{
  "best_metric": 0.009899638593196869,
  "best_model_checkpoint": "./experiments/unet_run_2026-06-23_22-07\\checkpoint-6534",
  "epoch": 100.0,
  "eval_steps": 500,
  "global_step": 6600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07575757575757576,
      "grad_norm": 3.0742416381835938,
      "learning_rate": 9.992424242424244e-05,
      "loss": 1.0652,
      "step": 5
    },
    {
      "epoch": 0.15151515151515152,
      "grad_norm": 2.8115806579589844,
      "learning_rate": 9.984848484848486e-05,
      "loss": 0.8972,
      "step": 10
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 2.210629940032959,
      "learning_rate": 9.977272727272728e-05,
      "loss": 0.7489,
      "step": 15
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 1.8707102537155151,
      "learning_rate": 9.96969696969697e-05,
      "loss": 0.6507,
      "step": 20
    },
    {
      "epoch": 0.3787878787878788,
      "grad_norm": 1.8910330533981323,
      "learning_rate": 9.962121212121213e-05,
      "loss": 0.6087,
      "step": 25
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 1.6793878078460693,
      "learning_rate": 9.954545454545455e-05,
      "loss": 0.542,
      "step": 30
    },
    {
      "epoch": 0.5303030303030303,
      "grad_norm": 1.7584264278411865,
      "learning_rate": 9.946969696969698e-05,
      "loss": 0.5201,
      "step": 35
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 1.5806080102920532,
      "learning_rate": 9.939393939393939e-05,
      "loss": 0.4802,
      "step": 40
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 1.532446265220642,
      "learning_rate": 9.931818181818182e-05,
      "loss": 0.4553,
      "step": 45
    },
    {
      "epoch": 0.7575757575757576,
      "grad_norm": 1.709405541419983,
      "learning_rate": 9.924242424242425e-05,
      "loss": 0.4325,
      "step": 50
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 1.5723271369934082,
      "learning_rate": 9.916666666666667e-05,
      "loss": 0.4013,
      "step": 55
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 1.564184308052063,
      "learning_rate": 9.909090909090911e-05,
      "loss": 0.3703,
      "step": 60
    },
    {
      "epoch": 0.9848484848484849,
      "grad_norm": 1.4000282287597656,
      "learning_rate": 9.901515151515151e-05,
      "loss": 0.3553,
      "step": 65
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.3434337079524994,
      "eval_mean_accuracy": 0.8901209122838848,
      "eval_mean_iou": 0.8450538274269488,
      "eval_runtime": 139.8754,
      "eval_samples_per_second": 1.68,
      "eval_steps_per_second": 0.214,
      "step": 66
    },
    {
      "epoch": 1.0606060606060606,
      "grad_norm": 1.5334577560424805,
      "learning_rate": 9.893939393939395e-05,
      "loss": 0.3417,
      "step": 70
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 1.3461956977844238,
      "learning_rate": 9.886363636363637e-05,
      "loss": 0.3295,
      "step": 75
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 1.2508584260940552,
      "learning_rate": 9.87878787878788e-05,
      "loss": 0.3065,
      "step": 80
    },
    {
      "epoch": 1.2878787878787878,
      "grad_norm": 1.306114673614502,
      "learning_rate": 9.871212121212122e-05,
      "loss": 0.2929,
      "step": 85
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 1.1709235906600952,
      "learning_rate": 9.863636363636364e-05,
      "loss": 0.2778,
      "step": 90
    },
    {
      "epoch": 1.4393939393939394,
      "grad_norm": 1.4133832454681396,
      "learning_rate": 9.856060606060607e-05,
      "loss": 0.2827,
      "step": 95
    },
    {
      "epoch": 1.5151515151515151,
      "grad_norm": 1.1168280839920044,
      "learning_rate": 9.848484848484849e-05,
      "loss": 0.2732,
      "step": 100
    },
    {
      "epoch": 1.5909090909090908,
      "grad_norm": 1.2507333755493164,
      "learning_rate": 9.840909090909092e-05,
      "loss": 0.2675,
      "step": 105
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 1.0351966619491577,
      "learning_rate": 9.833333333333333e-05,
      "loss": 0.2488,
      "step": 110
    },
    {
      "epoch": 1.7424242424242424,
      "grad_norm": 1.101205587387085,
      "learning_rate": 9.825757575757576e-05,
      "loss": 0.2253,
      "step": 115
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.9671285152435303,
      "learning_rate": 9.818181818181818e-05,
      "loss": 0.2193,
      "step": 120
    },
    {
      "epoch": 1.893939393939394,
      "grad_norm": 1.0657840967178345,
      "learning_rate": 9.810606060606061e-05,
      "loss": 0.2234,
      "step": 125
    },
    {
      "epoch": 1.9696969696969697,
      "grad_norm": 0.9236158728599548,
      "learning_rate": 9.803030303030303e-05,
      "loss": 0.2103,
      "step": 130
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.21199926733970642,
      "eval_mean_accuracy": 0.9370986730895982,
      "eval_mean_iou": 0.8852124293482566,
      "eval_runtime": 146.3976,
      "eval_samples_per_second": 1.605,
      "eval_steps_per_second": 0.205,
      "step": 132
    },
    {
      "epoch": 2.0454545454545454,
      "grad_norm": 0.8842264413833618,
      "learning_rate": 9.795454545454545e-05,
      "loss": 0.2016,
      "step": 135
    },
    {
      "epoch": 2.121212121212121,
      "grad_norm": 0.8960620760917664,
      "learning_rate": 9.787878787878789e-05,
      "loss": 0.1915,
      "step": 140
    },
    {
      "epoch": 2.196969696969697,
      "grad_norm": 0.7794420123100281,
      "learning_rate": 9.78030303030303e-05,
      "loss": 0.1887,
      "step": 145
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 0.7301771640777588,
      "learning_rate": 9.772727272727274e-05,
      "loss": 0.1841,
      "step": 150
    },
    {
      "epoch": 2.3484848484848486,
      "grad_norm": 0.9263225793838501,
      "learning_rate": 9.765151515151516e-05,
      "loss": 0.1755,
      "step": 155
    },
    {
      "epoch": 2.4242424242424243,
      "grad_norm": 1.0182371139526367,
      "learning_rate": 9.757575757575758e-05,
      "loss": 0.1776,
      "step": 160
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.7853384613990784,
      "learning_rate": 9.75e-05,
      "loss": 0.1568,
      "step": 165
    },
    {
      "epoch": 2.5757575757575757,
      "grad_norm": 0.9242074489593506,
      "learning_rate": 9.742424242424243e-05,
      "loss": 0.1555,
      "step": 170
    },
    {
      "epoch": 2.6515151515151514,
      "grad_norm": 0.9878806471824646,
      "learning_rate": 9.734848484848485e-05,
      "loss": 0.159,
      "step": 175
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 0.8724908828735352,
      "learning_rate": 9.727272727272728e-05,
      "loss": 0.1566,
      "step": 180
    },
    {
      "epoch": 2.8030303030303028,
      "grad_norm": 0.719180703163147,
      "learning_rate": 9.71969696969697e-05,
      "loss": 0.1415,
      "step": 185
    },
    {
      "epoch": 2.878787878787879,
      "grad_norm": 0.6156677007675171,
      "learning_rate": 9.712121212121212e-05,
      "loss": 0.1369,
      "step": 190
    },
    {
      "epoch": 2.9545454545454546,
      "grad_norm": 0.6489553451538086,
      "learning_rate": 9.704545454545456e-05,
      "loss": 0.134,
      "step": 195
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.13676267862319946,
      "eval_mean_accuracy": 0.8751923884848393,
      "eval_mean_iou": 0.8530271885547617,
      "eval_runtime": 134.302,
      "eval_samples_per_second": 1.75,
      "eval_steps_per_second": 0.223,
      "step": 198
    },
    {
      "epoch": 3.0303030303030303,
      "grad_norm": 0.6220528483390808,
      "learning_rate": 9.696969696969698e-05,
      "loss": 0.125,
      "step": 200
    },
    {
      "epoch": 3.106060606060606,
      "grad_norm": 0.7769432663917542,
      "learning_rate": 9.689393939393941e-05,
      "loss": 0.1289,
      "step": 205
    },
    {
      "epoch": 3.1818181818181817,
      "grad_norm": 0.6297140121459961,
      "learning_rate": 9.681818181818181e-05,
      "loss": 0.1144,
      "step": 210
    },
    {
      "epoch": 3.257575757575758,
      "grad_norm": 0.5939669013023376,
      "learning_rate": 9.674242424242425e-05,
      "loss": 0.1259,
      "step": 215
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.6643730998039246,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.1088,
      "step": 220
    },
    {
      "epoch": 3.409090909090909,
      "grad_norm": 0.5556369423866272,
      "learning_rate": 9.65909090909091e-05,
      "loss": 0.109,
      "step": 225
    },
    {
      "epoch": 3.484848484848485,
      "grad_norm": 0.5800299048423767,
      "learning_rate": 9.651515151515152e-05,
      "loss": 0.1026,
      "step": 230
    },
    {
      "epoch": 3.5606060606060606,
      "grad_norm": 0.6208093762397766,
      "learning_rate": 9.643939393939394e-05,
      "loss": 0.1073,
      "step": 235
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 0.43769362568855286,
      "learning_rate": 9.636363636363637e-05,
      "loss": 0.096,
      "step": 240
    },
    {
