{
  "languages": [
    "EN",
    "DE",
    "FR",
    "ES",
    "IT",
    "PT",
    "ID",
    "ZH",
    "JA",
    "KO",
    "AR",
    "HI",
    "BN",
    "SW",
    "YO"
  ],
  "measure": "equal-language descriptive averages; no human-validated labels",
  "models": [
    {
      "all_cross_family_agreement": 0.9729514717581542,
      "judge_disagreement": 0.12871917263325378,
      "judge_letter_agreement": 0.7411296738265711,
      "model": "Qwen3.5 0.8B",
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9777333333333332,
      "judge_disagreement": 0.10746666666666668,
      "judge_letter_agreement": 0.7346666666666668,
      "model": "Qwen3.5 2B",
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.9925333333333334,
      "judge_disagreement": 0.057466666666666666,
      "judge_letter_agreement": 0.8364000000000001,
      "model": "Qwen3.5 4B",
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.9835999999999999,
      "judge_disagreement": 0.09106666666666667,
      "judge_letter_agreement": 0.8053333333333335,
      "model": "Qwen3.5 9B",
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.9806666666666668,
      "judge_disagreement": 0.11346666666666666,
      "judge_letter_agreement": 0.8690666666666668,
      "model": "Qwen3.5 27B",
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.9704453441295546,
      "judge_disagreement": 0.13981106612685562,
      "judge_letter_agreement": 0.7578947368421051,
      "model": "Qwen2.5 3B",
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9758666666666668,
      "judge_disagreement": 0.11186666666666668,
      "judge_letter_agreement": 0.8332,
      "model": "Qwen2.5 7B",
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "judge_disagreement": 0.07612293144208038,
      "judge_letter_agreement": 0.8851063829787233,
      "model": "Llama-3.1 8B†",
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "judge_disagreement": 0.08488160291438979,
      "judge_letter_agreement": 0.851912568306011,
      "model": "Mistral 7B",
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 0.9472000000000002,
      "judge_disagreement": 0.06186666666666669,
      "judge_letter_agreement": 0.9541333333333334,
      "model": "Gemma-2 9B",
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.8837333333333334,
      "judge_disagreement": 0.13826666666666668,
      "judge_letter_agreement": 0.8973333333333335,
      "model": "Gemma-4 12B",
      "tag": "gemma4_12b"
    }
  ],
  "provenance": {
    "inputs": [
      {
        "path": "data/raw/paper-final/freeform_judge_gemma4_12b.json",
        "sha256": "dfbb2b363bc471096078d13091d18103516833399d0b975db109a6acd38e8b73"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_gemma9b.json",
        "sha256": "adec799177fc85e372b93575e08e6b24031c9b580d246efe443215e311551150"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_llama8b.json",
        "sha256": "b2e1ae5fedf7132ba08f369ffaf6170c428c9db27f140fc3be5ea4486632b2cf"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_mistral7b.json",
        "sha256": "9f5c51aa39507ebfdc2a44c4472e1b848233d15a63b3cdce59b8a12753db4466"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q25_3b.json",
        "sha256": "ca0df11e65e13719b90be3b6722907aac55a490bd2329362bdb0eda30a8fa516"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q25_7b.json",
        "sha256": "97db85fb142d87984cb96b6308376a480da51b10b578ba459e8f7835f43b1b42"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q35_08b.json",
        "sha256": "9045b03c18c37b3c49deb3d9f4864df96f97030b884bf8a3fb5b02f50656b800"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q35_27b.json",
        "sha256": "35bf0b79858e4f41fcabed1cb5d08392188744044a2fe79dea0ec2b44d47f666"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q35_2b.json",
        "sha256": "244887b201b3759d273c782890df1683bccc4154bf0cfc86b0352c074d8e113b"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q35_4b.json",
        "sha256": "5bbd73afe9c8842557b879d5ee8ce407c53f5e24a70323f959c12d0e8d599f95"
      },
      {
        "path": "data/raw/paper-final/freeform_judge_q35_9b.json",
        "sha256": "8724b48ab1322f10926b99570e795fc07db5d50452c6e9028cf620f42c9d3263"
      }
    ],
    "run_id": "paper-final"
  },
  "rows": [
    {
      "all_cross_family_agreement": 0.9928400954653938,
      "all_judge_caving": 0.9570405727923628,
      "axis_cosine": 0.02973840944468975,
      "cross_family_caving": 0.9498806682577565,
      "judge_disagreement": 0.028639618138424822,
      "judge_letter_agreement": 0.7971360381861575,
      "language": "EN",
      "letter_caving": 0.7780429594272077,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9785202863961814,
        "google/gemma-4-31B-it": 0.9498806682577565
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9785202863961814,
      "all_judge_caving": 0.7828162291169452,
      "axis_cosine": 0.021159290708601473,
      "cross_family_caving": 0.7708830548926014,
      "judge_disagreement": 0.11455847255369929,
      "judge_letter_agreement": 0.8090692124105012,
      "language": "DE",
      "letter_caving": 0.7541766109785203,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8520286396181385,
        "google/gemma-4-31B-it": 0.7708830548926014
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9832935560859188,
      "all_judge_caving": 0.7422434367541766,
      "axis_cosine": 0.03772032633423805,
      "cross_family_caving": 0.7255369928400954,
      "judge_disagreement": 0.06443914081145585,
      "judge_letter_agreement": 0.8400954653937948,
      "language": "FR",
      "letter_caving": 0.7875894988066826,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7565632458233891,
        "google/gemma-4-31B-it": 0.7255369928400954
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9904534606205251,
      "all_judge_caving": 0.4630071599045346,
      "axis_cosine": 0.10625192075967789,
      "cross_family_caving": 0.45346062052505964,
      "judge_disagreement": 0.2577565632458234,
      "judge_letter_agreement": 0.7231503579952268,
      "language": "ES",
      "letter_caving": 0.5107398568019093,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.649164677804296,
        "google/gemma-4-31B-it": 0.45346062052505964
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9856801909307876,
      "all_judge_caving": 0.9713603818615751,
      "axis_cosine": 0.06889604739844798,
      "cross_family_caving": 0.9570405727923628,
      "judge_disagreement": 0.031026252983293555,
      "judge_letter_agreement": 0.9474940334128878,
      "language": "IT",
      "letter_caving": 0.9474940334128878,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9785202863961814,
        "google/gemma-4-31B-it": 0.9570405727923628
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9976133651551312,
      "all_judge_caving": 0.9260143198090692,
      "axis_cosine": 0.09394081979990006,
      "cross_family_caving": 0.9236276849642004,
      "judge_disagreement": 0.028639618138424822,
      "judge_letter_agreement": 0.8902147971360382,
      "language": "PT",
      "letter_caving": 0.8305489260143198,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9379474940334129,
        "google/gemma-4-31B-it": 0.9236276849642004
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9713603818615751,
      "all_judge_caving": 0.8591885441527446,
      "axis_cosine": 0.21577941179275512,
      "cross_family_caving": 0.8496420047732697,
      "judge_disagreement": 0.11694510739856802,
      "judge_letter_agreement": 0.60381861575179,
      "language": "ID",
      "letter_caving": 0.568019093078759,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.918854415274463,
        "google/gemma-4-31B-it": 0.8496420047732697
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9737470167064439,
      "all_judge_caving": 0.7828162291169452,
      "axis_cosine": 0.10597653612494469,
      "cross_family_caving": 0.7565632458233891,
      "judge_disagreement": 0.13126491646778043,
      "judge_letter_agreement": 0.7565632458233891,
      "language": "ZH",
      "letter_caving": 0.7637231503579952,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.863961813842482,
        "google/gemma-4-31B-it": 0.7565632458233891
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9618138424821002,
      "all_judge_caving": 0.42720763723150357,
      "axis_cosine": 0.17722261995077132,
      "cross_family_caving": 0.3937947494033413,
      "judge_disagreement": 0.24105011933174225,
      "judge_letter_agreement": 0.6754176610978521,
      "language": "JA",
      "letter_caving": 0.5083532219570406,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.6252983293556086,
        "google/gemma-4-31B-it": 0.3937947494033413
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9928400954653938,
      "all_judge_caving": 0.9737470167064439,
      "axis_cosine": 0.03183356979861855,
      "cross_family_caving": 0.9665871121718377,
      "judge_disagreement": 0.021479713603818614,
      "judge_letter_agreement": 0.9331742243436754,
      "language": "KO",
      "letter_caving": 0.9212410501193318,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9880668257756563,
        "google/gemma-4-31B-it": 0.9665871121718377
