{
  "capid_evaluation": {
    "categories": [
      {
        "f1": 0.9771689497716894,
        "false_negative": 1,
        "false_positive": 4,
        "gold_support": 108,
        "label_id": "health",
        "precision": 0.963963963963964,
        "predicted": 111,
        "recall": 0.9907407407407407,
        "true_positive": 107
      },
      {
        "f1": 0.935064935064935,
        "false_negative": 3,
        "false_positive": 7,
        "gold_support": 75,
        "label_id": "location",
        "precision": 0.9113924050632911,
        "predicted": 79,
        "recall": 0.96,
        "true_positive": 72
      },
      {
        "f1": 1.0,
        "false_negative": 0,
        "false_positive": 0,
        "gold_support": 92,
        "label_id": "sexual orientation",
        "precision": 1.0,
        "predicted": 92,
        "recall": 1.0,
        "true_positive": 92
      },
      {
        "f1": 0.9081632653061223,
        "false_negative": 12,
        "false_positive": 6,
        "gold_support": 101,
        "label_id": "occupation",
        "precision": 0.9368421052631579,
        "predicted": 95,
        "recall": 0.8811881188118812,
        "true_positive": 89
      },
      {
        "f1": 0.9947089947089947,
        "false_negative": 0,
        "false_positive": 1,
        "gold_support": 94,
        "label_id": "age",
        "precision": 0.9894736842105263,
        "predicted": 95,
        "recall": 1.0,
        "true_positive": 94
      },
      {
        "f1": 1.0,
        "false_negative": 0,
        "false_positive": 0,
        "gold_support": 65,
        "label_id": "belief",
        "precision": 1.0,
        "predicted": 65,
        "recall": 1.0,
        "true_positive": 65
      },
      {
        "f1": 0.9577464788732394,
        "false_negative": 3,
        "false_positive": 3,
        "gold_support": 71,
        "label_id": "relationship",
        "precision": 0.9577464788732394,
        "predicted": 71,
        "recall": 0.9577464788732394,
        "true_positive": 68
      },
      {
        "f1": 0.9047619047619048,
        "false_negative": 6,
        "false_positive": 2,
        "gold_support": 44,
        "label_id": "name",
        "precision": 0.95,
        "predicted": 40,
        "recall": 0.8636363636363636,
        "true_positive": 38
      },
      {
        "f1": 0.9820359281437125,
        "false_negative": 2,
        "false_positive": 1,
        "gold_support": 84,
        "label_id": "education",
        "precision": 0.9879518072289156,
        "predicted": 83,
        "recall": 0.9761904761904762,
        "true_positive": 82
      },
      {
        "f1": 0.9702970297029702,
        "false_negative": 2,
        "false_positive": 1,
        "gold_support": 51,
        "label_id": "appearance",
        "precision": 0.98,
        "predicted": 50,
        "recall": 0.9607843137254902,
        "true_positive": 49
      },
      {
        "f1": 1.0,
        "false_negative": 0,
        "false_positive": 0,
        "gold_support": 33,
        "label_id": "code",
        "precision": 1.0,
        "predicted": 33,
        "recall": 1.0,
        "true_positive": 33
      },
      {
        "f1": 0.9305555555555556,
        "false_negative": 5,
        "false_positive": 5,
        "gold_support": 72,
        "label_id": "organization",
        "precision": 0.9305555555555556,
        "predicted": 72,
        "recall": 0.9305555555555556,
        "true_positive": 67
      },
      {
        "f1": 0.9805825242718447,
        "false_negative": 4,
        "false_positive": 0,
        "gold_support": 105,
        "label_id": "finance",
        "precision": 1.0,
        "predicted": 101,
        "recall": 0.9619047619047619,
        "true_positive": 101
      },
      {
        "f1": 0.9836065573770492,
        "false_negative": 2,
        "false_positive": 0,
        "gold_support": 62,
        "label_id": "datetime",
        "precision": 1.0,
        "predicted": 60,
        "recall": 0.967741935483871,
        "true_positive": 60
      },
      {
        "f1": 0.9651741293532337,
        "false_negative": 5,
        "false_positive": 2,
        "gold_support": 102,
        "label_id": "demographic",
        "precision": 0.9797979797979798,
        "predicted": 99,
        "recall": 0.9509803921568627,
        "true_positive": 97
      }
    ],
    "category_axis": "model_native_label",
    "dataset_id": "capid-test",
    "documents": 200,
    "matching": {
      "character_offsets_used": false,
      "name": "one_to_one_soft_surface_plus_exact_type",
      "surface_similarity": "token F1; single-token strings use character-multiset F1",
      "threshold_rule": "similarity > 0.2"
    },
    "model_id": "capid-reproduced-seed3407",
    "overall": {
      "f1": 0.9665943600867679,
      "false_negative": 45,
      "false_positive": 32,
      "gold_support": 1159,
      "precision": 0.9720767888307156,
      "predicted": 1146,
      "recall": 0.9611734253666955,
      "true_positive": 1114
    },
    "sources": {
      "frozen_overall": {
        "path": "outputs/evidence/lab2_full_seed3407/test_metrics_corrected_v4.json",
        "sha256": "7e25c8f4b5f76b7f38fb9e49f1872e2f0d1a9c7a9a3ecc8e66b6845cade6281c"
      },
      "predictions": {
        "path": "work/lab2_full_v2_20260731/test_predictions.jsonl",
        "sha256": "f629ecdbc87f440096065bcec3f02dab7bb44708e5847963c11a8ac37d364fb1"
      }
    },
    "status": "measured"
  },
  "datasets": [
    {
      "dataset_id": "pii-bench-zh-formal",
      "en": "PII Bench ZH formal",
      "zh": "PII Bench ZH 正式文本"
    },
    {
      "dataset_id": "pii-bench-zh-chat",
      "en": "PII Bench ZH chat",
      "zh": "PII Bench ZH 对话文本"
    },
    {
      "dataset_id": "nemotron-pii",
      "en": "NVIDIA Nemotron-PII test",
      "zh": "NVIDIA Nemotron-PII test"
    },
    {
      "dataset_id": "tab",
      "en": "TAB (ECHR legal)",
      "zh": "TAB（ECHR 法律）"
    },
    {
      "dataset_id": "meddocan",
      "en": "MEDDOCAN (medical)",
      "zh": "MEDDOCAN（医疗）"
    },
    {
      "dataset_id": "piilo",
      "en": "PIILO (education)",
      "zh": "PIILO（教育）"
    },
    {
      "dataset_id": "tw-short",
      "en": "Traditional Chinese short",
      "zh": "繁体中文短文本"
    },
    {
      "dataset_id": "tw-mid",
      "en": "Traditional Chinese medium",
      "zh": "繁体中文中等文本"
    },
    {
      "dataset_id": "tw-long",
      "en": "Traditional Chinese long",
      "zh": "繁体中文长文本"
    },
    {
      "dataset_id": "capid-test",
      "en": "CAPID official test (200 records)",
      "zh": "CAPID 官方 test（200 条）"
    }
  ],
  "frozen_at": "2026-08-14",
  "metric_definitions": {
    "capid_typed_soft_f1": "One-to-one soft surface matching at similarity > 0.2 plus exact normalized CAPID type; this is not a character-offset metric.",
    "static_category_axis": "Benchmark gold label. Predicted model labels are ignored; category rows therefore report gold-conditioned recall/leakage/coverage, not native-label F1.",
    "static_character_coverage": "Coverage definition is recorded per row as any_overlap or exact_matched_spans.",
    "static_exact_span_leakage": "1 - static exact-span recall",
    "static_exact_span_recall": "exact boundary TP / gold spans for this benchmark label"
  },
  "models": [
    {
      "label": "OpenAI Privacy Filter",
      "model_id": "openai-privacy-filter",
      "native_label_count": 8,
      "native_typed_prf_status": "not_measured"
    },
    {
      "label": "OpenMed Privacy Filter Nemotron v2",
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "native_label_count": 55,
      "native_typed_prf_status": "not_measured"
    },
    {
      "label": "OpenMed Privacy Filter Multilingual v2",
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "native_label_count": 54,
      "native_typed_prf_status": "not_measured"
    },
    {
      "label": "CAPID reproduced LoRA (seed 3407)",
      "model_id": "capid-reproduced-seed3407",
      "native_label_count": 15,
      "native_typed_prf_status": "measured"
    }
  ],
  "schema": "model-output-category-metrics/v1",
  "sources": {
    "taxonomy": {
      "path": "capid-reproduction-api/web/assets/model-output-taxonomy.json",
      "sha256": "c05447dd95d7276d8ba63d32ecd379396ca7a2bf728db8459860ff1d1adf3d62"
    }
  },
  "static_evaluations": [
    {
      "categories": [
        {
          "character_coverage": 0.782212538377736,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7898,
          "exact_span_leakage": 0.9636803874092009,
          "exact_span_recall": 0.03631961259079903,
          "false_negative": 398,
          "gold_characters": 10097,
          "gold_spans": 413,
          "label_id": "CALLE",
          "residual_pii_character_rate": 0.21778746162226403,
          "true_positive": 15
        },
        {
          "character_coverage": 0.3273809523809524,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 55,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 6,
          "gold_characters": 168,
          "gold_spans": 6,
          "label_id": "CENTRO_SALUD",
          "residual_pii_character_rate": 0.6726190476190477,
          "true_positive": 0
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5464,
          "exact_span_leakage": 0.02008032128514059,
          "exact_span_recall": 0.9799196787148594,
          "false_negative": 5,
          "gold_characters": 5464,
          "gold_spans": 249,
          "label_id": "CORREO_ELECTRONICO",
          "residual_pii_character_rate": 0.0,
          "true_positive": 244
        },
        {
          "character_coverage": 0.0005649717514124294,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 518,
          "gold_characters": 3540,
          "gold_spans": 518,
          "label_id": "EDAD_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.9994350282485875,
          "true_positive": 0
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 81,
          "gold_characters": 731,
          "gold_spans": 81,
          "label_id": "FAMILIARES_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8711315664180289,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5489,
          "exact_span_leakage": 0.1260229132569558,
          "exact_span_recall": 0.8739770867430442,
          "false_negative": 77,
          "gold_characters": 6301,
          "gold_spans": 611,
          "label_id": "FECHAS",
          "residual_pii_character_rate": 0.1288684335819711,
          "true_positive": 534
        },
        {
          "character_coverage": 0.29146619397315776,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1151,
          "exact_span_leakage": 0.9923076923076923,
          "exact_span_recall": 0.007692307692307693,
          "false_negative": 129,
          "gold_characters": 3949,
          "gold_spans": 130,
          "label_id": "HOSPITAL",
          "residual_pii_character_rate": 0.7085338060268422,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9989130434782608,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2757,
          "exact_span_leakage": 0.0505050505050505,
          "exact_span_recall": 0.9494949494949495,
          "false_negative": 10,
          "gold_characters": 2760,
          "gold_spans": 198,
          "label_id": "ID_ASEGURAMIENTO",
          "residual_pii_character_rate": 0.0010869565217391797,
          "true_positive": 188
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 304,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 304,
          "gold_spans": 39,
          "label_id": "ID_CONTACTO_ASISTENCIAL",
          "residual_pii_character_rate": 0.0,
          "true_positive": 39
        },
        {
          "character_coverage": 0.876008064516129,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1738,
          "exact_span_leakage": 0.18021201413427557,
          "exact_span_recall": 0.8197879858657244,
          "false_negative": 51,
          "gold_characters": 1984,
          "gold_spans": 283,
          "label_id": "ID_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.123991935483871,
          "true_positive": 232
        },
        {
          "character_coverage": 0.9656981436642453,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2393,
          "exact_span_leakage": 0.05128205128205132,
          "exact_span_recall": 0.9487179487179487,
          "false_negative": 12,
          "gold_characters": 2478,
          "gold_spans": 234,
          "label_id": "ID_TITULACION_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.034301856335754666,
          "true_positive": 222
        },
        {
          "character_coverage": 0.0766016713091922,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 110,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 67,
          "gold_characters": 1436,
          "gold_spans": 67,
          "label_id": "INSTITUCION",
          "residual_pii_character_rate": 0.9233983286908078,
          "true_positive": 0
        },
        {
          "character_coverage": 0.850748995250274,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 9314,
          "exact_span_leakage": 0.5229540918163673,
          "exact_span_recall": 0.47704590818363274,
          "false_negative": 262,
          "gold_characters": 10948,
          "gold_spans": 501,
          "label_id": "NOMBRE_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.14925100474972597,
          "true_positive": 239
        },
        {
          "character_coverage": 0.9854958839670718,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5028,
          "exact_span_leakage": 0.13346613545816732,
          "exact_span_recall": 0.8665338645418327,
          "false_negative": 67,
          "gold_characters": 5102,
          "gold_spans": 502,
          "label_id": "NOMBRE_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.014504116032928227,
          "true_positive": 435
        },
        {
          "character_coverage": 0.7307692307692307,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 57,
          "exact_span_leakage": 0.5714285714285714,
          "exact_span_recall": 0.42857142857142855,
          "false_negative": 4,
          "gold_characters": 78,
          "gold_spans": 7,
          "label_id": "NUMERO_FAX",
          "residual_pii_character_rate": 0.2692307692307693,
          "true_positive": 3
        },
        {
          "character_coverage": 0.9357142857142857,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 262,
          "exact_span_leakage": 0.15384615384615385,
          "exact_span_recall": 0.8461538461538461,
          "false_negative": 4,
          "gold_characters": 280,
          "gold_spans": 26,
          "label_id": "NUMERO_TELEFONO",
          "residual_pii_character_rate": 0.06428571428571428,
          "true_positive": 22
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 7,
          "gold_characters": 86,
          "gold_spans": 7,
          "label_id": "OTROS_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.048813255709807435,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 109,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 363,
          "gold_characters": 2233,
          "gold_spans": 363,
          "label_id": "PAIS",
          "residual_pii_character_rate": 0.9511867442901926,
          "true_positive": 0
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 9,
          "gold_characters": 160,
          "gold_spans": 9,
          "label_id": "PROFESION",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 461,
          "gold_characters": 1425,
          "gold_spans": 461,
          "label_id": "SEXO_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.5062019155283404,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3224,
          "exact_span_leakage": 0.7071129707112971,
          "exact_span_recall": 0.2928870292887029,
          "false_negative": 676,
          "gold_characters": 6369,
          "gold_spans": 956,
          "label_id": "TERRITORIO",
          "residual_pii_character_rate": 0.4937980844716596,
          "true_positive": 280
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "meddocan",
      "dataset_sha256": "68c972fd2fd5ffa202072c6d048c111af2a31622e1845ff999c83fe315558c2a",
      "documents": 250,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.5514606741573034,
        "false_negative": 3207,
        "false_positive": 785,
        "precision": 0.757641247298549,
        "recall": 0.43349231584525705,
        "true_positive": 2454
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-meddocan.json",
        "sha256": "79447898319fa86a477ec537240f1f5af0451f403be56ae4b1b2db9d1972fa25"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.9614833783847868,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 202748,
          "exact_span_leakage": 0.040076678847421054,
          "exact_span_recall": 0.959923321152579,
          "false_negative": 669,
          "gold_characters": 210870,
          "gold_spans": 16693,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.03851662161521319,
          "true_positive": 16024
        },
        {
          "character_coverage": 0.0052924791086350976,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 76,
          "exact_span_leakage": 0.99693037533138,
          "exact_span_recall": 0.0030696246686200642,
          "false_negative": 7145,
          "gold_characters": 14360,
          "gold_spans": 7167,
          "label_id": "age",
          "residual_pii_character_rate": 0.9947075208913649,
          "true_positive": 22
        },
        {
          "character_coverage": 0.9962389549055454,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 209258,
          "exact_span_leakage": 0.06449539318620101,
          "exact_span_recall": 0.935504606813799,
          "false_negative": 301,
          "gold_characters": 210048,
          "gold_spans": 4667,
          "label_id": "api_key",
          "residual_pii_character_rate": 0.003761045094454629,
          "true_positive": 4366
        },
        {
          "character_coverage": 0.8398137059214904,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 63112,
          "exact_span_leakage": 0.1608810150825951,
          "exact_span_recall": 0.8391189849174049,
          "false_negative": 1344,
          "gold_characters": 75150,
          "gold_spans": 8354,
          "label_id": "bank_routing_number",
          "residual_pii_character_rate": 0.1601862940785096,
          "true_positive": 7010
        },
        {
          "character_coverage": 0.9303493035607826,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 137327,
          "exact_span_leakage": 0.07909306617453205,
          "exact_span_recall": 0.920906933825468,
          "false_negative": 900,
          "gold_characters": 147608,
          "gold_spans": 11379,
          "label_id": "biometric_identifier",
          "residual_pii_character_rate": 0.0696506964392174,
          "true_positive": 10479
        },
        {
          "character_coverage": 0.008638520501277528,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 213,
          "exact_span_leakage": 0.9967503159415056,
          "exact_span_recall": 0.003249684058494313,
          "false_negative": 5521,
          "gold_characters": 24657,
          "gold_spans": 5539,
          "label_id": "blood_type",
          "residual_pii_character_rate": 0.9913614794987224,
          "true_positive": 18
        },
        {
          "character_coverage": 0.8866281706321614,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 29327,
          "exact_span_leakage": 0.11059293804130577,
          "exact_span_recall": 0.8894070619586942,
          "false_negative": 332,
          "gold_characters": 33077,
          "gold_spans": 3002,
          "label_id": "certificate_license_number",
          "residual_pii_character_rate": 0.11337182936783863,
          "true_positive": 2670
        },
        {
          "character_coverage": 0.3925611858438887,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 58978,
          "exact_span_leakage": 0.825966097999673,
          "exact_span_recall": 0.17403390200032703,
          "false_negative": 15154,
          "gold_characters": 150239,
          "gold_spans": 18347,
          "label_id": "city",
          "residual_pii_character_rate": 0.6074388141561113,
          "true_positive": 3193
        },
        {
          "character_coverage": 0.02014335374518721,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 20681,
          "exact_span_leakage": 0.9770957565147619,
          "exact_span_recall": 0.02290424348523807,
          "false_negative": 53581,
          "gold_characters": 1026691,
          "gold_spans": 54837,
          "label_id": "company_name",
          "residual_pii_character_rate": 0.9798566462548128,
          "true_positive": 1256
        },
        {
          "character_coverage": 0.6191471106424506,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 94532,
          "exact_span_leakage": 0.46598772685729206,
          "exact_span_recall": 0.5340122731427079,
          "false_negative": 3569,
          "gold_characters": 152681,
          "gold_spans": 7659,
          "label_id": "coordinate",
          "residual_pii_character_rate": 0.38085288935754935,
          "true_positive": 4090
        },
        {
          "character_coverage": 0.060395456699358134,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 8412,
          "exact_span_leakage": 0.9981256656017039,
          "exact_span_recall": 0.0018743343982960596,
          "false_negative": 23431,
          "gold_characters": 139282,
          "gold_spans": 23475,
          "label_id": "country",
          "residual_pii_character_rate": 0.9396045433006419,
          "true_positive": 44
        },
        {
          "character_coverage": 0.11384460174649379,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 13767,
          "exact_span_leakage": 0.9604252805670408,
          "exact_span_recall": 0.039574719432959246,
          "false_negative": 8130,
          "gold_characters": 120928,
          "gold_spans": 8465,
          "label_id": "county",
          "residual_pii_character_rate": 0.8861553982535062,
          "true_positive": 335
        },
        {
          "character_coverage": 0.8672141398140815,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 208036,
          "exact_span_leakage": 0.15287168726198808,
          "exact_span_recall": 0.8471283127380119,
          "false_negative": 1967,
          "gold_characters": 239890,
          "gold_spans": 12867,
          "label_id": "credit_debit_card",
          "residual_pii_character_rate": 0.13278586018591854,
          "true_positive": 10900
        },
        {
          "character_coverage": 0.8670480923602376,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 169579,
          "exact_span_leakage": 0.14671739342503165,
          "exact_span_recall": 0.8532826065749683,
          "false_negative": 3008,
          "gold_characters": 195582,
          "gold_spans": 20502,
          "label_id": "customer_id",
          "residual_pii_character_rate": 0.1329519076397624,
          "true_positive": 17494
        },
        {
          "character_coverage": 0.3082799917406566,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4479,
          "exact_span_leakage": 0.7136072682221763,
          "exact_span_recall": 0.28639273177782365,
          "false_negative": 3456,
          "gold_characters": 14529,
          "gold_spans": 4843,
          "label_id": "cvv",
          "residual_pii_character_rate": 0.6917200082593433,
          "true_positive": 1387
        },
        {
          "character_coverage": 0.7413600150410752,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 579639,
          "exact_span_leakage": 0.28915483233378914,
          "exact_span_recall": 0.7108451676662109,
          "false_negative": 21359,
          "gold_characters": 781859,
          "gold_spans": 73867,
          "label_id": "date",
          "residual_pii_character_rate": 0.2586399849589248,
          "true_positive": 52508
        },
        {
          "character_coverage": 0.987708613374489,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 178555,
          "exact_span_leakage": 0.013053819348415319,
          "exact_span_recall": 0.9869461806515847,
          "false_negative": 236,
          "gold_characters": 180777,
          "gold_spans": 18079,
          "label_id": "date_of_birth",
          "residual_pii_character_rate": 0.012291386625510992,
          "true_positive": 17843
        },
        {
          "character_coverage": 0.8937798012981006,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 224185,
          "exact_span_leakage": 0.1170108161258604,
          "exact_span_recall": 0.8829891838741396,
          "false_negative": 1547,
          "gold_characters": 250828,
          "gold_spans": 13221,
          "label_id": "date_time",
          "residual_pii_character_rate": 0.10622019870189936,
          "true_positive": 11674
        },
        {
          "character_coverage": 0.936257530120482,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 49734,
          "exact_span_leakage": 0.08695652173913049,
          "exact_span_recall": 0.9130434782608695,
          "false_negative": 218,
          "gold_characters": 53120,
          "gold_spans": 2507,
          "label_id": "device_identifier",
          "residual_pii_character_rate": 0.06374246987951804,
          "true_positive": 2289
        },
        {
          "character_coverage": 0.00045184235074626866,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 62,
          "exact_span_leakage": 0.9998933674557475,
          "exact_span_recall": 0.00010663254425250586,
          "false_negative": 9377,
          "gold_characters": 137216,
          "gold_spans": 9378,
          "label_id": "education_level",
          "residual_pii_character_rate": 0.9995481576492538,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9705632556346975,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1359895,
          "exact_span_leakage": 0.04791396254403857,
          "exact_span_recall": 0.9520860374559614,
          "false_negative": 2584,
          "gold_characters": 1401140,
          "gold_spans": 53930,
          "label_id": "email",
          "residual_pii_character_rate": 0.029436744365302525,
          "true_positive": 51346
        },
        {
          "character_coverage": 0.6577695271672153,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 42165,
          "exact_span_leakage": 0.3516557783284524,
          "exact_span_recall": 0.6483442216715476,
          "false_negative": 3122,
          "gold_characters": 64103,
          "gold_spans": 8878,
          "label_id": "employee_id",
          "residual_pii_character_rate": 0.34223047283278474,
          "true_positive": 5756
        },
        {
          "character_coverage": 0.0009826718946108716,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 101,
          "exact_span_leakage": 0.9992739154111454,
          "exact_span_recall": 0.0007260845888546016,
          "false_negative": 11010,
          "gold_characters": 102781,
          "gold_spans": 11018,
          "label_id": "employment_status",
          "residual_pii_character_rate": 0.9990173281053891,
          "true_positive": 8
        },
        {
          "character_coverage": 0.6428221059516024,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 55041,
          "exact_span_leakage": 0.3678966215164253,
          "exact_span_recall": 0.6321033784835747,
          "false_negative": 2363,
          "gold_characters": 85624,
          "gold_spans": 6423,
          "label_id": "fax_number",
          "residual_pii_character_rate": 0.3571778940483976,
          "true_positive": 4060
        },
        {
          "character_coverage": 0.8975699478860286,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 433336,
          "exact_span_leakage": 0.6358173791987435,
          "exact_span_recall": 0.3641826208012565,
          "false_negative": 53436,
          "gold_characters": 482788,
          "gold_spans": 84043,
          "label_id": "first_name",
          "residual_pii_character_rate": 0.10243005211397138,
          "true_positive": 30607
        },
        {
          "character_coverage": 0.002142962707522538,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 87,
          "exact_span_leakage": 0.9988227599738391,
          "exact_span_recall": 0.0011772400261608894,
          "false_negative": 7636,
          "gold_characters": 40598,
          "gold_spans": 7645,
          "label_id": "gender",
          "residual_pii_character_rate": 0.9978570372924774,
          "true_positive": 9
        },
        {
          "character_coverage": 0.9350812471290767,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 130283,
          "exact_span_leakage": 0.06923659784050007,
          "exact_span_recall": 0.9307634021594999,
          "false_negative": 731,
          "gold_characters": 139328,
          "gold_spans": 10558,
          "label_id": "health_plan_beneficiary_number",