      "epoch": 3.712121212121212,
      "grad_norm": 0.41298291087150574,
      "learning_rate": 9.628787878787879e-05,
      "loss": 0.0963,
      "step": 245
    },
    {
      "epoch": 3.787878787878788,
      "grad_norm": 0.5162907838821411,
      "learning_rate": 9.621212121212123e-05,
      "loss": 0.0985,
      "step": 250
    },
    {
      "epoch": 3.8636363636363638,
      "grad_norm": 0.5192968845367432,
      "learning_rate": 9.613636363636363e-05,
      "loss": 0.0997,
      "step": 255
    },
    {
      "epoch": 3.9393939393939394,
      "grad_norm": 0.4886101484298706,
      "learning_rate": 9.606060606060606e-05,
      "loss": 0.0885,
      "step": 260
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.08715613931417465,
      "eval_mean_accuracy": 0.9032919226316655,
      "eval_mean_iou": 0.8800149947276396,
      "eval_runtime": 136.3611,
      "eval_samples_per_second": 1.723,
      "eval_steps_per_second": 0.22,
      "step": 264
    },
    {
      "epoch": 4.015151515151516,
      "grad_norm": 0.4997844099998474,
      "learning_rate": 9.598484848484848e-05,
      "loss": 0.092,
      "step": 265
    },
    {
      "epoch": 4.090909090909091,
      "grad_norm": 0.520751953125,
      "learning_rate": 9.590909090909092e-05,
      "loss": 0.0916,
      "step": 270
    },
    {
      "epoch": 4.166666666666667,
      "grad_norm": 0.4164790213108063,
      "learning_rate": 9.583333333333334e-05,
      "loss": 0.0859,
      "step": 275
    },
    {
      "epoch": 4.242424242424242,
      "grad_norm": 0.441114604473114,
      "learning_rate": 9.575757575757576e-05,
      "loss": 0.0785,
      "step": 280
    },
    {
      "epoch": 4.318181818181818,
      "grad_norm": 0.4566723108291626,
      "learning_rate": 9.568181818181819e-05,
      "loss": 0.079,
      "step": 285
    },
    {
      "epoch": 4.393939393939394,
      "grad_norm": 0.35356175899505615,
      "learning_rate": 9.560606060606061e-05,
      "loss": 0.0781,
      "step": 290
    },
    {
      "epoch": 4.46969696969697,
      "grad_norm": 0.7153947949409485,
      "learning_rate": 9.553030303030304e-05,
      "loss": 0.0792,
      "step": 295
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 0.8675434589385986,
      "learning_rate": 9.545454545454546e-05,
      "loss": 0.0728,
      "step": 300
    },
    {
      "epoch": 4.621212121212121,
      "grad_norm": 0.3581145405769348,
      "learning_rate": 9.537878787878788e-05,
      "loss": 0.0726,
      "step": 305
    },
    {
      "epoch": 4.696969696969697,
      "grad_norm": 0.4127926826477051,
      "learning_rate": 9.53030303030303e-05,
      "loss": 0.0713,
      "step": 310
    },
    {
      "epoch": 4.7727272727272725,
      "grad_norm": 0.37551289796829224,
      "learning_rate": 9.522727272727273e-05,
      "loss": 0.0679,
      "step": 315
    },
    {
      "epoch": 4.848484848484849,
      "grad_norm": 0.39595314860343933,
      "learning_rate": 9.515151515151515e-05,
      "loss": 0.0662,
      "step": 320
    },
    {
      "epoch": 4.924242424242424,
      "grad_norm": 0.34127482771873474,
      "learning_rate": 9.507575757575759e-05,
      "loss": 0.0626,
      "step": 325
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.5234401226043701,
      "learning_rate": 9.5e-05,
      "loss": 0.0749,
      "step": 330
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.06678792834281921,
      "eval_mean_accuracy": 0.9453064048515515,
      "eval_mean_iou": 0.9068547892541426,
      "eval_runtime": 132.012,
      "eval_samples_per_second": 1.78,
      "eval_steps_per_second": 0.227,
      "step": 330
    },
    {
      "epoch": 5.075757575757576,
      "grad_norm": 0.39256536960601807,
      "learning_rate": 9.492424242424242e-05,
      "loss": 0.0667,
      "step": 335
    },
    {
      "epoch": 5.151515151515151,
      "grad_norm": 0.35619768500328064,
      "learning_rate": 9.484848484848486e-05,
      "loss": 0.0639,
      "step": 340
    },
    {
      "epoch": 5.2272727272727275,
      "grad_norm": 0.35817235708236694,
      "learning_rate": 9.477272727272728e-05,
      "loss": 0.0732,
      "step": 345
    },
    {
      "epoch": 5.303030303030303,
      "grad_norm": 0.2903057932853699,
      "learning_rate": 9.469696969696971e-05,
      "loss": 0.0601,
      "step": 350
    },
    {
      "epoch": 5.378787878787879,
      "grad_norm": 0.31652864813804626,
      "learning_rate": 9.462121212121212e-05,
      "loss": 0.0579,
      "step": 355
    },
    {
      "epoch": 5.454545454545454,
      "grad_norm": 0.8863661885261536,
      "learning_rate": 9.454545454545455e-05,
      "loss": 0.0637,
      "step": 360
    },
    {
      "epoch": 5.53030303030303,
      "grad_norm": 0.28541263937950134,
      "learning_rate": 9.446969696969697e-05,
      "loss": 0.0579,
      "step": 365
    },
    {
      "epoch": 5.606060606060606,
      "grad_norm": 0.2541329860687256,
      "learning_rate": 9.43939393939394e-05,
      "loss": 0.0572,
      "step": 370
    },
    {
      "epoch": 5.681818181818182,
      "grad_norm": 0.25615087151527405,
      "learning_rate": 9.431818181818182e-05,
      "loss": 0.0539,
      "step": 375
    },
    {
      "epoch": 5.757575757575758,
      "grad_norm": 0.2952459454536438,
      "learning_rate": 9.424242424242424e-05,
      "loss": 0.0541,
      "step": 380
    },
    {
      "epoch": 5.833333333333333,
      "grad_norm": 0.38196662068367004,
      "learning_rate": 9.416666666666667e-05,
      "loss": 0.0543,
      "step": 385
    },
    {
      "epoch": 5.909090909090909,
      "grad_norm": 0.46261489391326904,
      "learning_rate": 9.40909090909091e-05,
      "loss": 0.0519,
      "step": 390
    },
    {
      "epoch": 5.984848484848484,
      "grad_norm": 0.2180977165699005,
      "learning_rate": 9.401515151515153e-05,
      "loss": 0.0507,
      "step": 395
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.05079437419772148,
      "eval_mean_accuracy": 0.9489212804725593,
      "eval_mean_iou": 0.9132776641074732,
      "eval_runtime": 134.2982,
      "eval_samples_per_second": 1.75,
      "eval_steps_per_second": 0.223,
      "step": 396
    },
    {
      "epoch": 6.0606060606060606,
      "grad_norm": 0.26535189151763916,
      "learning_rate": 9.393939393939395e-05,
      "loss": 0.0548,
      "step": 400
    },
    {
      "epoch": 6.136363636363637,
      "grad_norm": 0.30551937222480774,
      "learning_rate": 9.386363636363637e-05,
      "loss": 0.0503,
      "step": 405
    },
    {
      "epoch": 6.212121212121212,
      "grad_norm": 0.2614859640598297,
      "learning_rate": 9.378787878787879e-05,
      "loss": 0.05,
      "step": 410
    },
    {
      "epoch": 6.287878787878788,
      "grad_norm": 0.24407653510570526,
      "learning_rate": 9.371212121212122e-05,
      "loss": 0.0553,
      "step": 415
    },
    {
      "epoch": 6.363636363636363,
      "grad_norm": 0.21342405676841736,
      "learning_rate": 9.363636363636364e-05,
      "loss": 0.0578,
      "step": 420
    },
    {
      "epoch": 6.4393939393939394,
      "grad_norm": 0.2092735469341278,
      "learning_rate": 9.356060606060606e-05,
      "loss": 0.0479,
      "step": 425
    },
    {
      "epoch": 6.515151515151516,
      "grad_norm": 2.144221067428589,
      "learning_rate": 9.348484848484849e-05,
      "loss": 0.0513,
      "step": 430
    },
    {
      "epoch": 6.590909090909091,
      "grad_norm": 0.24289515614509583,
      "learning_rate": 9.340909090909091e-05,
      "loss": 0.0488,
      "step": 435
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 0.23668228089809418,
      "learning_rate": 9.333333333333334e-05,
      "loss": 0.0521,
      "step": 440
    },
    {
      "epoch": 6.742424242424242,
      "grad_norm": 0.31861612200737,
      "learning_rate": 9.325757575757576e-05,
      "loss": 0.0451,
      "step": 445
    },
    {
      "epoch": 6.818181818181818,
      "grad_norm": 0.9867147207260132,
      "learning_rate": 9.318181818181818e-05,
      "loss": 0.0461,
      "step": 450
    },
    {
      "epoch": 6.893939393939394,
      "grad_norm": 0.43805214762687683,
      "learning_rate": 9.31060606060606e-05,
      "loss": 0.0427,
      "step": 455
    },
    {
      "epoch": 6.96969696969697,
      "grad_norm": 0.35387301445007324,
      "learning_rate": 9.303030303030303e-05,
      "loss": 0.0464,
      "step": 460
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.04212049022316933,
      "eval_mean_accuracy": 0.9516008510568517,
      "eval_mean_iou": 0.9125883959026395,
      "eval_runtime": 125.1991,
      "eval_samples_per_second": 1.877,
      "eval_steps_per_second": 0.24,
      "step": 462
    },
    {
      "epoch": 7.045454545454546,
      "grad_norm": 0.2569555342197418,
      "learning_rate": 9.295454545454545e-05,
      "loss": 0.0445,
      "step": 465
    },
    {
      "epoch": 7.121212121212121,
      "grad_norm": 0.41528481245040894,
      "learning_rate": 9.287878787878789e-05,
      "loss": 0.0397,
      "step": 470
    },
    {
      "epoch": 7.196969696969697,
      "grad_norm": 0.2989051938056946,
