{
  "run_id": "majority__sms_spam__20261006T043712Z",
  "backend": "majority",
  "model_version": "s1bench-majority-1",
  "suite_id": "sms_spam",
  "suite_version": "1.0.0",
  "suite_title": "SMS spam",
  "primitive": "noul",
  "protocol": "1.0.0",
  "split": "test",
  "limited": false,
  "limit": null,
  "prepare_ms": 0.25877799998852424,
  "latency_p50_ms": 0.0012189999551992514,
  "latency_p95_ms": 0.0016277500662909006,
  "latency_mean_ms": 0.0013608827752311285,
  "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": 0.0
    }
  },
  "warnings": [],
  "license": "The Hugging Face card lists the license as unknown. This is the public SMS Spam Collection (Almeida, Hidalgo, Yamakami), distributed for spam-filtering research.",
  "n": 836,
  "n_valid": 836,
  "schema_validity": 1.0,
  "accuracy": 0.8660287081339713,
  "accuracy_ci95": [
    0.8432715311004785,
    0.8899521531100478
  ],
  "macro_f1": 0.4641025641025641,
  "macro_f1_ci95": [
    0.4574863261333086,
    0.4708860759493671
  ],
  "ece": 6.253693971214336e-05,
  "ece_ci95": [
    6.253693971214336e-05,
    0.0274494247827759
  ],
  "brier": 0.23204597746528977,
  "brier_ci95": [
    0.19702529122660464,
    0.2653594052498389
  ],
  "brier_max": 2.0,
  "score_mae": null,
  "s1_score": 82.90081785344432,
  "s1_score_ci95": [
    81.79042598646262,
    83.39589075477805
  ],
  "components": {
    "accuracy": 0.8660287081339713,
    "macro_f1": 0.4641025641025641,
    "calibration": 0.9999374630602879,
    "brier_quality": 0.8839770112673551,
    "quality": 0.7557259693349189,
    "schema_validity": 1.0,
    "robustness": 1.0,
    "robustness_measured": true
  },
  "composite_id": "s1-v1",
  "reliability": [
    {
      "bin": 0,
      "low": 0.0,
      "high": 0.1,
      "count": 0,
      "confidence": null,
      "accuracy": null
    },
    {
      "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": 836,
      "confidence": 0.8659661711942591,
      "accuracy": 0.8660287081339713
    },
    {
      "bin": 9,
      "low": 0.9,
      "high": 1.0,
      "count": 0,
      "confidence": null,
      "accuracy": null
    }
  ],
  "selective": [
    {
      "threshold": 0.5,
      "n": 836,
      "coverage": 1.0,
      "risk": 0.13397129186602874
    },
    {
      "threshold": 0.6,
      "n": 836,
      "coverage": 1.0,
      "risk": 0.13397129186602874
    },
    {
      "threshold": 0.7,
      "n": 836,
      "coverage": 1.0,
      "risk": 0.13397129186602874
    },
    {
      "threshold": 0.8,
      "n": 836,
      "coverage": 1.0,
      "risk": 0.13397129186602874
    },
    {
      "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": 84,
      "risk": 0.08333333333333337
    },
    {
      "coverage": 0.2,
      "n": 167,
      "risk": 0.09580838323353291
    },
    {
      "coverage": 0.3,
      "n": 251,
      "risk": 0.09561752988047811
    },
    {
      "coverage": 0.4,
      "n": 334,
      "risk": 0.12574850299401197
    },
    {
      "coverage": 0.5,
      "n": 418,
      "risk": 0.13636363636363635
    },
    {
      "coverage": 0.6,
      "n": 502,
      "risk": 0.13545816733067728
    },
    {
      "coverage": 0.7,
      "n": 585,
      "risk": 0.1333333333333333
    },
    {
      "coverage": 0.8,
      "n": 669,
      "risk": 0.1360239162929746
    },
    {
      "coverage": 0.9,
      "n": 752,
      "risk": 0.13430851063829785
    },
    {
      "coverage": 1.0,
      "n": 836,
      "risk": 0.13397129186602874
    }
  ],
  "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."
  }
}