          "residual_pii_character_rate": 0.0649187528709233,
          "true_positive": 9827
        },
        {
          "character_coverage": 0.6016185765918868,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 248516,
          "exact_span_leakage": 0.830539656582175,
          "exact_span_recall": 0.169460343417825,
          "false_negative": 4063,
          "gold_characters": 413079,
          "gold_spans": 4892,
          "label_id": "http_cookie",
          "residual_pii_character_rate": 0.3983814234081132,
          "true_positive": 829
        },
        {
          "character_coverage": 0.8134747271075264,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 65357,
          "exact_span_leakage": 0.19233300427772293,
          "exact_span_recall": 0.8076669957222771,
          "false_negative": 1169,
          "gold_characters": 80343,
          "gold_spans": 6078,
          "label_id": "ipv4",
          "residual_pii_character_rate": 0.18652527289247356,
          "true_positive": 4909
        },
        {
          "character_coverage": 0.9656230270587502,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 108593,
          "exact_span_leakage": 0.06148454921949664,
          "exact_span_recall": 0.9385154507805034,
          "false_negative": 193,
          "gold_characters": 112459,
          "gold_spans": 3139,
          "label_id": "ipv6",
          "residual_pii_character_rate": 0.03437697294124975,
          "true_positive": 2946
        },
        {
          "character_coverage": 0.010359833723044947,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 638,
          "exact_span_leakage": 0.9889030169922552,
          "exact_span_recall": 0.01109698300774477,
          "false_negative": 8555,
          "gold_characters": 61584,
          "gold_spans": 8651,
          "label_id": "language",
          "residual_pii_character_rate": 0.9896401662769551,
          "true_positive": 96
        },
        {
          "character_coverage": 0.9458768675501305,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 366375,
          "exact_span_leakage": 0.8180750385931942,
          "exact_span_recall": 0.18192496140680584,
          "false_negative": 48754,
          "gold_characters": 387339,
          "gold_spans": 59596,
          "label_id": "last_name",
          "residual_pii_character_rate": 0.05412313244986955,
          "true_positive": 10842
        },
        {
          "character_coverage": 0.9523969363156449,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 31211,
          "exact_span_leakage": 0.05689063266307015,
          "exact_span_recall": 0.9431093673369298,
          "false_negative": 232,
          "gold_characters": 32771,
          "gold_spans": 4078,
          "label_id": "license_plate",
          "residual_pii_character_rate": 0.047603063684355074,
          "true_positive": 3846
        },
        {
          "character_coverage": 0.922498858242108,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 66657,
          "exact_span_leakage": 0.12682352941176467,
          "exact_span_recall": 0.8731764705882353,
          "false_negative": 539,
          "gold_characters": 72257,
          "gold_spans": 4250,
          "label_id": "mac_address",
          "residual_pii_character_rate": 0.07750114175789202,
          "true_positive": 3711
        },
        {
          "character_coverage": 0.8912612227447627,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 104233,
          "exact_span_leakage": 0.10848801585871326,
          "exact_span_recall": 0.8915119841412867,
          "false_negative": 1204,
          "gold_characters": 116950,
          "gold_spans": 11098,
          "label_id": "medical_record_number",
          "residual_pii_character_rate": 0.10873877725523728,
          "true_positive": 9894
        },
        {
          "character_coverage": 0.8420412158408508,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 31830,
          "exact_span_leakage": 0.18370214260625217,
          "exact_span_recall": 0.8162978573937478,
          "false_negative": 523,
          "gold_characters": 37801,
          "gold_spans": 2847,
          "label_id": "national_id",
          "residual_pii_character_rate": 0.1579587841591492,
          "true_positive": 2324
        },
        {
          "character_coverage": 0.004412069695359798,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3106,
          "exact_span_leakage": 0.9963402732595966,
          "exact_span_recall": 0.003659726740403383,
          "false_negative": 36753,
          "gold_characters": 703978,
          "gold_spans": 36888,
          "label_id": "occupation",
          "residual_pii_character_rate": 0.9955879303046402,
          "true_positive": 135
        },
        {
          "character_coverage": 0.8874128353622504,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 75084,
          "exact_span_leakage": 0.22576950608446666,
          "exact_span_recall": 0.7742304939155333,
          "false_negative": 1577,
          "gold_characters": 84610,
          "gold_spans": 6985,
          "label_id": "password",
          "residual_pii_character_rate": 0.11258716463774965,
          "true_positive": 5408
        },
        {
          "character_coverage": 0.8920589757696703,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 274020,
          "exact_span_leakage": 0.11249477643125783,
          "exact_span_recall": 0.8875052235687422,
          "false_negative": 2692,
          "gold_characters": 307177,
          "gold_spans": 23930,
          "label_id": "phone_number",
          "residual_pii_character_rate": 0.1079410242303297,
          "true_positive": 21238
        },
        {
          "character_coverage": 0.6142576480493965,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 21886,
          "exact_span_leakage": 0.4753717472118959,
          "exact_span_recall": 0.5246282527881041,
          "false_negative": 3069,
          "gold_characters": 35630,
          "gold_spans": 6456,
          "label_id": "pin",
          "residual_pii_character_rate": 0.38574235195060347,
          "true_positive": 3387
        },
        {
          "character_coverage": 0.033013252916435205,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2912,
          "exact_span_leakage": 0.9643861293345829,
          "exact_span_recall": 0.03561387066541706,
          "false_negative": 6174,
          "gold_characters": 88207,
          "gold_spans": 6402,
          "label_id": "political_view",
          "residual_pii_character_rate": 0.9669867470835648,
          "true_positive": 228
        },
        {
          "character_coverage": 0.6930270813659488,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 22750,
          "exact_span_leakage": 0.5918139831183309,
          "exact_span_recall": 0.40818601688166906,
          "false_negative": 3716,
          "gold_characters": 32827,
          "gold_spans": 6279,
          "label_id": "postcode",
          "residual_pii_character_rate": 0.3069729186340512,
          "true_positive": 2563
        },
        {
          "character_coverage": 0.0452522838409409,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2878,
          "exact_span_leakage": 0.9630190311418685,
          "exact_span_recall": 0.036980968858131485,
          "false_negative": 8906,
          "gold_characters": 63599,
          "gold_spans": 9248,
          "label_id": "race_ethnicity",
          "residual_pii_character_rate": 0.9547477161590591,
          "true_positive": 342
        },
        {
          "character_coverage": 0.019425212279555844,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1487,
          "exact_span_leakage": 0.9832187949496564,
          "exact_span_recall": 0.016781205050343615,
          "false_negative": 6152,
          "gold_characters": 76550,
          "gold_spans": 6257,
          "label_id": "religious_belief",
          "residual_pii_character_rate": 0.9805747877204442,
          "true_positive": 105
        },
        {
          "character_coverage": 0.0025315405046480743,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 61,
          "exact_span_leakage": 0.9969806763285024,
          "exact_span_recall": 0.0030193236714975845,
          "false_negative": 3302,
          "gold_characters": 24096,
          "gold_spans": 3312,
          "label_id": "sexuality",
          "residual_pii_character_rate": 0.9974684594953519,
          "true_positive": 10
        },
        {
          "character_coverage": 0.9661746617466175,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 64411,
          "exact_span_leakage": 0.047674034971956436,
          "exact_span_recall": 0.9523259650280436,
          "false_negative": 289,
          "gold_characters": 66666,
          "gold_spans": 6062,
          "label_id": "ssn",
          "residual_pii_character_rate": 0.03382533825338252,
          "true_positive": 5773
        },
        {
          "character_coverage": 0.20854648467274883,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 26539,
          "exact_span_leakage": 0.9657713724205699,
          "exact_span_recall": 0.03422862757943007,
          "false_negative": 17691,
          "gold_characters": 127257,
          "gold_spans": 18318,
          "label_id": "state",
          "residual_pii_character_rate": 0.7914535153272512,
          "true_positive": 627
        },
        {
          "character_coverage": 0.8264670670731094,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 246450,
          "exact_span_leakage": 0.5087979145683986,
          "exact_span_recall": 0.4912020854316014,
          "false_negative": 8588,
          "gold_characters": 298197,
          "gold_spans": 16879,
          "label_id": "street_address",
          "residual_pii_character_rate": 0.17353293292689065,
          "true_positive": 8291
        },
        {
          "character_coverage": 0.6530484291862948,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 38539,
          "exact_span_leakage": 0.5013491635186185,
          "exact_span_recall": 0.49865083648138153,
          "false_negative": 2787,
          "gold_characters": 59014,
          "gold_spans": 5559,
          "label_id": "swift_bic",
          "residual_pii_character_rate": 0.3469515708137052,
          "true_positive": 2772
        },
        {
          "character_coverage": 0.9015536917365599,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 13114,
          "exact_span_leakage": 0.10675883256528418,
          "exact_span_recall": 0.8932411674347158,
          "false_negative": 139,
          "gold_characters": 14546,
          "gold_spans": 1302,
          "label_id": "tax_id",
          "residual_pii_character_rate": 0.09844630826344014,
          "true_positive": 1163
        },
        {
          "character_coverage": 0.10067915803670344,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 20902,
          "exact_span_leakage": 0.9638455888353873,
          "exact_span_recall": 0.036154411164612744,
          "false_negative": 23620,
          "gold_characters": 207610,
          "gold_spans": 24506,
          "label_id": "time",
          "residual_pii_character_rate": 0.8993208419632965,
          "true_positive": 886
        },
        {
          "character_coverage": 0.9405164872779828,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 49458,
          "exact_span_leakage": 0.08047270680922902,
          "exact_span_recall": 0.919527293190771,
          "false_negative": 143,
          "gold_characters": 52586,
          "gold_spans": 1777,
          "label_id": "unique_id",
          "residual_pii_character_rate": 0.05948351272201724,
          "true_positive": 1634
        },
        {
          "character_coverage": 0.21625585530980246,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 426718,
          "exact_span_leakage": 0.8298147805638492,
          "exact_span_recall": 0.1701852194361508,
          "false_negative": 31406,
          "gold_characters": 1973209,
          "gold_spans": 37847,
          "label_id": "url",
          "residual_pii_character_rate": 0.7837441446901976,
          "true_positive": 6441
        },
        {
          "character_coverage": 0.9245596598752139,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 167447,
          "exact_span_leakage": 0.11822784810126585,
          "exact_span_recall": 0.8817721518987341,
          "false_negative": 1868,
          "gold_characters": 181110,
          "gold_spans": 15800,
          "label_id": "user_name",
          "residual_pii_character_rate": 0.07544034012478606,
          "true_positive": 13932
        },
        {
          "character_coverage": 0.9727495126833161,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 74356,
          "exact_span_leakage": 0.036176277132207835,
          "exact_span_recall": 0.9638237228677922,
          "false_negative": 165,
          "gold_characters": 76439,
          "gold_spans": 4561,
          "label_id": "vehicle_identifier",
          "residual_pii_character_rate": 0.02725048731668389,
          "true_positive": 4396
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "nemotron-pii",
      "dataset_sha256": "1a4b0512ecb5370f0992d29d0f9c07351e6de13f0d7ea33bb18cecb984780247",
      "documents": 100000,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.5868738144260561,
        "false_negative": 466396,
        "false_positive": 74154,
        "precision": 0.8381263397788247,
        "recall": 0.4515182162429146,
        "true_positive": 383944
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/nemotron-domain-slices-v1/openai-privacy-filter.json",
        "sha256": "7aa8065f23278d8f97385e0b89102b691720eab973598de600ca34adc23a7cdd"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.2775616083009079,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 642,
          "exact_span_leakage": 0.7294117647058824,
          "exact_span_recall": 0.27058823529411763,
          "false_negative": 186,
          "gold_characters": 2313,
          "gold_spans": 255,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 69
        },
        {
          "character_coverage": 0.3968253968253968,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 3200,
          "exact_span_leakage": 0.6031746031746033,
          "exact_span_recall": 0.3968253968253968,
          "false_negative": 304,
          "gold_characters": 8064,
          "gold_spans": 504,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 200
        },
        {
          "character_coverage": 0.7621834687052078,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 6381,
          "exact_span_leakage": 0.24137931034482762,
          "exact_span_recall": 0.7586206896551724,
          "false_negative": 119,
          "gold_characters": 8372,
          "gold_spans": 493,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 374
        },
        {
          "character_coverage": 0.3707347707766381,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 3542,
          "exact_span_leakage": 0.6340996168582376,
          "exact_span_recall": 0.36590038314176243,
          "false_negative": 331,
          "gold_characters": 9554,
          "gold_spans": 522,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 191
        },
        {
          "character_coverage": 0.3138075313807531,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 525,
          "exact_span_leakage": 0.6861924686192469,
          "exact_span_recall": 0.3138075313807531,
          "false_negative": 164,
          "gold_characters": 1673,
          "gold_spans": 239,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 75
        },
        {
          "character_coverage": 0.3153846153846154,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 738,
          "exact_span_leakage": 0.6846153846153846,
          "exact_span_recall": 0.3153846153846154,
          "false_negative": 178,
          "gold_characters": 2340,
          "gold_spans": 260,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 82
        },
        {
          "character_coverage": 0.29361100398508805,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 2284,
          "exact_span_leakage": 0.7242524916943522,
          "exact_span_recall": 0.2757475083056478,
          "false_negative": 2180,
          "gold_characters": 7779,
          "gold_spans": 3010,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 830
        },
        {
          "character_coverage": 0.5205924997061013,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 13285,
          "exact_span_leakage": 0.47722246793454226,
          "exact_span_recall": 0.5227775320654577,
          "false_negative": 1079,
          "gold_characters": 25519,
          "gold_spans": 2261,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 1182
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-chat",
      "dataset_sha256": "4ddacd78dfac49b723032de5e1c29c432a44049245e03a8ea960c86bbf3b765c",
      "documents": 3000,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.4935491823485908,
        "false_negative": 4541,
        "false_positive": 1622,
        "precision": 0.6492972972972973,
        "recall": 0.3980646871686108,
        "true_positive": 3003
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/openai-chat.json",
        "sha256": "4ba46e70234267957f498bf2cc326bd7fcd93a3bbb1d70cca892ffcf52b7b028"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.764881221208563,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 25404,
          "exact_span_leakage": 0.23512946116165145,
          "exact_span_recall": 0.7648705388383485,
          "false_negative": 336,
          "gold_characters": 33213,
          "gold_spans": 1429,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 1093
        },
        {
          "character_coverage": 0.7749419953596288,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 10688,
          "exact_span_leakage": 0.22505800464037118,
          "exact_span_recall": 0.7749419953596288,
          "false_negative": 194,
          "gold_characters": 13792,
          "gold_spans": 862,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 668
        },
        {
          "character_coverage": 0.9883068564341818,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 16735,
          "exact_span_leakage": 0.012889366272824887,
          "exact_span_recall": 0.9871106337271751,
          "false_negative": 12,
          "gold_characters": 16933,
          "gold_spans": 931,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 919
        },
        {
          "character_coverage": 0.23344370860927152,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 7614,
          "exact_span_leakage": 0.7665562913907285,
          "exact_span_recall": 0.23344370860927152,
          "false_negative": 1389,
          "gold_characters": 32616,
          "gold_spans": 1812,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 423
        },
        {
          "character_coverage": 0.4725274725274725,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1505,
          "exact_span_leakage": 0.5274725274725275,
          "exact_span_recall": 0.4725274725274725,
          "false_negative": 240,
          "gold_characters": 3185,
          "gold_spans": 455,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 215
        },
        {
          "character_coverage": 0.31101511879049676,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1296,
          "exact_span_leakage": 0.6889848812095032,
          "exact_span_recall": 0.31101511879049676,
          "false_negative": 319,
          "gold_characters": 4167,
          "gold_spans": 463,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 144
        },
        {
          "character_coverage": 0.628513030148186,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 8610,
          "exact_span_leakage": 0.3907721280602636,
          "exact_span_recall": 0.6092278719397364,
          "false_negative": 2075,
          "gold_characters": 13699,
          "gold_spans": 5310,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 3235
        },
        {
          "character_coverage": 0.8027272727272727,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 38852,
          "exact_span_leakage": 0.19727272727272727,
          "exact_span_recall": 0.8027272727272727,
          "false_negative": 868,
          "gold_characters": 48400,
          "gold_spans": 4400,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 3532
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-formal",
      "dataset_sha256": "bcaa54909fd69c6737b8e442c4d740ce1bf5b95b7d4af70b2f2692010e1ff7a5",
      "documents": 5000,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.6904255678174884,
        "false_negative": 5433,
        "false_positive": 3740,
        "precision": 0.7322642995203665,
        "recall": 0.6531094368535308,
        "true_positive": 10229
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/openai-formal.json",
        "sha256": "890383188d918b409a55102ae4dce03ed84be027ac379475d05be47bdf38b47e"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 761,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 761,
          "gold_spans": 39,
          "label_id": "EMAIL",
          "residual_pii_character_rate": 0.0,
          "true_positive": 39
        },
        {
          "character_coverage": 0.6288441145281018,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 593,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 39,
          "gold_characters": 943,
          "gold_spans": 78,
          "label_id": "ID_NUM",
          "residual_pii_character_rate": 0.37115588547189815,
          "true_positive": 39
        },
        {
          "character_coverage": 0.80236354139462,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 12289,
          "exact_span_leakage": 0.23663003663003668,
          "exact_span_recall": 0.7633699633699633,
          "false_negative": 323,
          "gold_characters": 15316,
          "gold_spans": 1365,
          "label_id": "NAME_STUDENT",
          "residual_pii_character_rate": 0.19763645860538004,
          "true_positive": 1042
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 95,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 95,
          "gold_spans": 6,
          "label_id": "PHONE_NUM",
          "residual_pii_character_rate": 0.0,
          "true_positive": 6
        },
        {
          "character_coverage": 0.5052631578947369,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 48,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 1,
          "gold_characters": 95,
          "gold_spans": 2,
          "label_id": "STREET_ADDRESS",
          "residual_pii_character_rate": 0.49473684210526314,
          "true_positive": 1
        },
        {
          "character_coverage": 0.2651499043163938,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1247,
          "exact_span_leakage": 0.7454545454545455,
          "exact_span_recall": 0.2545454545454545,
          "false_negative": 82,
          "gold_characters": 4703,
          "gold_spans": 110,
          "label_id": "URL_PERSONAL",
          "residual_pii_character_rate": 0.7348500956836062,
          "true_positive": 28
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 61,
          "exact_span_leakage": 0.16666666666666663,
          "exact_span_recall": 0.8333333333333334,
          "false_negative": 1,
          "gold_characters": 61,
          "gold_spans": 6,
          "label_id": "USERNAME",
          "residual_pii_character_rate": 0.0,
          "true_positive": 5
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "piilo",
      "dataset_sha256": "8276cd44f3b2eb357dfb405b3c5d8e9f821388e984cbf66e92e7df03f1b13117",
      "documents": 6807,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.5008635578583766,
        "false_negative": 446,
        "false_positive": 1866,
        "precision": 0.38334434897554526,
        "recall": 0.7222914072229141,
        "true_positive": 1160
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-piilo.json",
        "sha256": "d8fcc88bcad79f0a1fa9088c1ed7dab4c7c6b31ddb2ed84418d0b86fe8acceeb"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.012524461839530333,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 160,
          "exact_span_leakage": 0.9882133995037221,
          "exact_span_recall": 0.011786600496277916,
          "false_negative": 1593,
          "gold_characters": 12775,
          "gold_spans": 1612,
          "label_id": "CODE",
          "residual_pii_character_rate": 0.9874755381604696,
          "true_positive": 19
        },
        {
          "character_coverage": 0.26635886397804853,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32422,
          "exact_span_leakage": 0.7067953020134228,
          "exact_span_recall": 0.29320469798657717,
          "false_negative": 6740,
          "gold_characters": 121723,
          "gold_spans": 9536,
          "label_id": "DATETIME",
          "residual_pii_character_rate": 0.7336411360219515,
          "true_positive": 2796
        },
        {
          "character_coverage": 0.01969147005444646,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 217,
          "exact_span_leakage": 0.9727371864776445,
          "exact_span_recall": 0.027262813522355506,
          "false_negative": 892,
          "gold_characters": 11020,
          "gold_spans": 917,
          "label_id": "DEM",
          "residual_pii_character_rate": 0.9803085299455535,
          "true_positive": 25
        },
        {
          "character_coverage": 0.023629288998938805,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 334,
          "exact_span_leakage": 0.9904043865661412,
          "exact_span_recall": 0.009595613433858808,
          "false_negative": 1445,
          "gold_characters": 14135,
          "gold_spans": 1459,
          "label_id": "LOC",
          "residual_pii_character_rate": 0.9763707110010612,
          "true_positive": 14
        },
        {
          "character_coverage": 0.028014416775884666,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 342,
          "exact_span_leakage": 0.957169459962756,
          "exact_span_recall": 0.04283054003724395,
          "false_negative": 514,
          "gold_characters": 12208,
          "gold_spans": 537,
          "label_id": "MISC",
          "residual_pii_character_rate": 0.9719855832241153,
          "true_positive": 23
        },
        {
          "character_coverage": 0.008471892319873317,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 321,
          "exact_span_leakage": 0.9917737789203085,
          "exact_span_recall": 0.008226221079691516,
          "false_negative": 1929,
          "gold_characters": 37890,
          "gold_spans": 1945,
          "label_id": "ORG",
          "residual_pii_character_rate": 0.9915281076801267,
          "true_positive": 16
        },
        {
          "character_coverage": 0.7446089893478826,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 40124,
          "exact_span_leakage": 0.2984533591106815,
          "exact_span_recall": 0.7015466408893185,
          "false_negative": 1235,
          "gold_characters": 53886,
          "gold_spans": 4138,
          "label_id": "PERSON",
          "residual_pii_character_rate": 0.25539101065211745,
          "true_positive": 2903
        },
        {
          "character_coverage": 0.0013605442176870747,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 12,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 664,
          "gold_characters": 8820,
          "gold_spans": 664,
          "label_id": "QUANTITY",
          "residual_pii_character_rate": 0.998639455782313,
          "true_positive": 0
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tab",
      "dataset_sha256": "cd0f0f15f84a8739654c7cf30c6be8ce27b051ef73974d39d792a0cb8c846379",
      "documents": 555,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.4221566699442806,
        "false_negative": 15012,
        "false_positive": 855,
        "precision": 0.871447902571042,
        "recall": 0.27854671280276816,
        "true_positive": 5796
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-tab.json",
        "sha256": "80f23778702aeed66b440f352321b21add6533d263fb713802291762025ca740"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.7914561322921452,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3446,
          "exact_span_leakage": 0.25503355704697983,
          "exact_span_recall": 0.7449664429530202,
          "false_negative": 76,
          "gold_characters": 4354,
          "gold_spans": 298,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.20854386770785482,
          "true_positive": 222
        },
        {
          "character_coverage": 0.9419134396355353,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4962,
          "exact_span_leakage": 0.16838487972508587,
          "exact_span_recall": 0.8316151202749141,
          "false_negative": 49,
          "gold_characters": 5268,