      "learning_rate": 9.280303030303031e-05,
      "loss": 0.0393,
      "step": 475
    },
    {
      "epoch": 7.2727272727272725,
      "grad_norm": 0.2364627718925476,
      "learning_rate": 9.272727272727273e-05,
      "loss": 0.0425,
      "step": 480
    },
    {
      "epoch": 7.348484848484849,
      "grad_norm": 0.1845848113298416,
      "learning_rate": 9.265151515151516e-05,
      "loss": 0.0367,
      "step": 485
    },
    {
      "epoch": 7.424242424242424,
      "grad_norm": 0.20851445198059082,
      "learning_rate": 9.257575757575758e-05,
      "loss": 0.0375,
      "step": 490
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.21592620015144348,
      "learning_rate": 9.250000000000001e-05,
      "loss": 0.0419,
      "step": 495
    },
    {
      "epoch": 7.575757575757576,
      "grad_norm": 0.29624173045158386,
      "learning_rate": 9.242424242424242e-05,
      "loss": 0.0363,
      "step": 500
    },
    {
      "epoch": 7.651515151515151,
      "grad_norm": 0.16826100647449493,
      "learning_rate": 9.234848484848485e-05,
      "loss": 0.0392,
      "step": 505
    },
    {
      "epoch": 7.7272727272727275,
      "grad_norm": 0.17750632762908936,
      "learning_rate": 9.227272727272727e-05,
      "loss": 0.0402,
      "step": 510
    },
    {
      "epoch": 7.803030303030303,
      "grad_norm": 0.21508976817131042,
      "learning_rate": 9.21969696969697e-05,
      "loss": 0.0378,
      "step": 515
    },
    {
      "epoch": 7.878787878787879,
      "grad_norm": 0.24081674218177795,
      "learning_rate": 9.212121212121214e-05,
      "loss": 0.0414,
      "step": 520
    },
    {
      "epoch": 7.954545454545455,
      "grad_norm": 0.24248597025871277,
      "learning_rate": 9.204545454545454e-05,
      "loss": 0.0378,
      "step": 525
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.035655319690704346,
      "eval_mean_accuracy": 0.9592103038571963,
      "eval_mean_iou": 0.9154124286058577,
      "eval_runtime": 109.9401,
      "eval_samples_per_second": 2.138,
      "eval_steps_per_second": 0.273,
      "step": 528
    },
    {
      "epoch": 8.030303030303031,
      "grad_norm": 0.16671693325042725,
      "learning_rate": 9.196969696969698e-05,
      "loss": 0.0334,
      "step": 530
    },
    {
      "epoch": 8.106060606060606,
      "grad_norm": 0.20525330305099487,
      "learning_rate": 9.18939393939394e-05,
      "loss": 0.0373,
      "step": 535
    },
    {
      "epoch": 8.181818181818182,
      "grad_norm": 0.28408700227737427,
      "learning_rate": 9.181818181818183e-05,
      "loss": 0.0362,
      "step": 540
    },
    {
      "epoch": 8.257575757575758,
      "grad_norm": 0.1763504445552826,
      "learning_rate": 9.174242424242425e-05,
      "loss": 0.0321,
      "step": 545
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 0.1884240061044693,
      "learning_rate": 9.166666666666667e-05,
      "loss": 0.0371,
      "step": 550
    },
    {
      "epoch": 8.409090909090908,
      "grad_norm": 0.18748901784420013,
      "learning_rate": 9.159090909090909e-05,
      "loss": 0.0373,
      "step": 555
    },
    {
      "epoch": 8.484848484848484,
      "grad_norm": 0.24111661314964294,
      "learning_rate": 9.151515151515152e-05,
      "loss": 0.0332,
      "step": 560
    },
    {
      "epoch": 8.56060606060606,
      "grad_norm": 0.1503227800130844,
      "learning_rate": 9.143939393939395e-05,
      "loss": 0.0376,
      "step": 565
    },
    {
      "epoch": 8.636363636363637,
      "grad_norm": 0.14675354957580566,
      "learning_rate": 9.136363636363637e-05,
      "loss": 0.0327,
      "step": 570
    },
    {
      "epoch": 8.712121212121213,
      "grad_norm": 0.18998666107654572,
      "learning_rate": 9.128787878787879e-05,
      "loss": 0.0326,
      "step": 575
    },
    {
      "epoch": 8.787878787878787,
      "grad_norm": 0.20289573073387146,
      "learning_rate": 9.121212121212121e-05,
      "loss": 0.0309,
      "step": 580
    },
    {
      "epoch": 8.863636363636363,
      "grad_norm": 0.13536888360977173,
      "learning_rate": 9.113636363636365e-05,
      "loss": 0.0333,
      "step": 585
    },
    {
      "epoch": 8.93939393939394,
      "grad_norm": 0.2469329833984375,
      "learning_rate": 9.106060606060606e-05,
      "loss": 0.0351,
      "step": 590
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.032609354704618454,
      "eval_mean_accuracy": 0.9679423491949052,
      "eval_mean_iou": 0.906700960462588,
      "eval_runtime": 111.8114,
      "eval_samples_per_second": 2.102,
      "eval_steps_per_second": 0.268,
      "step": 594
    },
    {
      "epoch": 9.015151515151516,
      "grad_norm": 0.14683015644550323,
      "learning_rate": 9.098484848484848e-05,
      "loss": 0.028,
      "step": 595
    },
    {
      "epoch": 9.090909090909092,
      "grad_norm": 0.1612846404314041,
      "learning_rate": 9.090909090909092e-05,
      "loss": 0.0332,
      "step": 600
    },
    {
      "epoch": 9.166666666666666,
      "grad_norm": 0.18700768053531647,
      "learning_rate": 9.083333333333334e-05,
      "loss": 0.0296,
      "step": 605
    },
    {
      "epoch": 9.242424242424242,
      "grad_norm": 0.1474536508321762,
      "learning_rate": 9.075757575757577e-05,
      "loss": 0.0308,
      "step": 610
    },
    {
      "epoch": 9.318181818181818,
      "grad_norm": 0.25029927492141724,
      "learning_rate": 9.068181818181819e-05,
      "loss": 0.0359,
      "step": 615
    },
    {
      "epoch": 9.393939393939394,
      "grad_norm": 0.14850947260856628,
      "learning_rate": 9.060606060606061e-05,
      "loss": 0.0277,
      "step": 620
    },
    {
      "epoch": 9.469696969696969,
      "grad_norm": 0.20516231656074524,
      "learning_rate": 9.053030303030303e-05,
      "loss": 0.0278,
      "step": 625
    },
    {
      "epoch": 9.545454545454545,
      "grad_norm": 0.12035396695137024,
      "learning_rate": 9.045454545454546e-05,
      "loss": 0.0297,
      "step": 630
    },
    {
      "epoch": 9.621212121212121,
      "grad_norm": 0.5076970458030701,
      "learning_rate": 9.037878787878788e-05,
      "loss": 0.0334,
      "step": 635
    },
    {
      "epoch": 9.696969696969697,
      "grad_norm": 0.1543891280889511,
      "learning_rate": 9.030303030303031e-05,
      "loss": 0.0331,
      "step": 640
    },
    {
      "epoch": 9.772727272727273,
      "grad_norm": 0.7929154634475708,
      "learning_rate": 9.022727272727273e-05,
      "loss": 0.0367,
      "step": 645
    },
    {
      "epoch": 9.848484848484848,
      "grad_norm": 0.26206859946250916,
      "learning_rate": 9.015151515151515e-05,
      "loss": 0.0332,
      "step": 650
    },
    {
      "epoch": 9.924242424242424,
      "grad_norm": 0.2010592520236969,
      "learning_rate": 9.007575757575759e-05,
      "loss": 0.0315,
      "step": 655
    },
    {
      "epoch": 10.0,
      "grad_norm": 7.537333965301514,
      "learning_rate": 9e-05,
      "loss": 0.0728,
      "step": 660
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.027554607018828392,
      "eval_mean_accuracy": 0.9389347378035855,
      "eval_mean_iou": 0.9116466786413523,
      "eval_runtime": 109.878,
      "eval_samples_per_second": 2.139,
      "eval_steps_per_second": 0.273,
      "step": 660
    },
    {
      "epoch": 10.075757575757576,
      "grad_norm": 0.302475243806839,
      "learning_rate": 8.992424242424244e-05,
      "loss": 0.0323,
      "step": 665
    },
    {
      "epoch": 10.151515151515152,
      "grad_norm": 0.11345604807138443,
      "learning_rate": 8.984848484848484e-05,
      "loss": 0.0339,
      "step": 670
    },
    {
      "epoch": 10.227272727272727,
      "grad_norm": 0.11642608046531677,
      "learning_rate": 8.977272727272728e-05,
      "loss": 0.0356,
      "step": 675
    },
    {
      "epoch": 10.303030303030303,
      "grad_norm": 0.2661907970905304,
      "learning_rate": 8.96969696969697e-05,
      "loss": 0.0301,
      "step": 680
    },
    {
      "epoch": 10.378787878787879,
      "grad_norm": 0.10854136198759079,
      "learning_rate": 8.962121212121213e-05,
      "loss": 0.0274,
      "step": 685
    },
    {
      "epoch": 10.454545454545455,
      "grad_norm": 0.10771843045949936,
      "learning_rate": 8.954545454545455e-05,
      "loss": 0.0298,
      "step": 690
    },
    {
      "epoch": 10.530303030303031,
      "grad_norm": 0.1267024278640747,
      "learning_rate": 8.946969696969697e-05,
      "loss": 0.0295,
      "step": 695
    },
    {
      "epoch": 10.606060606060606,
      "grad_norm": 0.11347940564155579,
      "learning_rate": 8.93939393939394e-05,
      "loss": 0.0274,
      "step": 700
    },
    {
      "epoch": 10.681818181818182,
      "grad_norm": 0.12592707574367523,
      "learning_rate": 8.931818181818182e-05,
      "loss": 0.028,
      "step": 705
    },
    {
      "epoch": 10.757575757575758,
      "grad_norm": 0.39438164234161377,
      "learning_rate": 8.924242424242426e-05,
      "loss": 0.0286,
      "step": 710
    },
    {
      "epoch": 10.833333333333334,
      "grad_norm": 0.3030393123626709,
