[
  {
    "seed": 0,
    "method": "sft",
    "original": {
      "accuracy": 0.53125,
      "grounding_accuracy": 0.2916666666666667,
      "mean_iou": 0.1030899439102564,
      "answer_entropy": 0.8673810958862305,
      "mean_confidence": 0.6409214761418601
    },
    "no_image": {
      "accuracy": 0.4322916666666667,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.012920673076923073,
      "answer_entropy": 1.0480519533157349,
      "mean_confidence": 0.5843419010440508
    },
    "occluded": {
      "accuracy": 0.359375,
      "grounding_accuracy": 0.020833333333333332,
      "mean_iou": 0.01171875,
      "answer_entropy": 0.9877702792485555,
      "mean_confidence": 0.5906550952543815
    },
    "swapped": {
      "accuracy": 0.3020833333333333,
      "grounding_accuracy": 0.09375,
      "mean_iou": 0.04695012019230769,
      "answer_entropy": 0.8772192597389221,
      "mean_confidence": 0.640467556193471
    },
    "conflict": {
      "accuracy": 0.22916666666666666,
      "grounding_accuracy": 0.2552083333333333,
      "mean_iou": 0.08730329973741117,
      "answer_entropy": 0.9225612878799438,
      "mean_confidence": 0.6095357043668628
    },
    "visual_dependency_score": 0.21006944444444445
  },
  {
    "seed": 0,
    "method": "rl_answer_only",
    "original": {
      "accuracy": 0.46875,
      "grounding_accuracy": 0.4166666666666667,
      "mean_iou": 0.16499899839743587,
      "answer_entropy": 0.7259044647216797,
      "mean_confidence": 0.7020695880055428
    },
    "no_image": {
      "accuracy": 0.3645833333333333,
      "grounding_accuracy": 0.07291666666666667,
      "mean_iou": 0.014998998397435894,
      "answer_entropy": 0.9087217847506205,
      "mean_confidence": 0.634613250537465
    },
    "occluded": {
      "accuracy": 0.2552083333333333,
      "grounding_accuracy": 0.036458333333333336,
      "mean_iou": 0.02193008814102564,
      "answer_entropy": 0.8350757757822672,
      "mean_confidence": 0.6513754352927208
    },
    "swapped": {
      "accuracy": 0.21875,
      "grounding_accuracy": 0.07291666666666667,
      "mean_iou": 0.050279060955359896,
      "answer_entropy": 0.7250024080276489,
      "mean_confidence": 0.7077722117925683
    },
    "conflict": {
      "accuracy": 0.2552083333333333,
      "grounding_accuracy": 0.3958333333333333,
      "mean_iou": 0.16078086383997528,
      "answer_entropy": 0.7512796322504679,
      "mean_confidence": 0.6972069311887026
    },
    "visual_dependency_score": 0.18923611111111113
  },
  {
    "seed": 0,
    "method": "rl_grounding",
    "original": {
      "accuracy": 0.3958333333333333,
      "grounding_accuracy": 0.22395833333333334,
      "mean_iou": 0.0729417067307692,
      "answer_entropy": 0.6984402736028036,
      "mean_confidence": 0.7158503962370256
    },
    "no_image": {
      "accuracy": 0.3072916666666667,
      "grounding_accuracy": 0.10416666666666667,
      "mean_iou": 0.03257712339743591,
      "answer_entropy": 0.8837409814198812,
      "mean_confidence": 0.6385960350744426
    },
    "occluded": {
      "accuracy": 0.22395833333333334,
      "grounding_accuracy": 0.036458333333333336,
      "mean_iou": 0.025065104166666668,
      "answer_entropy": 0.8068359692891439,
      "mean_confidence": 0.6706525708238283
    },
    "swapped": {
      "accuracy": 0.21354166666666666,
      "grounding_accuracy": 0.078125,
      "mean_iou": 0.03638321314102565,
      "answer_entropy": 0.6912028789520264,
      "mean_confidence": 0.7256560465320945
    },
    "conflict": {
      "accuracy": 0.2760416666666667,
      "grounding_accuracy": 0.234375,
      "mean_iou": 0.0788010817307692,
      "answer_entropy": 0.7055357098579407,
      "mean_confidence": 0.7190291271544993
    },
    "visual_dependency_score": 0.1302083333333333
  },
  {
    "seed": 0,
    "method": "rl_weak",
    "original": {