          "gold_spans": 291,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.05808656036446469,
          "true_positive": 242
        },
        {
          "character_coverage": 0.18550791102958036,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 809,
          "exact_span_leakage": 0.8470319634703196,
          "exact_span_recall": 0.15296803652968036,
          "false_negative": 371,
          "gold_characters": 4361,
          "gold_spans": 438,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.8144920889704197,
          "true_positive": 67
        },
        {
          "character_coverage": 0.9993657237092477,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7878,
          "exact_span_leakage": 0.015974440894568676,
          "exact_span_recall": 0.9840255591054313,
          "false_negative": 5,
          "gold_characters": 7883,
          "gold_spans": 313,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.0006342762907523003,
          "true_positive": 308
        },
        {
          "character_coverage": 0.7587601078167115,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3378,
          "exact_span_leakage": 0.446685878962536,
          "exact_span_recall": 0.553314121037464,
          "false_negative": 620,
          "gold_characters": 4452,
          "gold_spans": 1388,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.24123989218328845,
          "true_positive": 768
        },
        {
          "character_coverage": 0.9765510503175379,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3998,
          "exact_span_leakage": 0.061046511627906974,
          "exact_span_recall": 0.938953488372093,
          "false_negative": 21,
          "gold_characters": 4094,
          "gold_spans": 344,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.02344894968246214,
          "true_positive": 323
        },
        {
          "character_coverage": 0.6354291962821214,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5811,
          "exact_span_leakage": 0.7089947089947091,
          "exact_span_recall": 0.291005291005291,
          "false_negative": 134,
          "gold_characters": 9145,
          "gold_spans": 189,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.36457080371787864,
          "true_positive": 55
        },
        {
          "character_coverage": 0.8751674256630056,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3267,
          "exact_span_leakage": 0.28936170212765955,
          "exact_span_recall": 0.7106382978723405,
          "false_negative": 68,
          "gold_characters": 3733,
          "gold_spans": 235,
          "label_id": "secret",
          "residual_pii_character_rate": 0.12483257433699435,
          "true_positive": 167
        },
        {
          "character_coverage": 0.22702702702702704,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 336,
          "exact_span_leakage": 0.772972972972973,
          "exact_span_recall": 0.22702702702702704,
          "false_negative": 143,
          "gold_characters": 1480,
          "gold_spans": 185,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.772972972972973,
          "true_positive": 42
        },
        {
          "character_coverage": 0.9047619047619048,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 950,
          "exact_span_leakage": 0.11428571428571432,
          "exact_span_recall": 0.8857142857142857,
          "false_negative": 12,
          "gold_characters": 1050,
          "gold_spans": 105,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.09523809523809523,
          "true_positive": 93
        },
        {
          "character_coverage": 0.6796812749003984,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 853,
          "exact_span_leakage": 0.31081081081081086,
          "exact_span_recall": 0.6891891891891891,
          "false_negative": 46,
          "gold_characters": 1255,
          "gold_spans": 148,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.3203187250996016,
          "true_positive": 102
        },
        {
          "character_coverage": 0.4597846738442052,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 726,
          "exact_span_leakage": 0.5606060606060606,
          "exact_span_recall": 0.4393939393939394,
          "false_negative": 111,
          "gold_characters": 1579,
          "gold_spans": 198,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.5402153261557948,
          "true_positive": 87
        },
        {
          "character_coverage": 0.7785209959123003,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2095,
          "exact_span_leakage": 0.25125628140703515,
          "exact_span_recall": 0.7487437185929648,
          "false_negative": 50,
          "gold_characters": 2691,
          "gold_spans": 199,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.22147900408769972,
          "true_positive": 149
        },
        {
          "character_coverage": 0.3410169491525424,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 503,
          "exact_span_leakage": 0.7123287671232876,
          "exact_span_recall": 0.2876712328767123,
          "false_negative": 104,
          "gold_characters": 1475,
          "gold_spans": 146,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.6589830508474577,
          "true_positive": 42
        },
        {
          "character_coverage": 0.6634615384615384,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 828,
          "exact_span_leakage": 0.3464566929133859,
          "exact_span_recall": 0.6535433070866141,
          "false_negative": 44,
          "gold_characters": 1248,
          "gold_spans": 127,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.33653846153846156,
          "true_positive": 83
        },
        {
          "character_coverage": 0.8814432989690721,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1710,
          "exact_span_leakage": 0.12371134020618557,
          "exact_span_recall": 0.8762886597938144,
          "false_negative": 24,
          "gold_characters": 1940,
          "gold_spans": 194,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.11855670103092786,
          "true_positive": 170
        },
        {
          "character_coverage": 0.9111111111111111,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1476,
          "exact_span_leakage": 0.0888888888888889,
          "exact_span_recall": 0.9111111111111111,
          "false_negative": 12,
          "gold_characters": 1620,
          "gold_spans": 135,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.0888888888888889,
          "true_positive": 123
        },
        {
          "character_coverage": 0.8928571428571429,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1125,
          "exact_span_leakage": 0.11428571428571432,
          "exact_span_recall": 0.8857142857142857,
          "false_negative": 16,
          "gold_characters": 1260,
          "gold_spans": 140,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.1071428571428571,
          "true_positive": 124
        },
        {
          "character_coverage": 0.45869947275922673,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 783,
          "exact_span_leakage": 0.5724137931034483,
          "exact_span_recall": 0.42758620689655175,
          "false_negative": 83,
          "gold_characters": 1707,
          "gold_spans": 145,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.5413005272407733,
          "true_positive": 62
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-long",
      "dataset_sha256": "b30ca2a1012254f164893d1aad491a778d2fe49cfef71239f8c934e0544883e2",
      "documents": 300,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.5626905985884812,
        "false_negative": 1989,
        "false_positive": 3030,
        "precision": 0.5158971081642435,
        "recall": 0.6188194710617094,
        "true_positive": 3229
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-tw-long.json",
        "sha256": "820452babdf2db65a330ef11cef7fda861ba2c74101c8dfb187611302d6935c3"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.8852361028093245,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1481,
          "exact_span_leakage": 0.18644067796610164,
          "exact_span_recall": 0.8135593220338984,
          "false_negative": 22,
          "gold_characters": 1673,
          "gold_spans": 118,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.11476389719067548,
          "true_positive": 96
        },
        {
          "character_coverage": 0.9676616915422885,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1945,
          "exact_span_leakage": 0.22608695652173916,
          "exact_span_recall": 0.7739130434782608,
          "false_negative": 26,
          "gold_characters": 2010,
          "gold_spans": 115,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.03233830845771146,
          "true_positive": 89
        },
        {
          "character_coverage": 0.10540184453227931,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 160,
          "exact_span_leakage": 0.9113924050632911,
          "exact_span_recall": 0.08860759493670886,
          "false_negative": 144,
          "gold_characters": 1518,
          "gold_spans": 158,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.8945981554677207,
          "true_positive": 14
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2939,
          "exact_span_leakage": 0.033333333333333326,
          "exact_span_recall": 0.9666666666666667,
          "false_negative": 4,
          "gold_characters": 2939,
          "gold_spans": 120,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.0,
          "true_positive": 116
        },
        {
          "character_coverage": 0.7612323491655969,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1186,
          "exact_span_leakage": 0.5077519379844961,
          "exact_span_recall": 0.49224806201550386,
          "false_negative": 262,
          "gold_characters": 1558,
          "gold_spans": 516,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.2387676508344031,
          "true_positive": 254
        },
        {
          "character_coverage": 0.9832946635730858,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2119,
          "exact_span_leakage": 0.02777777777777779,
          "exact_span_recall": 0.9722222222222222,
          "false_negative": 5,
          "gold_characters": 2155,
          "gold_spans": 180,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.016705336426914186,
          "true_positive": 175
        },
        {
          "character_coverage": 0.7208646616541353,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2301,
          "exact_span_leakage": 0.7671232876712328,
          "exact_span_recall": 0.2328767123287671,
          "false_negative": 56,
          "gold_characters": 3192,
          "gold_spans": 73,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.2791353383458647,
          "true_positive": 17
        },
        {
          "character_coverage": 0.8278301886792453,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 702,
          "exact_span_leakage": 0.3614457831325302,
          "exact_span_recall": 0.6385542168674698,
          "false_negative": 30,
          "gold_characters": 848,
          "gold_spans": 83,
          "label_id": "secret",
          "residual_pii_character_rate": 0.1721698113207547,
          "true_positive": 53
        },
        {
          "character_coverage": 0.4358974358974359,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 272,
          "exact_span_leakage": 0.5769230769230769,
          "exact_span_recall": 0.4230769230769231,
          "false_negative": 45,
          "gold_characters": 624,
          "gold_spans": 78,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.5641025641025641,
          "true_positive": 33
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 660,
          "exact_span_leakage": 0.015151515151515138,
          "exact_span_recall": 0.9848484848484849,
          "false_negative": 1,
          "gold_characters": 660,
          "gold_spans": 66,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.0,
          "true_positive": 65
        },
        {
          "character_coverage": 0.8868243243243243,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 525,
          "exact_span_leakage": 0.11428571428571432,
          "exact_span_recall": 0.8857142857142857,
          "false_negative": 8,
          "gold_characters": 592,
          "gold_spans": 70,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.11317567567567566,
          "true_positive": 62
        },
        {
          "character_coverage": 0.7165644171779141,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 584,
          "exact_span_leakage": 0.3137254901960784,
          "exact_span_recall": 0.6862745098039216,
          "false_negative": 32,
          "gold_characters": 815,
          "gold_spans": 102,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.2834355828220859,
          "true_positive": 70
        },
        {
          "character_coverage": 0.8842105263157894,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1092,
          "exact_span_leakage": 0.11827956989247312,
          "exact_span_recall": 0.8817204301075269,
          "false_negative": 11,
          "gold_characters": 1235,
          "gold_spans": 93,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.11578947368421055,
          "true_positive": 82
        },
        {
          "character_coverage": 0.3673469387755102,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 306,
          "exact_span_leakage": 0.7560975609756098,
          "exact_span_recall": 0.24390243902439024,
          "false_negative": 62,
          "gold_characters": 833,
          "gold_spans": 82,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.6326530612244898,
          "true_positive": 20
        },
        {
          "character_coverage": 0.892904953145917,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 667,
          "exact_span_leakage": 0.12,
          "exact_span_recall": 0.88,
          "false_negative": 9,
          "gold_characters": 747,
          "gold_spans": 75,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.107095046854083,
          "true_positive": 66
        },
        {
          "character_coverage": 0.9438202247191011,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 840,
          "exact_span_leakage": 0.0561797752808989,
          "exact_span_recall": 0.9438202247191011,
          "false_negative": 5,
          "gold_characters": 890,
          "gold_spans": 89,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.0561797752808989,
          "true_positive": 84
        },
        {
          "character_coverage": 0.961038961038961,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 888,
          "exact_span_leakage": 0.051948051948051965,
          "exact_span_recall": 0.948051948051948,
          "false_negative": 4,
          "gold_characters": 924,
          "gold_spans": 77,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.038961038961038974,
          "true_positive": 73
        },
        {
          "character_coverage": 0.9322033898305084,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 495,
          "exact_span_leakage": 0.0847457627118644,
          "exact_span_recall": 0.9152542372881356,
          "false_negative": 5,
          "gold_characters": 531,
          "gold_spans": 59,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.06779661016949157,
          "true_positive": 54
        },
        {
          "character_coverage": 0.5033025099075297,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 381,
          "exact_span_leakage": 0.5441176470588236,
          "exact_span_recall": 0.45588235294117646,
          "false_negative": 37,
          "gold_characters": 757,
          "gold_spans": 68,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.49669749009247033,
          "true_positive": 31
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-mid",
      "dataset_sha256": "5d05e1c7fc6adaf7c66591de0d76e5ac5d175fb182f67a19da6d23896b3b02e0",
      "documents": 300,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.6681985294117647,
        "false_negative": 768,
        "false_positive": 676,
        "precision": 0.6826291079812207,
        "recall": 0.6543654365436543,
        "true_positive": 1454
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-tw-mid.json",
        "sha256": "59faeb3b27777c60990de61121ddcc171839961f10a631fd0301488465b319c6"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.7162698412698413,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 361,
          "exact_span_leakage": 0.5454545454545454,
          "exact_span_recall": 0.45454545454545453,
          "false_negative": 24,
          "gold_characters": 504,
          "gold_spans": 44,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.2837301587301587,
          "true_positive": 20
        },
        {
          "character_coverage": 0.8970792767732962,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 645,
          "exact_span_leakage": 0.4054054054054054,
          "exact_span_recall": 0.5945945945945946,
          "false_negative": 15,
          "gold_characters": 719,
          "gold_spans": 37,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.10292072322670376,
          "true_positive": 22
        },
        {
          "character_coverage": 0.36774193548387096,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 114,
          "exact_span_leakage": 0.7894736842105263,
          "exact_span_recall": 0.21052631578947367,
          "false_negative": 30,
          "gold_characters": 310,
          "gold_spans": 38,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.6322580645161291,
          "true_positive": 8
        },
        {
          "character_coverage": 0.9670846394984326,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 617,
          "exact_span_leakage": 0.0714285714285714,
          "exact_span_recall": 0.9285714285714286,
          "false_negative": 2,
          "gold_characters": 638,
          "gold_spans": 28,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.0329153605015674,
          "true_positive": 26
        },
        {
          "character_coverage": 0.6231003039513677,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 205,
          "exact_span_leakage": 0.5909090909090908,
          "exact_span_recall": 0.4090909090909091,
          "false_negative": 52,
          "gold_characters": 329,
          "gold_spans": 88,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.37689969604863227,
          "true_positive": 36
        },
        {
          "character_coverage": 0.9760191846522782,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 407,
          "exact_span_leakage": 0.08108108108108103,
          "exact_span_recall": 0.918918918918919,
          "false_negative": 3,
          "gold_characters": 417,
          "gold_spans": 37,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.023980815347721784,
          "true_positive": 34
        },
        {
          "character_coverage": 0.6304985337243402,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 430,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 10,
          "gold_characters": 682,
          "gold_spans": 20,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.36950146627565983,
          "true_positive": 10
        },
        {
          "character_coverage": 0.5179640718562875,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 173,
          "exact_span_leakage": 0.6086956521739131,
          "exact_span_recall": 0.391304347826087,
          "false_negative": 14,
          "gold_characters": 334,
          "gold_spans": 23,
          "label_id": "secret",
          "residual_pii_character_rate": 0.48203592814371254,
          "true_positive": 9
        },
        {
          "character_coverage": 0.38461538461538464,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 40,
          "exact_span_leakage": 0.6923076923076923,
          "exact_span_recall": 0.3076923076923077,
          "false_negative": 9,
          "gold_characters": 104,
          "gold_spans": 13,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.6153846153846154,
          "true_positive": 4
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 110,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 110,
          "gold_spans": 11,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.0,
          "true_positive": 11
        },
        {
          "character_coverage": 0.6713286713286714,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 96,
          "exact_span_leakage": 0.4,
          "exact_span_recall": 0.6,
          "false_negative": 6,
          "gold_characters": 143,
          "gold_spans": 15,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.32867132867132864,
          "true_positive": 9
        },
        {
          "character_coverage": 0.6126760563380281,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 87,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 9,
          "gold_characters": 142,
          "gold_spans": 18,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.38732394366197187,
          "true_positive": 9
        },
        {
          "character_coverage": 0.6191950464396285,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 200,
          "exact_span_leakage": 0.30000000000000004,
          "exact_span_recall": 0.7,
          "false_negative": 6,
          "gold_characters": 323,
          "gold_spans": 20,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.3808049535603715,
          "true_positive": 14
        },
        {
          "character_coverage": 0.168,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 21,
          "exact_span_leakage": 0.8333333333333334,
          "exact_span_recall": 0.16666666666666666,
          "false_negative": 10,
          "gold_characters": 125,
          "gold_spans": 12,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.832,
          "true_positive": 2
        },
        {
          "character_coverage": 0.7372881355932204,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 87,
          "exact_span_leakage": 0.5454545454545454,
          "exact_span_recall": 0.45454545454545453,
          "false_negative": 6,
          "gold_characters": 118,
          "gold_spans": 11,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.2627118644067796,
          "true_positive": 5
        },
        {
          "character_coverage": 0.9523809523809523,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 200,
          "exact_span_leakage": 0.09523809523809523,
          "exact_span_recall": 0.9047619047619048,
          "false_negative": 2,
          "gold_characters": 210,
          "gold_spans": 21,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.04761904761904767,
          "true_positive": 19
        },
        {
          "character_coverage": 0.6923076923076923,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 108,
          "exact_span_leakage": 0.46153846153846156,
          "exact_span_recall": 0.5384615384615384,
          "false_negative": 6,
          "gold_characters": 156,
          "gold_spans": 13,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.3076923076923077,
          "true_positive": 7
        },
        {
          "character_coverage": 0.9166666666666666,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 99,
          "exact_span_leakage": 0.16666666666666663,
          "exact_span_recall": 0.8333333333333334,
          "false_negative": 2,
          "gold_characters": 108,
          "gold_spans": 12,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.08333333333333337,
          "true_positive": 10
        },
        {
          "character_coverage": 0.5066666666666667,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 76,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 7,
          "gold_characters": 150,
          "gold_spans": 14,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.4933333333333333,
          "true_positive": 7
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-short",
      "dataset_sha256": "03e4850a0a97a2b5cb046c1dcd3996b42c5eea3766ae2618d4bbf64391e73693",
      "documents": 310,
      "model_id": "openai-privacy-filter",
      "overall": {
        "f1": 0.6469135802469136,
        "false_negative": 213,
        "false_positive": 73,
        "precision": 0.7820895522388059,
        "recall": 0.5515789473684211,
        "true_positive": 262
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openai-privacy-filter-tw-short.json",
        "sha256": "e0642c9e02b253c21ca87856792b1484253c77e50f33859f4f9b66ade1ff4b1e"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.6805981974844013,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 6872,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 413,
          "gold_characters": 10097,
          "gold_spans": 413,
          "label_id": "CALLE",
          "residual_pii_character_rate": 0.3194018025155987,
          "true_positive": 0
        },
        {
          "character_coverage": 0.09523809523809523,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 16,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 6,
          "gold_characters": 168,
          "gold_spans": 6,
          "label_id": "CENTRO_SALUD",
          "residual_pii_character_rate": 0.9047619047619048,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9917642752562226,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5419,
          "exact_span_leakage": 0.9839357429718876,
          "exact_span_recall": 0.01606425702811245,
          "false_negative": 245,
          "gold_characters": 5464,
          "gold_spans": 249,
          "label_id": "CORREO_ELECTRONICO",
          "residual_pii_character_rate": 0.008235724743777428,
          "true_positive": 4
        },
        {
          "character_coverage": 0.2672316384180791,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 946,
          "exact_span_leakage": 0.9613899613899614,
          "exact_span_recall": 0.03861003861003861,
          "false_negative": 498,
          "gold_characters": 3540,
          "gold_spans": 518,
          "label_id": "EDAD_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.7327683615819209,
          "true_positive": 20
        },
        {
          "character_coverage": 0.03419972640218878,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 25,
          "exact_span_leakage": 0.9876543209876543,
          "exact_span_recall": 0.012345679012345678,
          "false_negative": 80,
          "gold_characters": 731,
          "gold_spans": 81,
          "label_id": "FAMILIARES_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.9658002735978112,
          "true_positive": 1
        },
        {
          "character_coverage": 0.8617679733375655,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5430,
          "exact_span_leakage": 0.1522094926350246,
          "exact_span_recall": 0.8477905073649754,
          "false_negative": 93,
          "gold_characters": 6301,
          "gold_spans": 611,
          "label_id": "FECHAS",
          "residual_pii_character_rate": 0.1382320266624345,
          "true_positive": 518
        },
        {
          "character_coverage": 0.2803241326918207,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1107,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 130,
          "gold_characters": 3949,
          "gold_spans": 130,
          "label_id": "HOSPITAL",
          "residual_pii_character_rate": 0.7196758673081793,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9681159420289855,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2672,
          "exact_span_leakage": 0.14646464646464652,
          "exact_span_recall": 0.8535353535353535,
          "false_negative": 29,
          "gold_characters": 2760,
          "gold_spans": 198,
          "label_id": "ID_ASEGURAMIENTO",
          "residual_pii_character_rate": 0.03188405797101446,
          "true_positive": 169
        },
        {
          "character_coverage": 0.6019736842105263,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 183,
          "exact_span_leakage": 0.4358974358974359,
          "exact_span_recall": 0.5641025641025641,
          "false_negative": 17,
          "gold_characters": 304,
          "gold_spans": 39,
          "label_id": "ID_CONTACTO_ASISTENCIAL",
          "residual_pii_character_rate": 0.3980263157894737,
          "true_positive": 22
        },
        {
          "character_coverage": 0.7409274193548387,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1470,
          "exact_span_leakage": 0.2826855123674912,
          "exact_span_recall": 0.7173144876325088,
          "false_negative": 80,
          "gold_characters": 1984,
          "gold_spans": 283,
          "label_id": "ID_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.25907258064516125,
          "true_positive": 203