      "learning_rate": 8.916666666666667e-05,
      "loss": 0.0311,
      "step": 715
    },
    {
      "epoch": 10.909090909090908,
      "grad_norm": 0.18854883313179016,
      "learning_rate": 8.90909090909091e-05,
      "loss": 0.057,
      "step": 720
    },
    {
      "epoch": 10.984848484848484,
      "grad_norm": 0.3189028799533844,
      "learning_rate": 8.901515151515151e-05,
      "loss": 0.0282,
      "step": 725
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.03061528503894806,
      "eval_mean_accuracy": 0.8918776151719268,
      "eval_mean_iou": 0.8709465457335627,
      "eval_runtime": 109.9569,
      "eval_samples_per_second": 2.137,
      "eval_steps_per_second": 0.273,
      "step": 726
    },
    {
      "epoch": 11.06060606060606,
      "grad_norm": 0.3581349849700928,
      "learning_rate": 8.893939393939395e-05,
      "loss": 0.0292,
      "step": 730
    },
    {
      "epoch": 11.136363636363637,
      "grad_norm": 0.12023809552192688,
      "learning_rate": 8.886363636363637e-05,
      "loss": 0.0245,
      "step": 735
    },
    {
      "epoch": 11.212121212121213,
      "grad_norm": 0.11607044190168381,
      "learning_rate": 8.87878787878788e-05,
      "loss": 0.0303,
      "step": 740
    },
    {
      "epoch": 11.287878787878787,
      "grad_norm": 0.17514030635356903,
      "learning_rate": 8.871212121212122e-05,
      "loss": 0.0271,
      "step": 745
    },
    {
      "epoch": 11.363636363636363,
      "grad_norm": 0.2238277941942215,
      "learning_rate": 8.863636363636364e-05,
      "loss": 0.0285,
      "step": 750
    },
    {
      "epoch": 11.43939393939394,
      "grad_norm": 0.15197783708572388,
      "learning_rate": 8.856060606060607e-05,
      "loss": 0.024,
      "step": 755
    },
    {
      "epoch": 11.515151515151516,
      "grad_norm": 0.27303916215896606,
      "learning_rate": 8.848484848484849e-05,
      "loss": 0.0275,
      "step": 760
    },
    {
      "epoch": 11.590909090909092,
      "grad_norm": 0.17839168012142181,
      "learning_rate": 8.840909090909091e-05,
      "loss": 0.0258,
      "step": 765
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 0.09125503152608871,
      "learning_rate": 8.833333333333333e-05,
      "loss": 0.0242,
      "step": 770
    },
    {
      "epoch": 11.742424242424242,
      "grad_norm": 0.22240027785301208,
      "learning_rate": 8.825757575757576e-05,
      "loss": 0.0228,
      "step": 775
    },
    {
      "epoch": 11.818181818181818,
      "grad_norm": 0.13228470087051392,
      "learning_rate": 8.818181818181818e-05,
      "loss": 0.0249,
      "step": 780
    },
    {
      "epoch": 11.893939393939394,
      "grad_norm": 0.14466527104377747,
      "learning_rate": 8.810606060606062e-05,
      "loss": 0.029,
      "step": 785
    },
    {
      "epoch": 11.969696969696969,
      "grad_norm": 0.1271480917930603,
      "learning_rate": 8.803030303030304e-05,
      "loss": 0.0247,
      "step": 790
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.024765823036432266,
      "eval_mean_accuracy": 0.9168595632876241,
      "eval_mean_iou": 0.8958466371074029,
      "eval_runtime": 110.2709,
      "eval_samples_per_second": 2.131,
      "eval_steps_per_second": 0.272,
      "step": 792
    },
    {
      "epoch": 12.045454545454545,
      "grad_norm": 0.13231201469898224,
      "learning_rate": 8.795454545454545e-05,
      "loss": 0.0283,
      "step": 795
    },
    {
      "epoch": 12.121212121212121,
      "grad_norm": 0.22508250176906586,
      "learning_rate": 8.787878787878789e-05,
      "loss": 0.0276,
      "step": 800
    },
    {
      "epoch": 12.196969696969697,
      "grad_norm": 0.12359216809272766,
      "learning_rate": 8.780303030303031e-05,
      "loss": 0.0239,
      "step": 805
    },
    {
      "epoch": 12.272727272727273,
      "grad_norm": 0.10091531276702881,
      "learning_rate": 8.772727272727274e-05,
      "loss": 0.022,
      "step": 810
    },
    {
      "epoch": 12.348484848484848,
      "grad_norm": 0.1036147028207779,
      "learning_rate": 8.765151515151515e-05,
      "loss": 0.023,
      "step": 815
    },
    {
      "epoch": 12.424242424242424,
      "grad_norm": 0.09251093864440918,
      "learning_rate": 8.757575757575758e-05,
      "loss": 0.0245,
      "step": 820
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.1625688523054123,
      "learning_rate": 8.75e-05,
      "loss": 0.0266,
      "step": 825
    },
    {
      "epoch": 12.575757575757576,
      "grad_norm": 0.114060178399086,
      "learning_rate": 8.742424242424243e-05,
      "loss": 0.0241,
      "step": 830
    },
    {
      "epoch": 12.651515151515152,
      "grad_norm": 0.08975587040185928,
      "learning_rate": 8.734848484848485e-05,
      "loss": 0.022,
      "step": 835
    },
    {
      "epoch": 12.727272727272727,
      "grad_norm": 0.12875740230083466,
      "learning_rate": 8.727272727272727e-05,
      "loss": 0.0198,
      "step": 840
    },
    {
      "epoch": 12.803030303030303,
      "grad_norm": 0.22359319031238556,
      "learning_rate": 8.71969696969697e-05,
      "loss": 0.0262,
      "step": 845
    },
    {
      "epoch": 12.878787878787879,
      "grad_norm": 0.10468384623527527,
      "learning_rate": 8.712121212121212e-05,
      "loss": 0.0221,
      "step": 850
    },
    {
      "epoch": 12.954545454545455,
      "grad_norm": 0.09065494686365128,
      "learning_rate": 8.704545454545456e-05,
      "loss": 0.0203,
      "step": 855
    },
    {
      "epoch": 13.0,
      "eval_loss": 0.022203492000699043,
      "eval_mean_accuracy": 0.9325883068492704,
      "eval_mean_iou": 0.9097222385998064,
      "eval_runtime": 109.9147,
      "eval_samples_per_second": 2.138,
      "eval_steps_per_second": 0.273,
      "step": 858
    },
    {
      "epoch": 13.030303030303031,
      "grad_norm": 0.18725955486297607,
      "learning_rate": 8.696969696969698e-05,
      "loss": 0.0206,
      "step": 860
    },
    {
      "epoch": 13.106060606060606,
      "grad_norm": 0.17102640867233276,
      "learning_rate": 8.68939393939394e-05,
      "loss": 0.0242,
      "step": 865
    },
    {
      "epoch": 13.181818181818182,
      "grad_norm": 0.11914201080799103,
      "learning_rate": 8.681818181818182e-05,
      "loss": 0.0181,
      "step": 870
    },
    {
      "epoch": 13.257575757575758,
      "grad_norm": 0.08897839486598969,
      "learning_rate": 8.674242424242425e-05,
      "loss": 0.0221,
      "step": 875
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 0.13805468380451202,
      "learning_rate": 8.666666666666667e-05,
      "loss": 0.022,
      "step": 880
    },
    {
      "epoch": 13.409090909090908,
      "grad_norm": 0.10788358002901077,
      "learning_rate": 8.65909090909091e-05,
      "loss": 0.0304,
      "step": 885
    },
    {
      "epoch": 13.484848484848484,
      "grad_norm": 0.28546327352523804,
      "learning_rate": 8.651515151515152e-05,
      "loss": 0.0252,
      "step": 890
    },
    {
      "epoch": 13.56060606060606,
      "grad_norm": 0.2713989317417145,
      "learning_rate": 8.643939393939394e-05,
      "loss": 0.0221,
      "step": 895
    },
    {
      "epoch": 13.636363636363637,
      "grad_norm": 0.11129526793956757,
      "learning_rate": 8.636363636363637e-05,
      "loss": 0.0212,
      "step": 900
    },
    {
      "epoch": 13.712121212121213,
      "grad_norm": 0.1384202092885971,
      "learning_rate": 8.628787878787879e-05,
      "loss": 0.0179,
      "step": 905
    },
    {
      "epoch": 13.787878787878787,
      "grad_norm": 0.08507904410362244,
      "learning_rate": 8.621212121212121e-05,
      "loss": 0.0209,
      "step": 910
    },
    {
      "epoch": 13.863636363636363,
      "grad_norm": 0.11744596064090729,
      "learning_rate": 8.613636363636363e-05,
      "loss": 0.0198,
      "step": 915
    },
    {
      "epoch": 13.93939393939394,
      "grad_norm": 0.09632156789302826,
      "learning_rate": 8.606060606060606e-05,
      "loss": 0.0215,
      "step": 920
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.02144075743854046,
      "eval_mean_accuracy": 0.9329142370069784,
      "eval_mean_iou": 0.9107059081089334,
      "eval_runtime": 122.6782,
      "eval_samples_per_second": 1.916,
      "eval_steps_per_second": 0.245,
      "step": 924
    },
    {
      "epoch": 14.015151515151516,
      "grad_norm": 0.21378231048583984,
      "learning_rate": 8.598484848484848e-05,
      "loss": 0.0183,
      "step": 925
    },
    {
      "epoch": 14.090909090909092,
      "grad_norm": 0.09192424267530441,
      "learning_rate": 8.590909090909092e-05,
      "loss": 0.0192,
      "step": 930
    },
    {
      "epoch": 14.166666666666666,
      "grad_norm": 0.12242019921541214,
      "learning_rate": 8.583333333333334e-05,
      "loss": 0.0173,
      "step": 935
    },
    {
      "epoch": 14.242424242424242,
      "grad_norm": 0.11881183832883835,
      "learning_rate": 8.575757575757576e-05,
      "loss": 0.0188,
      "step": 940
    },
    {
      "epoch": 14.318181818181818,
      "grad_norm": 0.07136339694261551,
      "learning_rate": 8.568181818181819e-05,
