "timeout_rate",
      "label": "Failure or timeout outcome rate",
      "status": "OBSERVED",
      "value": 0.766667,
      "unit": "probability",
      "direction": "minimize",
      "numerator": 115,
      "denominator": 150,
      "ci95": [0.7, 0.826667],
      "estimator": "observed ratio with layout-cluster bootstrap CI",
      "evidence_ref": "native_score150_audit"
    },
    {
      "id": "mean_normalized_max_latched_checkpoint",
      "label": "Mean normalized maximum verified milestone",
      "status": "DERIVED",
      "value": 0.653333,
      "unit": "fraction_of_four_required_milestones",
      "direction": "maximize",
      "numerator": 392,
      "denominator": 600,
      "ci95": null,
      "estimator": "mean(K_max / 4)",
      "evidence_ref": "native_score150_audit",
      "warning": "This is verified milestone progress, not reset recoverability."
    },
    {
      "id": "critical_transition_progress",
      "label": "Observed P3 to P4 progression",
      "status": "OBSERVED",
      "value": 0.380435,
      "unit": "probability",
      "direction": "maximize",
      "numerator": 35,
      "denominator": 92,
      "ci95": [0.290698, 0.475248],
      "estimator": "observed ratio with layout-cluster bootstrap CI",
      "evidence_ref": "observed_landscape"
    },
    {
      "id": "reset_recoverability_auc",
      "label": "Reset recoverability AUC",
      "status": "UNKNOWN",
      "value": null,
      "unit": "probability_area",
      "direction": "maximize",
      "numerator": null,
      "denominator": null,
      "ci95": null,
      "estimator": null,
      "evidence_ref": null
    },
    {
      "id": "identity_specific_coverage",
      "label": "Identity-specific task-graph coverage",
      "status": "UNKNOWN",
      "value": null,
      "unit": "fraction",
      "direction": "maximize",
      "numerator": null,
      "denominator": 16,
      "ci95": null,
      "estimator": null,
      "evidence_ref": "normative_task_graph"
    },
    {
      "id": "success_vs_budget_auc",
      "label": "Success versus budget AUC",
      "status": "UNKNOWN",
      "value": null,
      "unit": "probability_steps",
      "direction": "maximize",
      "numerator": null,
      "denominator": null,
      "ci95": null,
      "estimator": null,
      "evidence_ref": null
    }
  ],
  "distributions": {
    "canonical_exit_stage": { "P0": 6, "P1": 23, "P2": 29, "P3": 57, "P4_ACCEPTED": 35 },
    "terminal_outcome": { "SUCCESS": 35, "FAILURE_OR_TIMEOUT": 115 }
  },
  "curves": {
    "success_vs_budget": {
      "status": "SINGLE_POINT_ONLY",
      "points": [{ "budget_steps": 1100, "success_rate": 0.233333, "ci95": [0.173333, 0.3] }],
      "interpolation_allowed": false,
      "auc": null
    },
    "progress_over_steps": {
      "status": "UNKNOWN",
      "points": null,
      "reason": "The aggregate audit does not publish episode-level first-hit timestamps."
    }
  },
  "layer_coverage": {
    "01_high_dimensional_reality": "AGGREGATE_NATIVE_RESULT_AUDIT",
    "02_semantic_task_graph": "NORMATIVE_COMPLETE_IDENTITY_EMPIRICAL_UNKNOWN",
    "03_task_landscape": "OBSERVATIONAL_CANONICAL_STAGE_PROJECTION",
    "04_measured_coordinates": "EVIDENCE_BOUNDED_PRIMARY"
  },
  "diagnoses": [
    {
      "status": "EVIDENCE_BOUNDED_DIAGNOSTIC",
      "claim": "P3->P4 is the largest observed canonical-stage loss under this cohort condition.",
      "evidence": "35/92 advanced; 57/92 exited; global observed failure mass 57/150 = 0.38.",
      "causal": false
    }
  ],
  "limitations": [
    "Natural-start survivor conditioning is not reset-based recoverability.",
    "The web artifact publishes verified aggregate counts, not raw episode RunBundles and content digests.",
    "Target identity is not reconstructed by the canonical P0-P4 projection.",
    "One 1100-step operating point cannot define a budget curve or AUC."
  ],
  "evidence_refs": [
    { "evidence_id": "native_score150_audit", "kind": "json", "path": "../fill-pen-holder-score-audit-official150-v1.json" },
    { "evidence_id": "observed_landscape", "kind": "json", "path": "../fill-pen-holder-observed-task-landscape-v1.json" },
    { "evidence_id": "canonical_stage_contract", "kind": "json", "path": "../canonical-progress-stage-contract-v1.json" },
    { "evidence_id": "task_adapter", "kind": "json", "path": "../fill-pen-holder-stage-adapter-v1.json" },
    { "evidence_id": "normative_task_graph", "kind": "json", "path": "../fill-pen-holder-task-graph-v6.json" }
  ]
}