        },
        {
          "character_coverage": 0.9451170298627926,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2342,
          "exact_span_leakage": 0.12393162393162394,
          "exact_span_recall": 0.8760683760683761,
          "false_negative": 29,
          "gold_characters": 2478,
          "gold_spans": 234,
          "label_id": "ID_TITULACION_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.05488297013720744,
          "true_positive": 205
        },
        {
          "character_coverage": 0.25348189415041783,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 364,
          "exact_span_leakage": 0.9253731343283582,
          "exact_span_recall": 0.07462686567164178,
          "false_negative": 62,
          "gold_characters": 1436,
          "gold_spans": 67,
          "label_id": "INSTITUCION",
          "residual_pii_character_rate": 0.7465181058495822,
          "true_positive": 5
        },
        {
          "character_coverage": 0.9712276214833759,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 10633,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 501,
          "gold_characters": 10948,
          "gold_spans": 501,
          "label_id": "NOMBRE_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.028772378516624064,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9919639357114857,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5061,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 502,
          "gold_characters": 5102,
          "gold_spans": 502,
          "label_id": "NOMBRE_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.008036064288514333,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8461538461538461,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 66,
          "exact_span_leakage": 0.2857142857142857,
          "exact_span_recall": 0.7142857142857143,
          "false_negative": 2,
          "gold_characters": 78,
          "gold_spans": 7,
          "label_id": "NUMERO_FAX",
          "residual_pii_character_rate": 0.15384615384615385,
          "true_positive": 5
        },
        {
          "character_coverage": 0.975,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 273,
          "exact_span_leakage": 0.15384615384615385,
          "exact_span_recall": 0.8461538461538461,
          "false_negative": 4,
          "gold_characters": 280,
          "gold_spans": 26,
          "label_id": "NUMERO_TELEFONO",
          "residual_pii_character_rate": 0.025000000000000022,
          "true_positive": 22
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 7,
          "gold_characters": 86,
          "gold_spans": 7,
          "label_id": "OTROS_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.02328705776981639,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 52,
          "exact_span_leakage": 0.9972451790633609,
          "exact_span_recall": 0.0027548209366391185,
          "false_negative": 362,
          "gold_characters": 2233,
          "gold_spans": 363,
          "label_id": "PAIS",
          "residual_pii_character_rate": 0.9767129422301836,
          "true_positive": 1
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 9,
          "gold_characters": 160,
          "gold_spans": 9,
          "label_id": "PROFESION",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.2007017543859649,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 286,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 461,
          "gold_characters": 1425,
          "gold_spans": 461,
          "label_id": "SEXO_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.7992982456140351,
          "true_positive": 0
        },
        {
          "character_coverage": 0.7244465379180405,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4614,
          "exact_span_leakage": 0.5784518828451883,
          "exact_span_recall": 0.4215481171548117,
          "false_negative": 553,
          "gold_characters": 6369,
          "gold_spans": 956,
          "label_id": "TERRITORIO",
          "residual_pii_character_rate": 0.2755534620819595,
          "true_positive": 403
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "meddocan",
      "dataset_sha256": "68c972fd2fd5ffa202072c6d048c111af2a31622e1845ff999c83fe315558c2a",
      "documents": 250,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.2547215496368039,
        "false_negative": 4083,
        "false_positive": 5151,
        "precision": 0.2345073562193491,
        "recall": 0.27874933757286696,
        "true_positive": 1578
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-meddocan.json",
        "sha256": "6e4d6550eaaa9490ef277e28902dace8b8f87ea94c8ea1e4bda9c1f06aee2519"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.9955138236828378,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 209924,
          "exact_span_leakage": 0.29455460372611275,
          "exact_span_recall": 0.7054453962738872,
          "false_negative": 4917,
          "gold_characters": 210870,
          "gold_spans": 16693,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.004486176317162194,
          "true_positive": 11776
        },
        {
          "character_coverage": 0.966016713091922,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 13872,
          "exact_span_leakage": 0.06139249337240127,
          "exact_span_recall": 0.9386075066275987,
          "false_negative": 440,
          "gold_characters": 14360,
          "gold_spans": 7167,
          "label_id": "age",
          "residual_pii_character_rate": 0.03398328690807795,
          "true_positive": 6727
        },
        {
          "character_coverage": 0.9986622105423523,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 209767,
          "exact_span_leakage": 0.8322262695521748,
          "exact_span_recall": 0.16777373044782515,
          "false_negative": 3884,
          "gold_characters": 210048,
          "gold_spans": 4667,
          "label_id": "api_key",
          "residual_pii_character_rate": 0.0013377894576477312,
          "true_positive": 783
        },
        {
          "character_coverage": 0.9975515635395875,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 74966,
          "exact_span_leakage": 0.00251376586066554,
          "exact_span_recall": 0.9974862341393345,
          "false_negative": 21,
          "gold_characters": 75150,
          "gold_spans": 8354,
          "label_id": "bank_routing_number",
          "residual_pii_character_rate": 0.002448436460412484,
          "true_positive": 8333
        },
        {
          "character_coverage": 0.9992683323397106,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 147500,
          "exact_span_leakage": 0.9760963177783637,
          "exact_span_recall": 0.023903682221636347,
          "false_negative": 11107,
          "gold_characters": 147608,
          "gold_spans": 11379,
          "label_id": "biometric_identifier",
          "residual_pii_character_rate": 0.0007316676602894256,
          "true_positive": 272
        },
        {
          "character_coverage": 0.9525084154601128,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 23486,
          "exact_span_leakage": 0.9976530059577541,
          "exact_span_recall": 0.002346994042245893,
          "false_negative": 5526,
          "gold_characters": 24657,
          "gold_spans": 5539,
          "label_id": "blood_type",
          "residual_pii_character_rate": 0.04749158453988722,
          "true_positive": 13
        },
        {
          "character_coverage": 0.9956162892644436,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32932,
          "exact_span_leakage": 0.6442371752165224,
          "exact_span_recall": 0.3557628247834777,
          "false_negative": 1934,
          "gold_characters": 33077,
          "gold_spans": 3002,
          "label_id": "certificate_license_number",
          "residual_pii_character_rate": 0.004383710735556434,
          "true_positive": 1068
        },
        {
          "character_coverage": 0.9551847389825545,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 143506,
          "exact_span_leakage": 0.9673516106175396,
          "exact_span_recall": 0.03264838938246035,
          "false_negative": 17748,
          "gold_characters": 150239,
          "gold_spans": 18347,
          "label_id": "city",
          "residual_pii_character_rate": 0.044815261017445485,
          "true_positive": 599
        },
        {
          "character_coverage": 0.8685904522392813,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 891774,
          "exact_span_leakage": 0.7760818425515619,
          "exact_span_recall": 0.2239181574484381,
          "false_negative": 42558,
          "gold_characters": 1026691,
          "gold_spans": 54837,
          "label_id": "company_name",
          "residual_pii_character_rate": 0.13140954776071867,
          "true_positive": 12279
        },
        {
          "character_coverage": 0.99922059719284,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 152562,
          "exact_span_leakage": 0.23710667188928058,
          "exact_span_recall": 0.7628933281107194,
          "false_negative": 1816,
          "gold_characters": 152681,
          "gold_spans": 7659,
          "label_id": "coordinate",
          "residual_pii_character_rate": 0.0007794028071600456,
          "true_positive": 5843
        },
        {
          "character_coverage": 0.9801625479243549,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 136519,
          "exact_span_leakage": 0.9781043663471779,
          "exact_span_recall": 0.02189563365282215,
          "false_negative": 22961,
          "gold_characters": 139282,
          "gold_spans": 23475,
          "label_id": "country",
          "residual_pii_character_rate": 0.019837452075645134,
          "true_positive": 514
        },
        {
          "character_coverage": 0.9477044191585076,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 114604,
          "exact_span_leakage": 0.9669226225634967,
          "exact_span_recall": 0.03307737743650325,
          "false_negative": 8185,
          "gold_characters": 120928,
          "gold_spans": 8465,
          "label_id": "county",
          "residual_pii_character_rate": 0.05229558084149244,
          "true_positive": 280
        },
        {
          "character_coverage": 0.9994914335737213,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 239768,
          "exact_span_leakage": 0.0019429548457293722,
          "exact_span_recall": 0.9980570451542706,
          "false_negative": 25,
          "gold_characters": 239890,
          "gold_spans": 12867,
          "label_id": "credit_debit_card",
          "residual_pii_character_rate": 0.0005085664262787226,
          "true_positive": 12842
        },
        {
          "character_coverage": 0.9971469767156487,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 195024,
          "exact_span_leakage": 0.4682470002926544,
          "exact_span_recall": 0.5317529997073456,
          "false_negative": 9600,
          "gold_characters": 195582,
          "gold_spans": 20502,
          "label_id": "customer_id",
          "residual_pii_character_rate": 0.0028530232843513303,
          "true_positive": 10902
        },
        {
          "character_coverage": 0.9739830683460665,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14151,
          "exact_span_leakage": 0.026016931653933484,
          "exact_span_recall": 0.9739830683460665,
          "false_negative": 126,
          "gold_characters": 14529,
          "gold_spans": 4843,
          "label_id": "cvv",
          "residual_pii_character_rate": 0.026016931653933484,
          "true_positive": 4717
        },
        {
          "character_coverage": 0.9467845225289981,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 740252,
          "exact_span_leakage": 0.14277011385327687,
          "exact_span_recall": 0.8572298861467231,
          "false_negative": 10546,
          "gold_characters": 781859,
          "gold_spans": 73867,
          "label_id": "date",
          "residual_pii_character_rate": 0.05321547747100186,
          "true_positive": 63321
        },
        {
          "character_coverage": 0.9998783031027177,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 180755,
          "exact_span_leakage": 0.0003871895569445005,
          "exact_span_recall": 0.9996128104430555,
          "false_negative": 7,
          "gold_characters": 180777,
          "gold_spans": 18079,
          "label_id": "date_of_birth",
          "residual_pii_character_rate": 0.00012169689728225297,
          "true_positive": 18072
        },
        {
          "character_coverage": 0.9790174940596743,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 245565,
          "exact_span_leakage": 0.033280387262688116,
          "exact_span_recall": 0.9667196127373119,
          "false_negative": 440,
          "gold_characters": 250828,
          "gold_spans": 13221,
          "label_id": "date_time",
          "residual_pii_character_rate": 0.02098250594032569,
          "true_positive": 12781
        },
        {
          "character_coverage": 0.9982492469879518,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 53027,
          "exact_span_leakage": 0.44475468687674513,
          "exact_span_recall": 0.5552453131232549,
          "false_negative": 1115,
          "gold_characters": 53120,
          "gold_spans": 2507,
          "label_id": "device_identifier",
          "residual_pii_character_rate": 0.0017507530120481896,
          "true_positive": 1392
        },
        {
          "character_coverage": 0.8061377681902985,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 110615,
          "exact_span_leakage": 0.943911281723182,
          "exact_span_recall": 0.056088718276818085,
          "false_negative": 8852,
          "gold_characters": 137216,
          "gold_spans": 9378,
          "label_id": "education_level",
          "residual_pii_character_rate": 0.19386223180970152,
          "true_positive": 526
        },
        {
          "character_coverage": 0.9970352712791013,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1396986,
          "exact_span_leakage": 0.9824587428147599,
          "exact_span_recall": 0.017541257185240126,
          "false_negative": 52984,
          "gold_characters": 1401140,
          "gold_spans": 53930,
          "label_id": "email",
          "residual_pii_character_rate": 0.0029647287208987017,
          "true_positive": 946
        },
        {
          "character_coverage": 0.9934168447654556,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 63681,
          "exact_span_leakage": 0.5295111511601712,
          "exact_span_recall": 0.4704888488398288,
          "false_negative": 4701,
          "gold_characters": 64103,
          "gold_spans": 8878,
          "label_id": "employee_id",
          "residual_pii_character_rate": 0.006583155234544447,
          "true_positive": 4177
        },
        {
          "character_coverage": 0.8697327327035153,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 89392,
          "exact_span_leakage": 0.9950989290252314,
          "exact_span_recall": 0.0049010709747685606,
          "false_negative": 10964,
          "gold_characters": 102781,
          "gold_spans": 11018,
          "label_id": "employment_status",
          "residual_pii_character_rate": 0.13026726729648475,
          "true_positive": 54
        },
        {
          "character_coverage": 0.9995678781650005,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 85587,
          "exact_span_leakage": 0.40697493383154293,
          "exact_span_recall": 0.5930250661684571,
          "false_negative": 2614,
          "gold_characters": 85624,
          "gold_spans": 6423,
          "label_id": "fax_number",
          "residual_pii_character_rate": 0.0004321218349995215,
          "true_positive": 3809
        },
        {
          "character_coverage": 0.9883965632948624,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 477186,
          "exact_span_leakage": 0.8659495734326476,
          "exact_span_recall": 0.13405042656735242,
          "false_negative": 72777,
          "gold_characters": 482788,
          "gold_spans": 84043,
          "label_id": "first_name",
          "residual_pii_character_rate": 0.011603436705137637,
          "true_positive": 11266
        },
        {
          "character_coverage": 0.9453913985910636,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 38381,
          "exact_span_leakage": 0.9897972531066056,
          "exact_span_recall": 0.010202746893394376,
          "false_negative": 7567,
          "gold_characters": 40598,
          "gold_spans": 7645,
          "label_id": "gender",
          "residual_pii_character_rate": 0.05460860140893642,
          "true_positive": 78
        },
        {
          "character_coverage": 0.9967343247588425,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 138873,
          "exact_span_leakage": 0.7485319189240387,
          "exact_span_recall": 0.25146808107596136,
          "false_negative": 7903,
          "gold_characters": 139328,
          "gold_spans": 10558,
          "label_id": "health_plan_beneficiary_number",
          "residual_pii_character_rate": 0.0032656752411575063,
          "true_positive": 2655
        },
        {
          "character_coverage": 0.9930086012602917,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 410191,
          "exact_span_leakage": 0.9562551103843009,
          "exact_span_recall": 0.0437448896156991,
          "false_negative": 4678,
          "gold_characters": 413079,
          "gold_spans": 4892,
          "label_id": "http_cookie",
          "residual_pii_character_rate": 0.006991398739708332,
          "true_positive": 214
        },
        {
          "character_coverage": 0.9977471590555493,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 80162,
          "exact_span_leakage": 0.0032905561039815456,
          "exact_span_recall": 0.9967094438960185,
          "false_negative": 20,
          "gold_characters": 80343,
          "gold_spans": 6078,
          "label_id": "ipv4",
          "residual_pii_character_rate": 0.0022528409444506625,
          "true_positive": 6058
        },
        {
          "character_coverage": 0.9989329444508666,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 112339,
          "exact_span_leakage": 0.07136030582988218,
          "exact_span_recall": 0.9286396941701178,
          "false_negative": 224,
          "gold_characters": 112459,
          "gold_spans": 3139,
          "label_id": "ipv6",
          "residual_pii_character_rate": 0.0010670555491334444,
          "true_positive": 2915
        },
        {
          "character_coverage": 0.9256300337750065,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 57004,
          "exact_span_leakage": 0.9780372211305052,
          "exact_span_recall": 0.021962778869494857,
          "false_negative": 8461,
          "gold_characters": 61584,
          "gold_spans": 8651,
          "label_id": "language",
          "residual_pii_character_rate": 0.07436996622499348,
          "true_positive": 190
        },
        {
          "character_coverage": 0.9908194114199707,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 383783,
          "exact_span_leakage": 0.9978354251963218,
          "exact_span_recall": 0.002164574803678099,
          "false_negative": 59467,
          "gold_characters": 387339,
          "gold_spans": 59596,
          "label_id": "last_name",
          "residual_pii_character_rate": 0.009180588580029347,
          "true_positive": 129
        },
        {
          "character_coverage": 0.9920966708370206,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32512,
          "exact_span_leakage": 0.832270720941638,
          "exact_span_recall": 0.16772927905836194,
          "false_negative": 3394,
          "gold_characters": 32771,
          "gold_spans": 4078,
          "label_id": "license_plate",
          "residual_pii_character_rate": 0.007903329162979422,
          "true_positive": 684
        },
        {
          "character_coverage": 0.9993633834784174,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 72211,
          "exact_span_leakage": 0.14117647058823535,
          "exact_span_recall": 0.8588235294117647,
          "false_negative": 600,
          "gold_characters": 72257,
          "gold_spans": 4250,
          "label_id": "mac_address",
          "residual_pii_character_rate": 0.0006366165215826447,
          "true_positive": 3650
        },
        {
          "character_coverage": 0.9984779820436084,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 116772,
          "exact_span_leakage": 0.4131375022526581,
          "exact_span_recall": 0.5868624977473419,
          "false_negative": 4585,
          "gold_characters": 116950,
          "gold_spans": 11098,
          "label_id": "medical_record_number",
          "residual_pii_character_rate": 0.0015220179563916192,
          "true_positive": 6513
        },
        {
          "character_coverage": 0.9859262982460781,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 37269,
          "exact_span_leakage": 0.18370214260625217,
          "exact_span_recall": 0.8162978573937478,
          "false_negative": 523,
          "gold_characters": 37801,
          "gold_spans": 2847,
          "label_id": "national_id",
          "residual_pii_character_rate": 0.014073701753921863,
          "true_positive": 2324
        },
        {
          "character_coverage": 0.5968837094341016,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 420193,
          "exact_span_leakage": 0.9646497505963999,
          "exact_span_recall": 0.035350249403600084,
          "false_negative": 35584,
          "gold_characters": 703978,
          "gold_spans": 36888,
          "label_id": "occupation",
          "residual_pii_character_rate": 0.40311629056589837,
          "true_positive": 1304
        },
        {
          "character_coverage": 0.9655950833234842,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 81699,
          "exact_span_leakage": 0.9430207587687902,
          "exact_span_recall": 0.05697924123120973,
          "false_negative": 6587,
          "gold_characters": 84610,
          "gold_spans": 6985,
          "label_id": "password",
          "residual_pii_character_rate": 0.034404916676515795,
          "true_positive": 398
        },
        {
          "character_coverage": 0.9995865575873194,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 307050,
          "exact_span_leakage": 0.2718763058921856,
          "exact_span_recall": 0.7281236941078144,
          "false_negative": 6506,
          "gold_characters": 307177,
          "gold_spans": 23930,
          "label_id": "phone_number",
          "residual_pii_character_rate": 0.000413442412680598,
          "true_positive": 17424
        },
        {
          "character_coverage": 0.9819534100477126,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 34987,
          "exact_span_leakage": 0.019981412639405227,
          "exact_span_recall": 0.9800185873605948,
          "false_negative": 129,
          "gold_characters": 35630,
          "gold_spans": 6456,
          "label_id": "pin",
          "residual_pii_character_rate": 0.018046589952287362,
          "true_positive": 6327
        },
        {
          "character_coverage": 0.8088020225152199,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 71342,
          "exact_span_leakage": 0.9909403311465167,
          "exact_span_recall": 0.009059668853483286,
          "false_negative": 6344,
          "gold_characters": 88207,
          "gold_spans": 6402,
          "label_id": "political_view",
          "residual_pii_character_rate": 0.19119797748478007,
          "true_positive": 58
        },
        {
          "character_coverage": 0.9825448563682335,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32254,
          "exact_span_leakage": 0.06051919095397351,
          "exact_span_recall": 0.9394808090460265,
          "false_negative": 380,
          "gold_characters": 32827,
          "gold_spans": 6279,
          "label_id": "postcode",
          "residual_pii_character_rate": 0.01745514363176648,
          "true_positive": 5899
        },
        {
          "character_coverage": 0.9227660812276922,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 58687,
          "exact_span_leakage": 0.990484429065744,
          "exact_span_recall": 0.009515570934256055,
          "false_negative": 9160,
          "gold_characters": 63599,
          "gold_spans": 9248,
          "label_id": "race_ethnicity",
          "residual_pii_character_rate": 0.07723391877230779,
          "true_positive": 88
        },
        {
          "character_coverage": 0.8846636185499673,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 67721,
          "exact_span_leakage": 0.9880134249640403,
          "exact_span_recall": 0.011986575035959726,
          "false_negative": 6182,
          "gold_characters": 76550,
          "gold_spans": 6257,
          "label_id": "religious_belief",
          "residual_pii_character_rate": 0.11533638145003267,
          "true_positive": 75
        },
        {
          "character_coverage": 0.9257968127490039,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 22308,
          "exact_span_leakage": 0.9966787439613527,
          "exact_span_recall": 0.003321256038647343,
          "false_negative": 3301,
          "gold_characters": 24096,
          "gold_spans": 3312,
          "label_id": "sexuality",
          "residual_pii_character_rate": 0.07420318725099606,
          "true_positive": 11
        },
        {
          "character_coverage": 0.9975249752497525,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 66501,
          "exact_span_leakage": 0.0024744308808973825,
          "exact_span_recall": 0.9975255691191026,
          "false_negative": 15,
          "gold_characters": 66666,
          "gold_spans": 6062,
          "label_id": "ssn",
          "residual_pii_character_rate": 0.002475024750247523,
          "true_positive": 6047
        },
        {
          "character_coverage": 0.9301492255828756,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 118368,
          "exact_span_leakage": 0.9854787640572115,
          "exact_span_recall": 0.014521235942788514,
          "false_negative": 18052,
          "gold_characters": 127257,
          "gold_spans": 18318,
          "label_id": "state",
          "residual_pii_character_rate": 0.0698507744171244,
          "true_positive": 266
        },
        {
          "character_coverage": 0.9865458069665356,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 294185,
          "exact_span_leakage": 0.29101250074056517,
          "exact_span_recall": 0.7089874992594348,
          "false_negative": 4912,
          "gold_characters": 298197,
          "gold_spans": 16879,
          "label_id": "street_address",
          "residual_pii_character_rate": 0.013454193033464401,
          "true_positive": 11967
        },
        {
          "character_coverage": 0.9984071576236148,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 58920,
          "exact_span_leakage": 0.9901061341967979,
          "exact_span_recall": 0.009893865803202015,
          "false_negative": 5504,
          "gold_characters": 59014,
          "gold_spans": 5559,
          "label_id": "swift_bic",
          "residual_pii_character_rate": 0.0015928423763852395,
          "true_positive": 55
        },
        {
          "character_coverage": 0.9887941702186168,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14383,
          "exact_span_leakage": 0.09984639016897079,
          "exact_span_recall": 0.9001536098310292,
          "false_negative": 130,
          "gold_characters": 14546,
          "gold_spans": 1302,
          "label_id": "tax_id",
          "residual_pii_character_rate": 0.01120582978138318,
          "true_positive": 1172
        },
        {
          "character_coverage": 0.7644140455662058,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 158700,
          "exact_span_leakage": 0.26226230310944254,
          "exact_span_recall": 0.7377376968905575,
          "false_negative": 6427,
          "gold_characters": 207610,
          "gold_spans": 24506,
          "label_id": "time",
          "residual_pii_character_rate": 0.23558595443379415,
          "true_positive": 18079
        },
        {
          "character_coverage": 0.9739094055452021,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 51214,
          "exact_span_leakage": 0.4220596510973551,
          "exact_span_recall": 0.5779403489026449,
          "false_negative": 750,
          "gold_characters": 52586,
          "gold_spans": 1777,
          "label_id": "unique_id",
          "residual_pii_character_rate": 0.02609059445479789,
          "true_positive": 1027
        },
        {
          "character_coverage": 0.9981127189263783,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1969485,