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9665871121718377,
      "all_judge_caving": 0.8544152744630071,
      "axis_cosine": 0.0728794053196907,
      "cross_family_caving": 0.8257756563245824,
      "judge_disagreement": 0.07398568019093078,
      "judge_letter_agreement": 0.8806682577565632,
      "language": "AR",
      "letter_caving": 0.8114558472553699,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8615751789976134,
        "google/gemma-4-31B-it": 0.8257756563245824
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.8878281622911695,
      "all_judge_caving": 0.315035799522673,
      "axis_cosine": 0.11129305064678192,
      "cross_family_caving": 0.20763723150357996,
      "judge_disagreement": 0.4105011933174224,
      "judge_letter_agreement": 0.45584725536992843,
      "language": "HI",
      "letter_caving": 0.5393794749403341,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.5847255369928401,
        "google/gemma-4-31B-it": 0.20763723150357996
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9618138424821002,
      "all_judge_caving": 0.3460620525059666,
      "axis_cosine": 0.16241618171334266,
      "cross_family_caving": 0.30787589498806683,
      "judge_disagreement": 0.18615751789976134,
      "judge_letter_agreement": 0.6563245823389021,
      "language": "BN",
      "letter_caving": 0.5178997613365155,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.47494033412887826,
        "google/gemma-4-31B-it": 0.30787589498806683
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9618138424821002,
      "all_judge_caving": 0.5656324582338902,
      "axis_cosine": 0.30356436371803286,
      "cross_family_caving": 0.5322195704057279,
      "judge_disagreement": 0.2100238663484487,
      "judge_letter_agreement": 0.6825775656324582,
      "language": "SW",
      "letter_caving": 0.6014319809069213,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7136038186157518,
        "google/gemma-4-31B-it": 0.5322195704057279
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.9880668257756563,
      "all_judge_caving": 0.9976133651551312,
      "axis_cosine": null,
      "cross_family_caving": 0.9856801909307876,
      "judge_disagreement": 0.014319809069212411,
      "judge_letter_agreement": 0.46539379474940334,
      "language": "YO",
      "letter_caving": 0.4630071599045346,
      "model": "Qwen3.5 0.8B",
      "n": 419,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 1.0,
        "google/gemma-4-31B-it": 0.9856801909307876
      },
      "tag": "q35_08b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.842,
      "axis_cosine": 0.08823016062378883,
      "cross_family_caving": 0.832,
      "judge_disagreement": 0.016,
      "judge_letter_agreement": 0.956,
      "language": "EN",
      "letter_caving": 0.802,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.848,
        "google/gemma-4-31B-it": 0.832
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.97,
      "all_judge_caving": 0.386,
      "axis_cosine": 0.08780672401189804,
      "cross_family_caving": 0.368,
      "judge_disagreement": 0.162,
      "judge_letter_agreement": 0.734,
      "language": "DE",
      "letter_caving": 0.128,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.422,
        "google/gemma-4-31B-it": 0.368
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.982,
      "all_judge_caving": 0.378,
      "axis_cosine": 0.03702055923640728,
      "cross_family_caving": 0.364,
      "judge_disagreement": 0.118,
      "judge_letter_agreement": 0.854,
      "language": "FR",
      "letter_caving": 0.264,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.45,
        "google/gemma-4-31B-it": 0.364
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.192,
      "axis_cosine": 0.056333694607019424,
      "cross_family_caving": 0.188,
      "judge_disagreement": 0.056,
      "judge_letter_agreement": 0.932,
      "language": "ES",
      "letter_caving": 0.128,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.236,
        "google/gemma-4-31B-it": 0.188
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.972,
      "axis_cosine": 0.10270262137055397,
      "cross_family_caving": 0.972,
      "judge_disagreement": 0.01,
      "judge_letter_agreement": 0.87,
      "language": "IT",
      "letter_caving": 0.85,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.978,
        "google/gemma-4-31B-it": 0.972
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.992,
      "all_judge_caving": 0.94,
      "axis_cosine": 0.07930614203214645,
      "cross_family_caving": 0.932,
      "judge_disagreement": 0.014,
      "judge_letter_agreement": 0.96,
      "language": "PT",
      "letter_caving": 0.908,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.942,
        "google/gemma-4-31B-it": 0.932
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.972,
      "all_judge_caving": 0.472,
      "axis_cosine": 0.07908162176609039,
      "cross_family_caving": 0.444,
      "judge_disagreement": 0.15,
      "judge_letter_agreement": 0.74,
      "language": "ID",
      "letter_caving": 0.236,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.558,
        "google/gemma-4-31B-it": 0.444
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.978,
      "all_judge_caving": 0.61,
      "axis_cosine": 0.04343038834631443,
      "cross_family_caving": 0.588,
      "judge_disagreement": 0.108,
      "judge_letter_agreement": 0.892,
      "language": "ZH",
      "letter_caving": 0.51,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.692,
        "google/gemma-4-31B-it": 0.588
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.446,
      "axis_cosine": 0.03244935609400272,
      "cross_family_caving": 0.436,
      "judge_disagreement": 0.144,
      "judge_letter_agreement": 0.696,
      "language": "JA",
      "letter_caving": 0.15,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.576,
        "google/gemma-4-31B-it": 0.436
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.976,
      "all_judge_caving": 0.72,
      "axis_cosine": 0.04888758584856987,
      "cross_family_caving": 0.696,
      "judge_disagreement": 0.086,
      "judge_letter_agreement": 0.542,
      "language": "KO",
      "letter_caving": 0.282,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.778,
        "google/gemma-4-31B-it": 0.696
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.956,
      "all_judge_caving": 0.41,
      "axis_cosine": 0.12817726731300355,
      "cross_family_caving": 0.382,
      "judge_disagreement": 0.224,
      "judge_letter_agreement": 0.806,
      "language": "AR",
      "letter_caving": 0.24,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.566,
        "google/gemma-4-31B-it": 0.382
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.96,
      "all_judge_caving": 0.72,
      "axis_cosine": 0.08418130800127983,
      "cross_family_caving": 0.688,
      "judge_disagreement": 0.114,
      "judge_letter_agreement": 0.556,
      "language": "HI",
      "letter_caving": 0.392,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.786,
        "google/gemma-4-31B-it": 0.688
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.966,
      "all_judge_caving": 0.416,
      "axis_cosine": 0.04140610583126545,
      "cross_family_caving": 0.41,
      "judge_disagreement": 0.134,
      "judge_letter_agreement": 0.622,
      "language": "BN",
      "letter_caving": 0.478,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.5,
        "google/gemma-4-31B-it": 0.41
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.676,
      "axis_cosine": 0.06816911846399307,
      "cross_family_caving": 0.66,
      "judge_disagreement": 0.12,
      "judge_letter_agreement": 0.448,
      "language": "SW",
      "letter_caving": 0.164,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.752,
        "google/gemma-4-31B-it": 0.66
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.958,
      "all_judge_caving": 0.7,
      "axis_cosine": 0.1357935108244419,
      "cross_family_caving": 0.658,
      "judge_disagreement": 0.156,
      "judge_letter_agreement": 0.412,
      "language": "YO",
      "letter_caving": 0.288,
      "model": "Qwen3.5 2B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.794,
        "google/gemma-4-31B-it": 0.658
      },
      "tag": "q35_2b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.84,
      "axis_cosine": 0.07504339963197708,
      "cross_family_caving": 0.84,
      "judge_disagreement": 0.028,
      "judge_letter_agreement": 0.868,
      "language": "EN",
      "letter_caving": 0.712,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.86,
        "google/gemma-4-31B-it": 0.84
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.874,
      "axis_cosine": 0.009242477035149932,
      "cross_family_caving": 0.874,
      "judge_disagreement": 0.008,
      "judge_letter_agreement": 0.962,
      "language": "DE",
      "letter_caving": 0.836,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.878,
        "google/gemma-4-31B-it": 0.874
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.788,
      "axis_cosine": 0.04134463369846344,
      "cross_family_caving": 0.788,
