{
  "run_id": "random__banking77__20261006T043913Z",
  "backend": "random",
  "model_version": "s1bench-random-1",
  "suite_id": "banking77",
  "suite_version": "1.0.0",
  "suite_title": "Banking77 intent",
  "primitive": "choice",
  "protocol": "1.0.0",
  "split": "test",
  "limited": false,
  "limit": null,
  "prepare_ms": 0.0,
  "latency_p50_ms": 0.03683949989863322,
  "latency_p95_ms": 0.046885899848803085,
  "latency_mean_ms": 0.039189049349399825,
  "batch_size": 1,
  "concurrency": 1,
  "cost_usd_total": 0.0,
  "cost_per_1k_usd": 0.0,
  "hardware": {
    "platform": "Linux-6.12.94+-x86_64-with-glibc2.39",
    "python": "3.12.3",
    "machine": "x86_64",
    "processor": "x86_64",
    "cpu_count": 4
  },
  "robustness_probes": {
    "order_reverse": {
      "n": 200,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "order_shuffle": {
      "n": 200,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "paraphrase": {
      "n": 200,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "irrelevant_context": {
      "n": 200,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "adversarial": {
      "n": 200,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": 0.025
    }
  },
  "warnings": [],
  "license": "CC-BY-4.0",
  "n": 3080,
  "n_valid": 3080,
  "schema_validity": 1.0,
  "accuracy": 0.012662337662337663,
  "accuracy_ci95": [
    0.008766233766233766,
    0.016883116883116882
  ],
  "macro_f1": 0.012479506953319494,
  "macro_f1_ci95": [
    0.008531953764441252,
    0.016281076720026204
  ],
  "ece": 0.01307478332379311,
  "ece_ci95": [
    0.008865677853896395,
    0.01697780379626258
  ],
  "brier": 0.9912425508444199,
  "brier_ci95": [
    0.9906830466405868,
    0.9917621241163754
  ],
  "brier_max": 2.0,
  "score_mae": null,
  "s1_score": 51.405594007003195,
  "s1_score_ci95": [
    51.18161659377937,
    51.630986376186456
  ],
  "components": {
    "accuracy": 0.012662337662337663,
    "macro_f1": 0.012479506953319494,
    "calibration": 0.9869252166762069,
    "brier_quality": 0.50437872457779,
    "quality": 0.30579420010004565,
    "schema_validity": 1.0,
    "robustness": 1.0,
    "robustness_measured": true
  },
  "composite_id": "s1-v1",
  "reliability": [
    {
      "bin": 0,
      "low": 0.0,
      "high": 0.1,
      "count": 3080,
      "confidence": 0.025737120986130773,
      "accuracy": 0.012662337662337663
    },
    {
      "bin": 1,
      "low": 0.1,
      "high": 0.2,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 2,
      "low": 0.2,
      "high": 0.3,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 3,
      "low": 0.3,
      "high": 0.4,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 4,
      "low": 0.4,
      "high": 0.5,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 5,
      "low": 0.5,
      "high": 0.6,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 6,
      "low": 0.6,
      "high": 0.7,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 7,
      "low": 0.7,
      "high": 0.8,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 8,
      "low": 0.8,
      "high": 0.9,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "bin": 9,
      "low": 0.9,
      "high": 1.0,
      "count": 0,
      "confidence": null,
      "accuracy": null
    }
  ],
  "selective": [
    {
      "threshold": 0.5,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    },
    {
      "threshold": 0.6,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    },
    {
      "threshold": 0.7,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    },
    {
      "threshold": 0.8,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    },
    {
      "threshold": 0.9,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    },
    {
      "threshold": 0.95,
      "n": 0,
      "coverage": 0.0,
      "risk": null
    }
  ],
  "risk_coverage": [
    {
      "coverage": 0.1,
      "n": 308,
      "risk": 0.9805194805194806
    },
    {
      "coverage": 0.2,
      "n": 616,
      "risk": 0.9837662337662337
    },
    {
      "coverage": 0.3,
      "n": 924,
      "risk": 0.987012987012987
    },
    {
      "coverage": 0.4,
      "n": 1232,
      "risk": 0.9878246753246753
    },
    {
      "coverage": 0.5,
      "n": 1540,
      "risk": 0.9876623376623377
    },
    {
      "coverage": 0.6,
      "n": 1848,
      "risk": 0.9886363636363636
    },
    {
      "coverage": 0.7,
      "n": 2156,
      "risk": 0.9888682745825603
    },
    {
      "coverage": 0.8,
      "n": 2464,
      "risk": 0.9874188311688312
    },
    {
      "coverage": 0.9,
      "n": 2772,
      "risk": 0.9877344877344877
    },
    {
      "coverage": 1.0,
      "n": 3080,
      "risk": 0.9873376623376623
    }
  ],
  "bootstrap": {
    "n": 1000,
    "seed": 20261006,
    "interval": "percentile 2.5/97.5"
  },
  "notes": {
    "s1_score_interval": "The interval resamples test items and keeps the robustness term fixed at its point estimate. It is not a bootstrap of the probe subset."
  }
}