          "exact_span_leakage": 0.9684783470288266,
          "exact_span_recall": 0.03152165297117341,
          "false_negative": 36654,
          "gold_characters": 1973209,
          "gold_spans": 37847,
          "label_id": "url",
          "residual_pii_character_rate": 0.0018872810736216827,
          "true_positive": 1193
        },
        {
          "character_coverage": 0.9904533156645133,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 179381,
          "exact_span_leakage": 0.9335443037974683,
          "exact_span_recall": 0.06645569620253164,
          "false_negative": 14750,
          "gold_characters": 181110,
          "gold_spans": 15800,
          "label_id": "user_name",
          "residual_pii_character_rate": 0.009546684335486666,
          "true_positive": 1050
        },
        {
          "character_coverage": 0.9963892777247216,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 76163,
          "exact_span_leakage": 0.7285682964262223,
          "exact_span_recall": 0.27143170357377766,
          "false_negative": 3323,
          "gold_characters": 76439,
          "gold_spans": 4561,
          "label_id": "vehicle_identifier",
          "residual_pii_character_rate": 0.0036107222752783663,
          "true_positive": 1238
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "nemotron-pii",
      "dataset_sha256": "1a4b0512ecb5370f0992d29d0f9c07351e6de13f0d7ea33bb18cecb984780247",
      "documents": 100000,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.34884803073508885,
        "false_negative": 557960,
        "false_positive": 533540,
        "precision": 0.354005230530802,
        "recall": 0.34383893501422963,
        "true_positive": 292380
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/nemotron-domain-slices-v1/openmed-multilingual-v2.json",
        "sha256": "8e466225ab7271b64a1cbdbc985d6d9c4a3d59d89ef5ad1b3e9cfeca32e2d02f"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.16817985300475574,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 389,
          "exact_span_leakage": 0.8235294117647058,
          "exact_span_recall": 0.17647058823529413,
          "false_negative": 210,
          "gold_characters": 2313,
          "gold_spans": 255,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 45
        },
        {
          "character_coverage": 0.9940476190476191,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 8016,
          "exact_span_leakage": 0.005952380952380931,
          "exact_span_recall": 0.9940476190476191,
          "false_negative": 3,
          "gold_characters": 8064,
          "gold_spans": 504,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 501
        },
        {
          "character_coverage": 0.3840181557572862,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 3215,
          "exact_span_leakage": 0.6389452332657201,
          "exact_span_recall": 0.36105476673427994,
          "false_negative": 315,
          "gold_characters": 8372,
          "gold_spans": 493,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 178
        },
        {
          "character_coverage": 0.843207033703161,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 8056,
          "exact_span_leakage": 0.15708812260536398,
          "exact_span_recall": 0.842911877394636,
          "false_negative": 82,
          "gold_characters": 9554,
          "gold_spans": 522,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 440
        },
        {
          "character_coverage": 0.4393305439330544,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 735,
          "exact_span_leakage": 0.5606694560669456,
          "exact_span_recall": 0.4393305439330544,
          "false_negative": 134,
          "gold_characters": 1673,
          "gold_spans": 239,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 105
        },
        {
          "character_coverage": 0.6807692307692308,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1593,
          "exact_span_leakage": 0.3192307692307692,
          "exact_span_recall": 0.6807692307692308,
          "false_negative": 83,
          "gold_characters": 2340,
          "gold_spans": 260,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 177
        },
        {
          "character_coverage": 0.09808458670780305,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 763,
          "exact_span_leakage": 0.8956810631229236,
          "exact_span_recall": 0.10431893687707641,
          "false_negative": 2696,
          "gold_characters": 7779,
          "gold_spans": 3010,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 314
        },
        {
          "character_coverage": 0.9139464712567107,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 23323,
          "exact_span_leakage": 0.08359133126934981,
          "exact_span_recall": 0.9164086687306502,
          "false_negative": 189,
          "gold_characters": 25519,
          "gold_spans": 2261,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 2072
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-chat",
      "dataset_sha256": "4ddacd78dfac49b723032de5e1c29c432a44049245e03a8ea960c86bbf3b765c",
      "documents": 3000,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.4847258237935615,
        "false_negative": 3712,
        "false_positive": 4435,
        "precision": 0.46352969638321034,
        "recall": 0.5079533404029692,
        "true_positive": 3832
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/multilingual-chat.json",
        "sha256": "486fcc65f28cfa99fdf6a7ee99c23f95204d0c123a79d0bbb784af5463f56069"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.032547496462228646,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1081,
          "exact_span_leakage": 0.967109867039888,
          "exact_span_recall": 0.032890132960111965,
          "false_negative": 1382,
          "gold_characters": 33213,
          "gold_spans": 1429,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 47
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 13792,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 13792,
          "gold_spans": 862,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 862
        },
        {
          "character_coverage": 0.98807063131164,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 16731,
          "exact_span_leakage": 0.011815252416756183,
          "exact_span_recall": 0.9881847475832438,
          "false_negative": 11,
          "gold_characters": 16933,
          "gold_spans": 931,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 920
        },
        {
          "character_coverage": 0.9988962472406181,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 32580,
          "exact_span_leakage": 0.001103752759381904,
          "exact_span_recall": 0.9988962472406181,
          "false_negative": 2,
          "gold_characters": 32616,
          "gold_spans": 1812,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 1810
        },
        {
          "character_coverage": 0.4703296703296703,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1498,
          "exact_span_leakage": 0.5296703296703297,
          "exact_span_recall": 0.4703296703296703,
          "false_negative": 241,
          "gold_characters": 3185,
          "gold_spans": 455,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 214
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 4167,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 4167,
          "gold_spans": 463,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 463
        },
        {
          "character_coverage": 0.06635520840937295,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 909,
          "exact_span_leakage": 0.9267419962335217,
          "exact_span_recall": 0.07325800376647834,
          "false_negative": 4921,
          "gold_characters": 13699,
          "gold_spans": 5310,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 389
        },
        {
          "character_coverage": 0.9972727272727273,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 48268,
          "exact_span_leakage": 0.0027272727272726893,
          "exact_span_recall": 0.9972727272727273,
          "false_negative": 12,
          "gold_characters": 48400,
          "gold_spans": 4400,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 4388
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-formal",
      "dataset_sha256": "bcaa54909fd69c6737b8e442c4d740ce1bf5b95b7d4af70b2f2692010e1ff7a5",
      "documents": 5000,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.4679274411424161,
        "false_negative": 6569,
        "false_positive": 14110,
        "precision": 0.39188897987329224,
        "recall": 0.580577193206487,
        "true_positive": 9093
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/multilingual-formal.json",
        "sha256": "a4751242ee567c1b68a77cc80912fbe485bf0f575391e4dded6b46f57c372a73"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 761,
          "exact_span_leakage": 0.5384615384615384,
          "exact_span_recall": 0.46153846153846156,
          "false_negative": 21,
          "gold_characters": 761,
          "gold_spans": 39,
          "label_id": "EMAIL",
          "residual_pii_character_rate": 0.0,
          "true_positive": 18
        },
        {
          "character_coverage": 0.926829268292683,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 874,
          "exact_span_leakage": 0.23076923076923073,
          "exact_span_recall": 0.7692307692307693,
          "false_negative": 18,
          "gold_characters": 943,
          "gold_spans": 78,
          "label_id": "ID_NUM",
          "residual_pii_character_rate": 0.07317073170731703,
          "true_positive": 60
        },
        {
          "character_coverage": 0.9664403238443458,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14802,
          "exact_span_leakage": 0.9509157509157509,
          "exact_span_recall": 0.04908424908424908,
          "false_negative": 1298,
          "gold_characters": 15316,
          "gold_spans": 1365,
          "label_id": "NAME_STUDENT",
          "residual_pii_character_rate": 0.03355967615565425,
          "true_positive": 67
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 95,
          "exact_span_leakage": 0.8333333333333334,
          "exact_span_recall": 0.16666666666666666,
          "false_negative": 5,
          "gold_characters": 95,
          "gold_spans": 6,
          "label_id": "PHONE_NUM",
          "residual_pii_character_rate": 0.0,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9052631578947369,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 86,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 2,
          "gold_characters": 95,
          "gold_spans": 2,
          "label_id": "STREET_ADDRESS",
          "residual_pii_character_rate": 0.09473684210526312,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9955347650435892,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4682,
          "exact_span_leakage": 0.6090909090909091,
          "exact_span_recall": 0.39090909090909093,
          "false_negative": 67,
          "gold_characters": 4703,
          "gold_spans": 110,
          "label_id": "URL_PERSONAL",
          "residual_pii_character_rate": 0.004465234956410824,
          "true_positive": 43
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 61,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 3,
          "gold_characters": 61,
          "gold_spans": 6,
          "label_id": "USERNAME",
          "residual_pii_character_rate": 0.0,
          "true_positive": 3
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "piilo",
      "dataset_sha256": "8276cd44f3b2eb357dfb405b3c5d8e9f821388e984cbf66e92e7df03f1b13117",
      "documents": 6807,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.017070460102244944,
        "false_negative": 1414,
        "false_positive": 20697,
        "precision": 0.009191440471061324,
        "recall": 0.11955168119551682,
        "true_positive": 192
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-piilo.json",
        "sha256": "1a521755c2dcdc7aca2be88336043ab805889ea9c47bddb247babd1c059b9976"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.6490802348336595,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 8292,
          "exact_span_leakage": 0.4137717121588089,
          "exact_span_recall": 0.5862282878411911,
          "false_negative": 667,
          "gold_characters": 12775,
          "gold_spans": 1612,
          "label_id": "CODE",
          "residual_pii_character_rate": 0.3509197651663405,
          "true_positive": 945
        },
        {
          "character_coverage": 0.8975707138338687,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 109255,
          "exact_span_leakage": 0.2371015100671141,
          "exact_span_recall": 0.7628984899328859,
          "false_negative": 2261,
          "gold_characters": 121723,
          "gold_spans": 9536,
          "label_id": "DATETIME",
          "residual_pii_character_rate": 0.10242928616613134,
          "true_positive": 7275
        },
        {
          "character_coverage": 0.3297640653357532,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3634,
          "exact_span_leakage": 0.9705561613958561,
          "exact_span_recall": 0.029443838604143947,
          "false_negative": 890,
          "gold_characters": 11020,
          "gold_spans": 917,
          "label_id": "DEM",
          "residual_pii_character_rate": 0.6702359346642468,
          "true_positive": 27
        },
        {
          "character_coverage": 0.8640962150689777,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 12214,
          "exact_span_leakage": 0.9883481836874571,
          "exact_span_recall": 0.011651816312542838,
          "false_negative": 1442,
          "gold_characters": 14135,
          "gold_spans": 1459,
          "label_id": "LOC",
          "residual_pii_character_rate": 0.1359037849310223,
          "true_positive": 17
        },
        {
          "character_coverage": 0.08043905635648754,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 982,
          "exact_span_leakage": 0.9776536312849162,
          "exact_span_recall": 0.0223463687150838,
          "false_negative": 525,
          "gold_characters": 12208,
          "gold_spans": 537,
          "label_id": "MISC",
          "residual_pii_character_rate": 0.9195609436435125,
          "true_positive": 12
        },
        {
          "character_coverage": 0.40612298759567167,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 15388,
          "exact_span_leakage": 0.9691516709511568,
          "exact_span_recall": 0.030848329048843187,
          "false_negative": 1885,
          "gold_characters": 37890,
          "gold_spans": 1945,
          "label_id": "ORG",
          "residual_pii_character_rate": 0.5938770124043283,
          "true_positive": 60
        },
        {
          "character_coverage": 0.8742344950450952,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 47109,
          "exact_span_leakage": 0.9987916868052199,
          "exact_span_recall": 0.001208313194780087,
          "false_negative": 4133,
          "gold_characters": 53886,
          "gold_spans": 4138,
          "label_id": "PERSON",
          "residual_pii_character_rate": 0.1257655049549048,
          "true_positive": 5
        },
        {
          "character_coverage": 0.03458049886621315,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 305,
          "exact_span_leakage": 0.9909638554216867,
          "exact_span_recall": 0.009036144578313253,
          "false_negative": 658,
          "gold_characters": 8820,
          "gold_spans": 664,
          "label_id": "QUANTITY",
          "residual_pii_character_rate": 0.9654195011337868,
          "true_positive": 6
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tab",
      "dataset_sha256": "cd0f0f15f84a8739654c7cf30c6be8ce27b051ef73974d39d792a0cb8c846379",
      "documents": 555,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.34611158335579373,
        "false_negative": 12461,
        "false_positive": 19078,
        "precision": 0.3043573381950775,
        "recall": 0.4011437908496732,
        "true_positive": 8347
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-tab.json",
        "sha256": "117d002a1a649dfa6ef3f80675f6d52ddadd0f457dc646334e332a884af3eb2e"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4354,
          "exact_span_leakage": 0.05033557046979864,
          "exact_span_recall": 0.9496644295302014,
          "false_negative": 15,
          "gold_characters": 4354,
          "gold_spans": 298,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0,
          "true_positive": 283
        },
        {
          "character_coverage": 0.6807137433561123,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3586,
          "exact_span_leakage": 0.9896907216494846,
          "exact_span_recall": 0.010309278350515464,
          "false_negative": 288,
          "gold_characters": 5268,
          "gold_spans": 291,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.31928625664388766,
          "true_positive": 3
        },
        {
          "character_coverage": 0.8656271497362991,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 3775,
          "exact_span_leakage": 0.3744292237442922,
          "exact_span_recall": 0.6255707762557078,
          "false_negative": 164,
          "gold_characters": 4361,
          "gold_spans": 438,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.13437285026370094,
          "true_positive": 274
        },
        {
          "character_coverage": 0.9936572370924774,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7833,
          "exact_span_leakage": 0.24281150159744413,
          "exact_span_recall": 0.7571884984025559,
          "false_negative": 76,
          "gold_characters": 7883,
          "gold_spans": 313,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.006342762907522559,
          "true_positive": 237
        },
        {
          "character_coverage": 0.5916442048517521,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2634,
          "exact_span_leakage": 0.9135446685878963,
          "exact_span_recall": 0.08645533141210375,
          "false_negative": 1268,
          "gold_characters": 4452,
          "gold_spans": 1388,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.40835579514824794,
          "true_positive": 120
        },
        {
          "character_coverage": 0.9990229604298974,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4090,
          "exact_span_leakage": 0.0029069767441860517,
          "exact_span_recall": 0.997093023255814,
          "false_negative": 1,
          "gold_characters": 4094,
          "gold_spans": 344,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.0009770395701026446,
          "true_positive": 343
        },
        {
          "character_coverage": 0.88726079825041,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 8114,
          "exact_span_leakage": 0.8624338624338624,
          "exact_span_recall": 0.13756613756613756,
          "false_negative": 163,
          "gold_characters": 9145,
          "gold_spans": 189,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.11273920174958996,
          "true_positive": 26
        },
        {
          "character_coverage": 0.650415215644254,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2428,
          "exact_span_leakage": 0.44680851063829785,
          "exact_span_recall": 0.5531914893617021,
          "false_negative": 105,
          "gold_characters": 3733,
          "gold_spans": 235,
          "label_id": "secret",
          "residual_pii_character_rate": 0.349584784355746,
          "true_positive": 130
        },
        {
          "character_coverage": 0.9783783783783784,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1448,
          "exact_span_leakage": 0.021621621621621623,
          "exact_span_recall": 0.9783783783783784,
          "false_negative": 4,
          "gold_characters": 1480,
          "gold_spans": 185,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.021621621621621623,
          "true_positive": 181
        },
        {
          "character_coverage": 0.9904761904761905,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1040,
          "exact_span_leakage": 0.5047619047619047,
          "exact_span_recall": 0.49523809523809526,
          "false_negative": 53,
          "gold_characters": 1050,
          "gold_spans": 105,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.00952380952380949,
          "true_positive": 52
        },
        {
          "character_coverage": 0.9936254980079682,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1247,
          "exact_span_leakage": 0.3581081081081081,
          "exact_span_recall": 0.6418918918918919,
          "false_negative": 53,
          "gold_characters": 1255,
          "gold_spans": 148,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.0063745019920318224,
          "true_positive": 95
        },
        {
          "character_coverage": 0.9594680177327423,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1515,
          "exact_span_leakage": 0.5656565656565656,
          "exact_span_recall": 0.43434343434343436,
          "false_negative": 112,
          "gold_characters": 1579,
          "gold_spans": 198,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.04053198226725774,
          "true_positive": 86
        },
        {
          "character_coverage": 0.8145670754366406,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2192,
          "exact_span_leakage": 0.6331658291457287,
          "exact_span_recall": 0.36683417085427134,
          "false_negative": 126,
          "gold_characters": 2691,
          "gold_spans": 199,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.18543292456335936,
          "true_positive": 73
        },
        {
          "character_coverage": 0.09491525423728814,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 140,
          "exact_span_leakage": 0.9794520547945206,
          "exact_span_recall": 0.02054794520547945,
          "false_negative": 143,
          "gold_characters": 1475,
          "gold_spans": 146,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.9050847457627118,
          "true_positive": 3
        },
        {
          "character_coverage": 0.9855769230769231,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1230,
          "exact_span_leakage": 0.4173228346456693,
          "exact_span_recall": 0.5826771653543307,
          "false_negative": 53,
          "gold_characters": 1248,
          "gold_spans": 127,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.014423076923076872,
          "true_positive": 74
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1940,
          "exact_span_leakage": 0.2731958762886598,
          "exact_span_recall": 0.7268041237113402,
          "false_negative": 53,
          "gold_characters": 1940,
          "gold_spans": 194,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 141
        },
        {
          "character_coverage": 0.9925925925925926,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1608,
          "exact_span_leakage": 0.007407407407407418,
          "exact_span_recall": 0.9925925925925926,
          "false_negative": 1,
          "gold_characters": 1620,
          "gold_spans": 135,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.007407407407407418,
          "true_positive": 134
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1260,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 1260,
          "gold_spans": 140,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.0,
          "true_positive": 140
        },
        {
          "character_coverage": 0.7545401288810779,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1288,
          "exact_span_leakage": 0.7862068965517242,
          "exact_span_recall": 0.21379310344827587,
          "false_negative": 114,
          "gold_characters": 1707,
          "gold_spans": 145,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.2454598711189221,
          "true_positive": 31
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-long",
      "dataset_sha256": "b30ca2a1012254f164893d1aad491a778d2fe49cfef71239f8c934e0544883e2",
      "documents": 300,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.2141879662737827,
        "false_negative": 2792,
        "false_positive": 15009,
        "precision": 0.13914539718956123,
        "recall": 0.4649290916059793,
        "true_positive": 2426
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-tw-long.json",
        "sha256": "878fb363f28b421268a951fe23ccea34fa832021d7129dbd7e4e80b8b4425949"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1673,
          "exact_span_leakage": 0.05932203389830504,
          "exact_span_recall": 0.940677966101695,
          "false_negative": 7,
          "gold_characters": 1673,
          "gold_spans": 118,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0,
          "true_positive": 111
        },
        {
          "character_coverage": 0.7164179104477612,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1440,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 115,
          "gold_characters": 2010,
          "gold_spans": 115,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.28358208955223885,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8432147562582345,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1280,
          "exact_span_leakage": 0.4683544303797469,
          "exact_span_recall": 0.5316455696202531,
          "false_negative": 74,
          "gold_characters": 1518,
          "gold_spans": 158,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.1567852437417655,
          "true_positive": 84
        },
        {
          "character_coverage": 0.9853691731881592,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2896,
          "exact_span_leakage": 0.44999999999999996,
          "exact_span_recall": 0.55,
          "false_negative": 54,
          "gold_characters": 2939,
          "gold_spans": 120,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.014630826811840802,
          "true_positive": 66
        },
        {
          "character_coverage": 0.5673940949935815,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 884,
          "exact_span_leakage": 0.939922480620155,
          "exact_span_recall": 0.060077519379844964,
          "false_negative": 485,
          "gold_characters": 1558,
          "gold_spans": 516,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.4326059050064185,
          "true_positive": 31
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2155,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 2155,
          "gold_spans": 180,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.0,
          "true_positive": 180
        },
        {
          "character_coverage": 0.8988095238095238,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2869,
          "exact_span_leakage": 0.8493150684931507,
          "exact_span_recall": 0.1506849315068493,
          "false_negative": 62,
          "gold_characters": 3192,
          "gold_spans": 73,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.10119047619047616,
          "true_positive": 11
        },
        {
          "character_coverage": 0.8419811320754716,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 714,
          "exact_span_leakage": 0.2530120481927711,
          "exact_span_recall": 0.7469879518072289,
          "false_negative": 21,
          "gold_characters": 848,
          "gold_spans": 83,
          "label_id": "secret",
          "residual_pii_character_rate": 0.15801886792452835,
          "true_positive": 62
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 624,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 624,
          "gold_spans": 78,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 78
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 660,
          "exact_span_leakage": 0.6515151515151515,
          "exact_span_recall": 0.3484848484848485,
          "false_negative": 43,
          "gold_characters": 660,
          "gold_spans": 66,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.0,
          "true_positive": 23
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 592,
          "exact_span_leakage": 0.4285714285714286,
          "exact_span_recall": 0.5714285714285714,
          "false_negative": 30,
          "gold_characters": 592,
          "gold_spans": 70,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.0,
          "true_positive": 40
        },
        {
          "character_coverage": 0.9815950920245399,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 800,
          "exact_span_leakage": 0.5882352941176471,
          "exact_span_recall": 0.4117647058823529,
          "false_negative": 60,
          "gold_characters": 815,