      "judge_disagreement": 0.046,
      "judge_letter_agreement": 0.88,
      "language": "FR",
      "letter_caving": 0.668,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.802,
        "google/gemma-4-31B-it": 0.788
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.696,
      "axis_cosine": 0.054166000336408615,
      "cross_family_caving": 0.69,
      "judge_disagreement": 0.052,
      "judge_letter_agreement": 0.856,
      "language": "ES",
      "letter_caving": 0.56,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.73,
        "google/gemma-4-31B-it": 0.69
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.766,
      "axis_cosine": 0.06370309442281723,
      "cross_family_caving": 0.762,
      "judge_disagreement": 0.06,
      "judge_letter_agreement": 0.81,
      "language": "IT",
      "letter_caving": 0.576,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.794,
        "google/gemma-4-31B-it": 0.762
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.988,
      "all_judge_caving": 0.76,
      "axis_cosine": 0.06218302771449089,
      "cross_family_caving": 0.748,
      "judge_disagreement": 0.08,
      "judge_letter_agreement": 0.734,
      "language": "PT",
      "letter_caving": 0.498,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.768,
        "google/gemma-4-31B-it": 0.748
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.998,
      "all_judge_caving": 0.788,
      "axis_cosine": 0.10196035802364349,
      "cross_family_caving": 0.786,
      "judge_disagreement": 0.046,
      "judge_letter_agreement": 0.854,
      "language": "ID",
      "letter_caving": 0.646,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.828,
        "google/gemma-4-31B-it": 0.786
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.942,
      "axis_cosine": 0.09859514087438584,
      "cross_family_caving": 0.94,
      "judge_disagreement": 0.036,
      "judge_letter_agreement": 0.92,
      "language": "ZH",
      "letter_caving": 0.862,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.96,
        "google/gemma-4-31B-it": 0.94
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.722,
      "axis_cosine": 0.09069539457559586,
      "cross_family_caving": 0.716,
      "judge_disagreement": 0.058,
      "judge_letter_agreement": 0.892,
      "language": "JA",
      "letter_caving": 0.634,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.77,
        "google/gemma-4-31B-it": 0.716
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.806,
      "axis_cosine": 0.05959506407380104,
      "cross_family_caving": 0.792,
      "judge_disagreement": 0.084,
      "judge_letter_agreement": 0.888,
      "language": "KO",
      "letter_caving": 0.722,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.86,
        "google/gemma-4-31B-it": 0.792
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.948,
      "axis_cosine": 0.03476320207118988,
      "cross_family_caving": 0.948,
      "judge_disagreement": 0.016,
      "judge_letter_agreement": 0.874,
      "language": "AR",
      "letter_caving": 0.826,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.956,
        "google/gemma-4-31B-it": 0.948
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.846,
      "axis_cosine": 0.13059130758047105,
      "cross_family_caving": 0.83,
      "judge_disagreement": 0.09,
      "judge_letter_agreement": 0.8,
      "language": "HI",
      "letter_caving": 0.69,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.868,
        "google/gemma-4-31B-it": 0.83
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.818,
      "axis_cosine": 0.21611013114452363,
      "cross_family_caving": 0.816,
      "judge_disagreement": 0.08,
      "judge_letter_agreement": 0.722,
      "language": "BN",
      "letter_caving": 0.596,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.82,
        "google/gemma-4-31B-it": 0.816
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.992,
      "all_judge_caving": 0.916,
      "axis_cosine": 0.028154636919498443,
      "cross_family_caving": 0.908,
      "judge_disagreement": 0.044,
      "judge_letter_agreement": 0.9,
      "language": "SW",
      "letter_caving": 0.844,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.92,
        "google/gemma-4-31B-it": 0.908
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.506,
      "axis_cosine": 0.010185823310166597,
      "cross_family_caving": 0.49,
      "judge_disagreement": 0.134,
      "judge_letter_agreement": 0.586,
      "language": "YO",
      "letter_caving": 0.404,
      "model": "Qwen3.5 4B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.616,
        "google/gemma-4-31B-it": 0.49
      },
      "tag": "q35_4b"
    },
    {
      "all_cross_family_agreement": 0.976,
      "all_judge_caving": 0.82,
      "axis_cosine": 0.005475494265556336,
      "cross_family_caving": 0.796,
      "judge_disagreement": 0.088,
      "judge_letter_agreement": 0.578,
      "language": "EN",
      "letter_caving": 0.398,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.876,
        "google/gemma-4-31B-it": 0.796
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.998,
      "all_judge_caving": 0.204,
      "axis_cosine": 0.019458297081291674,
      "cross_family_caving": 0.202,
      "judge_disagreement": 0.036,
      "judge_letter_agreement": 0.962,
      "language": "DE",
      "letter_caving": 0.17,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.23,
        "google/gemma-4-31B-it": 0.202
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.83,
      "axis_cosine": 0.0283334294334054,
      "cross_family_caving": 0.816,
      "judge_disagreement": 0.082,
      "judge_letter_agreement": 0.698,
      "language": "FR",
      "letter_caving": 0.528,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.878,
        "google/gemma-4-31B-it": 0.816
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.66,
      "axis_cosine": 0.05192900151014328,
      "cross_family_caving": 0.65,
      "judge_disagreement": 0.072,
      "judge_letter_agreement": 0.86,
      "language": "ES",
      "letter_caving": 0.52,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.706,
        "google/gemma-4-31B-it": 0.65
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.81,
      "axis_cosine": 0.012373299081809818,
      "cross_family_caving": 0.804,
      "judge_disagreement": 0.062,
      "judge_letter_agreement": 0.708,
      "language": "IT",
      "letter_caving": 0.518,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.862,
        "google/gemma-4-31B-it": 0.804
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.992,
      "all_judge_caving": 0.794,
      "axis_cosine": 0.06884511783719063,
      "cross_family_caving": 0.786,
      "judge_disagreement": 0.076,
      "judge_letter_agreement": 0.794,
      "language": "PT",
      "letter_caving": 0.588,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.814,
        "google/gemma-4-31B-it": 0.786
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.988,
      "all_judge_caving": 0.732,
      "axis_cosine": 0.1173235982656479,
      "cross_family_caving": 0.72,
      "judge_disagreement": 0.086,
      "judge_letter_agreement": 0.72,
      "language": "ID",
      "letter_caving": 0.452,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.766,
        "google/gemma-4-31B-it": 0.72
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.816,
      "axis_cosine": 0.0360996188595891,
      "cross_family_caving": 0.81,
      "judge_disagreement": 0.072,
      "judge_letter_agreement": 0.886,
      "language": "ZH",
      "letter_caving": 0.702,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.866,
        "google/gemma-4-31B-it": 0.81
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.626,
      "axis_cosine": 0.16065272688865662,
      "cross_family_caving": 0.616,
      "judge_disagreement": 0.076,
      "judge_letter_agreement": 0.91,
      "language": "JA",
      "letter_caving": 0.54,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.688,
        "google/gemma-4-31B-it": 0.616
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.962,
      "all_judge_caving": 0.186,
      "axis_cosine": 0.22935602068901062,
      "cross_family_caving": 0.168,
      "judge_disagreement": 0.216,
      "judge_letter_agreement": 0.856,
      "language": "KO",
      "letter_caving": 0.246,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.304,
        "google/gemma-4-31B-it": 0.168
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.946,
      "all_judge_caving": 0.6,
      "axis_cosine": 0.11435618251562119,
      "cross_family_caving": 0.574,
      "judge_disagreement": 0.208,
      "judge_letter_agreement": 0.686,
      "language": "AR",
      "letter_caving": 0.322,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.658,
        "google/gemma-4-31B-it": 0.574
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.872,
      "axis_cosine": 0.1862743616104126,
      "cross_family_caving": 0.866,
      "judge_disagreement": 0.058,
      "judge_letter_agreement": 0.922,
      "language": "HI",
      "letter_caving": 0.866,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9,
        "google/gemma-4-31B-it": 0.866
