{
  "experiment": "expG_decision_stability",
  "author": "Simon-Pierre Boucher",
  "contact": "contact@spboucher.ai",
  "manifest": {
    "author": "Simon-Pierre Boucher",
    "contact": "contact@spboucher.ai",
    "project": "localvm-research",
    "collected_utc": "2026-08-12T04:26:28.220460+00:00",
    "chip": {
      "brand": "Apple M5 Max",
      "arch": "arm64",
      "cores_total": 18,
      "cores_performance": 6,
      "cores_efficiency": 12,
      "gpu_cores": 40
    },
    "memory": {
      "unified_bytes": 51539607552,
      "unified_gb": 48.0,
      "pagesize": 16384
    },
    "ssd": {
      "model": "APPLE SSD AP2048Z",
      "size": "2 TB",
      "smart_status": "Verified"
    },
    "os": {
      "product": "macOS",
      "version": "27.0",
      "build": "26A5388g",
      "kernel": "27.0.0"
    },
    "software": {
      "python": "3.14.4",
      "mlx": "0.32.0",
      "mlx_lm": "0.31.3",
      "torch": null,
      "numpy": "2.5.2"
    },
    "git": {
      "commit": "2b7d23b36999ff09c3e6ac03a730575e71e79650",
      "dirty_tree": false
    },
    "thermal_level_at_collect": null
  },
  "config": {
    "model": "mlx-community/Qwen3-1.7B-bf16",
    "gen_tokens": 48,
    "bits": "4",
    "per_domain": 2,
    "group_size": 64
  },
  "n_trajectories": 12,
  "results_by_bits": {
    "4": {
      "bits": 4,
      "n_positions": 576,
      "agreement_rate": 0.9236111111111112,
      "mean_kl_ref_q": 0.11886843293905258,
      "auroc_margin_predicts_disagreement": 0.9429895762132604,
      "median_margin_agree": 8.8125,
      "median_margin_disagree": 0.875,
      "escalation_curve": [
        {
          "tau": 0.0,
          "escalated_frac": 0.0,
          "residual_disagree": 0.0763888888888889
        },
        {
          "tau": 0.11118090452261309,
          "escalated_frac": 0.005208333333333333,
          "residual_disagree": 0.07291666666666667
        },
        {
          "tau": 0.22236180904522618,
          "escalated_frac": 0.010416666666666666,
          "residual_disagree": 0.06944444444444445
        },
        {
          "tau": 0.25,
          "escalated_frac": 0.010416666666666666,
          "residual_disagree": 0.06944444444444445
        },
        {
          "tau": 0.31972361809045236,
          "escalated_frac": 0.020833333333333332,
          "residual_disagree": 0.0625
        },
        {
          "tau": 0.375,
          "escalated_frac": 0.020833333333333332,
          "residual_disagree": 0.0625
        },
        {
          "tau": 0.5,
          "escalated_frac": 0.027777777777777776,
          "residual_disagree": 0.059027777777777776
        },
        {
          "tau": 0.5,
          "escalated_frac": 0.027777777777777776,
          "residual_disagree": 0.059027777777777776
        },
        {
          "tau": 0.5,
          "escalated_frac": 0.027777777777777776,
          "residual_disagree": 0.059027777777777776
        },
        {
          "tau": 0.5,
          "escalated_frac": 0.027777777777777776,
          "residual_disagree": 0.059027777777777776
        },
        {
          "tau": 0.5,
          "escalated_frac": 0.027777777777777776,
          "residual_disagree": 0.059027777777777776
        },
        {
          "tau": 0.75,
          "escalated_frac": 0.052083333333333336,
          "residual_disagree": 0.04513888888888889
        },
        {
          "tau": 0.75,
          "escalated_frac": 0.052083333333333336,
          "residual_disagree": 0.04513888888888889
        },
        {
          "tau": 0.8203517587939704,
          "escalated_frac": 0.06597222222222222,
          "residual_disagree": 0.03993055555555555
        },
        {
          "tau": 0.875,
          "escalated_frac": 0.06597222222222222,
          "residual_disagree": 0.03993055555555555
        },
        {
          "tau": 0.9177135678391961,
          "escalated_frac": 0.0763888888888889,
          "residual_disagree": 0.034722222222222224
        },
        {
          "tau": 1.0,
          "escalated_frac": 0.0763888888888889,
          "residual_disagree": 0.034722222222222224
        },
        {
          "tau": 1.0150753768844218,
          "escalated_frac": 0.08680555555555555,
          "residual_disagree": 0.027777777777777776
        },
        {
          "tau": 1.125,
          "escalated_frac": 0.08680555555555555,
          "residual_disagree": 0.027777777777777776
        },
        {
          "tau": 1.25,
          "escalated_frac": 0.09375,
          "residual_disagree": 0.026041666666666668
        },
        {
          "tau": 1.25,
          "escalated_frac": 0.09375,
          "residual_disagree": 0.026041666666666668
        },
        {
          "tau": 1.25,
          "escalated_frac": 0.09375,
          "residual_disagree": 0.026041666666666668
        },
        {
          "tau": 1.25,
          "escalated_frac": 0.09375,
          "residual_disagree": 0.026041666666666668
        },
        {
          "tau": 1.375,
          "escalated_frac": 0.11458333333333333,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.5,
          "escalated_frac": 0.11979166666666667,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.5,
          "escalated_frac": 0.11979166666666667,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.5,
          "escalated_frac": 0.11979166666666667,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.5,
          "escalated_frac": 0.11979166666666667,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.5,
          "escalated_frac": 0.11979166666666667,
          "residual_disagree": 0.019097222222222224
        },
        {
          "tau": 1.625,
          "escalated_frac": 0.1440972222222222,
          "residual_disagree": 0.013888888888888888
        },
        {
          "tau": 1.75,
          "escalated_frac": 0.14930555555555555,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 1.75,
          "escalated_frac": 0.14930555555555555,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 1.75,