      "loss": 0.0179,
      "step": 945
    },
    {
      "epoch": 14.393939393939394,
      "grad_norm": 0.12548968195915222,
      "learning_rate": 8.560606060606061e-05,
      "loss": 0.0214,
      "step": 950
    },
    {
      "epoch": 14.469696969696969,
      "grad_norm": 0.17797769606113434,
      "learning_rate": 8.553030303030304e-05,
      "loss": 0.0222,
      "step": 955
    },
    {
      "epoch": 14.545454545454545,
      "grad_norm": 0.08581986278295517,
      "learning_rate": 8.545454545454545e-05,
      "loss": 0.0198,
      "step": 960
    },
    {
      "epoch": 14.621212121212121,
      "grad_norm": 0.07558993250131607,
      "learning_rate": 8.537878787878788e-05,
      "loss": 0.0215,
      "step": 965
    },
    {
      "epoch": 14.696969696969697,
      "grad_norm": 0.11741326749324799,
      "learning_rate": 8.53030303030303e-05,
      "loss": 0.0188,
      "step": 970
    },
    {
      "epoch": 14.772727272727273,
      "grad_norm": 0.9199545979499817,
      "learning_rate": 8.522727272727273e-05,
      "loss": 0.0174,
      "step": 975
    },
    {
      "epoch": 14.848484848484848,
      "grad_norm": 0.07153519243001938,
      "learning_rate": 8.515151515151515e-05,
      "loss": 0.0163,
      "step": 980
    },
    {
      "epoch": 14.924242424242424,
      "grad_norm": 0.12593400478363037,
      "learning_rate": 8.507575757575757e-05,
      "loss": 0.0208,
      "step": 985
    },
    {
      "epoch": 15.0,
      "grad_norm": 2.7577977180480957,
      "learning_rate": 8.5e-05,
      "loss": 0.0211,
      "step": 990
    },
    {
      "epoch": 15.0,
      "eval_loss": 0.019323177635669708,
      "eval_mean_accuracy": 0.9464577147472207,
      "eval_mean_iou": 0.9193393690353729,
      "eval_runtime": 128.4809,
      "eval_samples_per_second": 1.829,
      "eval_steps_per_second": 0.233,
      "step": 990
    },
    {
      "epoch": 15.075757575757576,
      "grad_norm": 0.1450919657945633,
      "learning_rate": 8.492424242424243e-05,
      "loss": 0.0173,
      "step": 995
    },
    {
      "epoch": 15.151515151515152,
      "grad_norm": 0.0656338483095169,
      "learning_rate": 8.484848484848486e-05,
      "loss": 0.0158,
      "step": 1000
    },
    {
      "epoch": 15.227272727272727,
      "grad_norm": 0.09024962037801743,
      "learning_rate": 8.477272727272728e-05,
      "loss": 0.0238,
      "step": 1005
    },
    {
      "epoch": 15.303030303030303,
      "grad_norm": 0.06770208477973938,
      "learning_rate": 8.46969696969697e-05,
      "loss": 0.0164,
      "step": 1010
    },
    {
      "epoch": 15.378787878787879,
      "grad_norm": 0.09075728058815002,
      "learning_rate": 8.462121212121212e-05,
      "loss": 0.016,
      "step": 1015
    },
    {
      "epoch": 15.454545454545455,
      "grad_norm": 0.05959879234433174,
      "learning_rate": 8.454545454545455e-05,
      "loss": 0.0189,
      "step": 1020
    },
    {
      "epoch": 15.530303030303031,
      "grad_norm": 0.11655189841985703,
      "learning_rate": 8.446969696969697e-05,
      "loss": 0.0196,
      "step": 1025
    },
    {
      "epoch": 15.606060606060606,
      "grad_norm": 0.09038227796554565,
      "learning_rate": 8.43939393939394e-05,
      "loss": 0.0175,
      "step": 1030
    },
    {
      "epoch": 15.681818181818182,
      "grad_norm": 0.19900625944137573,
      "learning_rate": 8.431818181818182e-05,
      "loss": 0.0209,
      "step": 1035
    },
    {
      "epoch": 15.757575757575758,
      "grad_norm": 0.09526349604129791,
      "learning_rate": 8.424242424242424e-05,
      "loss": 0.0191,
      "step": 1040
    },
    {
      "epoch": 15.833333333333334,
      "grad_norm": 0.12076755613088608,
      "learning_rate": 8.416666666666668e-05,
      "loss": 0.0239,
      "step": 1045
    },
    {
      "epoch": 15.909090909090908,
      "grad_norm": 0.093843013048172,
      "learning_rate": 8.40909090909091e-05,
      "loss": 0.0177,
      "step": 1050
    },
    {
      "epoch": 15.984848484848484,
      "grad_norm": 0.06355836987495422,
      "learning_rate": 8.401515151515153e-05,
      "loss": 0.0206,
      "step": 1055
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.018249332904815674,
      "eval_mean_accuracy": 0.9605840301041596,
      "eval_mean_iou": 0.9193997899635725,
      "eval_runtime": 130.0143,
      "eval_samples_per_second": 1.807,
      "eval_steps_per_second": 0.231,
      "step": 1056
    },
    {
      "epoch": 16.060606060606062,
      "grad_norm": 0.24040895700454712,
      "learning_rate": 8.393939393939393e-05,
      "loss": 0.0209,
      "step": 1060
    },
    {
      "epoch": 16.136363636363637,
      "grad_norm": 0.29702943563461304,
      "learning_rate": 8.386363636363637e-05,
      "loss": 0.0179,
      "step": 1065
    },
    {
      "epoch": 16.21212121212121,
      "grad_norm": 0.08094320446252823,
      "learning_rate": 8.378787878787879e-05,
      "loss": 0.0203,
      "step": 1070
    },
    {
      "epoch": 16.28787878787879,
      "grad_norm": 0.1057291030883789,
      "learning_rate": 8.371212121212122e-05,
      "loss": 0.0153,
      "step": 1075
    },
    {
      "epoch": 16.363636363636363,
      "grad_norm": 0.16128107905387878,
      "learning_rate": 8.363636363636364e-05,
      "loss": 0.0192,
      "step": 1080
    },
    {
      "epoch": 16.439393939393938,
      "grad_norm": 0.11210722476243973,
      "learning_rate": 8.356060606060606e-05,
      "loss": 0.0175,
      "step": 1085
    },
    {
      "epoch": 16.515151515151516,
      "grad_norm": 0.21033479273319244,
      "learning_rate": 8.348484848484849e-05,
      "loss": 0.0198,
      "step": 1090
    },
    {
      "epoch": 16.59090909090909,
      "grad_norm": 0.10092729330062866,
      "learning_rate": 8.340909090909091e-05,
      "loss": 0.0156,
      "step": 1095
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.25405940413475037,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.022,
      "step": 1100
    },
    {
      "epoch": 16.742424242424242,
      "grad_norm": 0.0710429847240448,
      "learning_rate": 8.325757575757575e-05,
      "loss": 0.0161,
      "step": 1105
    },
    {
      "epoch": 16.818181818181817,
      "grad_norm": 0.07577074319124222,
      "learning_rate": 8.318181818181818e-05,
      "loss": 0.0172,
      "step": 1110
    },
    {
      "epoch": 16.893939393939394,
      "grad_norm": 0.10891614109277725,
      "learning_rate": 8.310606060606062e-05,
      "loss": 0.0151,
      "step": 1115
    },
    {
      "epoch": 16.96969696969697,
      "grad_norm": 0.08066507428884506,
      "learning_rate": 8.303030303030304e-05,
      "loss": 0.0141,
      "step": 1120
    },
    {
      "epoch": 17.0,
      "eval_loss": 0.017201829701662064,
      "eval_mean_accuracy": 0.9479456301608019,
      "eval_mean_iou": 0.9224179200381082,
      "eval_runtime": 126.8333,
      "eval_samples_per_second": 1.853,
      "eval_steps_per_second": 0.237,
      "step": 1122
    },
    {
      "epoch": 17.045454545454547,
      "grad_norm": 0.11866308003664017,
      "learning_rate": 8.295454545454547e-05,
      "loss": 0.0184,
      "step": 1125
    },
    {
      "epoch": 17.12121212121212,
      "grad_norm": 0.1006651371717453,
      "learning_rate": 8.287878787878787e-05,
      "loss": 0.018,
      "step": 1130
    },
    {
      "epoch": 17.196969696969695,
      "grad_norm": 0.19091796875,
      "learning_rate": 8.280303030303031e-05,
      "loss": 0.0158,
      "step": 1135
    },
    {
      "epoch": 17.272727272727273,
      "grad_norm": 0.06820344179868698,
      "learning_rate": 8.272727272727273e-05,
      "loss": 0.0188,
      "step": 1140
    },
    {
      "epoch": 17.348484848484848,
      "grad_norm": 0.2098952978849411,
      "learning_rate": 8.265151515151516e-05,
      "loss": 0.0178,
      "step": 1145
    },
    {
      "epoch": 17.424242424242426,
      "grad_norm": 0.06817901134490967,
      "learning_rate": 8.257575757575758e-05,
      "loss": 0.0172,
      "step": 1150
    },
    {
      "epoch": 17.5,
      "grad_norm": 0.15966415405273438,
      "learning_rate": 8.25e-05,
      "loss": 0.0188,
      "step": 1155
    },
    {
      "epoch": 17.575757575757574,
      "grad_norm": 0.16438080370426178,
      "learning_rate": 8.242424242424243e-05,
      "loss": 0.015,
      "step": 1160
    },
    {
      "epoch": 17.651515151515152,
      "grad_norm": 0.07232612371444702,
      "learning_rate": 8.234848484848485e-05,
      "loss": 0.0141,
      "step": 1165
    },
    {
      "epoch": 17.727272727272727,
      "grad_norm": 0.056016601622104645,
      "learning_rate": 8.227272727272729e-05,
      "loss": 0.0143,
      "step": 1170
    },
    {
      "epoch": 17.803030303030305,
      "grad_norm": 0.05460710823535919,
      "learning_rate": 8.21969696969697e-05,
      "loss": 0.0152,
      "step": 1175
    },
    {
      "epoch": 17.87878787878788,
      "grad_norm": 0.05740410089492798,
      "learning_rate": 8.212121212121212e-05,
      "loss": 0.0133,
      "step": 1180
    },
    {
      "epoch": 17.954545454545453,
      "grad_norm": 0.14944913983345032,