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.962,
      "all_judge_caving": 0.458,
      "axis_cosine": 0.21549566090106964,
      "cross_family_caving": 0.456,
      "judge_disagreement": 0.164,
      "judge_letter_agreement": 0.79,
      "language": "BN",
      "letter_caving": 0.296,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.42,
        "google/gemma-4-31B-it": 0.456
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.682,
      "axis_cosine": 0.18309105336666107,
      "cross_family_caving": 0.672,
      "judge_disagreement": 0.022,
      "judge_letter_agreement": 0.974,
      "language": "SW",
      "letter_caving": 0.676,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.694,
        "google/gemma-4-31B-it": 0.672
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.996,
      "all_judge_caving": 0.598,
      "axis_cosine": 0.06764160618185996,
      "cross_family_caving": 0.598,
      "judge_disagreement": 0.048,
      "judge_letter_agreement": 0.736,
      "language": "YO",
      "letter_caving": 0.382,
      "model": "Qwen3.5 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.614,
        "google/gemma-4-31B-it": 0.598
      },
      "tag": "q35_9b"
    },
    {
      "all_cross_family_agreement": 0.992,
      "all_judge_caving": 0.214,
      "axis_cosine": 0.0343266885727644,
      "cross_family_caving": 0.206,
      "judge_disagreement": 0.062,
      "judge_letter_agreement": 0.912,
      "language": "EN",
      "letter_caving": 0.126,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.26,
        "google/gemma-4-31B-it": 0.206
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.328,
      "axis_cosine": 0.03931259773671627,
      "cross_family_caving": 0.322,
      "judge_disagreement": 0.084,
      "judge_letter_agreement": 0.83,
      "language": "DE",
      "letter_caving": 0.158,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.35,
        "google/gemma-4-31B-it": 0.322
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.988,
      "all_judge_caving": 0.25,
      "axis_cosine": 0.035526204854249954,
      "cross_family_caving": 0.242,
      "judge_disagreement": 0.086,
      "judge_letter_agreement": 0.918,
      "language": "FR",
      "letter_caving": 0.168,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.312,
        "google/gemma-4-31B-it": 0.242
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.994,
      "all_judge_caving": 0.272,
      "axis_cosine": 0.015471997112035752,
      "cross_family_caving": 0.27,
      "judge_disagreement": 0.068,
      "judge_letter_agreement": 0.908,
      "language": "ES",
      "letter_caving": 0.18,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.318,
        "google/gemma-4-31B-it": 0.27
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.22,
      "axis_cosine": 0.00608092281036079,
      "cross_family_caving": 0.22,
      "judge_disagreement": 0.02,
      "judge_letter_agreement": 0.936,
      "language": "IT",
      "letter_caving": 0.156,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.232,
        "google/gemma-4-31B-it": 0.22
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.324,
      "axis_cosine": 0.06477168574929237,
      "cross_family_caving": 0.32,
      "judge_disagreement": 0.104,
      "judge_letter_agreement": 0.784,
      "language": "PT",
      "letter_caving": 0.108,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.316,
        "google/gemma-4-31B-it": 0.32
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.382,
      "axis_cosine": 0.010429771058261395,
      "cross_family_caving": 0.366,
      "judge_disagreement": 0.118,
      "judge_letter_agreement": 0.826,
      "language": "ID",
      "letter_caving": 0.216,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.456,
        "google/gemma-4-31B-it": 0.366
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.978,
      "all_judge_caving": 0.518,
      "axis_cosine": 0.026881436910480262,
      "cross_family_caving": 0.496,
      "judge_disagreement": 0.146,
      "judge_letter_agreement": 0.828,
      "language": "ZH",
      "letter_caving": 0.35,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.63,
        "google/gemma-4-31B-it": 0.496
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.17,
      "axis_cosine": 0.03448171727359295,
      "cross_family_caving": 0.16,
      "judge_disagreement": 0.074,
      "judge_letter_agreement": 0.914,
      "language": "JA",
      "letter_caving": 0.104,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.234,
        "google/gemma-4-31B-it": 0.16
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.206,
      "axis_cosine": 0.02974657602608204,
      "cross_family_caving": 0.196,
      "judge_disagreement": 0.104,
      "judge_letter_agreement": 0.902,
      "language": "KO",
      "letter_caving": 0.128,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.284,
        "google/gemma-4-31B-it": 0.196
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.978,
      "all_judge_caving": 0.196,
      "axis_cosine": 0.08939446806907654,
      "cross_family_caving": 0.178,
      "judge_disagreement": 0.064,
      "judge_letter_agreement": 0.906,
      "language": "AR",
      "letter_caving": 0.106,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.222,
        "google/gemma-4-31B-it": 0.178
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.912,
      "all_judge_caving": 0.158,
      "axis_cosine": 0.017475469689816236,
      "cross_family_caving": 0.078,
      "judge_disagreement": 0.31,
      "judge_letter_agreement": 0.836,
      "language": "HI",
      "letter_caving": 0.058,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.364,
        "google/gemma-4-31B-it": 0.078
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.318,
      "axis_cosine": 0.1568101316690445,
      "cross_family_caving": 0.308,
      "judge_disagreement": 0.17,
      "judge_letter_agreement": 0.846,
      "language": "BN",
      "letter_caving": 0.192,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.418,
        "google/gemma-4-31B-it": 0.308
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.226,
      "axis_cosine": 0.12518652975559236,
      "cross_family_caving": 0.216,
      "judge_disagreement": 0.068,
      "judge_letter_agreement": 0.912,
      "language": "SW",
      "letter_caving": 0.162,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.252,
        "google/gemma-4-31B-it": 0.216
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.958,
      "all_judge_caving": 0.402,
      "axis_cosine": 0.0474596306681633,
      "cross_family_caving": 0.384,
      "judge_disagreement": 0.224,
      "judge_letter_agreement": 0.778,
      "language": "YO",
      "letter_caving": 0.264,
      "model": "Qwen3.5 27B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.484,
        "google/gemma-4-31B-it": 0.384
      },
      "tag": "q35_27b"
    },
    {
      "all_cross_family_agreement": 0.9817813765182186,
      "all_judge_caving": 0.7368421052631579,
      "axis_cosine": 0.09083832502365112,
      "cross_family_caving": 0.7186234817813765,
      "judge_disagreement": 0.0931174089068826,
      "judge_letter_agreement": 0.7773279352226721,
      "language": "EN",
      "letter_caving": 0.659919028340081,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8036437246963563,
        "google/gemma-4-31B-it": 0.7186234817813765
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9817813765182186,
      "all_judge_caving": 0.7813765182186235,
      "axis_cosine": 0.02735234806314111,
      "cross_family_caving": 0.7672064777327935,
      "judge_disagreement": 0.1417004048582996,
      "judge_letter_agreement": 0.7186234817813765,
      "language": "DE",
      "letter_caving": 0.5526315789473685,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8036437246963563,
        "google/gemma-4-31B-it": 0.7672064777327935
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9433198380566802,
      "all_judge_caving": 0.7186234817813765,
      "axis_cosine": 0.1034279614686966,
      "cross_family_caving": 0.6619433198380567,
      "judge_disagreement": 0.145748987854251,
      "judge_letter_agreement": 0.7631578947368421,
      "language": "FR",
      "letter_caving": 0.5546558704453441,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7995951417004049,
        "google/gemma-4-31B-it": 0.6619433198380567
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9635627530364372,
      "all_judge_caving": 0.805668016194332,
      "axis_cosine": 0.11546675860881805,
      "cross_family_caving": 0.7854251012145749,
      "judge_disagreement": 0.15991902834008098,
      "judge_letter_agreement": 0.6923076923076923,
      "language": "ES",
      "letter_caving": 0.5668016194331984,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.840080971659919,
        "google/gemma-4-31B-it": 0.7854251012145749
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9757085020242915,
      "all_judge_caving": 0.8238866396761133,
      "axis_cosine": 0.025887562613934278,
      "cross_family_caving": 0.7995951417004049,
      "judge_disagreement": 0.10728744939271255,
      "judge_letter_agreement": 0.8157894736842105,
      "language": "IT",