          "escalated_frac": 0.14930555555555555,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 1.7939698492462313,
          "escalated_frac": 0.16666666666666666,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.0,
          "escalated_frac": 0.1684027777777778,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.0,
          "escalated_frac": 0.1684027777777778,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.125,
          "escalated_frac": 0.17881944444444445,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.25,
          "escalated_frac": 0.1840277777777778,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.25,
          "escalated_frac": 0.1840277777777778,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.25,
          "escalated_frac": 0.1840277777777778,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.3222361809045236,
          "escalated_frac": 0.2013888888888889,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.375,
          "escalated_frac": 0.2013888888888889,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.5,
          "escalated_frac": 0.20833333333333334,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.5,
          "escalated_frac": 0.20833333333333334,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.5,
          "escalated_frac": 0.20833333333333334,
          "residual_disagree": 0.010416666666666666
        },
        {
          "tau": 2.5031407035175874,
          "escalated_frac": 0.22743055555555555,
          "residual_disagree": 0.008680555555555556
        },
        {
          "tau": 2.625,
          "escalated_frac": 0.22743055555555555,
          "residual_disagree": 0.008680555555555556
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 2.75,
          "escalated_frac": 0.2326388888888889,
          "residual_disagree": 0.006944444444444444
        },
        {
          "tau": 3.0,
          "escalated_frac": 0.265625,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.0037688442211063,
          "escalated_frac": 0.2725694444444444,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.25,
          "escalated_frac": 0.2743055555555556,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.25,
          "escalated_frac": 0.2743055555555556,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.375,
          "escalated_frac": 0.2847222222222222,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.448492462311556,
          "escalated_frac": 0.2916666666666667,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.5,
          "escalated_frac": 0.2916666666666667,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.5,
          "escalated_frac": 0.2916666666666667,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.75,
          "escalated_frac": 0.3055555555555556,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.75,
          "escalated_frac": 0.3055555555555556,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 3.8793969849246217,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.0,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.0,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.0,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.0,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.0,
          "escalated_frac": 0.3177083333333333,
          "residual_disagree": 0.003472222222222222
        },
        {
          "tau": 4.125,
          "escalated_frac": 0.34375,
          "residual_disagree": 0.001736111111111111
        },
        {
          "tau": 4.282663316582916,
          "escalated_frac": 0.3524305555555556,
          "residual_disagree": 0.001736111111111111
        },
        {
          "tau": 4.5,
          "escalated_frac": 0.3541666666666667,
          "residual_disagree": 0.001736111111111111
        },
        {
          "tau": 4.75,
          "escalated_frac": 0.3611111111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 4.75,
          "escalated_frac": 0.3611111111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 4.75,
          "escalated_frac": 0.3611111111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 4.963567839195981,
          "escalated_frac": 0.3767361111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.125,
          "escalated_frac": 0.3802083333333333,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.25,
          "escalated_frac": 0.3836805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.25,
          "escalated_frac": 0.3836805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.408291457286431,
          "escalated_frac": 0.3975694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.5,
          "escalated_frac": 0.3975694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.505653266331656,
          "escalated_frac": 0.4079861111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.625,
          "escalated_frac": 0.4079861111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.75,
          "escalated_frac": 0.4131944444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.839195979899497,
          "escalated_frac": 0.421875,
          "residual_disagree": 0.0
        },
        {
          "tau": 5.950376884422113,
          "escalated_frac": 0.4270833333333333,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.0,
          "escalated_frac": 0.4270833333333333,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.172738693467338,
          "escalated_frac": 0.4375,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.25,
          "escalated_frac": 0.4375,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.5,
          "escalated_frac": 0.4461805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.5,
          "escalated_frac": 0.4461805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.5,
          "escalated_frac": 0.4461805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 6.75,
          "escalated_frac": 0.4600694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.089824120603019,
          "escalated_frac": 0.4670138888888889,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.25,