      "accuracy": 0.5572916666666666,
      "grounding_accuracy": 0.16666666666666666,
      "mean_iou": 0.046466032398826074,
      "answer_entropy": 0.8063065409660339,
      "mean_confidence": 0.6729031540453434
    },
    "no_image": {
      "accuracy": 0.390625,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.04171843624497992,
      "answer_entropy": 0.9964831471443176,
      "mean_confidence": 0.6108073061332107
    },
    "occluded": {
      "accuracy": 0.3333333333333333,
      "grounding_accuracy": 0.052083333333333336,
      "mean_iou": 0.033016440763052206,
      "answer_entropy": 0.9345340728759766,
      "mean_confidence": 0.6183945471420884
    },
    "swapped": {
      "accuracy": 0.2916666666666667,
      "grounding_accuracy": 0.06770833333333333,
      "mean_iou": 0.04692676957831326,
      "answer_entropy": 0.8294246594111124,
      "mean_confidence": 0.6611194717697799
    },
    "conflict": {
      "accuracy": 0.22395833333333334,
      "grounding_accuracy": 0.171875,
      "mean_iou": 0.04661627278344146,
      "answer_entropy": 0.8624780178070068,
      "mean_confidence": 0.6432130585114161
    },
    "visual_dependency_score": 0.25520833333333326
  },
  {
    "seed": 0,
    "method": "best_of_4",
    "original": {
      "accuracy": 0.7239583333333334,
      "grounding_accuracy": 0.4479166666666667,
      "mean_iou": 0.18051028806158698
    }
  },
  {
    "seed": 1,
    "method": "sft",
    "original": {
      "accuracy": 0.5520833333333334,
      "grounding_accuracy": 0.7604166666666666,
      "mean_iou": 0.3794568598914889,
      "answer_entropy": 0.8352936108907064,
      "mean_confidence": 0.6750688031315804
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.078125,
      "mean_iou": 0.02737102255174544,
      "answer_entropy": 1.010390281677246,
      "mean_confidence": 0.6421572375111282
    },
    "occluded": {
      "accuracy": 0.390625,
      "grounding_accuracy": 0.041666666666666664,
      "mean_iou": 0.02383759798810627,
      "answer_entropy": 0.9669130047162374,
      "mean_confidence": 0.643779297824949
    },
    "swapped": {
      "accuracy": 0.3645833333333333,
      "grounding_accuracy": 0.08333333333333333,
      "mean_iou": 0.052416939392184114,
      "answer_entropy": 0.8736680746078491,
      "mean_confidence": 0.65008807539319
    },
    "conflict": {
      "accuracy": 0.19270833333333334,
      "grounding_accuracy": 0.75,
      "mean_iou": 0.3774809273150292,
      "answer_entropy": 0.9107796748479208,
      "mean_confidence": 0.6321283016974727
    },
    "visual_dependency_score": 0.22048611111111116
  },
  {
    "seed": 1,
    "method": "rl_answer_only",
    "original": {
      "accuracy": 0.578125,
      "grounding_accuracy": 0.7395833333333334,
      "mean_iou": 0.36919154985325914,
      "answer_entropy": 0.7980014085769653,
      "mean_confidence": 0.6999201074553033
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.08854166666666667,
      "mean_iou": 0.039021892859901146,
      "answer_entropy": 0.9581242402394613,
      "mean_confidence": 0.6784803831639389
    },
    "occluded": {
      "accuracy": 0.3802083333333333,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.03605714927015755,
      "answer_entropy": 0.925250768661499,
      "mean_confidence": 0.6731527036366364
    },
    "swapped": {
      "accuracy": 0.375,
      "grounding_accuracy": 0.08854166666666667,
      "mean_iou": 0.05306798105885078,
      "answer_entropy": 0.8428923686345419,
      "mean_confidence": 0.6731951371766627
    },
    "conflict": {
      "accuracy": 0.23958333333333334,
      "grounding_accuracy": 0.734375,
      "mean_iou": 0.3613676460650292,
      "answer_entropy": 0.8914381464322408,
      "mean_confidence": 0.6452935134681562
    },
    "visual_dependency_score": 0.22743055555555555
  },
  {
    "seed": 1,