      "learning_rate": 8.204545454545454e-05,
      "loss": 0.016,
      "step": 1185
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.01630590483546257,
      "eval_mean_accuracy": 0.9588449580060451,
      "eval_mean_iou": 0.9254050524385349,
      "eval_runtime": 128.1471,
      "eval_samples_per_second": 1.834,
      "eval_steps_per_second": 0.234,
      "step": 1188
    },
    {
      "epoch": 18.03030303030303,
      "grad_norm": 0.054342471063137054,
      "learning_rate": 8.196969696969698e-05,
      "loss": 0.0173,
      "step": 1190
    },
    {
      "epoch": 18.106060606060606,
      "grad_norm": 0.0768555998802185,
      "learning_rate": 8.18939393939394e-05,
      "loss": 0.016,
      "step": 1195
    },
    {
      "epoch": 18.181818181818183,
      "grad_norm": 0.07927267253398895,
      "learning_rate": 8.181818181818183e-05,
      "loss": 0.0169,
      "step": 1200
    },
    {
      "epoch": 18.257575757575758,
      "grad_norm": 0.0649801567196846,
      "learning_rate": 8.174242424242425e-05,
      "loss": 0.0167,
      "step": 1205
    },
    {
      "epoch": 18.333333333333332,
      "grad_norm": 0.06333769112825394,
      "learning_rate": 8.166666666666667e-05,
      "loss": 0.0135,
      "step": 1210
    },
    {
      "epoch": 18.40909090909091,
      "grad_norm": 0.07222408801317215,
      "learning_rate": 8.15909090909091e-05,
      "loss": 0.0189,
      "step": 1215
    },
    {
      "epoch": 18.484848484848484,
      "grad_norm": 0.07573547214269638,
      "learning_rate": 8.151515151515152e-05,
      "loss": 0.0173,
      "step": 1220
    },
    {
      "epoch": 18.560606060606062,
      "grad_norm": 0.12189855426549911,
      "learning_rate": 8.143939393939395e-05,
      "loss": 0.0189,
      "step": 1225
    },
    {
      "epoch": 18.636363636363637,
      "grad_norm": 0.06855466216802597,
      "learning_rate": 8.136363636363636e-05,
      "loss": 0.0132,
      "step": 1230
    },
    {
      "epoch": 18.71212121212121,
      "grad_norm": 0.08273063600063324,
      "learning_rate": 8.12878787878788e-05,
      "loss": 0.0149,
      "step": 1235
    },
    {
      "epoch": 18.78787878787879,
      "grad_norm": 0.06296434998512268,
      "learning_rate": 8.121212121212121e-05,
      "loss": 0.0137,
      "step": 1240
    },
    {
      "epoch": 18.863636363636363,
      "grad_norm": 0.11089876294136047,
      "learning_rate": 8.113636363636365e-05,
      "loss": 0.0149,
      "step": 1245
    },
    {
      "epoch": 18.939393939393938,
      "grad_norm": 0.1310448795557022,
      "learning_rate": 8.106060606060607e-05,
      "loss": 0.0138,
      "step": 1250
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.016173426061868668,
      "eval_mean_accuracy": 0.9503850292035511,
      "eval_mean_iou": 0.9227455805816746,
      "eval_runtime": 127.3113,
      "eval_samples_per_second": 1.846,
      "eval_steps_per_second": 0.236,
      "step": 1254
    },
    {
      "epoch": 19.015151515151516,
      "grad_norm": 0.04924704506993294,
      "learning_rate": 8.098484848484848e-05,
      "loss": 0.0133,
      "step": 1255
    },
    {
      "epoch": 19.09090909090909,
      "grad_norm": 0.08565647900104523,
      "learning_rate": 8.090909090909092e-05,
      "loss": 0.0174,
      "step": 1260
    },
    {
      "epoch": 19.166666666666668,
      "grad_norm": 0.2731133699417114,
      "learning_rate": 8.083333333333334e-05,
      "loss": 0.0159,
      "step": 1265
    },
    {
      "epoch": 19.242424242424242,
      "grad_norm": 0.050038594752550125,
      "learning_rate": 8.075757575757577e-05,
      "loss": 0.014,
      "step": 1270
    },
    {
      "epoch": 19.318181818181817,
      "grad_norm": 0.049511101096868515,
      "learning_rate": 8.068181818181818e-05,
      "loss": 0.0149,
      "step": 1275
    },
    {
      "epoch": 19.393939393939394,
      "grad_norm": 0.10031841695308685,
      "learning_rate": 8.060606060606061e-05,
      "loss": 0.0164,
      "step": 1280
    },
    {
      "epoch": 19.46969696969697,
      "grad_norm": 0.2157675176858902,
      "learning_rate": 8.053030303030303e-05,
      "loss": 0.0177,
      "step": 1285
    },
    {
      "epoch": 19.545454545454547,
      "grad_norm": 0.11044749617576599,
      "learning_rate": 8.045454545454546e-05,
      "loss": 0.0126,
      "step": 1290
    },
    {
      "epoch": 19.62121212121212,
      "grad_norm": 0.05526359751820564,
      "learning_rate": 8.037878787878788e-05,
      "loss": 0.0135,
      "step": 1295
    },
    {
      "epoch": 19.696969696969695,
      "grad_norm": 0.06311521679162979,
      "learning_rate": 8.03030303030303e-05,
      "loss": 0.0174,
      "step": 1300
    },
    {
      "epoch": 19.772727272727273,
      "grad_norm": 0.07769162207841873,
      "learning_rate": 8.022727272727273e-05,
      "loss": 0.0176,
      "step": 1305
    },
    {
      "epoch": 19.848484848484848,
      "grad_norm": 0.6201515793800354,
      "learning_rate": 8.015151515151515e-05,
      "loss": 0.0152,
      "step": 1310
    },
    {
      "epoch": 19.924242424242426,
      "grad_norm": 0.19936323165893555,
      "learning_rate": 8.007575757575759e-05,
      "loss": 0.0218,
      "step": 1315
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.08982124924659729,
      "learning_rate": 8e-05,
      "loss": 0.0161,
      "step": 1320
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.016703521832823753,
      "eval_mean_accuracy": 0.9478463648237497,
      "eval_mean_iou": 0.9169158564335724,
      "eval_runtime": 129.886,
      "eval_samples_per_second": 1.809,
      "eval_steps_per_second": 0.231,
      "step": 1320
    },
    {
      "epoch": 20.075757575757574,
      "grad_norm": 0.10852258652448654,
      "learning_rate": 7.992424242424243e-05,
      "loss": 0.0159,
      "step": 1325
    },
    {
      "epoch": 20.151515151515152,
      "grad_norm": 0.07423771172761917,
      "learning_rate": 7.984848484848485e-05,
      "loss": 0.0164,
      "step": 1330
    },
    {
      "epoch": 20.227272727272727,
      "grad_norm": 0.0760967805981636,
      "learning_rate": 7.977272727272728e-05,
      "loss": 0.0164,
      "step": 1335
    },
    {
      "epoch": 20.303030303030305,
      "grad_norm": 0.1271897554397583,
      "learning_rate": 7.96969696969697e-05,
      "loss": 0.0137,
      "step": 1340
    },
    {
      "epoch": 20.37878787878788,
      "grad_norm": 0.05332360789179802,
      "learning_rate": 7.962121212121213e-05,
      "loss": 0.0126,
      "step": 1345
    },
    {
      "epoch": 20.454545454545453,
      "grad_norm": 0.10887204855680466,
      "learning_rate": 7.954545454545455e-05,
      "loss": 0.0134,
      "step": 1350
    },
    {
      "epoch": 20.53030303030303,
      "grad_norm": 0.08481290936470032,
      "learning_rate": 7.946969696969697e-05,
      "loss": 0.0135,
      "step": 1355
    },
    {
      "epoch": 20.606060606060606,
      "grad_norm": 0.12505367398262024,
      "learning_rate": 7.93939393939394e-05,
      "loss": 0.0218,
      "step": 1360
    },
    {
      "epoch": 20.681818181818183,
      "grad_norm": 0.17305514216423035,
      "learning_rate": 7.931818181818182e-05,
      "loss": 0.0207,
      "step": 1365
    },
    {
      "epoch": 20.757575757575758,
      "grad_norm": 0.1940699964761734,
      "learning_rate": 7.924242424242426e-05,
      "loss": 0.0147,
      "step": 1370
    },
    {
      "epoch": 20.833333333333332,
      "grad_norm": 0.06477572023868561,
      "learning_rate": 7.916666666666666e-05,
      "loss": 0.0143,
      "step": 1375
    },
    {
      "epoch": 20.90909090909091,
      "grad_norm": 0.10808603465557098,
      "learning_rate": 7.90909090909091e-05,
      "loss": 0.0163,
      "step": 1380
    },
    {
      "epoch": 20.984848484848484,
      "grad_norm": 0.34553155303001404,
      "learning_rate": 7.901515151515151e-05,
      "loss": 0.0203,
      "step": 1385
    },
    {
      "epoch": 21.0,
      "eval_loss": 0.015599638223648071,
      "eval_mean_accuracy": 0.952480574089721,
      "eval_mean_iou": 0.9212913181682456,
      "eval_runtime": 128.8278,
      "eval_samples_per_second": 1.824,
      "eval_steps_per_second": 0.233,
      "step": 1386
    },
    {
      "epoch": 21.060606060606062,
      "grad_norm": 0.1115134209394455,
      "learning_rate": 7.893939393939395e-05,
      "loss": 0.0161,
      "step": 1390
    },
    {
      "epoch": 21.136363636363637,
      "grad_norm": 0.07223149389028549,
      "learning_rate": 7.886363636363637e-05,
      "loss": 0.016,
      "step": 1395
    },
    {
      "epoch": 21.21212121212121,
      "grad_norm": 0.11839142441749573,
      "learning_rate": 7.878787878787879e-05,
      "loss": 0.0152,
      "step": 1400
    },
    {
      "epoch": 21.28787878787879,
      "grad_norm": 0.04929115250706673,
      "learning_rate": 7.871212121212122e-05,
      "loss": 0.0169,
      "step": 1405
    },
    {
      "epoch": 21.363636363636363,
      "grad_norm": 0.0751696452498436,
      "learning_rate": 7.863636363636364e-05,
      "loss": 0.0133,
      "step": 1410
    },
    {
