{
  "run_id": "random__tool_selection__20261006T043915Z",
  "backend": "random",
  "model_version": "s1bench-random-1",
  "suite_id": "tool_selection",
  "suite_version": "1.0.0",
  "suite_title": "Tool selection",
  "primitive": "choice",
  "protocol": "1.0.0",
  "split": "test",
  "limited": false,
  "limit": null,
  "prepare_ms": 0.0,
  "latency_p50_ms": 0.015415500001836335,
  "latency_p95_ms": 0.02148325006601226,
  "latency_mean_ms": 0.01639437499534324,
  "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": 16,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "order_shuffle": {
      "n": 16,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "paraphrase": {
      "n": 16,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "irrelevant_context": {
      "n": 16,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": null
    },
    "adversarial": {
      "n": 16,
      "invalid_pairs": 0,
      "flip_rate": 0.0,
      "probability_drift": 0.0,
      "attack_success_rate": 0.25
    }
  },
  "warnings": [],
  "license": "CC0-1.0",
  "n": 16,
  "n_valid": 16,
  "schema_validity": 1.0,
  "accuracy": 0.125,
  "accuracy_ci95": [
    0.0,
    0.3125
  ],
  "macro_f1": 0.1,
  "macro_f1_ci95": [
    0.0,
    0.1875
  ],
  "ece": 0.12707801483871403,
  "ece_ci95": [
    0.09893411091124102,
    0.2908014331974305
  ],
  "brier": 0.9327420032695181,
  "brier_ci95": [
    0.8460977820161019,
    1.020109715783119
  ],
  "brier_max": 2.0,
  "score_mae": null,
  "s1_score": 54.32921376937138,
  "s1_score_ci95": [
    47.486332965303255,
    58.11257420263406
  ],
  "components": {
    "accuracy": 0.125,
    "macro_f1": 0.1,
    "calibration": 0.8729219851612859,
    "brier_quality": 0.5336289983652409,
    "quality": 0.34756019670530536,
    "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": 4,
      "confidence": 0.18068971304047485,
      "accuracy": 0.0
    },
    {
      "bin": 2,
      "low": 0.2,
      "high": 0.3,
      "count": 8,
      "confidence": 0.2408495777636237,
      "accuracy": 0.25
    },
    {
      "bin": 3,
      "low": 0.3,
      "high": 0.4,
      "count": 4,
      "confidence": 0.30932150184162865,
      "accuracy": 0.0
    },
    {
      "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": 2,
      "risk": 1.0
    },
    {
      "coverage": 0.2,
      "n": 3,
      "risk": 1.0
    },
    {
      "coverage": 0.3,
      "n": 5,
      "risk": 1.0
    },
    {
      "coverage": 0.4,
      "n": 6,
      "risk": 1.0
    },
    {
      "coverage": 0.5,
      "n": 8,
      "risk": 0.875
    },
    {
      "coverage": 0.6,
      "n": 10,
      "risk": 0.9
    },
    {
      "coverage": 0.7,
      "n": 11,
      "risk": 0.9090909090909091
    },
    {
      "coverage": 0.8,
      "n": 13,
      "risk": 0.8461538461538461
    },
    {
      "coverage": 0.9,
      "n": 14,
      "risk": 0.8571428571428572
    },
    {
      "coverage": 1.0,
      "n": 16,
      "risk": 0.875
    }
  ],
  "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."
  }
}