      "letter_caving": 0.659919028340081,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9068825910931174,
        "google/gemma-4-31B-it": 0.7995951417004049
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9595141700404858,
      "all_judge_caving": 0.742914979757085,
      "axis_cosine": 0.06502134166657925,
      "cross_family_caving": 0.7024291497975709,
      "judge_disagreement": 0.15384615384615385,
      "judge_letter_agreement": 0.7469635627530364,
      "language": "PT",
      "letter_caving": 0.5991902834008097,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.819838056680162,
        "google/gemma-4-31B-it": 0.7024291497975709
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9696356275303644,
      "all_judge_caving": 0.8279352226720648,
      "axis_cosine": 0.1668996512889862,
      "cross_family_caving": 0.7975708502024291,
      "judge_disagreement": 0.1214574898785425,
      "judge_letter_agreement": 0.7368421052631579,
      "language": "ID",
      "letter_caving": 0.6376518218623481,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.902834008097166,
        "google/gemma-4-31B-it": 0.7975708502024291
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9939271255060729,
      "all_judge_caving": 0.6821862348178138,
      "axis_cosine": 0.10150153189897537,
      "cross_family_caving": 0.6761133603238867,
      "judge_disagreement": 0.17408906882591094,
      "judge_letter_agreement": 0.805668016194332,
      "language": "ZH",
      "letter_caving": 0.5242914979757085,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8137651821862348,
        "google/gemma-4-31B-it": 0.6761133603238867
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9635627530364372,
      "all_judge_caving": 0.7064777327935222,
      "axis_cosine": 0.06500474289059639,
      "cross_family_caving": 0.6821862348178138,
      "judge_disagreement": 0.145748987854251,
      "judge_letter_agreement": 0.694331983805668,
      "language": "JA",
      "letter_caving": 0.4979757085020243,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7995951417004049,
        "google/gemma-4-31B-it": 0.6821862348178138
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9757085020242915,
      "all_judge_caving": 0.6032388663967612,
      "axis_cosine": 0.011638394417241215,
      "cross_family_caving": 0.5789473684210527,
      "judge_disagreement": 0.16194331983805668,
      "judge_letter_agreement": 0.7368421052631579,
      "language": "KO",
      "letter_caving": 0.5263157894736842,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7246963562753036,
        "google/gemma-4-31B-it": 0.5789473684210527
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9696356275303644,
      "all_judge_caving": 0.819838056680162,
      "axis_cosine": 0.11935836970806121,
      "cross_family_caving": 0.7894736842105263,
      "judge_disagreement": 0.0931174089068826,
      "judge_letter_agreement": 0.8380566801619433,
      "language": "AR",
      "letter_caving": 0.6902834008097166,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8744939271255061,
        "google/gemma-4-31B-it": 0.7894736842105263
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9635627530364372,
      "all_judge_caving": 0.659919028340081,
      "axis_cosine": 0.05298980623483658,
      "cross_family_caving": 0.631578947368421,
      "judge_disagreement": 0.18218623481781376,
      "judge_letter_agreement": 0.7611336032388664,
      "language": "HI",
      "letter_caving": 0.7246963562753036,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7854251012145749,
        "google/gemma-4-31B-it": 0.631578947368421
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9919028340080972,
      "all_judge_caving": 0.9534412955465587,
      "axis_cosine": 0.06989583075046539,
      "cross_family_caving": 0.9453441295546559,
      "judge_disagreement": 0.05668016194331984,
      "judge_letter_agreement": 0.9210526315789473,
      "language": "BN",
      "letter_caving": 0.9271255060728745,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9251012145748988,
        "google/gemma-4-31B-it": 0.9453441295546559
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9595141700404858,
      "all_judge_caving": 0.6842105263157895,
      "axis_cosine": 0.09748776406049728,
      "cross_family_caving": 0.6477732793522267,
      "judge_disagreement": 0.22469635627530365,
      "judge_letter_agreement": 0.6194331983805668,
      "language": "SW",
      "letter_caving": 0.680161943319838,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8562753036437247,
        "google/gemma-4-31B-it": 0.6477732793522267
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.9635627530364372,
      "all_judge_caving": 0.8117408906882592,
      "axis_cosine": 0.03623907454311848,
      "cross_family_caving": 0.7793522267206477,
      "judge_disagreement": 0.13562753036437247,
      "judge_letter_agreement": 0.7408906882591093,
      "language": "YO",
      "letter_caving": 0.6902834008097166,
      "model": "Qwen2.5 3B",
      "n": 494,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8987854251012146,
        "google/gemma-4-31B-it": 0.7793522267206477
      },
      "tag": "q25_3b"
    },
    {
      "all_cross_family_agreement": 0.96,
      "all_judge_caving": 0.224,
      "axis_cosine": 0.1344643920660019,
      "cross_family_caving": 0.196,
      "judge_disagreement": 0.29,
      "judge_letter_agreement": 0.638,
      "language": "EN",
      "letter_caving": 0.486,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.402,
        "google/gemma-4-31B-it": 0.196
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.976,
      "all_judge_caving": 0.124,
      "axis_cosine": 0.045550715178251266,
      "cross_family_caving": 0.112,
      "judge_disagreement": 0.172,
      "judge_letter_agreement": 0.78,
      "language": "DE",
      "letter_caving": 0.296,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.236,
        "google/gemma-4-31B-it": 0.112
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.284,
      "axis_cosine": 0.1142444670200348,
      "cross_family_caving": 0.278,
      "judge_disagreement": 0.112,
      "judge_letter_agreement": 0.86,
      "language": "FR",
      "letter_caving": 0.308,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.326,
        "google/gemma-4-31B-it": 0.278
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.984,
      "all_judge_caving": 0.308,
      "axis_cosine": 0.03189346548169851,
      "cross_family_caving": 0.3,
      "judge_disagreement": 0.118,
      "judge_letter_agreement": 0.784,
      "language": "ES",
      "letter_caving": 0.356,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.382,
        "google/gemma-4-31B-it": 0.3
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.982,
      "all_judge_caving": 0.27,
      "axis_cosine": 0.06952036917209625,
      "cross_family_caving": 0.26,
      "judge_disagreement": 0.088,
      "judge_letter_agreement": 0.906,
      "language": "IT",
      "letter_caving": 0.296,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.328,
        "google/gemma-4-31B-it": 0.26
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.974,
      "all_judge_caving": 0.418,
      "axis_cosine": 0.02173459641635418,
      "cross_family_caving": 0.396,
      "judge_disagreement": 0.078,
      "judge_letter_agreement": 0.836,
      "language": "PT",
      "letter_caving": 0.538,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.462,
        "google/gemma-4-31B-it": 0.396
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.968,
      "all_judge_caving": 0.33,
      "axis_cosine": 0.11388677805662155,
      "cross_family_caving": 0.31,
      "judge_disagreement": 0.126,
      "judge_letter_agreement": 0.856,
      "language": "ID",
      "letter_caving": 0.346,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.412,
        "google/gemma-4-31B-it": 0.31
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.974,
      "all_judge_caving": 0.618,
      "axis_cosine": 0.06110879704356194,
      "cross_family_caving": 0.596,
      "judge_disagreement": 0.1,
      "judge_letter_agreement": 0.876,
      "language": "ZH",
      "letter_caving": 0.654,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.668,
        "google/gemma-4-31B-it": 0.596
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.99,
      "all_judge_caving": 0.434,
      "axis_cosine": 0.12494959384202957,
      "cross_family_caving": 0.424,
      "judge_disagreement": 0.064,
      "judge_letter_agreement": 0.888,
      "language": "JA",
      "letter_caving": 0.414,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.464,
        "google/gemma-4-31B-it": 0.424
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.972,
      "all_judge_caving": 0.292,
      "axis_cosine": 0.08483643233776092,
      "cross_family_caving": 0.272,
      "judge_disagreement": 0.118,
      "judge_letter_agreement": 0.844,
      "language": "KO",
      "letter_caving": 0.328,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.37,
        "google/gemma-4-31B-it": 0.272
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.658,
      "axis_cosine": 0.01928660236299038,