    "method": "rl_grounding",
    "original": {
      "accuracy": 0.578125,
      "grounding_accuracy": 0.7291666666666666,
      "mean_iou": 0.3585773381989496,
      "answer_entropy": 0.765130877494812,
      "mean_confidence": 0.7115564763856431
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.08333333333333333,
      "mean_iou": 0.02135917467948718,
      "answer_entropy": 0.9643959204355875,
      "mean_confidence": 0.6663039770598213
    },
    "occluded": {
      "accuracy": 0.3802083333333333,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.017641841404077846,
      "answer_entropy": 0.9179961085319519,
      "mean_confidence": 0.6677540742481748
    },
    "swapped": {
      "accuracy": 0.3541666666666667,
      "grounding_accuracy": 0.078125,
      "mean_iou": 0.04492996022551745,
      "answer_entropy": 0.8094725608825684,
      "mean_confidence": 0.6868884492044648
    },
    "conflict": {
      "accuracy": 0.265625,
      "grounding_accuracy": 0.7395833333333334,
      "mean_iou": 0.3583608774038461,
      "answer_entropy": 0.8560169140497843,
      "mean_confidence": 0.6551371382859846
    },
    "visual_dependency_score": 0.22569444444444442
  },
  {
    "seed": 1,
    "method": "rl_weak",
    "original": {
      "accuracy": 0.5729166666666666,
      "grounding_accuracy": 0.3489583333333333,
      "mean_iou": 0.15134884489882608,
      "answer_entropy": 0.7963104248046875,
      "mean_confidence": 0.7007405835514268
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.04171843624497992,
      "answer_entropy": 0.9648158947626749,
      "mean_confidence": 0.6786933749293288
    },
    "occluded": {
      "accuracy": 0.3854166666666667,
      "grounding_accuracy": 0.07291666666666667,
      "mean_iou": 0.037936541840438674,
      "answer_entropy": 0.9245290557543436,
      "mean_confidence": 0.6756930711368719
    },
    "swapped": {
      "accuracy": 0.359375,
      "grounding_accuracy": 0.08854166666666667,
      "mean_iou": 0.047295130039388324,
      "answer_entropy": 0.8420508702596029,
      "mean_confidence": 0.6742713133183619
    },
    "conflict": {
      "accuracy": 0.2604166666666667,
      "grounding_accuracy": 0.2760416666666667,
      "mean_iou": 0.11545641701421068,
      "answer_entropy": 0.8966805537541708,
      "mean_confidence": 0.6400055664901932
    },
    "visual_dependency_score": 0.22048611111111108
  },
  {
    "seed": 1,
    "method": "best_of_4",
    "original": {
      "accuracy": 0.7708333333333334,
      "grounding_accuracy": 0.6614583333333334,
      "mean_iou": 0.3213502748976675
    }
  },
  {
    "seed": 2,
    "method": "sft",
    "original": {
      "accuracy": 0.5729166666666666,
      "grounding_accuracy": 0.515625,
      "mean_iou": 0.21350732600732594,
      "answer_entropy": 0.7576161623001099,
      "mean_confidence": 0.7039865516126156
    },
    "no_image": {
      "accuracy": 0.4427083333333333,
      "grounding_accuracy": 0.10416666666666667,
      "mean_iou": 0.029091632670793063,
      "answer_entropy": 0.9131559332211813,
      "mean_confidence": 0.662523442103217
    },
    "occluded": {
      "accuracy": 0.359375,
      "grounding_accuracy": 0.036458333333333336,
      "mean_iou": 0.017728365384615384,
      "answer_entropy": 0.8531242807706197,
      "mean_confidence": 0.663184716211011
    },
    "swapped": {
      "accuracy": 0.359375,
      "grounding_accuracy": 0.09375,
      "mean_iou": 0.04978912042400371,
      "answer_entropy": 0.7707332571347555,
      "mean_confidence": 0.7003636394316951
    },
    "conflict": {
      "accuracy": 0.2864583333333333,
      "grounding_accuracy": 0.4895833333333333,
      "mean_iou": 0.19300666781135525,
      "answer_entropy": 0.8183489839235941,
      "mean_confidence": 0.6719991839490831
    },
    "visual_dependency_score": 0.21006944444444442
  },
  {
    "seed": 2,