          "escalated_frac": 0.4704861111111111,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.312185929648244,
          "escalated_frac": 0.4774305555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.423366834170849,
          "escalated_frac": 0.4826388888888889,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.5,
          "escalated_frac": 0.4826388888888889,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.75,
          "escalated_frac": 0.4895833333333333,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.756909547738694,
          "escalated_frac": 0.4982638888888889,
          "residual_disagree": 0.0
        },
        {
          "tau": 7.875,
          "escalated_frac": 0.4982638888888889,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.104271356783919,
          "escalated_frac": 0.5069444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.125,
          "escalated_frac": 0.5069444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.32663316582915,
          "escalated_frac": 0.5173611111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.375,
          "escalated_frac": 0.5173611111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.5,
          "escalated_frac": 0.5260416666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.75,
          "escalated_frac": 0.53125,
          "residual_disagree": 0.0
        },
        {
          "tau": 8.7713567839196,
          "escalated_frac": 0.5381944444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.0,
          "escalated_frac": 0.5399305555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.0,
          "escalated_frac": 0.5399305555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.125,
          "escalated_frac": 0.5503472222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.34108040201005,
          "escalated_frac": 0.5572916666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.5,
          "escalated_frac": 0.5590277777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.5,
          "escalated_frac": 0.5590277777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.625,
          "escalated_frac": 0.5694444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.75,
          "escalated_frac": 0.5763888888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 9.896984924623112,
          "escalated_frac": 0.5833333333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.125,
          "escalated_frac": 0.5850694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.25,
          "escalated_frac": 0.5902777777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.375,
          "escalated_frac": 0.59375,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.5,
          "escalated_frac": 0.5989583333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.577889447236181,
          "escalated_frac": 0.6076388888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.625,
          "escalated_frac": 0.6076388888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.75,
          "escalated_frac": 0.6163194444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 10.875,
          "escalated_frac": 0.6197916666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.0,
          "escalated_frac": 0.6267361111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.125,
          "escalated_frac": 0.6319444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.244974874371856,
          "escalated_frac": 0.6371527777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.481155778894475,
          "escalated_frac": 0.6423611111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.684673366834176,
          "escalated_frac": 0.6475694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 11.75,
          "escalated_frac": 0.6475694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 12.0,
          "escalated_frac": 0.65625,
          "residual_disagree": 0.0
        },
        {
          "tau": 12.25,
          "escalated_frac": 0.6614583333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 12.5,
          "escalated_frac": 0.6666666666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 12.7964824120603,
          "escalated_frac": 0.6736111111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.00942211055277,
          "escalated_frac": 0.6788194444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.370603015075375,
          "escalated_frac": 0.6822916666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.5,
          "escalated_frac": 0.6857638888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.625,
          "escalated_frac": 0.6892361111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.75,
          "escalated_frac": 0.6944444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.875,
          "escalated_frac": 0.6996527777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.875,
          "escalated_frac": 0.6996527777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 13.912688442211056,
          "escalated_frac": 0.7135416666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.148869346733669,
          "escalated_frac": 0.71875,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.375,
          "escalated_frac": 0.7222222222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.375,
          "escalated_frac": 0.7222222222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.375,
          "escalated_frac": 0.7222222222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.5,
          "escalated_frac": 0.734375,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.75,
          "escalated_frac": 0.7413194444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 14.875,
          "escalated_frac": 0.7447916666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.0,
          "escalated_frac": 0.7517361111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.144158291457288,
          "escalated_frac": 0.7586805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.375,