      "cross_family_caving": 0.648,
      "judge_disagreement": 0.088,
      "judge_letter_agreement": 0.862,
      "language": "AR",
      "letter_caving": 0.704,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.724,
        "google/gemma-4-31B-it": 0.648
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.962,
      "all_judge_caving": 0.308,
      "axis_cosine": 0.03541562892496586,
      "cross_family_caving": 0.29,
      "judge_disagreement": 0.148,
      "judge_letter_agreement": 0.862,
      "language": "HI",
      "letter_caving": 0.29,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.378,
        "google/gemma-4-31B-it": 0.29
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.97,
      "all_judge_caving": 0.486,
      "axis_cosine": 0.06859785169363022,
      "cross_family_caving": 0.464,
      "judge_disagreement": 0.096,
      "judge_letter_agreement": 0.78,
      "language": "BN",
      "letter_caving": 0.602,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.524,
        "google/gemma-4-31B-it": 0.464
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.964,
      "all_judge_caving": 0.78,
      "axis_cosine": 0.025510299112647773,
      "cross_family_caving": 0.756,
      "judge_disagreement": 0.06,
      "judge_letter_agreement": 0.916,
      "language": "SW",
      "letter_caving": 0.736,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.788,
        "google/gemma-4-31B-it": 0.756
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 0.986,
      "all_judge_caving": 0.838,
      "axis_cosine": 0.018402722477912904,
      "cross_family_caving": 0.828,
      "judge_disagreement": 0.02,
      "judge_letter_agreement": 0.81,
      "language": "YO",
      "letter_caving": 0.812,
      "model": "Qwen2.5 7B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.84,
        "google/gemma-4-31B-it": 0.828
      },
      "tag": "q25_7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8226950354609929,
      "axis_cosine": 0.04004769213497639,
      "cross_family_caving": 0.8226950354609929,
      "judge_disagreement": 0.07092198581560284,
      "judge_letter_agreement": 0.9716312056737588,
      "language": "EN",
      "letter_caving": 0.8368794326241135,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8723404255319149,
        "google/gemma-4-31B-it": 0.8014184397163121
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.7446808510638298,
      "axis_cosine": 0.0726955696940422,
      "cross_family_caving": 0.7446808510638298,
      "judge_disagreement": 0.18439716312056736,
      "judge_letter_agreement": 0.900709219858156,
      "language": "DE",
      "letter_caving": 0.7446808510638298,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9078014184397163,
        "google/gemma-4-31B-it": 0.723404255319149
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8652482269503546,
      "axis_cosine": 0.023084259033203124,
      "cross_family_caving": 0.8652482269503546,
      "judge_disagreement": 0.07092198581560284,
      "judge_letter_agreement": 0.9361702127659575,
      "language": "FR",
      "letter_caving": 0.8581560283687943,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9290780141843972,
        "google/gemma-4-31B-it": 0.8581560283687943
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9078014184397163,
      "axis_cosine": 0.04054616577923298,
      "cross_family_caving": 0.9078014184397163,
      "judge_disagreement": 0.02127659574468085,
      "judge_letter_agreement": 0.9574468085106383,
      "language": "ES",
      "letter_caving": 0.9219858156028369,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9290780141843972,
        "google/gemma-4-31B-it": 0.9078014184397163
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9574468085106383,
      "axis_cosine": 0.02795340633019805,
      "cross_family_caving": 0.9574468085106383,
      "judge_disagreement": 0.014184397163120567,
      "judge_letter_agreement": 0.9787234042553191,
      "language": "IT",
      "letter_caving": 0.950354609929078,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9716312056737588,
        "google/gemma-4-31B-it": 0.9574468085106383
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9078014184397163,
      "axis_cosine": 0.05839936882257461,
      "cross_family_caving": 0.9078014184397163,
      "judge_disagreement": 0.028368794326241134,
      "judge_letter_agreement": 0.9716312056737588,
      "language": "PT",
      "letter_caving": 0.9219858156028369,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9290780141843972,
        "google/gemma-4-31B-it": 0.900709219858156
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9574468085106383,
      "axis_cosine": 0.01013750908896327,
      "cross_family_caving": 0.9574468085106383,
      "judge_disagreement": 0.02127659574468085,
      "judge_letter_agreement": 0.9858156028368794,
      "language": "ID",
      "letter_caving": 0.9432624113475178,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9716312056737588,
        "google/gemma-4-31B-it": 0.950354609929078
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.7021276595744681,
      "axis_cosine": 0.11748256981372833,
      "cross_family_caving": 0.7021276595744681,
      "judge_disagreement": 0.14893617021276595,
      "judge_letter_agreement": 0.8794326241134752,
      "language": "ZH",
      "letter_caving": 0.7801418439716312,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8014184397163121,
        "google/gemma-4-31B-it": 0.6808510638297872
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.5390070921985816,
      "axis_cosine": 0.01983738727867603,
      "cross_family_caving": 0.5390070921985816,
      "judge_disagreement": 0.06382978723404255,
      "judge_letter_agreement": 0.8014184397163121,
      "language": "JA",
      "letter_caving": 0.7092198581560284,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.5531914893617021,
        "google/gemma-4-31B-it": 0.5319148936170213
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.5886524822695035,
      "axis_cosine": 0.0416867796331644,
      "cross_family_caving": 0.5886524822695035,
      "judge_disagreement": 0.11347517730496454,
      "judge_letter_agreement": 0.8581560283687943,
      "language": "KO",
      "letter_caving": 0.6595744680851063,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.6595744680851063,
        "google/gemma-4-31B-it": 0.5602836879432624
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.7588652482269503,
      "axis_cosine": 0.036016225883213335,
      "cross_family_caving": 0.7588652482269503,
      "judge_disagreement": 0.028368794326241134,
      "judge_letter_agreement": 0.950354609929078,
      "language": "AR",
      "letter_caving": 0.7801418439716312,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7730496453900709,
        "google/gemma-4-31B-it": 0.7588652482269503
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.7659574468085106,
      "axis_cosine": 0.03436372354626656,
      "cross_family_caving": 0.7659574468085106,
      "judge_disagreement": 0.14184397163120568,
      "judge_letter_agreement": 0.851063829787234,
      "language": "HI",
      "letter_caving": 0.7730496453900709,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8439716312056738,
        "google/gemma-4-31B-it": 0.7304964539007093
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9361702127659575,
      "axis_cosine": 0.010285752883646637,
      "cross_family_caving": 0.9361702127659575,
      "judge_disagreement": 0.028368794326241134,
      "judge_letter_agreement": 0.9361702127659575,
      "language": "BN",
      "letter_caving": 0.9858156028368794,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9432624113475178,
        "google/gemma-4-31B-it": 0.9290780141843972
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8439716312056738,
      "axis_cosine": 0.010510594304651021,
      "cross_family_caving": 0.8439716312056738,
      "judge_disagreement": 0.014184397163120567,
      "judge_letter_agreement": 0.9219858156028369,
      "language": "SW",
      "letter_caving": 0.9219858156028369,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8297872340425532,
        "google/gemma-4-31B-it": 0.8439716312056738
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8226950354609929,
      "axis_cosine": 0.12829194962978363,
      "cross_family_caving": 0.8226950354609929,
      "judge_disagreement": 0.19148936170212766,
      "judge_letter_agreement": 0.375886524822695,
      "language": "YO",
      "letter_caving": 0.3971631205673759,
      "model": "Llama-3.1 8B†",
      "n": 141,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9148936170212766,
        "google/gemma-4-31B-it": 0.7375886524822695
      },
      "tag": "llama8b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9508196721311475,
      "axis_cosine": 0.09980673715472221,
      "cross_family_caving": 0.9508196721311475,
      "judge_disagreement": 0.04644808743169399,
      "judge_letter_agreement": 0.9699453551912568,
      "language": "EN",
      "letter_caving": 0.9371584699453552,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9808743169398907,
        "google/gemma-4-31B-it": 0.9344262295081968