          "gold_spans": 102,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.018404907975460127,
          "true_positive": 42
        },
        {
          "character_coverage": 0.8736842105263158,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1079,
          "exact_span_leakage": 0.6774193548387097,
          "exact_span_recall": 0.3225806451612903,
          "false_negative": 63,
          "gold_characters": 1235,
          "gold_spans": 93,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.12631578947368416,
          "true_positive": 30
        },
        {
          "character_coverage": 0.19207683073229292,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 160,
          "exact_span_leakage": 0.9878048780487805,
          "exact_span_recall": 0.012195121951219513,
          "false_negative": 81,
          "gold_characters": 833,
          "gold_spans": 82,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.8079231692677071,
          "true_positive": 1
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 747,
          "exact_span_leakage": 0.48,
          "exact_span_recall": 0.52,
          "false_negative": 36,
          "gold_characters": 747,
          "gold_spans": 75,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 39
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 890,
          "exact_span_leakage": 0.4831460674157303,
          "exact_span_recall": 0.5168539325842697,
          "false_negative": 43,
          "gold_characters": 890,
          "gold_spans": 89,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 46
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 924,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 924,
          "gold_spans": 77,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.0,
          "true_positive": 77
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 531,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 531,
          "gold_spans": 59,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.0,
          "true_positive": 59
        },
        {
          "character_coverage": 0.7952443857331571,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 602,
          "exact_span_leakage": 0.8970588235294118,
          "exact_span_recall": 0.10294117647058823,
          "false_negative": 61,
          "gold_characters": 757,
          "gold_spans": 68,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.20475561426684286,
          "true_positive": 7
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-mid",
      "dataset_sha256": "5d05e1c7fc6adaf7c66591de0d76e5ac5d175fb182f67a19da6d23896b3b02e0",
      "documents": 300,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.3230240549828179,
        "false_negative": 1235,
        "false_positive": 2902,
        "precision": 0.2537927487786063,
        "recall": 0.4441944194419442,
        "true_positive": 987
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-tw-mid.json",
        "sha256": "4d1b4460f24a3fcd6beacee6806166d710d67e0028d3c1a98a8d7166b83d1ecb"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 504,
          "exact_span_leakage": 0.11363636363636365,
          "exact_span_recall": 0.8863636363636364,
          "false_negative": 5,
          "gold_characters": 504,
          "gold_spans": 44,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0,
          "true_positive": 39
        },
        {
          "character_coverage": 0.7872044506258693,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 566,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 37,
          "gold_characters": 719,
          "gold_spans": 37,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.2127955493741307,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8516129032258064,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 264,
          "exact_span_leakage": 0.368421052631579,
          "exact_span_recall": 0.631578947368421,
          "false_negative": 14,
          "gold_characters": 310,
          "gold_spans": 38,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.1483870967741936,
          "true_positive": 24
        },
        {
          "character_coverage": 0.9905956112852664,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 632,
          "exact_span_leakage": 0.8214285714285714,
          "exact_span_recall": 0.17857142857142858,
          "false_negative": 23,
          "gold_characters": 638,
          "gold_spans": 28,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.009404388714733591,
          "true_positive": 5
        },
        {
          "character_coverage": 0.6504559270516718,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 214,
          "exact_span_leakage": 0.9318181818181819,
          "exact_span_recall": 0.06818181818181818,
          "false_negative": 82,
          "gold_characters": 329,
          "gold_spans": 88,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.34954407294832823,
          "true_positive": 6
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 417,
          "exact_span_leakage": 0.027027027027026973,
          "exact_span_recall": 0.972972972972973,
          "false_negative": 1,
          "gold_characters": 417,
          "gold_spans": 37,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.0,
          "true_positive": 36
        },
        {
          "character_coverage": 0.9941348973607038,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 678,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 20,
          "gold_characters": 682,
          "gold_spans": 20,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.0058651026392961825,
          "true_positive": 0
        },
        {
          "character_coverage": 0.6167664670658682,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 206,
          "exact_span_leakage": 0.34782608695652173,
          "exact_span_recall": 0.6521739130434783,
          "false_negative": 8,
          "gold_characters": 334,
          "gold_spans": 23,
          "label_id": "secret",
          "residual_pii_character_rate": 0.38323353293413176,
          "true_positive": 15
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 104,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 104,
          "gold_spans": 13,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 13
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 110,
          "exact_span_leakage": 0.8181818181818181,
          "exact_span_recall": 0.18181818181818182,
          "false_negative": 9,
          "gold_characters": 110,
          "gold_spans": 11,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.0,
          "true_positive": 2
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 143,
          "exact_span_leakage": 0.06666666666666665,
          "exact_span_recall": 0.9333333333333333,
          "false_negative": 1,
          "gold_characters": 143,
          "gold_spans": 15,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.0,
          "true_positive": 14
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 142,
          "exact_span_leakage": 0.7222222222222222,
          "exact_span_recall": 0.2777777777777778,
          "false_negative": 13,
          "gold_characters": 142,
          "gold_spans": 18,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.0,
          "true_positive": 5
        },
        {
          "character_coverage": 0.8761609907120743,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 283,
          "exact_span_leakage": 0.25,
          "exact_span_recall": 0.75,
          "false_negative": 5,
          "gold_characters": 323,
          "gold_spans": 20,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.12383900928792568,
          "true_positive": 15
        },
        {
          "character_coverage": 0.368,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 46,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 12,
          "gold_characters": 125,
          "gold_spans": 12,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.632,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8135593220338984,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 96,
          "exact_span_leakage": 0.7272727272727273,
          "exact_span_recall": 0.2727272727272727,
          "false_negative": 8,
          "gold_characters": 118,
          "gold_spans": 11,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.18644067796610164,
          "true_positive": 3
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 210,
          "exact_span_leakage": 0.23809523809523814,
          "exact_span_recall": 0.7619047619047619,
          "false_negative": 5,
          "gold_characters": 210,
          "gold_spans": 21,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 16
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 156,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 156,
          "gold_spans": 13,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.0,
          "true_positive": 13
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 108,
          "exact_span_leakage": 0.0,
          "exact_span_recall": 1.0,
          "false_negative": 0,
          "gold_characters": 108,
          "gold_spans": 12,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.0,
          "true_positive": 12
        },
        {
          "character_coverage": 0.6866666666666666,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 103,
          "exact_span_leakage": 0.7857142857142857,
          "exact_span_recall": 0.21428571428571427,
          "false_negative": 11,
          "gold_characters": 150,
          "gold_spans": 14,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.31333333333333335,
          "true_positive": 3
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-short",
      "dataset_sha256": "03e4850a0a97a2b5cb046c1dcd3996b42c5eea3766ae2618d4bbf64391e73693",
      "documents": 310,
      "model_id": "openmed-privacy-filter-multilingual-v2",
      "overall": {
        "f1": 0.3646864686468647,
        "false_negative": 254,
        "false_positive": 516,
        "precision": 0.29986431478968795,
        "recall": 0.4652631578947368,
        "true_positive": 221
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-multilingual-v2-tw-short.json",
        "sha256": "637f039b70075d80cc33a81bb9614e585e1ddaeea0260e744dea7b9c68169541"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.7007031791621273,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7075,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 413,
          "gold_characters": 10097,
          "gold_spans": 413,
          "label_id": "CALLE",
          "residual_pii_character_rate": 0.2992968208378727,
          "true_positive": 0
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 6,
          "gold_characters": 168,
          "gold_spans": 6,
          "label_id": "CENTRO_SALUD",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9908491947291361,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5414,
          "exact_span_leakage": 0.9799196787148594,
          "exact_span_recall": 0.020080321285140562,
          "false_negative": 244,
          "gold_characters": 5464,
          "gold_spans": 249,
          "label_id": "CORREO_ELECTRONICO",
          "residual_pii_character_rate": 0.009150805270863871,
          "true_positive": 5
        },
        {
          "character_coverage": 0.27175141242937856,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 962,
          "exact_span_leakage": 0.9594594594594594,
          "exact_span_recall": 0.04054054054054054,
          "false_negative": 497,
          "gold_characters": 3540,
          "gold_spans": 518,
          "label_id": "EDAD_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.7282485875706215,
          "true_positive": 21
        },
        {
          "character_coverage": 0.013679890560875513,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 10,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 81,
          "gold_characters": 731,
          "gold_spans": 81,
          "label_id": "FAMILIARES_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.9863201094391245,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8489128709728615,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5349,
          "exact_span_leakage": 0.1669394435351882,
          "exact_span_recall": 0.8330605564648118,
          "false_negative": 102,
          "gold_characters": 6301,
          "gold_spans": 611,
          "label_id": "FECHAS",
          "residual_pii_character_rate": 0.1510871290271385,
          "true_positive": 509
        },
        {
          "character_coverage": 0.3317295517852621,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1310,
          "exact_span_leakage": 0.9923076923076923,
          "exact_span_recall": 0.007692307692307693,
          "false_negative": 129,
          "gold_characters": 3949,
          "gold_spans": 130,
          "label_id": "HOSPITAL",
          "residual_pii_character_rate": 0.6682704482147379,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9630434782608696,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2658,
          "exact_span_leakage": 0.1767676767676768,
          "exact_span_recall": 0.8232323232323232,
          "false_negative": 35,
          "gold_characters": 2760,
          "gold_spans": 198,
          "label_id": "ID_ASEGURAMIENTO",
          "residual_pii_character_rate": 0.036956521739130443,
          "true_positive": 163
        },
        {
          "character_coverage": 0.9375,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 285,
          "exact_span_leakage": 0.07692307692307687,
          "exact_span_recall": 0.9230769230769231,
          "false_negative": 3,
          "gold_characters": 304,
          "gold_spans": 39,
          "label_id": "ID_CONTACTO_ASISTENCIAL",
          "residual_pii_character_rate": 0.0625,
          "true_positive": 36
        },
        {
          "character_coverage": 0.9128024193548387,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1811,
          "exact_span_leakage": 0.13427561837455826,
          "exact_span_recall": 0.8657243816254417,
          "false_negative": 38,
          "gold_characters": 1984,
          "gold_spans": 283,
          "label_id": "ID_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.08719758064516125,
          "true_positive": 245
        },
        {
          "character_coverage": 0.9556093623890234,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2368,
          "exact_span_leakage": 0.08119658119658124,
          "exact_span_recall": 0.9188034188034188,
          "false_negative": 19,
          "gold_characters": 2478,
          "gold_spans": 234,
          "label_id": "ID_TITULACION_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.04439063761097661,
          "true_positive": 215
        },
        {
          "character_coverage": 0.3008356545961003,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 432,
          "exact_span_leakage": 0.9104477611940298,
          "exact_span_recall": 0.08955223880597014,
          "false_negative": 61,
          "gold_characters": 1436,
          "gold_spans": 67,
          "label_id": "INSTITUCION",
          "residual_pii_character_rate": 0.6991643454038997,
          "true_positive": 6
        },
        {
          "character_coverage": 0.9877603215199123,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 10814,
          "exact_span_leakage": 0.998003992015968,
          "exact_span_recall": 0.001996007984031936,
          "false_negative": 500,
          "gold_characters": 10948,
          "gold_spans": 501,
          "label_id": "NOMBRE_PERSONAL_SANITARIO",
          "residual_pii_character_rate": 0.012239678480087668,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9982359858878871,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5093,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 502,
          "gold_characters": 5102,
          "gold_spans": 502,
          "label_id": "NOMBRE_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.0017640141121129105,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8461538461538461,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 66,
          "exact_span_leakage": 0.2857142857142857,
          "exact_span_recall": 0.7142857142857143,
          "false_negative": 2,
          "gold_characters": 78,
          "gold_spans": 7,
          "label_id": "NUMERO_FAX",
          "residual_pii_character_rate": 0.15384615384615385,
          "true_positive": 5
        },
        {
          "character_coverage": 0.9571428571428572,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 268,
          "exact_span_leakage": 0.07692307692307687,
          "exact_span_recall": 0.9230769230769231,
          "false_negative": 2,
          "gold_characters": 280,
          "gold_spans": 26,
          "label_id": "NUMERO_TELEFONO",
          "residual_pii_character_rate": 0.042857142857142816,
          "true_positive": 24
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 7,
          "gold_characters": 86,
          "gold_spans": 7,
          "label_id": "OTROS_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.05329153605015674,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 119,
          "exact_span_leakage": 0.9944903581267217,
          "exact_span_recall": 0.005509641873278237,
          "false_negative": 361,
          "gold_characters": 2233,
          "gold_spans": 363,
          "label_id": "PAIS",
          "residual_pii_character_rate": 0.9467084639498433,
          "true_positive": 2
        },
        {
          "character_coverage": 0.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 0,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 9,
          "gold_characters": 160,
          "gold_spans": 9,
          "label_id": "PROFESION",
          "residual_pii_character_rate": 1.0,
          "true_positive": 0
        },
        {
          "character_coverage": 0.20771929824561403,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 296,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 461,
          "gold_characters": 1425,
          "gold_spans": 461,
          "label_id": "SEXO_SUJETO_ASISTENCIA",
          "residual_pii_character_rate": 0.792280701754386,
          "true_positive": 0
        },
        {
          "character_coverage": 0.8965300675145235,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 5710,
          "exact_span_leakage": 0.5763598326359833,
          "exact_span_recall": 0.4236401673640167,
          "false_negative": 551,
          "gold_characters": 6369,
          "gold_spans": 956,
          "label_id": "TERRITORIO",
          "residual_pii_character_rate": 0.10346993248547653,
          "true_positive": 405
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "meddocan",
      "dataset_sha256": "68c972fd2fd5ffa202072c6d048c111af2a31622e1845ff999c83fe315558c2a",
      "documents": 250,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.258931394245969,
        "false_negative": 4023,
        "false_positive": 5353,
        "precision": 0.2343012444571592,
        "recall": 0.2893481717011129,
        "true_positive": 1638
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-meddocan.json",
        "sha256": "8697dc62c1650297acfecd47bc4cd977f0ec6f43a69ad4efef848fa8716c284e"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.9971546450419689,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 210270,
          "exact_span_leakage": 0.2931767806865153,
          "exact_span_recall": 0.7068232193134847,
          "false_negative": 4894,
          "gold_characters": 210870,
          "gold_spans": 16693,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0028453549580310655,
          "true_positive": 11799
        },
        {
          "character_coverage": 0.9252785515320334,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 13287,
          "exact_span_leakage": 0.07855448583786806,
          "exact_span_recall": 0.9214455141621319,
          "false_negative": 563,
          "gold_characters": 14360,
          "gold_spans": 7167,
          "label_id": "age",
          "residual_pii_character_rate": 0.07472144846796658,
          "true_positive": 6604
        },
        {
          "character_coverage": 0.99457742992078,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 208909,
          "exact_span_leakage": 0.8317977287336619,
          "exact_span_recall": 0.16820227126633813,
          "false_negative": 3882,
          "gold_characters": 210048,
          "gold_spans": 4667,
          "label_id": "api_key",
          "residual_pii_character_rate": 0.005422570079219979,
          "true_positive": 785
        },
        {
          "character_coverage": 0.9989221556886227,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 75069,
          "exact_span_leakage": 0.0011970313622217121,
          "exact_span_recall": 0.9988029686377783,
          "false_negative": 10,
          "gold_characters": 75150,
          "gold_spans": 8354,
          "label_id": "bank_routing_number",
          "residual_pii_character_rate": 0.0010778443113772518,
          "true_positive": 8344
        },
        {
          "character_coverage": 0.9991328383285458,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 147480,
          "exact_span_leakage": 0.9760963177783637,
          "exact_span_recall": 0.023903682221636347,
          "false_negative": 11107,
          "gold_characters": 147608,
          "gold_spans": 11379,
          "label_id": "biometric_identifier",
          "residual_pii_character_rate": 0.0008671616714541752,
          "true_positive": 272
        },
        {
          "character_coverage": 0.9672709575374133,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 23850,
          "exact_span_leakage": 0.9963892399350063,
          "exact_span_recall": 0.003610760064993681,
          "false_negative": 5519,
          "gold_characters": 24657,
          "gold_spans": 5539,
          "label_id": "blood_type",
          "residual_pii_character_rate": 0.03272904246258668,
          "true_positive": 20
        },
        {
          "character_coverage": 0.9971883786316776,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32984,
          "exact_span_leakage": 0.6449033977348435,
          "exact_span_recall": 0.35509660226515655,
          "false_negative": 1936,
          "gold_characters": 33077,
          "gold_spans": 3002,
          "label_id": "certificate_license_number",
          "residual_pii_character_rate": 0.002811621368322359,
          "true_positive": 1066
        },
        {
          "character_coverage": 0.9558902814848341,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 143612,
          "exact_span_leakage": 0.96822368779637,
          "exact_span_recall": 0.031776312203630024,
          "false_negative": 17764,
          "gold_characters": 150239,
          "gold_spans": 18347,
          "label_id": "city",
          "residual_pii_character_rate": 0.044109718515165874,
          "true_positive": 583
        },
        {
          "character_coverage": 0.9102339457538831,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 934529,
          "exact_span_leakage": 0.7547458832540073,
          "exact_span_recall": 0.24525411674599268,
          "false_negative": 41388,
          "gold_characters": 1026691,
          "gold_spans": 54837,
          "label_id": "company_name",
          "residual_pii_character_rate": 0.08976605424611694,
          "true_positive": 13449
        },
        {
          "character_coverage": 0.9982054086625055,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 152407,
          "exact_span_leakage": 0.23723723723723722,
          "exact_span_recall": 0.7627627627627628,
          "false_negative": 1817,
          "gold_characters": 152681,
          "gold_spans": 7659,
          "label_id": "coordinate",
          "residual_pii_character_rate": 0.0017945913374944888,
          "true_positive": 5842
        },
        {
          "character_coverage": 0.974117258511509,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 135677,
          "exact_span_leakage": 0.9773801916932907,
          "exact_span_recall": 0.022619808306709264,
          "false_negative": 22944,
          "gold_characters": 139282,
          "gold_spans": 23475,
          "label_id": "country",
          "residual_pii_character_rate": 0.025882741488491012,
          "true_positive": 531
        },
        {
          "character_coverage": 0.9599431066419688,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 116084,
          "exact_span_leakage": 0.9612522150029533,
          "exact_span_recall": 0.03874778499704666,
          "false_negative": 8137,
          "gold_characters": 120928,
          "gold_spans": 8465,
          "label_id": "county",
          "residual_pii_character_rate": 0.0400568933580312,
          "true_positive": 328
        },
        {
          "character_coverage": 0.9994414106465463,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 239756,
          "exact_span_leakage": 0.0025647003963628068,
          "exact_span_recall": 0.9974352996036372,
          "false_negative": 33,
          "gold_characters": 239890,
          "gold_spans": 12867,
          "label_id": "credit_debit_card",
          "residual_pii_character_rate": 0.0005585893534536535,
          "true_positive": 12834
        },
        {
          "character_coverage": 0.9983434058348928,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 195258,
          "exact_span_leakage": 0.46707638279192276,
          "exact_span_recall": 0.5329236172080772,
          "false_negative": 9576,
          "gold_characters": 195582,
          "gold_spans": 20502,
          "label_id": "customer_id",
          "residual_pii_character_rate": 0.001656594165107217,
          "true_positive": 10926
        },
        {
          "character_coverage": 0.9741895519306215,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14154,
          "exact_span_leakage": 0.02581044806937849,
          "exact_span_recall": 0.9741895519306215,
          "false_negative": 125,
          "gold_characters": 14529,
          "gold_spans": 4843,
          "label_id": "cvv",
          "residual_pii_character_rate": 0.02581044806937849,
          "true_positive": 4718
        },
        {
          "character_coverage": 0.942261967950743,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 736716,
          "exact_span_leakage": 0.14624934003005396,
          "exact_span_recall": 0.853750659969946,
          "false_negative": 10803,
          "gold_characters": 781859,
          "gold_spans": 73867,
          "label_id": "date",
          "residual_pii_character_rate": 0.05773803204925698,
          "true_positive": 63064
        },
        {
          "character_coverage": 0.9999336198742097,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 180765,
          "exact_span_leakage": 0.00022125117539684158,
          "exact_span_recall": 0.9997787488246032,
          "false_negative": 4,
          "gold_characters": 180777,
          "gold_spans": 18079,
          "label_id": "date_of_birth",
          "residual_pii_character_rate": 6.638012579029962e-05,
          "true_positive": 18075
        },
        {
          "character_coverage": 0.9722000733570415,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 243855,
          "exact_span_leakage": 0.04023901369034111,
          "exact_span_recall": 0.9597609863096589,
          "false_negative": 532,
          "gold_characters": 250828,
          "gold_spans": 13221,
          "label_id": "date_time",
          "residual_pii_character_rate": 0.027799926642958495,
          "true_positive": 12689
        },
        {
          "character_coverage": 0.9978727409638555,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 53007,
          "exact_span_leakage": 0.4455524531312326,
          "exact_span_recall": 0.5544475468687674,
          "false_negative": 1117,
          "gold_characters": 53120,
          "gold_spans": 2507,
          "label_id": "device_identifier",
          "residual_pii_character_rate": 0.0021272590361445243,
          "true_positive": 1390
        },
        {
          "character_coverage": 0.8929716651119403,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 122530,
          "exact_span_leakage": 0.9397526124973342,
          "exact_span_recall": 0.06024738750266581,
          "false_negative": 8813,
          "gold_characters": 137216,
          "gold_spans": 9378,
          "label_id": "education_level",
          "residual_pii_character_rate": 0.10702833488805974,
          "true_positive": 565
        },
        {
          "character_coverage": 0.9970281342335526,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1396976,
          "exact_span_leakage": 0.9824587428147599,
          "exact_span_recall": 0.017541257185240126,
          "false_negative": 52984,
          "gold_characters": 1401140,
          "gold_spans": 53930,
          "label_id": "email",