      "epoch": 21.439393939393938,
      "grad_norm": 0.10838822275400162,
      "learning_rate": 7.856060606060607e-05,
      "loss": 0.015,
      "step": 1415
    },
    {
      "epoch": 21.515151515151516,
      "grad_norm": 0.12898345291614532,
      "learning_rate": 7.848484848484848e-05,
      "loss": 0.0159,
      "step": 1420
    },
    {
      "epoch": 21.59090909090909,
      "grad_norm": 0.060182053595781326,
      "learning_rate": 7.840909090909091e-05,
      "loss": 0.0138,
      "step": 1425
    },
    {
      "epoch": 21.666666666666668,
      "grad_norm": 0.08941216766834259,
      "learning_rate": 7.833333333333333e-05,
      "loss": 0.0145,
      "step": 1430
    },
    {
      "epoch": 21.742424242424242,
      "grad_norm": 0.13711506128311157,
      "learning_rate": 7.825757575757576e-05,
      "loss": 0.016,
      "step": 1435
    },
    {
      "epoch": 21.818181818181817,
      "grad_norm": 0.19674240052700043,
      "learning_rate": 7.818181818181818e-05,
      "loss": 0.0146,
      "step": 1440
    },
    {
      "epoch": 21.893939393939394,
      "grad_norm": 0.09153243899345398,
      "learning_rate": 7.81060606060606e-05,
      "loss": 0.0165,
      "step": 1445
    },
    {
      "epoch": 21.96969696969697,
      "grad_norm": 0.055770087987184525,
      "learning_rate": 7.803030303030304e-05,
      "loss": 0.0126,
      "step": 1450
    },
    {
      "epoch": 22.0,
      "eval_loss": 0.0154061084613204,
      "eval_mean_accuracy": 0.9362286443627124,
      "eval_mean_iou": 0.9136977942344185,
      "eval_runtime": 112.5965,
      "eval_samples_per_second": 2.087,
      "eval_steps_per_second": 0.266,
      "step": 1452
    },
    {
      "epoch": 22.045454545454547,
      "grad_norm": 0.11982175707817078,
      "learning_rate": 7.795454545454546e-05,
      "loss": 0.0133,
      "step": 1455
    },
    {
      "epoch": 22.12121212121212,
      "grad_norm": 0.12388788163661957,
      "learning_rate": 7.787878787878789e-05,
      "loss": 0.0128,
      "step": 1460
    },
    {
      "epoch": 22.196969696969695,
      "grad_norm": 0.04308038949966431,
      "learning_rate": 7.780303030303031e-05,
      "loss": 0.0125,
      "step": 1465
    },
    {
      "epoch": 22.272727272727273,
      "grad_norm": 0.04750819131731987,
      "learning_rate": 7.772727272727273e-05,
      "loss": 0.0126,
      "step": 1470
    },
    {
      "epoch": 22.348484848484848,
      "grad_norm": 0.06720277667045593,
      "learning_rate": 7.765151515151515e-05,
      "loss": 0.0139,
      "step": 1475
    },
    {
      "epoch": 22.424242424242426,
      "grad_norm": 0.05976364016532898,
      "learning_rate": 7.757575757575758e-05,
      "loss": 0.0127,
      "step": 1480
    },
    {
      "epoch": 22.5,
      "grad_norm": 0.15385450422763824,
      "learning_rate": 7.75e-05,
      "loss": 0.0124,
      "step": 1485
    },
    {
      "epoch": 22.575757575757574,
      "grad_norm": 0.06428530812263489,
      "learning_rate": 7.742424242424243e-05,
      "loss": 0.0119,
      "step": 1490
    },
    {
      "epoch": 22.651515151515152,
      "grad_norm": 0.12518718838691711,
      "learning_rate": 7.734848484848485e-05,
      "loss": 0.0156,
      "step": 1495
    },
    {
      "epoch": 22.727272727272727,
      "grad_norm": 0.040064405649900436,
      "learning_rate": 7.727272727272727e-05,
      "loss": 0.0134,
      "step": 1500
    },
    {
      "epoch": 22.803030303030305,
      "grad_norm": 0.19167010486125946,
      "learning_rate": 7.71969696969697e-05,
      "loss": 0.0171,
      "step": 1505
    },
    {
      "epoch": 22.87878787878788,
      "grad_norm": 0.08281899988651276,
      "learning_rate": 7.712121212121212e-05,
      "loss": 0.0155,
      "step": 1510
    },
    {
      "epoch": 22.954545454545453,
      "grad_norm": 0.05278579145669937,
      "learning_rate": 7.704545454545456e-05,
      "loss": 0.0157,
      "step": 1515
    },
    {
      "epoch": 23.0,
      "eval_loss": 0.015154623426496983,
      "eval_mean_accuracy": 0.9302574244120768,
      "eval_mean_iou": 0.9103396758234199,
      "eval_runtime": 128.1274,
      "eval_samples_per_second": 1.834,
      "eval_steps_per_second": 0.234,
      "step": 1518
    },
    {
      "epoch": 23.03030303030303,
      "grad_norm": 0.050781186670064926,
      "learning_rate": 7.696969696969696e-05,
      "loss": 0.0146,
      "step": 1520
    },
    {
      "epoch": 23.106060606060606,
      "grad_norm": 0.07075611501932144,
      "learning_rate": 7.68939393939394e-05,
      "loss": 0.0161,
      "step": 1525
    },
    {
      "epoch": 23.181818181818183,
      "grad_norm": 0.2788224518299103,
      "learning_rate": 7.681818181818182e-05,
      "loss": 0.017,
      "step": 1530
    },
    {
      "epoch": 23.257575757575758,
      "grad_norm": 0.06572628766298294,
      "learning_rate": 7.674242424242425e-05,
      "loss": 0.0125,
      "step": 1535
    },
    {
      "epoch": 23.333333333333332,
      "grad_norm": 0.05212420970201492,
      "learning_rate": 7.666666666666667e-05,
      "loss": 0.0128,
      "step": 1540
    },
    {
      "epoch": 23.40909090909091,
      "grad_norm": 0.10534659028053284,
      "learning_rate": 7.659090909090909e-05,
      "loss": 0.0129,
      "step": 1545
    },
    {
      "epoch": 23.484848484848484,
      "grad_norm": 0.07555500417947769,
      "learning_rate": 7.651515151515152e-05,
      "loss": 0.0139,
      "step": 1550
    },
    {
      "epoch": 23.560606060606062,
      "grad_norm": 0.09692145138978958,
      "learning_rate": 7.643939393939394e-05,
      "loss": 0.0122,
      "step": 1555
    },
    {
      "epoch": 23.636363636363637,
      "grad_norm": 0.09686421602964401,
      "learning_rate": 7.636363636363637e-05,
      "loss": 0.0112,
      "step": 1560
    },
    {
      "epoch": 23.71212121212121,
      "grad_norm": 0.05772320553660393,
      "learning_rate": 7.62878787878788e-05,
      "loss": 0.0128,
      "step": 1565
    },
    {
      "epoch": 23.78787878787879,
      "grad_norm": 0.10238448530435562,
      "learning_rate": 7.621212121212121e-05,
      "loss": 0.0144,
      "step": 1570
    },
    {
      "epoch": 23.863636363636363,
      "grad_norm": 0.21881021559238434,
      "learning_rate": 7.613636363636363e-05,
      "loss": 0.0151,
      "step": 1575
    },
    {
      "epoch": 23.939393939393938,
      "grad_norm": 0.14404788613319397,
      "learning_rate": 7.606060606060607e-05,
      "loss": 0.015,
      "step": 1580
    },
    {
      "epoch": 24.0,
      "eval_loss": 0.01371687836945057,
      "eval_mean_accuracy": 0.9522964731777362,
      "eval_mean_iou": 0.9264947368779796,
      "eval_runtime": 128.2377,
      "eval_samples_per_second": 1.833,
      "eval_steps_per_second": 0.234,
      "step": 1584
    },
    {
      "epoch": 24.015151515151516,
      "grad_norm": 0.08882619440555573,
      "learning_rate": 7.598484848484849e-05,
      "loss": 0.0215,
      "step": 1585
    },
    {
      "epoch": 24.09090909090909,
      "grad_norm": 0.23932959139347076,
      "learning_rate": 7.59090909090909e-05,
      "loss": 0.0141,
      "step": 1590
    },
    {
      "epoch": 24.166666666666668,
      "grad_norm": 0.12718889117240906,
      "learning_rate": 7.583333333333334e-05,
      "loss": 0.014,
      "step": 1595
    },
    {
      "epoch": 24.242424242424242,
      "grad_norm": 0.13312387466430664,
      "learning_rate": 7.575757575757576e-05,
      "loss": 0.0117,
      "step": 1600
    },
    {
      "epoch": 24.318181818181817,
      "grad_norm": 0.055063847452402115,
      "learning_rate": 7.568181818181819e-05,
      "loss": 0.0134,
      "step": 1605
    },
    {
      "epoch": 24.393939393939394,
      "grad_norm": 0.07437331974506378,
      "learning_rate": 7.560606060606061e-05,
      "loss": 0.0121,
      "step": 1610
    },
    {
      "epoch": 24.46969696969697,
      "grad_norm": 0.10028072446584702,
      "learning_rate": 7.553030303030303e-05,
      "loss": 0.0131,
      "step": 1615
    },
    {
      "epoch": 24.545454545454547,
      "grad_norm": 0.03968743979930878,
      "learning_rate": 7.545454545454545e-05,
      "loss": 0.0111,
      "step": 1620
    },
    {
      "epoch": 24.62121212121212,
      "grad_norm": 0.32868048548698425,
      "learning_rate": 7.537878787878788e-05,
      "loss": 0.0169,
      "step": 1625
    },
    {
      "epoch": 24.696969696969695,
      "grad_norm": 0.10447657853364944,
      "learning_rate": 7.530303030303032e-05,
      "loss": 0.0189,
      "step": 1630
    },
    {
      "epoch": 24.772727272727273,
      "grad_norm": 0.11356969177722931,
      "learning_rate": 7.522727272727273e-05,
      "loss": 0.0191,
      "step": 1635
    },
    {
      "epoch": 24.848484848484848,
      "grad_norm": 0.10910993069410324,
      "learning_rate": 7.515151515151515e-05,
      "loss": 0.0121,
      "step": 1640
    },
    {
      "epoch": 24.924242424242426,
      "grad_norm": 0.07475501298904419,
      "learning_rate": 7.507575757575757e-05,
      "loss": 0.0121,
      "step": 1645
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.1641620248556137,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.0166,