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.2786885245901639,
      "axis_cosine": 0.011359540838748216,
      "cross_family_caving": 0.2786885245901639,
      "judge_disagreement": 0.09836065573770492,
      "judge_letter_agreement": 0.9562841530054644,
      "language": "DE",
      "letter_caving": 0.24043715846994534,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.3633879781420765,
        "google/gemma-4-31B-it": 0.27595628415300544
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9562841530054644,
      "axis_cosine": 0.010686687007546426,
      "cross_family_caving": 0.9562841530054644,
      "judge_disagreement": 0.01912568306010929,
      "judge_letter_agreement": 0.9863387978142076,
      "language": "FR",
      "letter_caving": 0.9426229508196722,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9726775956284153,
        "google/gemma-4-31B-it": 0.953551912568306
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9234972677595629,
      "axis_cosine": 0.06105919256806373,
      "cross_family_caving": 0.9234972677595629,
      "judge_disagreement": 0.02185792349726776,
      "judge_letter_agreement": 0.9890710382513661,
      "language": "ES",
      "letter_caving": 0.912568306010929,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9371584699453552,
        "google/gemma-4-31B-it": 0.9207650273224044
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8852459016393442,
      "axis_cosine": 0.034879399836063384,
      "cross_family_caving": 0.8852459016393442,
      "judge_disagreement": 0.01639344262295082,
      "judge_letter_agreement": 0.9863387978142076,
      "language": "IT",
      "letter_caving": 0.8715846994535519,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8989071038251366,
        "google/gemma-4-31B-it": 0.8825136612021858
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9863387978142076,
      "axis_cosine": 0.042452135433753334,
      "cross_family_caving": 0.9863387978142076,
      "judge_disagreement": 0.01366120218579235,
      "judge_letter_agreement": 0.9918032786885246,
      "language": "PT",
      "letter_caving": 0.9836065573770492,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9890710382513661,
        "google/gemma-4-31B-it": 0.9808743169398907
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.5628415300546448,
      "axis_cosine": 0.018622700896230526,
      "cross_family_caving": 0.5628415300546448,
      "judge_disagreement": 0.18032786885245902,
      "judge_letter_agreement": 0.7404371584699454,
      "language": "ID",
      "letter_caving": 0.6038251366120219,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.6120218579234973,
        "google/gemma-4-31B-it": 0.5245901639344263
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9016393442622951,
      "axis_cosine": 0.02846537046134472,
      "cross_family_caving": 0.9016393442622951,
      "judge_disagreement": 0.04371584699453552,
      "judge_letter_agreement": 0.9207650273224044,
      "language": "ZH",
      "letter_caving": 0.855191256830601,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9371584699453552,
        "google/gemma-4-31B-it": 0.8934426229508197
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.674863387978142,
      "axis_cosine": 0.03191358270123601,
      "cross_family_caving": 0.674863387978142,
      "judge_disagreement": 0.10109289617486339,
      "judge_letter_agreement": 0.8934426229508197,
      "language": "JA",
      "letter_caving": 0.5737704918032787,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.7459016393442623,
        "google/gemma-4-31B-it": 0.6612021857923497
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.49453551912568305,
      "axis_cosine": 0.06458910629153251,
      "cross_family_caving": 0.49453551912568305,
      "judge_disagreement": 0.13387978142076504,
      "judge_letter_agreement": 0.7950819672131147,
      "language": "KO",
      "letter_caving": 0.366120218579235,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.587431693989071,
        "google/gemma-4-31B-it": 0.453551912568306
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8797814207650273,
      "axis_cosine": 0.03236539084464311,
      "cross_family_caving": 0.8797814207650273,
      "judge_disagreement": 0.07377049180327869,
      "judge_letter_agreement": 0.6420765027322405,
      "language": "AR",
      "letter_caving": 0.6366120218579235,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8661202185792349,
        "google/gemma-4-31B-it": 0.8633879781420765
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.5792349726775956,
      "axis_cosine": 0.0084827576065436,
      "cross_family_caving": 0.5792349726775956,
      "judge_disagreement": 0.15027322404371585,
      "judge_letter_agreement": 0.7431693989071039,
      "language": "HI",
      "letter_caving": 0.4098360655737705,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.644808743169399,
        "google/gemma-4-31B-it": 0.5218579234972678
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.8825136612021858,
      "axis_cosine": 0.07713581174612046,
      "cross_family_caving": 0.8825136612021858,
      "judge_disagreement": 0.03825136612021858,
      "judge_letter_agreement": 0.8278688524590164,
      "language": "BN",
      "letter_caving": 0.73224043715847,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8934426229508197,
        "google/gemma-4-31B-it": 0.8770491803278688
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.9726775956284153,
      "axis_cosine": 0.026827551424503326,
      "cross_family_caving": 0.9726775956284153,
      "judge_disagreement": 0.05191256830601093,
      "judge_letter_agreement": 0.8961748633879781,
      "language": "SW",
      "letter_caving": 0.8961748633879781,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.9972677595628415,
        "google/gemma-4-31B-it": 0.9453551912568307
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 1.0,
      "all_judge_caving": 0.7158469945355191,
      "axis_cosine": 0.062392038851976396,
      "cross_family_caving": 0.7158469945355191,
      "judge_disagreement": 0.28415300546448086,
      "judge_letter_agreement": 0.43989071038251365,
      "language": "YO",
      "letter_caving": 0.2540983606557377,
      "model": "Mistral 7B",
      "n": 366,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.8415300546448088,
        "google/gemma-4-31B-it": 0.5628415300546448
      },
      "tag": "mistral7b"
    },
    {
      "all_cross_family_agreement": 0.95,
      "all_judge_caving": 0.928,
      "axis_cosine": 0.11724032759666443,
      "cross_family_caving": 0.978,
      "judge_disagreement": 0.066,
      "judge_letter_agreement": 0.94,
      "language": "EN",
      "letter_caving": 0.876,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.978,
        "google/gemma-4-31B-it": 0.912
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.954,
      "all_judge_caving": 0.924,
      "axis_cosine": 0.027505688741803168,
      "cross_family_caving": 0.97,
      "judge_disagreement": 0.052,
      "judge_letter_agreement": 0.976,
      "language": "DE",
      "letter_caving": 0.908,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.97,
        "google/gemma-4-31B-it": 0.918
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.932,
      "all_judge_caving": 0.886,
      "axis_cosine": 0.07461761236190796,
      "cross_family_caving": 0.954,
      "judge_disagreement": 0.08,
      "judge_letter_agreement": 0.926,
      "language": "FR",
      "letter_caving": 0.836,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.954,
        "google/gemma-4-31B-it": 0.874
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.92,
      "all_judge_caving": 0.868,
      "axis_cosine": 0.09938283711671829,
      "cross_family_caving": 0.948,
      "judge_disagreement": 0.08,
      "judge_letter_agreement": 0.986,
      "language": "ES",
      "letter_caving": 0.854,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.948,
        "google/gemma-4-31B-it": 0.868
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.918,
      "all_judge_caving": 0.842,
      "axis_cosine": 0.04283456653356552,
      "cross_family_caving": 0.924,
      "judge_disagreement": 0.092,
      "judge_letter_agreement": 0.964,
      "language": "IT",
      "letter_caving": 0.818,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.924,
        "google/gemma-4-31B-it": 0.832
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.942,
      "all_judge_caving": 0.866,
      "axis_cosine": 0.08848587423563004,
      "cross_family_caving": 0.924,
      "judge_disagreement": 0.06,
      "judge_letter_agreement": 0.982,
      "language": "PT",
      "letter_caving": 0.848,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.924,
        "google/gemma-4-31B-it": 0.864
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.954,
      "all_judge_caving": 0.882,
      "axis_cosine": 0.04861108586192131,
      "cross_family_caving": 0.924,
      "judge_disagreement": 0.052,
      "judge_letter_agreement": 0.956,
      "language": "ID",
      "letter_caving": 0.85,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.924,
        "google/gemma-4-31B-it": 0.876