          "residual_pii_character_rate": 0.002971865766447368,
          "true_positive": 946
        },
        {
          "character_coverage": 0.9937288426438701,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 63701,
          "exact_span_leakage": 0.5284974093264249,
          "exact_span_recall": 0.47150259067357514,
          "false_negative": 4692,
          "gold_characters": 64103,
          "gold_spans": 8878,
          "label_id": "employee_id",
          "residual_pii_character_rate": 0.0062711573561299305,
          "true_positive": 4186
        },
        {
          "character_coverage": 0.9590683102908125,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 98574,
          "exact_span_leakage": 0.9915592666545653,
          "exact_span_recall": 0.008440733345434744,
          "false_negative": 10925,
          "gold_characters": 102781,
          "gold_spans": 11018,
          "label_id": "employment_status",
          "residual_pii_character_rate": 0.04093168970918748,
          "true_positive": 93
        },
        {
          "character_coverage": 0.9996613099131084,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 85595,
          "exact_span_leakage": 0.4068192433442317,
          "exact_span_recall": 0.5931807566557683,
          "false_negative": 2613,
          "gold_characters": 85624,
          "gold_spans": 6423,
          "label_id": "fax_number",
          "residual_pii_character_rate": 0.00033869008689157987,
          "true_positive": 3810
        },
        {
          "character_coverage": 0.9843989494353629,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 475256,
          "exact_span_leakage": 0.8668300750806135,
          "exact_span_recall": 0.1331699249193865,
          "false_negative": 72851,
          "gold_characters": 482788,
          "gold_spans": 84043,
          "label_id": "first_name",
          "residual_pii_character_rate": 0.01560105056463712,
          "true_positive": 11192
        },
        {
          "character_coverage": 0.969899995073649,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 39376,
          "exact_span_leakage": 0.9916285153695226,
          "exact_span_recall": 0.008371484630477436,
          "false_negative": 7581,
          "gold_characters": 40598,
          "gold_spans": 7645,
          "label_id": "gender",
          "residual_pii_character_rate": 0.03010000492635101,
          "true_positive": 64
        },
        {
          "character_coverage": 0.9984640560404227,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 139114,
          "exact_span_leakage": 0.7471111953021405,
          "exact_span_recall": 0.25288880469785946,
          "false_negative": 7888,
          "gold_characters": 139328,
          "gold_spans": 10558,
          "label_id": "health_plan_beneficiary_number",
          "residual_pii_character_rate": 0.001535943959577346,
          "true_positive": 2670
        },
        {
          "character_coverage": 0.994344907390596,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 410743,
          "exact_span_leakage": 0.9499182338511856,
          "exact_span_recall": 0.05008176614881439,
          "false_negative": 4647,
          "gold_characters": 413079,
          "gold_spans": 4892,
          "label_id": "http_cookie",
          "residual_pii_character_rate": 0.005655092609404,
          "true_positive": 245
        },
        {
          "character_coverage": 0.997610246069975,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 80151,
          "exact_span_leakage": 0.0034550839091807006,
          "exact_span_recall": 0.9965449160908193,
          "false_negative": 21,
          "gold_characters": 80343,
          "gold_spans": 6078,
          "label_id": "ipv4",
          "residual_pii_character_rate": 0.0023897539300250514,
          "true_positive": 6057
        },
        {
          "character_coverage": 0.9997154518535644,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 112427,
          "exact_span_leakage": 0.06913029627269829,
          "exact_span_recall": 0.9308697037273017,
          "false_negative": 217,
          "gold_characters": 112459,
          "gold_spans": 3139,
          "label_id": "ipv6",
          "residual_pii_character_rate": 0.00028454814643563697,
          "true_positive": 2922
        },
        {
          "character_coverage": 0.9294134840218239,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 57237,
          "exact_span_leakage": 0.9773436596925211,
          "exact_span_recall": 0.022656340307478904,
          "false_negative": 8455,
          "gold_characters": 61584,
          "gold_spans": 8651,
          "label_id": "language",
          "residual_pii_character_rate": 0.07058651597817611,
          "true_positive": 196
        },
        {
          "character_coverage": 0.9898590123896638,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 383411,
          "exact_span_leakage": 0.9979193234445265,
          "exact_span_recall": 0.002080676555473522,
          "false_negative": 59472,
          "gold_characters": 387339,
          "gold_spans": 59596,
          "label_id": "last_name",
          "residual_pii_character_rate": 0.010140987610336216,
          "true_positive": 124
        },
        {
          "character_coverage": 0.9954227823380428,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32621,
          "exact_span_leakage": 0.8310446297204512,
          "exact_span_recall": 0.1689553702795488,
          "false_negative": 3389,
          "gold_characters": 32771,
          "gold_spans": 4078,
          "label_id": "license_plate",
          "residual_pii_character_rate": 0.004577217661957178,
          "true_positive": 689
        },
        {
          "character_coverage": 0.9991142726656241,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 72193,
          "exact_span_leakage": 0.14188235294117646,
          "exact_span_recall": 0.8581176470588235,
          "false_negative": 603,
          "gold_characters": 72257,
          "gold_spans": 4250,
          "label_id": "mac_address",
          "residual_pii_character_rate": 0.0008857273343758632,
          "true_positive": 3647
        },
        {
          "character_coverage": 0.9987772552372809,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 116807,
          "exact_span_leakage": 0.41295728960173006,
          "exact_span_recall": 0.5870427103982699,
          "false_negative": 4583,
          "gold_characters": 116950,
          "gold_spans": 11098,
          "label_id": "medical_record_number",
          "residual_pii_character_rate": 0.001222744762719108,
          "true_positive": 6515
        },
        {
          "character_coverage": 0.9866934737176265,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 37298,
          "exact_span_leakage": 0.1865121180189674,
          "exact_span_recall": 0.8134878819810326,
          "false_negative": 531,
          "gold_characters": 37801,
          "gold_spans": 2847,
          "label_id": "national_id",
          "residual_pii_character_rate": 0.013306526282373499,
          "true_positive": 2316
        },
        {
          "character_coverage": 0.6512575677080817,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 458471,
          "exact_span_leakage": 0.9630232053784429,
          "exact_span_recall": 0.036976794621557146,
          "false_negative": 35524,
          "gold_characters": 703978,
          "gold_spans": 36888,
          "label_id": "occupation",
          "residual_pii_character_rate": 0.34874243229191826,
          "true_positive": 1364
        },
        {
          "character_coverage": 0.9657014537288736,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 81708,
          "exact_span_leakage": 0.9431639226914817,
          "exact_span_recall": 0.05683607730851825,
          "false_negative": 6588,
          "gold_characters": 84610,
          "gold_spans": 6985,
          "label_id": "password",
          "residual_pii_character_rate": 0.034298546271126384,
          "true_positive": 397
        },
        {
          "character_coverage": 0.9994107631756284,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 306996,
          "exact_span_leakage": 0.27175094024237356,
          "exact_span_recall": 0.7282490597576264,
          "false_negative": 6503,
          "gold_characters": 307177,
          "gold_spans": 23930,
          "label_id": "phone_number",
          "residual_pii_character_rate": 0.0005892368243716195,
          "true_positive": 17427
        },
        {
          "character_coverage": 0.9831321919730565,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 35029,
          "exact_span_leakage": 0.015334572490706355,
          "exact_span_recall": 0.9846654275092936,
          "false_negative": 99,
          "gold_characters": 35630,
          "gold_spans": 6456,
          "label_id": "pin",
          "residual_pii_character_rate": 0.016867808026943543,
          "true_positive": 6357
        },
        {
          "character_coverage": 0.9392678585599782,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 82850,
          "exact_span_leakage": 0.9900031240237426,
          "exact_span_recall": 0.00999687597625742,
          "false_negative": 6338,
          "gold_characters": 88207,
          "gold_spans": 6402,
          "label_id": "political_view",
          "residual_pii_character_rate": 0.060732141440021814,
          "true_positive": 64
        },
        {
          "character_coverage": 0.9828799463855972,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 32265,
          "exact_span_leakage": 0.059245102723363585,
          "exact_span_recall": 0.9407548972766364,
          "false_negative": 372,
          "gold_characters": 32827,
          "gold_spans": 6279,
          "label_id": "postcode",
          "residual_pii_character_rate": 0.01712005361440283,
          "true_positive": 5907
        },
        {
          "character_coverage": 0.9409581911665278,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 59844,
          "exact_span_leakage": 0.9924307958477508,
          "exact_span_recall": 0.007569204152249135,
          "false_negative": 9178,
          "gold_characters": 63599,
          "gold_spans": 9248,
          "label_id": "race_ethnicity",
          "residual_pii_character_rate": 0.05904180883347221,
          "true_positive": 70
        },
        {
          "character_coverage": 0.961097322011757,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 73572,
          "exact_span_leakage": 0.9894518139683555,
          "exact_span_recall": 0.010548186031644558,
          "false_negative": 6191,
          "gold_characters": 76550,
          "gold_spans": 6257,
          "label_id": "religious_belief",
          "residual_pii_character_rate": 0.038902677988242984,
          "true_positive": 66
        },
        {
          "character_coverage": 0.973605577689243,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 23460,
          "exact_span_leakage": 0.9963768115942029,
          "exact_span_recall": 0.0036231884057971015,
          "false_negative": 3300,
          "gold_characters": 24096,
          "gold_spans": 3312,
          "label_id": "sexuality",
          "residual_pii_character_rate": 0.02639442231075695,
          "true_positive": 12
        },
        {
          "character_coverage": 0.9979449794497945,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 66529,
          "exact_span_leakage": 0.0021445067634443538,
          "exact_span_recall": 0.9978554932365556,
          "false_negative": 13,
          "gold_characters": 66666,
          "gold_spans": 6062,
          "label_id": "ssn",
          "residual_pii_character_rate": 0.0020550205502054553,
          "true_positive": 6049
        },
        {
          "character_coverage": 0.9472327651916986,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 120542,
          "exact_span_leakage": 0.9806201550387597,
          "exact_span_recall": 0.01937984496124031,
          "false_negative": 17963,
          "gold_characters": 127257,
          "gold_spans": 18318,
          "label_id": "state",
          "residual_pii_character_rate": 0.052767234808301366,
          "true_positive": 355
        },
        {
          "character_coverage": 0.9903687830528141,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 295325,
          "exact_span_leakage": 0.2562355589786125,
          "exact_span_recall": 0.7437644410213875,
          "false_negative": 4325,
          "gold_characters": 298197,
          "gold_spans": 16879,
          "label_id": "street_address",
          "residual_pii_character_rate": 0.009631216947185939,
          "true_positive": 12554
        },
        {
          "character_coverage": 0.9955773206357813,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 58753,
          "exact_span_leakage": 0.9902860226659471,
          "exact_span_recall": 0.009713977334052886,
          "false_negative": 5505,
          "gold_characters": 59014,
          "gold_spans": 5559,
          "label_id": "swift_bic",
          "residual_pii_character_rate": 0.004422679364218696,
          "true_positive": 54
        },
        {
          "character_coverage": 0.9926440258490307,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14439,
          "exact_span_leakage": 0.09677419354838712,
          "exact_span_recall": 0.9032258064516129,
          "false_negative": 126,
          "gold_characters": 14546,
          "gold_spans": 1302,
          "label_id": "tax_id",
          "residual_pii_character_rate": 0.0073559741509693355,
          "true_positive": 1176
        },
        {
          "character_coverage": 0.7880352584172245,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 163604,
          "exact_span_leakage": 0.23961478821513094,
          "exact_span_recall": 0.7603852117848691,
          "false_negative": 5872,
          "gold_characters": 207610,
          "gold_spans": 24506,
          "label_id": "time",
          "residual_pii_character_rate": 0.21196474158277545,
          "true_positive": 18634
        },
        {
          "character_coverage": 0.9960635910698665,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 52379,
          "exact_span_leakage": 0.40686550365785035,
          "exact_span_recall": 0.5931344963421497,
          "false_negative": 723,
          "gold_characters": 52586,
          "gold_spans": 1777,
          "label_id": "unique_id",
          "residual_pii_character_rate": 0.003936408930133473,
          "true_positive": 1054
        },
        {
          "character_coverage": 0.998291108544508,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1969837,
          "exact_span_leakage": 0.967315771395355,
          "exact_span_recall": 0.03268422860464502,
          "false_negative": 36610,
          "gold_characters": 1973209,
          "gold_spans": 37847,
          "label_id": "url",
          "residual_pii_character_rate": 0.0017088914554920231,
          "true_positive": 1237
        },
        {
          "character_coverage": 0.9896416542432775,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 179234,
          "exact_span_leakage": 0.9312658227848101,
          "exact_span_recall": 0.06873417721518987,
          "false_negative": 14714,
          "gold_characters": 181110,
          "gold_spans": 15800,
          "label_id": "user_name",
          "residual_pii_character_rate": 0.010358345756722454,
          "true_positive": 1086
        },
        {
          "character_coverage": 0.9967425005559989,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 76190,
          "exact_span_leakage": 0.7281297960973471,
          "exact_span_recall": 0.27187020390265293,
          "false_negative": 3321,
          "gold_characters": 76439,
          "gold_spans": 4561,
          "label_id": "vehicle_identifier",
          "residual_pii_character_rate": 0.0032574994440011107,
          "true_positive": 1240
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "nemotron-pii",
      "dataset_sha256": "1a4b0512ecb5370f0992d29d0f9c07351e6de13f0d7ea33bb18cecb984780247",
      "documents": 100000,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.3522120307842918,
        "false_negative": 555681,
        "false_positive": 528192,
        "precision": 0.35809520800242084,
        "recall": 0.3465190394430463,
        "true_positive": 294659
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/nemotron-domain-slices-v1/openmed-nemotron-v2.json",
        "sha256": "663c4f93a4656f7cff4ca3c7f31cacf1055a3688824c5f58d852eaefa3dce29c"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.09770860354517942,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 226,
          "exact_span_leakage": 0.9019607843137255,
          "exact_span_recall": 0.09803921568627451,
          "false_negative": 230,
          "gold_characters": 2313,
          "gold_spans": 255,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 25
        },
        {
          "character_coverage": 0.9880952380952381,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 7968,
          "exact_span_leakage": 0.011904761904761862,
          "exact_span_recall": 0.9880952380952381,
          "false_negative": 6,
          "gold_characters": 8064,
          "gold_spans": 504,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 498
        },
        {
          "character_coverage": 0.2594362159579551,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 2172,
          "exact_span_leakage": 0.7525354969574036,
          "exact_span_recall": 0.24746450304259635,
          "false_negative": 371,
          "gold_characters": 8372,
          "gold_spans": 493,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 122
        },
        {
          "character_coverage": 0.7031609796943689,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 6718,
          "exact_span_leakage": 0.2931034482758621,
          "exact_span_recall": 0.7068965517241379,
          "false_negative": 153,
          "gold_characters": 9554,
          "gold_spans": 522,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 369
        },
        {
          "character_coverage": 0.5355648535564853,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 896,
          "exact_span_leakage": 0.4644351464435147,
          "exact_span_recall": 0.5355648535564853,
          "false_negative": 111,
          "gold_characters": 1673,
          "gold_spans": 239,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 128
        },
        {
          "character_coverage": 0.6692307692307692,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1566,
          "exact_span_leakage": 0.3307692307692308,
          "exact_span_recall": 0.6692307692307692,
          "false_negative": 86,
          "gold_characters": 2340,
          "gold_spans": 260,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 174
        },
        {
          "character_coverage": 0.18897030466640957,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1470,
          "exact_span_leakage": 0.8132890365448505,
          "exact_span_recall": 0.1867109634551495,
          "false_negative": 2448,
          "gold_characters": 7779,
          "gold_spans": 3010,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 562
        },
        {
          "character_coverage": 0.8616325091108585,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 21988,
          "exact_span_leakage": 0.13710747456877492,
          "exact_span_recall": 0.8628925254312251,
          "false_negative": 310,
          "gold_characters": 25519,
          "gold_spans": 2261,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 1951
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-chat",
      "dataset_sha256": "4ddacd78dfac49b723032de5e1c29c432a44049245e03a8ea960c86bbf3b765c",
      "documents": 3000,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.5601228788765361,
        "false_negative": 3715,
        "false_positive": 2299,
        "precision": 0.6248368146214099,
        "recall": 0.5075556733828208,
        "true_positive": 3829
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/nemotron-chat.json",
        "sha256": "456a0573fba9e39db607f2a8aca8cf6d4d9a9dda4fac6d1cb7705c0b54d670a2"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.28780899045554453,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 9559,
          "exact_span_leakage": 0.7109867039888034,
          "exact_span_recall": 0.28901329601119663,
          "false_negative": 1016,
          "gold_characters": 33213,
          "gold_spans": 1429,
          "label_id": "address",
          "residual_pii_character_rate": null,
          "true_positive": 413
        },
        {
          "character_coverage": 0.9791183294663574,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 13504,
          "exact_span_leakage": 0.02088167053364265,
          "exact_span_recall": 0.9791183294663574,
          "false_negative": 18,
          "gold_characters": 13792,
          "gold_spans": 862,
          "label_id": "bank_card",
          "residual_pii_character_rate": null,
          "true_positive": 844
        },
        {
          "character_coverage": 0.9154904624106773,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 15502,
          "exact_span_leakage": 0.08807733619763691,
          "exact_span_recall": 0.9119226638023631,
          "false_negative": 82,
          "gold_characters": 16933,
          "gold_spans": 931,
          "label_id": "email",
          "residual_pii_character_rate": null,
          "true_positive": 849
        },
        {
          "character_coverage": 0.9828918322295805,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 32058,
          "exact_span_leakage": 0.017108167770419458,
          "exact_span_recall": 0.9828918322295805,
          "false_negative": 31,
          "gold_characters": 32616,
          "gold_spans": 1812,
          "label_id": "id_number",
          "residual_pii_character_rate": null,
          "true_positive": 1781
        },
        {
          "character_coverage": 0.5538461538461539,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 1764,
          "exact_span_leakage": 0.4461538461538461,
          "exact_span_recall": 0.5538461538461539,
          "false_negative": 203,
          "gold_characters": 3185,
          "gold_spans": 455,
          "label_id": "license_plate",
          "residual_pii_character_rate": null,
          "true_positive": 252
        },
        {
          "character_coverage": 0.9935205183585313,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 4140,
          "exact_span_leakage": 0.006479481641468721,
          "exact_span_recall": 0.9935205183585313,
          "false_negative": 3,
          "gold_characters": 4167,
          "gold_spans": 463,
          "label_id": "passport",
          "residual_pii_character_rate": null,
          "true_positive": 460
        },
        {
          "character_coverage": 0.42375355865391634,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 5805,
          "exact_span_leakage": 0.580225988700565,
          "exact_span_recall": 0.41977401129943503,
          "false_negative": 3081,
          "gold_characters": 13699,
          "gold_spans": 5310,
          "label_id": "person",
          "residual_pii_character_rate": null,
          "true_positive": 2229
        },
        {
          "character_coverage": 0.9604545454545454,
          "character_coverage_metric": "exact_matched_spans",
          "covered_characters": 46486,
          "exact_span_leakage": 0.03954545454545455,
          "exact_span_recall": 0.9604545454545454,
          "false_negative": 174,
          "gold_characters": 48400,
          "gold_spans": 4400,
          "label_id": "phone",
          "residual_pii_character_rate": null,
          "true_positive": 4226
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "pii-bench-zh-formal",
      "dataset_sha256": "bcaa54909fd69c6737b8e442c4d740ce1bf5b95b7d4af70b2f2692010e1ff7a5",
      "documents": 5000,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.6974132492113565,
        "false_negative": 4608,
        "false_positive": 4984,
        "precision": 0.68923805960843,
        "recall": 0.7057847018260759,
        "true_positive": 11054
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/chinese-slices-v1/nemotron-formal.json",
        "sha256": "4f3e63ce18384cac3a122fc21485f4de28262e2020408bd5d1d1c2ea4c6489b5"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 761,
          "exact_span_leakage": 0.5384615384615384,
          "exact_span_recall": 0.46153846153846156,
          "false_negative": 21,
          "gold_characters": 761,
          "gold_spans": 39,
          "label_id": "EMAIL",
          "residual_pii_character_rate": 0.0,
          "true_positive": 18
        },
        {
          "character_coverage": 0.8780487804878049,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 828,
          "exact_span_leakage": 0.2692307692307693,
          "exact_span_recall": 0.7307692307692307,
          "false_negative": 21,
          "gold_characters": 943,
          "gold_spans": 78,
          "label_id": "ID_NUM",
          "residual_pii_character_rate": 0.12195121951219512,
          "true_positive": 57
        },
        {
          "character_coverage": 0.9514233481326717,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 14572,
          "exact_span_leakage": 0.9465201465201465,
          "exact_span_recall": 0.05347985347985348,
          "false_negative": 1292,
          "gold_characters": 15316,
          "gold_spans": 1365,
          "label_id": "NAME_STUDENT",
          "residual_pii_character_rate": 0.04857665186732829,
          "true_positive": 73
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 95,
          "exact_span_leakage": 0.8333333333333334,
          "exact_span_recall": 0.16666666666666666,
          "false_negative": 5,
          "gold_characters": 95,
          "gold_spans": 6,
          "label_id": "PHONE_NUM",
          "residual_pii_character_rate": 0.0,
          "true_positive": 1
        },
        {
          "character_coverage": 0.9368421052631579,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 89,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 2,
          "gold_characters": 95,
          "gold_spans": 2,
          "label_id": "STREET_ADDRESS",
          "residual_pii_character_rate": 0.06315789473684208,
          "true_positive": 0
        },
        {
          "character_coverage": 0.9559855411439506,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4496,
          "exact_span_leakage": 0.6545454545454545,
          "exact_span_recall": 0.34545454545454546,
          "false_negative": 72,
          "gold_characters": 4703,
          "gold_spans": 110,
          "label_id": "URL_PERSONAL",
          "residual_pii_character_rate": 0.04401445885604938,
          "true_positive": 38
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 61,
          "exact_span_leakage": 0.5,
          "exact_span_recall": 0.5,
          "false_negative": 3,
          "gold_characters": 61,
          "gold_spans": 6,
          "label_id": "USERNAME",
          "residual_pii_character_rate": 0.0,
          "true_positive": 3
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "piilo",
      "dataset_sha256": "8276cd44f3b2eb357dfb405b3c5d8e9f821388e984cbf66e92e7df03f1b13117",
      "documents": 6807,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.020703933747412008,
        "false_negative": 1416,
        "false_positive": 16558,
        "precision": 0.011344638165751134,
        "recall": 0.11830635118306351,
        "true_positive": 190
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-piilo.json",
        "sha256": "3eac8f9ba4e7376ed45315048ed1a5cc0008140a497815a8260c7d49fcc0c1d3"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.5697847358121331,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7279,
          "exact_span_leakage": 0.46401985111662536,
          "exact_span_recall": 0.5359801488833746,
          "false_negative": 748,
          "gold_characters": 12775,
          "gold_spans": 1612,
          "label_id": "CODE",
          "residual_pii_character_rate": 0.43021526418786693,
          "true_positive": 864
        },
        {
          "character_coverage": 0.8850258373520206,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 107728,
          "exact_span_leakage": 0.2477978187919463,
          "exact_span_recall": 0.7522021812080537,
          "false_negative": 2363,
          "gold_characters": 121723,
          "gold_spans": 9536,
          "label_id": "DATETIME",
          "residual_pii_character_rate": 0.11497416264797944,
          "true_positive": 7173
        },
        {
          "character_coverage": 0.22813067150635208,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2514,
          "exact_span_leakage": 0.9781897491821155,
          "exact_span_recall": 0.021810250817884406,
          "false_negative": 897,
          "gold_characters": 11020,