          "escalated_frac": 0.7621527777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.3963567839196,
          "escalated_frac": 0.7690972222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.746859296482413,
          "escalated_frac": 0.7725694444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 15.983040201005025,
          "escalated_frac": 0.7777777777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.0,
          "escalated_frac": 0.7777777777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.08040201005025,
          "escalated_frac": 0.7881944444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.25,
          "escalated_frac": 0.7916666666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.427763819095482,
          "escalated_frac": 0.7986111111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.625,
          "escalated_frac": 0.8003472222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.75,
          "escalated_frac": 0.8055555555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 16.886306532663312,
          "escalated_frac": 0.8142361111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.0,
          "escalated_frac": 0.8142361111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.125,
          "escalated_frac": 0.8211805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.219849246231156,
          "escalated_frac": 0.828125,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.5,
          "escalated_frac": 0.8298611111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.625,
          "escalated_frac": 0.8368055555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.625,
          "escalated_frac": 0.8368055555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 17.875,
          "escalated_frac": 0.8472222222222222,
          "residual_disagree": 0.0
        },
        {
          "tau": 18.05150753768845,
          "escalated_frac": 0.8541666666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 18.375,
          "escalated_frac": 0.8559027777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 18.5,
          "escalated_frac": 0.8611111111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 18.75,
          "escalated_frac": 0.8663194444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 18.970477386934675,
          "escalated_frac": 0.8732638888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.0,
          "escalated_frac": 0.8732638888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.125,
          "escalated_frac": 0.8802083333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.179020100502512,
          "escalated_frac": 0.8888888888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.375,
          "escalated_frac": 0.8923611111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.526381909547737,
          "escalated_frac": 0.8993055555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 19.875,
          "escalated_frac": 0.9027777777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.123743718592962,
          "escalated_frac": 0.9079861111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.234924623115575,
          "escalated_frac": 0.9131944444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.25,
          "escalated_frac": 0.9131944444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.5822864321608,
          "escalated_frac": 0.9236111111111112,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.693467336683426,
          "escalated_frac": 0.9288194444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 20.80464824120604,
          "escalated_frac": 0.9340277777777778,
          "residual_disagree": 0.0
        },
        {
          "tau": 21.0,
          "escalated_frac": 0.9375,
          "residual_disagree": 0.0
        },
        {
          "tau": 21.081030150753747,
          "escalated_frac": 0.9444444444444444,
          "residual_disagree": 0.0
        },
        {
          "tau": 21.625,
          "escalated_frac": 0.9461805555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 21.75,
          "escalated_frac": 0.9513888888888888,
          "residual_disagree": 0.0
        },
        {
          "tau": 22.1105527638191,
          "escalated_frac": 0.9583333333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 22.25,
          "escalated_frac": 0.9618055555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 23.123743718592976,
          "escalated_frac": 0.96875,
          "residual_disagree": 0.0
        },
        {
          "tau": 23.638190954773876,
          "escalated_frac": 0.9739583333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 24.858040201005053,
          "escalated_frac": 0.9791666666666666,
          "residual_disagree": 0.0
        },
        {
          "tau": 27.289572864321627,
          "escalated_frac": 0.984375,
          "residual_disagree": 0.0
        },
        {
          "tau": 29.402638190954775,
          "escalated_frac": 0.9895833333333334,
          "residual_disagree": 0.0
        },
        {
          "tau": 31.5,
          "escalated_frac": 0.9930555555555556,
          "residual_disagree": 0.0
        },
        {
          "tau": 31.75,
          "escalated_frac": 0.9982638888888888,
          "residual_disagree": 0.0
        }
      ],
      "per_domain": {
        "code": {
          "agreement_rate": 0.90625,
          "auroc": 0.9086845466155811
        },
        "math": {
          "agreement_rate": 0.9895833333333334,
          "auroc": 0.8842105263157894
        },
        "chat": {
          "agreement_rate": 0.9375,
          "auroc": 0.9462962962962963
        },
        "french": {
          "agreement_rate": 0.8854166666666666,
          "auroc": 0.9368983957219251
        },
        "english": {
          "agreement_rate": 0.8958333333333334,
          "auroc": 0.986046511627907
        },
        "reasoning": {
          "agreement_rate": 0.9270833333333334,
          "auroc": 0.9630818619582665
        }
      },
      "escalation_frac_for_99pct": 0.22743055555555555
    }
  }
}