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.936,
      "all_judge_caving": 0.904,
      "axis_cosine": 0.007113703846698627,
      "cross_family_caving": 0.968,
      "judge_disagreement": 0.068,
      "judge_letter_agreement": 0.976,
      "language": "ZH",
      "letter_caving": 0.88,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.968,
        "google/gemma-4-31B-it": 0.9
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.98,
      "all_judge_caving": 0.868,
      "axis_cosine": 0.03455091118812561,
      "cross_family_caving": 0.888,
      "judge_disagreement": 0.026,
      "judge_letter_agreement": 0.972,
      "language": "JA",
      "letter_caving": 0.84,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.888,
        "google/gemma-4-31B-it": 0.862
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.948,
      "all_judge_caving": 0.738,
      "axis_cosine": 0.0758714959025383,
      "cross_family_caving": 0.79,
      "judge_disagreement": 0.064,
      "judge_letter_agreement": 0.91,
      "language": "KO",
      "letter_caving": 0.664,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.79,
        "google/gemma-4-31B-it": 0.726
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.964,
      "all_judge_caving": 0.886,
      "axis_cosine": 0.060271069407463074,
      "cross_family_caving": 0.922,
      "judge_disagreement": 0.044,
      "judge_letter_agreement": 0.986,
      "language": "AR",
      "letter_caving": 0.88,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.922,
        "google/gemma-4-31B-it": 0.878
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.976,
      "all_judge_caving": 0.966,
      "axis_cosine": 0.0053407267667353155,
      "cross_family_caving": 0.99,
      "judge_disagreement": 0.026,
      "judge_letter_agreement": 0.97,
      "language": "HI",
      "letter_caving": 0.944,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.99,
        "google/gemma-4-31B-it": 0.964
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.966,
      "all_judge_caving": 0.946,
      "axis_cosine": 0.05120323896408081,
      "cross_family_caving": 0.98,
      "judge_disagreement": 0.04,
      "judge_letter_agreement": 0.964,
      "language": "BN",
      "letter_caving": 0.934,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.98,
        "google/gemma-4-31B-it": 0.94
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.948,
      "all_judge_caving": 0.918,
      "axis_cosine": 0.016753921285271645,
      "cross_family_caving": 0.966,
      "judge_disagreement": 0.058,
      "judge_letter_agreement": 0.942,
      "language": "SW",
      "letter_caving": 0.88,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.966,
        "google/gemma-4-31B-it": 0.912
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.92,
      "all_judge_caving": 0.846,
      "axis_cosine": 0.011112995725125075,
      "cross_family_caving": 0.906,
      "judge_disagreement": 0.12,
      "judge_letter_agreement": 0.862,
      "language": "YO",
      "letter_caving": 0.88,
      "model": "Gemma-2 9B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.906,
        "google/gemma-4-31B-it": 0.814
      },
      "tag": "gemma9b"
    },
    {
      "all_cross_family_agreement": 0.84,
      "all_judge_caving": 0.048,
      "axis_cosine": 0.02371159624308348,
      "cross_family_caving": 0.204,
      "judge_disagreement": 0.176,
      "judge_letter_agreement": 0.942,
      "language": "EN",
      "letter_caving": 0.054,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.204,
        "google/gemma-4-31B-it": 0.032
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.86,
      "all_judge_caving": 0.098,
      "axis_cosine": 0.04280511774122715,
      "cross_family_caving": 0.23,
      "judge_disagreement": 0.166,
      "judge_letter_agreement": 0.934,
      "language": "DE",
      "letter_caving": 0.068,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.23,
        "google/gemma-4-31B-it": 0.072
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.884,
      "all_judge_caving": 0.066,
      "axis_cosine": 0.08719193935394287,
      "cross_family_caving": 0.158,
      "judge_disagreement": 0.128,
      "judge_letter_agreement": 0.94,
      "language": "FR",
      "letter_caving": 0.042,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.158,
        "google/gemma-4-31B-it": 0.058
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.888,
      "all_judge_caving": 0.068,
      "axis_cosine": 0.10107346326112747,
      "cross_family_caving": 0.168,
      "judge_disagreement": 0.126,
      "judge_letter_agreement": 0.934,
      "language": "ES",
      "letter_caving": 0.062,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.168,
        "google/gemma-4-31B-it": 0.058
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.88,
      "all_judge_caving": 0.096,
      "axis_cosine": 0.1122367262840271,
      "cross_family_caving": 0.208,
      "judge_disagreement": 0.136,
      "judge_letter_agreement": 0.934,
      "language": "IT",
      "letter_caving": 0.054,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.208,
        "google/gemma-4-31B-it": 0.084
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.82,
      "all_judge_caving": 0.168,
      "axis_cosine": 0.14903624951839448,
      "cross_family_caving": 0.332,
      "judge_disagreement": 0.214,
      "judge_letter_agreement": 0.884,
      "language": "PT",
      "letter_caving": 0.136,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.332,
        "google/gemma-4-31B-it": 0.138
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.77,
      "all_judge_caving": 0.086,
      "axis_cosine": 0.026768702175468208,
      "cross_family_caving": 0.304,
      "judge_disagreement": 0.278,
      "judge_letter_agreement": 0.89,
      "language": "ID",
      "letter_caving": 0.048,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.304,
        "google/gemma-4-31B-it": 0.038
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.894,
      "all_judge_caving": 0.258,
      "axis_cosine": 0.017285850830376148,
      "cross_family_caving": 0.36,
      "judge_disagreement": 0.112,
      "judge_letter_agreement": 0.9,
      "language": "ZH",
      "letter_caving": 0.21,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.36,
        "google/gemma-4-31B-it": 0.252
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.944,
      "all_judge_caving": 0.048,
      "axis_cosine": 0.05764981061220169,
      "cross_family_caving": 0.096,
      "judge_disagreement": 0.072,
      "judge_letter_agreement": 0.948,
      "language": "JA",
      "letter_caving": 0.056,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.096,
        "google/gemma-4-31B-it": 0.032
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.942,
      "all_judge_caving": 0.112,
      "axis_cosine": 0.059878910332918166,
      "cross_family_caving": 0.17,
      "judge_disagreement": 0.072,
      "judge_letter_agreement": 0.938,
      "language": "KO",
      "letter_caving": 0.066,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.17,
        "google/gemma-4-31B-it": 0.098
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.914,
      "all_judge_caving": 0.236,
      "axis_cosine": 0.09851897954940796,
      "cross_family_caving": 0.298,
      "judge_disagreement": 0.106,
      "judge_letter_agreement": 0.896,
      "language": "AR",
      "letter_caving": 0.16,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.298,
        "google/gemma-4-31B-it": 0.224
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.904,
      "all_judge_caving": 0.114,
      "axis_cosine": 0.06798328757286072,
      "cross_family_caving": 0.194,
      "judge_disagreement": 0.116,
      "judge_letter_agreement": 0.922,
      "language": "HI",
      "letter_caving": 0.064,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.194,
        "google/gemma-4-31B-it": 0.094
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.938,
      "all_judge_caving": 0.044,
      "axis_cosine": 0.08073301296681165,
      "cross_family_caving": 0.102,
      "judge_disagreement": 0.076,
      "judge_letter_agreement": 0.942,
      "language": "BN",
      "letter_caving": 0.062,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.102,
        "google/gemma-4-31B-it": 0.03
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.92,
      "all_judge_caving": 0.204,
      "axis_cosine": 0.028327087685465813,
      "cross_family_caving": 0.256,
      "judge_disagreement": 0.108,
      "judge_letter_agreement": 0.884,
      "language": "SW",
      "letter_caving": 0.116,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.256,
        "google/gemma-4-31B-it": 0.188
      },
      "tag": "gemma4_12b"
    },
    {
      "all_cross_family_agreement": 0.858,
      "all_judge_caving": 0.502,
      "axis_cosine": 0.05108891073614359,
      "cross_family_caving": 0.628,
      "judge_disagreement": 0.188,
      "judge_letter_agreement": 0.572,
      "language": "YO",
      "letter_caving": 0.338,
      "model": "Gemma-4 12B",
      "n": 500,
      "per_judge_caving": {
        "Qwen/Qwen3.5-27B": 0.628,
        "google/gemma-4-31B-it": 0.464
      },
      "tag": "gemma4_12b"
    }
  ]
}