          "gold_spans": 917,
          "label_id": "DEM",
          "residual_pii_character_rate": 0.771869328493648,
          "true_positive": 20
        },
        {
          "character_coverage": 0.8482490272373541,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 11990,
          "exact_span_leakage": 0.9869773817683345,
          "exact_span_recall": 0.013022618231665525,
          "false_negative": 1440,
          "gold_characters": 14135,
          "gold_spans": 1459,
          "label_id": "LOC",
          "residual_pii_character_rate": 0.15175097276264593,
          "true_positive": 19
        },
        {
          "character_coverage": 0.060615989515072084,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 740,
          "exact_span_leakage": 0.9981378026070763,
          "exact_span_recall": 0.00186219739292365,
          "false_negative": 536,
          "gold_characters": 12208,
          "gold_spans": 537,
          "label_id": "MISC",
          "residual_pii_character_rate": 0.9393840104849279,
          "true_positive": 1
        },
        {
          "character_coverage": 0.3177091580892056,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 12038,
          "exact_span_leakage": 0.9943444730077121,
          "exact_span_recall": 0.005655526992287918,
          "false_negative": 1934,
          "gold_characters": 37890,
          "gold_spans": 1945,
          "label_id": "ORG",
          "residual_pii_character_rate": 0.6822908419107945,
          "true_positive": 11
        },
        {
          "character_coverage": 0.8786883420554504,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 47349,
          "exact_span_leakage": 0.9990333494441759,
          "exact_span_recall": 0.0009666505558240696,
          "false_negative": 4134,
          "gold_characters": 53886,
          "gold_spans": 4138,
          "label_id": "PERSON",
          "residual_pii_character_rate": 0.12131165794454957,
          "true_positive": 4
        },
        {
          "character_coverage": 0.009750566893424037,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 86,
          "exact_span_leakage": 0.9924698795180723,
          "exact_span_recall": 0.007530120481927711,
          "false_negative": 659,
          "gold_characters": 8820,
          "gold_spans": 664,
          "label_id": "QUANTITY",
          "residual_pii_character_rate": 0.990249433106576,
          "true_positive": 5
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tab",
      "dataset_sha256": "cd0f0f15f84a8739654c7cf30c6be8ce27b051ef73974d39d792a0cb8c846379",
      "documents": 555,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.35629579107170356,
        "false_negative": 12711,
        "false_positive": 16546,
        "precision": 0.32857200827821287,
        "recall": 0.38912918108419836,
        "true_positive": 8097
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-tab.json",
        "sha256": "93886d3ad165fe0a447bb7740a43a600d2f29380a3b49bbbad704d27fa697007"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.9795590261828204,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4265,
          "exact_span_leakage": 0.06040268456375841,
          "exact_span_recall": 0.9395973154362416,
          "false_negative": 18,
          "gold_characters": 4354,
          "gold_spans": 298,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.020440973817179575,
          "true_positive": 280
        },
        {
          "character_coverage": 0.8208048595292331,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4324,
          "exact_span_leakage": 0.7869415807560137,
          "exact_span_recall": 0.21305841924398625,
          "false_negative": 229,
          "gold_characters": 5268,
          "gold_spans": 291,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.1791951404707669,
          "true_positive": 62
        },
        {
          "character_coverage": 0.28869525338225177,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1259,
          "exact_span_leakage": 0.860730593607306,
          "exact_span_recall": 0.13926940639269406,
          "false_negative": 377,
          "gold_characters": 4361,
          "gold_spans": 438,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.7113047466177482,
          "true_positive": 61
        },
        {
          "character_coverage": 0.9772929087910693,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7704,
          "exact_span_leakage": 0.2843450479233227,
          "exact_span_recall": 0.7156549520766773,
          "false_negative": 89,
          "gold_characters": 7883,
          "gold_spans": 313,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.022707091208930663,
          "true_positive": 224
        },
        {
          "character_coverage": 0.6623989218328841,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2949,
          "exact_span_leakage": 0.7161383285302594,
          "exact_span_recall": 0.2838616714697406,
          "false_negative": 994,
          "gold_characters": 4452,
          "gold_spans": 1388,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.33760107816711593,
          "true_positive": 394
        },
        {
          "character_coverage": 0.9865657059110894,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 4039,
          "exact_span_leakage": 0.023255813953488413,
          "exact_span_recall": 0.9767441860465116,
          "false_negative": 8,
          "gold_characters": 4094,
          "gold_spans": 344,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.013434294088910614,
          "true_positive": 336
        },
        {
          "character_coverage": 0.8572990705303445,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 7840,
          "exact_span_leakage": 0.8518518518518519,
          "exact_span_recall": 0.14814814814814814,
          "false_negative": 161,
          "gold_characters": 9145,
          "gold_spans": 189,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.1427009294696555,
          "true_positive": 28
        },
        {
          "character_coverage": 0.6581837664077149,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2457,
          "exact_span_leakage": 0.4936170212765958,
          "exact_span_recall": 0.5063829787234042,
          "false_negative": 116,
          "gold_characters": 3733,
          "gold_spans": 235,
          "label_id": "secret",
          "residual_pii_character_rate": 0.3418162335922851,
          "true_positive": 119
        },
        {
          "character_coverage": 0.9945945945945946,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1472,
          "exact_span_leakage": 0.01621621621621616,
          "exact_span_recall": 0.9837837837837838,
          "false_negative": 3,
          "gold_characters": 1480,
          "gold_spans": 185,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.00540540540540535,
          "true_positive": 182
        },
        {
          "character_coverage": 0.8371428571428572,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 879,
          "exact_span_leakage": 0.6190476190476191,
          "exact_span_recall": 0.38095238095238093,
          "false_negative": 65,
          "gold_characters": 1050,
          "gold_spans": 105,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.1628571428571428,
          "true_positive": 40
        },
        {
          "character_coverage": 0.9450199203187251,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1186,
          "exact_span_leakage": 0.4054054054054054,
          "exact_span_recall": 0.5945945945945946,
          "false_negative": 60,
          "gold_characters": 1255,
          "gold_spans": 148,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.05498007968127494,
          "true_positive": 88
        },
        {
          "character_coverage": 0.7245091830272324,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1144,
          "exact_span_leakage": 0.7525252525252525,
          "exact_span_recall": 0.2474747474747475,
          "false_negative": 149,
          "gold_characters": 1579,
          "gold_spans": 198,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.27549081697276756,
          "true_positive": 49
        },
        {
          "character_coverage": 0.4388703084355258,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1181,
          "exact_span_leakage": 0.8341708542713568,
          "exact_span_recall": 0.1658291457286432,
          "false_negative": 166,
          "gold_characters": 2691,
          "gold_spans": 199,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.5611296915644741,
          "true_positive": 33
        },
        {
          "character_coverage": 0.17627118644067796,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 260,
          "exact_span_leakage": 0.9794520547945206,
          "exact_span_recall": 0.02054794520547945,
          "false_negative": 143,
          "gold_characters": 1475,
          "gold_spans": 146,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.823728813559322,
          "true_positive": 3
        },
        {
          "character_coverage": 0.9262820512820513,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1156,
          "exact_span_leakage": 0.4645669291338582,
          "exact_span_recall": 0.5354330708661418,
          "false_negative": 59,
          "gold_characters": 1248,
          "gold_spans": 127,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.07371794871794868,
          "true_positive": 68
        },
        {
          "character_coverage": 0.9432989690721649,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1830,
          "exact_span_leakage": 0.31443298969072164,
          "exact_span_recall": 0.6855670103092784,
          "false_negative": 61,
          "gold_characters": 1940,
          "gold_spans": 194,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.05670103092783507,
          "true_positive": 133
        },
        {
          "character_coverage": 0.9925925925925926,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1608,
          "exact_span_leakage": 0.007407407407407418,
          "exact_span_recall": 0.9925925925925926,
          "false_negative": 1,
          "gold_characters": 1620,
          "gold_spans": 135,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.007407407407407418,
          "true_positive": 134
        },
        {
          "character_coverage": 0.9857142857142858,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1242,
          "exact_span_leakage": 0.02857142857142858,
          "exact_span_recall": 0.9714285714285714,
          "false_negative": 4,
          "gold_characters": 1260,
          "gold_spans": 140,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.014285714285714235,
          "true_positive": 136
        },
        {
          "character_coverage": 0.5383714118336262,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 919,
          "exact_span_leakage": 0.8344827586206897,
          "exact_span_recall": 0.16551724137931034,
          "false_negative": 121,
          "gold_characters": 1707,
          "gold_spans": 145,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.46162858816637375,
          "true_positive": 24
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-long",
      "dataset_sha256": "b30ca2a1012254f164893d1aad491a778d2fe49cfef71239f8c934e0544883e2",
      "documents": 300,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.3242584315318976,
        "false_negative": 2824,
        "false_positive": 7154,
        "precision": 0.250733137829912,
        "recall": 0.45879647374472976,
        "true_positive": 2394
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-tw-long.json",
        "sha256": "ddd59cdb7dc9aa147226331eaf7a0b11cee449b7dc0132e429dc26381bbef262"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1673,
          "exact_span_leakage": 0.0423728813559322,
          "exact_span_recall": 0.9576271186440678,
          "false_negative": 5,
          "gold_characters": 1673,
          "gold_spans": 118,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0,
          "true_positive": 113
        },
        {
          "character_coverage": 0.7572139303482587,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 1522,
          "exact_span_leakage": 0.8695652173913043,
          "exact_span_recall": 0.13043478260869565,
          "false_negative": 100,
          "gold_characters": 2010,
          "gold_spans": 115,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.24278606965174132,
          "true_positive": 15
        },
        {
          "character_coverage": 0.24703557312252963,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 375,
          "exact_span_leakage": 0.9430379746835443,
          "exact_span_recall": 0.056962025316455694,
          "false_negative": 149,
          "gold_characters": 1518,
          "gold_spans": 158,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.7529644268774703,
          "true_positive": 9
        },
        {
          "character_coverage": 0.9758421231711466,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2868,
          "exact_span_leakage": 0.42500000000000004,
          "exact_span_recall": 0.575,
          "false_negative": 51,
          "gold_characters": 2939,
          "gold_spans": 120,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.024157876828853397,
          "true_positive": 69
        },
        {
          "character_coverage": 0.5930680359435173,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 924,
          "exact_span_leakage": 0.7461240310077519,
          "exact_span_recall": 0.25387596899224807,
          "false_negative": 385,
          "gold_characters": 1558,
          "gold_spans": 516,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.4069319640564827,
          "true_positive": 131
        },
        {
          "character_coverage": 0.9888631090487239,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2131,
          "exact_span_leakage": 0.033333333333333326,
          "exact_span_recall": 0.9666666666666667,
          "false_negative": 6,
          "gold_characters": 2155,
          "gold_spans": 180,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.011136890951276124,
          "true_positive": 174
        },
        {
          "character_coverage": 0.7271303258145363,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 2321,
          "exact_span_leakage": 0.8767123287671232,
          "exact_span_recall": 0.1232876712328767,
          "false_negative": 64,
          "gold_characters": 3192,
          "gold_spans": 73,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.2728696741854637,
          "true_positive": 9
        },
        {
          "character_coverage": 0.8773584905660378,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 744,
          "exact_span_leakage": 0.28915662650602414,
          "exact_span_recall": 0.7108433734939759,
          "false_negative": 24,
          "gold_characters": 848,
          "gold_spans": 83,
          "label_id": "secret",
          "residual_pii_character_rate": 0.12264150943396224,
          "true_positive": 59
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 624,
          "exact_span_leakage": 0.012820512820512775,
          "exact_span_recall": 0.9871794871794872,
          "false_negative": 1,
          "gold_characters": 624,
          "gold_spans": 78,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 77
        },
        {
          "character_coverage": 0.9848484848484849,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 650,
          "exact_span_leakage": 0.6666666666666667,
          "exact_span_recall": 0.3333333333333333,
          "false_negative": 44,
          "gold_characters": 660,
          "gold_spans": 66,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.015151515151515138,
          "true_positive": 22
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 592,
          "exact_span_leakage": 0.4285714285714286,
          "exact_span_recall": 0.5714285714285714,
          "false_negative": 30,
          "gold_characters": 592,
          "gold_spans": 70,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.0,
          "true_positive": 40
        },
        {
          "character_coverage": 0.852760736196319,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 695,
          "exact_span_leakage": 0.6568627450980392,
          "exact_span_recall": 0.3431372549019608,
          "false_negative": 67,
          "gold_characters": 815,
          "gold_spans": 102,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.14723926380368102,
          "true_positive": 35
        },
        {
          "character_coverage": 0.5263157894736842,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 650,
          "exact_span_leakage": 0.8279569892473118,
          "exact_span_recall": 0.17204301075268819,
          "false_negative": 77,
          "gold_characters": 1235,
          "gold_spans": 93,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.4736842105263158,
          "true_positive": 16
        },
        {
          "character_coverage": 0.15366146458583432,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 128,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 82,
          "gold_characters": 833,
          "gold_spans": 82,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.8463385354141657,
          "true_positive": 0
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 747,
          "exact_span_leakage": 0.48,
          "exact_span_recall": 0.52,
          "false_negative": 36,
          "gold_characters": 747,
          "gold_spans": 75,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 39
        },
        {
          "character_coverage": 0.9887640449438202,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 880,
          "exact_span_leakage": 0.4943820224719101,
          "exact_span_recall": 0.5056179775280899,
          "false_negative": 44,
          "gold_characters": 890,
          "gold_spans": 89,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.011235955056179803,
          "true_positive": 45
        },
        {
          "character_coverage": 0.987012987012987,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 912,
          "exact_span_leakage": 0.012987012987012991,
          "exact_span_recall": 0.987012987012987,
          "false_negative": 1,
          "gold_characters": 924,
          "gold_spans": 77,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.012987012987012991,
          "true_positive": 76
        },
        {
          "character_coverage": 0.9830508474576272,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 522,
          "exact_span_leakage": 0.016949152542372836,
          "exact_span_recall": 0.9830508474576272,
          "false_negative": 1,
          "gold_characters": 531,
          "gold_spans": 59,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.016949152542372836,
          "true_positive": 58
        },
        {
          "character_coverage": 0.5746367239101717,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 435,
          "exact_span_leakage": 0.8823529411764706,
          "exact_span_recall": 0.11764705882352941,
          "false_negative": 60,
          "gold_characters": 757,
          "gold_spans": 68,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.42536327608982827,
          "true_positive": 8
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-mid",
      "dataset_sha256": "5d05e1c7fc6adaf7c66591de0d76e5ac5d175fb182f67a19da6d23896b3b02e0",
      "documents": 300,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.4118377483443708,
        "false_negative": 1227,
        "false_positive": 1615,
        "precision": 0.38122605363984674,
        "recall": 0.44779477947794777,
        "true_positive": 995
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-tw-mid.json",
        "sha256": "af63adfd8f08846c7c5238b29fddc33ac110e85faccb7b9d26244dc6cea3c73b"
      },
      "status": "measured"
    },
    {
      "categories": [
        {
          "character_coverage": 0.9285714285714286,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 468,
          "exact_span_leakage": 0.25,
          "exact_span_recall": 0.75,
          "false_negative": 11,
          "gold_characters": 504,
          "gold_spans": 44,
          "label_id": "account_number",
          "residual_pii_character_rate": 0.0714285714285714,
          "true_positive": 33
        },
        {
          "character_coverage": 0.6717663421418637,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 483,
          "exact_span_leakage": 0.972972972972973,
          "exact_span_recall": 0.02702702702702703,
          "false_negative": 36,
          "gold_characters": 719,
          "gold_spans": 37,
          "label_id": "private_address",
          "residual_pii_character_rate": 0.3282336578581363,
          "true_positive": 1
        },
        {
          "character_coverage": 0.3032258064516129,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 94,
          "exact_span_leakage": 0.7894736842105263,
          "exact_span_recall": 0.21052631578947367,
          "false_negative": 30,
          "gold_characters": 310,
          "gold_spans": 38,
          "label_id": "private_date",
          "residual_pii_character_rate": 0.6967741935483871,
          "true_positive": 8
        },
        {
          "character_coverage": 0.9529780564263323,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 608,
          "exact_span_leakage": 0.8571428571428572,
          "exact_span_recall": 0.14285714285714285,
          "false_negative": 24,
          "gold_characters": 638,
          "gold_spans": 28,
          "label_id": "private_email",
          "residual_pii_character_rate": 0.047021943573667735,
          "true_positive": 4
        },
        {
          "character_coverage": 0.6838905775075987,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 225,
          "exact_span_leakage": 0.6136363636363636,
          "exact_span_recall": 0.38636363636363635,
          "false_negative": 54,
          "gold_characters": 329,
          "gold_spans": 88,
          "label_id": "private_person",
          "residual_pii_character_rate": 0.31610942249240126,
          "true_positive": 34
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 417,
          "exact_span_leakage": 0.08108108108108103,
          "exact_span_recall": 0.918918918918919,
          "false_negative": 3,
          "gold_characters": 417,
          "gold_spans": 37,
          "label_id": "private_phone",
          "residual_pii_character_rate": 0.0,
          "true_positive": 34
        },
        {
          "character_coverage": 0.7771260997067448,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 530,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 20,
          "gold_characters": 682,
          "gold_spans": 20,
          "label_id": "private_url",
          "residual_pii_character_rate": 0.22287390029325516,
          "true_positive": 0
        },
        {
          "character_coverage": 0.31736526946107785,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 106,
          "exact_span_leakage": 0.6086956521739131,
          "exact_span_recall": 0.391304347826087,
          "false_negative": 14,
          "gold_characters": 334,
          "gold_spans": 23,
          "label_id": "secret",
          "residual_pii_character_rate": 0.6826347305389222,
          "true_positive": 9
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 104,
          "exact_span_leakage": 0.3076923076923077,
          "exact_span_recall": 0.6923076923076923,
          "false_negative": 4,
          "gold_characters": 104,
          "gold_spans": 13,
          "label_id": "tw_company_id",
          "residual_pii_character_rate": 0.0,
          "true_positive": 9
        },
        {
          "character_coverage": 0.8181818181818182,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 90,
          "exact_span_leakage": 0.9090909090909091,
          "exact_span_recall": 0.09090909090909091,
          "false_negative": 10,
          "gold_characters": 110,
          "gold_spans": 11,
          "label_id": "tw_driver_license",
          "residual_pii_character_rate": 0.18181818181818177,
          "true_positive": 1
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 143,
          "exact_span_leakage": 0.1333333333333333,
          "exact_span_recall": 0.8666666666666667,
          "false_negative": 2,
          "gold_characters": 143,
          "gold_spans": 15,
          "label_id": "tw_household_no",
          "residual_pii_character_rate": 0.0,
          "true_positive": 13
        },
        {
          "character_coverage": 0.6126760563380281,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 87,
          "exact_span_leakage": 0.9444444444444444,
          "exact_span_recall": 0.05555555555555555,
          "false_negative": 17,
          "gold_characters": 142,
          "gold_spans": 18,
          "label_id": "tw_license_plate",
          "residual_pii_character_rate": 0.38732394366197187,
          "true_positive": 1
        },
        {
          "character_coverage": 0.43962848297213625,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 142,
          "exact_span_leakage": 0.65,
          "exact_span_recall": 0.35,
          "false_negative": 13,
          "gold_characters": 323,
          "gold_spans": 20,
          "label_id": "tw_line_id",
          "residual_pii_character_rate": 0.5603715170278638,
          "true_positive": 7
        },
        {
          "character_coverage": 0.304,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 38,
          "exact_span_leakage": 1.0,
          "exact_span_recall": 0.0,
          "false_negative": 12,
          "gold_characters": 125,
          "gold_spans": 12,
          "label_id": "tw_medical_license",
          "residual_pii_character_rate": 0.696,
          "true_positive": 0
        },
        {
          "character_coverage": 0.7288135593220338,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 86,
          "exact_span_leakage": 0.8181818181818181,
          "exact_span_recall": 0.18181818181818182,
          "false_negative": 9,
          "gold_characters": 118,
          "gold_spans": 11,
          "label_id": "tw_military_id",
          "residual_pii_character_rate": 0.27118644067796616,
          "true_positive": 2
        },
        {
          "character_coverage": 0.9523809523809523,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 200,
          "exact_span_leakage": 0.2857142857142857,
          "exact_span_recall": 0.7142857142857143,
          "false_negative": 6,
          "gold_characters": 210,
          "gold_spans": 21,
          "label_id": "tw_national_id",
          "residual_pii_character_rate": 0.04761904761904767,
          "true_positive": 15
        },
        {
          "character_coverage": 0.8269230769230769,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 129,
          "exact_span_leakage": 0.23076923076923073,
          "exact_span_recall": 0.7692307692307693,
          "false_negative": 3,
          "gold_characters": 156,
          "gold_spans": 13,
          "label_id": "tw_nhi_card",
          "residual_pii_character_rate": 0.17307692307692313,
          "true_positive": 10
        },
        {
          "character_coverage": 1.0,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 108,
          "exact_span_leakage": 0.08333333333333337,
          "exact_span_recall": 0.9166666666666666,
          "false_negative": 1,
          "gold_characters": 108,
          "gold_spans": 12,
          "label_id": "tw_passport",
          "residual_pii_character_rate": 0.0,
          "true_positive": 11
        },
        {
          "character_coverage": 0.4066666666666667,
          "character_coverage_metric": "any_overlap",
          "covered_characters": 61,
          "exact_span_leakage": 0.9285714285714286,
          "exact_span_recall": 0.07142857142857142,
          "false_negative": 13,
          "gold_characters": 150,
          "gold_spans": 14,
          "label_id": "tw_ptt_id",
          "residual_pii_character_rate": 0.5933333333333333,
          "true_positive": 1
        }
      ],
      "category_axis": "dataset_gold_label",
      "dataset_id": "tw-short",
      "dataset_sha256": "03e4850a0a97a2b5cb046c1dcd3996b42c5eea3766ae2618d4bbf64391e73693",
      "documents": 310,
      "model_id": "openmed-privacy-filter-nemotron-v2",
      "overall": {
        "f1": 0.4016649323621228,
        "false_negative": 282,
        "false_positive": 293,
        "precision": 0.39711934156378603,
        "recall": 0.4063157894736842,
        "true_positive": 193
      },
      "prediction_type_used": false,
      "source": {
        "path": "outputs/evidence/research-20260813-static/slices-final-v3/openmed-nemotron-v2-tw-short.json",
        "sha256": "c37a8c6bcca54e35295fc520d2ee09ec3b4ab0c1b025087879f04101d8ad516f"
      },
      "status": "measured"
    }
  ]
}