    "method": "rl_answer_only",
    "original": {
      "accuracy": 0.5625,
      "grounding_accuracy": 0.40625,
      "mean_iou": 0.1583984856416254,
      "answer_entropy": 0.7247044245402018,
      "mean_confidence": 0.7532920623198152
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.057291666666666664,
      "mean_iou": 0.04379148203390485,
      "answer_entropy": 0.8461999495824178,
      "mean_confidence": 0.7354081793067356
    },
    "occluded": {
      "accuracy": 0.3541666666666667,
      "grounding_accuracy": 0.03125,
      "mean_iou": 0.022135416666666668,
      "answer_entropy": 0.8102388779322306,
      "mean_confidence": 0.7310393713414669
    },
    "swapped": {
      "accuracy": 0.390625,
      "grounding_accuracy": 0.08854166666666667,
      "mean_iou": 0.05611924862913192,
      "answer_entropy": 0.7753022114435831,
      "mean_confidence": 0.7229288515324394
    },
    "conflict": {
      "accuracy": 0.21875,
      "grounding_accuracy": 0.4010416666666667,
      "mean_iou": 0.16397463639849905,
      "answer_entropy": 0.8182175755500793,
      "mean_confidence": 0.7111034579575062
    },
    "visual_dependency_score": 0.21354166666666666
  },
  {
    "seed": 2,
    "method": "rl_grounding",
    "original": {
      "accuracy": 0.5729166666666666,
      "grounding_accuracy": 0.4166666666666667,
      "mean_iou": 0.16097756410256406,
      "answer_entropy": 0.7098053693771362,
      "mean_confidence": 0.7147684072454771
    },
    "no_image": {
      "accuracy": 0.40625,
      "grounding_accuracy": 0.140625,
      "mean_iou": 0.042838541666666674,
      "answer_entropy": 0.8572784264882406,
      "mean_confidence": 0.6658683282633623
    },
    "occluded": {
      "accuracy": 0.2864583333333333,
      "grounding_accuracy": 0.06770833333333333,
      "mean_iou": 0.039863782051282055,
      "answer_entropy": 0.7930487593015035,
      "mean_confidence": 0.675526717522492
    },
    "swapped": {
      "accuracy": 0.3125,
      "grounding_accuracy": 0.109375,
      "mean_iou": 0.05663646171223355,
      "answer_entropy": 0.7178334395090739,
      "mean_confidence": 0.7139262435957789
    },
    "conflict": {
      "accuracy": 0.2916666666666667,
      "grounding_accuracy": 0.390625,
      "mean_iou": 0.15460236378205125,
      "answer_entropy": 0.755301296710968,
      "mean_confidence": 0.6918392443719009
    },
    "visual_dependency_score": 0.23611111111111108
  },
  {
    "seed": 2,
    "method": "rl_weak",
    "original": {
      "accuracy": 0.5572916666666666,
      "grounding_accuracy": 0.15104166666666666,
      "mean_iou": 0.045579614129595304,
      "answer_entropy": 0.7284555435180664,
      "mean_confidence": 0.7501393497611085
    },
    "no_image": {
      "accuracy": 0.4375,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.04171843624497992,
      "answer_entropy": 0.8530399799346924,
      "mean_confidence": 0.7253938351447383
    },
    "occluded": {
      "accuracy": 0.359375,
      "grounding_accuracy": 0.0625,
      "mean_iou": 0.03418831576305221,
      "answer_entropy": 0.8153330485026041,
      "mean_confidence": 0.7233919283996025
    },
    "swapped": {
      "accuracy": 0.375,
      "grounding_accuracy": 0.06770833333333333,
      "mean_iou": 0.04692676957831326,
      "answer_entropy": 0.7783402601877848,
      "mean_confidence": 0.721021327500542
    },
    "conflict": {
      "accuracy": 0.22395833333333334,
      "grounding_accuracy": 0.11979166666666667,
      "mean_iou": 0.04574849880290393,
      "answer_entropy": 0.8166558742523193,
      "mean_confidence": 0.7105455303875109
    },
    "visual_dependency_score": 0.2118055555555555
  },
  {
    "seed": 2,
    "method": "best_of_4",
    "original": {
      "accuracy": 0.734375,
      "grounding_accuracy": 0.4479166666666667,
      "mean_iou": 0.17884400755494498
    }
  }
]