      "step": 1650
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.013781994581222534,
      "eval_mean_accuracy": 0.9631466596618333,
      "eval_mean_iou": 0.927330067250793,
      "eval_runtime": 128.0473,
      "eval_samples_per_second": 1.835,
      "eval_steps_per_second": 0.234,
      "step": 1650
    },
    {
      "epoch": 25.075757575757574,
      "grad_norm": 0.07435738295316696,
      "learning_rate": 7.492424242424243e-05,
      "loss": 0.0125,
      "step": 1655
    },
    {
      "epoch": 25.151515151515152,
      "grad_norm": 0.06095406413078308,
      "learning_rate": 7.484848484848486e-05,
      "loss": 0.0132,
      "step": 1660
    },
    {
      "epoch": 25.227272727272727,
      "grad_norm": 0.05284012109041214,
      "learning_rate": 7.477272727272727e-05,
      "loss": 0.0103,
      "step": 1665
    },
    {
      "epoch": 25.303030303030305,
      "grad_norm": 0.22933153808116913,
      "learning_rate": 7.46969696969697e-05,
      "loss": 0.0175,
      "step": 1670
    },
    {
      "epoch": 25.37878787878788,
      "grad_norm": 0.05187980458140373,
      "learning_rate": 7.462121212121213e-05,
      "loss": 0.0141,
      "step": 1675
    },
    {
      "epoch": 25.454545454545453,
      "grad_norm": 0.08226528763771057,
      "learning_rate": 7.454545454545455e-05,
      "loss": 0.0092,
      "step": 1680
    },
    {
      "epoch": 25.53030303030303,
      "grad_norm": 0.05373677983880043,
      "learning_rate": 7.446969696969698e-05,
      "loss": 0.0175,
      "step": 1685
    },
    {
      "epoch": 25.606060606060606,
      "grad_norm": 0.0743103101849556,
      "learning_rate": 7.439393939393939e-05,
      "loss": 0.0145,
      "step": 1690
    },
    {
      "epoch": 25.681818181818183,
      "grad_norm": 0.06081731617450714,
      "learning_rate": 7.431818181818182e-05,
      "loss": 0.0118,
      "step": 1695
    },
    {
      "epoch": 25.757575757575758,
      "grad_norm": 0.1899833381175995,
      "learning_rate": 7.424242424242424e-05,
      "loss": 0.0139,
      "step": 1700
    },
    {
      "epoch": 25.833333333333332,
      "grad_norm": 0.07575638592243195,
      "learning_rate": 7.416666666666668e-05,
      "loss": 0.0144,
      "step": 1705
    },
    {
      "epoch": 25.90909090909091,
      "grad_norm": 0.08208439499139786,
      "learning_rate": 7.40909090909091e-05,
      "loss": 0.0114,
      "step": 1710
    },
    {
      "epoch": 25.984848484848484,
      "grad_norm": 0.06419762223958969,
      "learning_rate": 7.401515151515152e-05,
      "loss": 0.0166,
      "step": 1715
    },
    {
      "epoch": 26.0,
      "eval_loss": 0.014123682864010334,
      "eval_mean_accuracy": 0.947887529251595,
      "eval_mean_iou": 0.9217069167158415,
      "eval_runtime": 151.3215,
      "eval_samples_per_second": 1.553,
      "eval_steps_per_second": 0.198,
      "step": 1716
    },
    {
      "epoch": 26.060606060606062,
      "grad_norm": 0.055003661662340164,
      "learning_rate": 7.393939393939395e-05,
      "loss": 0.0136,
      "step": 1720
    },
    {
      "epoch": 26.136363636363637,
      "grad_norm": 0.04427693039178848,
      "learning_rate": 7.386363636363637e-05,
      "loss": 0.0128,
      "step": 1725
    },
    {
      "epoch": 26.21212121212121,
      "grad_norm": 0.0662568062543869,
      "learning_rate": 7.37878787878788e-05,
      "loss": 0.0112,
      "step": 1730
    },
    {
      "epoch": 26.28787878787879,
      "grad_norm": 0.12903353571891785,
      "learning_rate": 7.37121212121212e-05,
      "loss": 0.0115,
      "step": 1735
    },
    {
      "epoch": 26.363636363636363,
      "grad_norm": 0.07952788472175598,
      "learning_rate": 7.363636363636364e-05,
      "loss": 0.0109,
      "step": 1740
    },
    {
      "epoch": 26.439393939393938,
      "grad_norm": 0.22697339951992035,
      "learning_rate": 7.356060606060606e-05,
      "loss": 0.0124,
      "step": 1745
    },
    {
      "epoch": 26.515151515151516,
      "grad_norm": 0.15446186065673828,
      "learning_rate": 7.348484848484849e-05,
      "loss": 0.0196,
      "step": 1750
    },
    {
      "epoch": 26.59090909090909,
      "grad_norm": 0.05486298352479935,
      "learning_rate": 7.340909090909091e-05,
      "loss": 0.0129,
      "step": 1755
    },
    {
      "epoch": 26.666666666666668,
      "grad_norm": 0.05585933104157448,
      "learning_rate": 7.333333333333333e-05,
      "loss": 0.0115,
      "step": 1760
    },
    {
      "epoch": 26.742424242424242,
      "grad_norm": 0.05741039663553238,
      "learning_rate": 7.325757575757576e-05,
      "loss": 0.0129,
      "step": 1765
    },
    {
      "epoch": 26.818181818181817,
      "grad_norm": 0.07074657827615738,
      "learning_rate": 7.318181818181818e-05,
      "loss": 0.0125,
      "step": 1770
    },
    {
      "epoch": 26.893939393939394,
      "grad_norm": 0.04262099415063858,
      "learning_rate": 7.310606060606062e-05,
      "loss": 0.0127,
      "step": 1775
    },
    {
      "epoch": 26.96969696969697,
      "grad_norm": 0.057138580828905106,
      "learning_rate": 7.303030303030304e-05,
      "loss": 0.0124,
      "step": 1780
    },
    {
      "epoch": 27.0,
      "eval_loss": 0.014159897342324257,
      "eval_mean_accuracy": 0.9375892670870511,
      "eval_mean_iou": 0.9165066042813446,
      "eval_runtime": 129.0563,
      "eval_samples_per_second": 1.821,
      "eval_steps_per_second": 0.232,
      "step": 1782
    },
    {
      "epoch": 27.045454545454547,
      "grad_norm": 0.03430837020277977,
      "learning_rate": 7.295454545454546e-05,
      "loss": 0.0127,
      "step": 1785
    },
    {
      "epoch": 27.12121212121212,
      "grad_norm": 0.16391168534755707,
      "learning_rate": 7.287878787878788e-05,
      "loss": 0.0165,
      "step": 1790
    },
    {
      "epoch": 27.196969696969695,
      "grad_norm": 0.0755462646484375,
      "learning_rate": 7.280303030303031e-05,
      "loss": 0.013,
      "step": 1795
    },
    {
      "epoch": 27.272727272727273,
      "grad_norm": 0.1624448597431183,
      "learning_rate": 7.272727272727273e-05,
      "loss": 0.0107,
      "step": 1800
    },
    {
      "epoch": 27.348484848484848,
      "grad_norm": 0.0843912735581398,
      "learning_rate": 7.265151515151516e-05,
      "loss": 0.0103,
      "step": 1805
    },
    {
      "epoch": 27.424242424242426,
      "grad_norm": 0.05027702823281288,
      "learning_rate": 7.257575757575758e-05,
      "loss": 0.0129,
      "step": 1810
    },
    {
      "epoch": 27.5,
      "grad_norm": 0.1020628809928894,
      "learning_rate": 7.25e-05,
      "loss": 0.0123,
      "step": 1815
    },
    {
      "epoch": 27.575757575757574,
      "grad_norm": 0.05663345009088516,
      "learning_rate": 7.242424242424243e-05,
      "loss": 0.0118,
      "step": 1820
    },
    {
      "epoch": 27.651515151515152,
      "grad_norm": 0.08548890054225922,
      "learning_rate": 7.234848484848485e-05,
      "loss": 0.0144,
      "step": 1825
    },
    {
      "epoch": 27.727272727272727,
      "grad_norm": 0.08186863362789154,
      "learning_rate": 7.227272727272729e-05,
      "loss": 0.0131,
      "step": 1830
    },
    {
      "epoch": 27.803030303030305,
      "grad_norm": 0.09840793162584305,
      "learning_rate": 7.219696969696969e-05,
      "loss": 0.0107,
      "step": 1835
    },
    {
      "epoch": 27.87878787878788,
      "grad_norm": 0.10851320624351501,
      "learning_rate": 7.212121212121213e-05,
      "loss": 0.013,
      "step": 1840
    },
    {
      "epoch": 27.954545454545453,
      "grad_norm": 0.206655353307724,
      "learning_rate": 7.204545454545454e-05,
      "loss": 0.0121,
      "step": 1845
    },
    {
      "epoch": 28.0,
      "eval_loss": 0.01330589223653078,
      "eval_mean_accuracy": 0.9512454264194404,
      "eval_mean_iou": 0.926253433623855,
      "eval_runtime": 129.2331,
      "eval_samples_per_second": 1.818,
      "eval_steps_per_second": 0.232,
      "step": 1848
    },
    {
      "epoch": 28.03030303030303,
      "grad_norm": 0.09899212419986725,
      "learning_rate": 7.196969696969698e-05,
      "loss": 0.0107,
      "step": 1850
    },
    {
      "epoch": 28.106060606060606,
      "grad_norm": 0.0908733680844307,
      "learning_rate": 7.18939393939394e-05,
      "loss": 0.0095,
      "step": 1855
    },
    {
      "epoch": 28.181818181818183,
      "grad_norm": 0.143727108836174,
      "learning_rate": 7.181818181818182e-05,
      "loss": 0.0138,
      "step": 1860
    },
    {
      "epoch": 28.257575757575758,
      "grad_norm": 0.044411592185497284,
      "learning_rate": 7.174242424242425e-05,
      "loss": 0.0119,
      "step": 1865
    },
    {
      "epoch": 28.333333333333332,
      "grad_norm": 0.11762754619121552,
      "learning_rate": 7.166666666666667e-05,
      "loss": 0.0127,
      "step": 1870
    },
    {
      "epoch": 28.40909090909091,
      "grad_norm": 0.025273283943533897,
      "learning_rate": 7.15909090909091e-05,
      "loss": 0.0084,
      "step": 1875
    },
    {
      "epoch": 28.484848484848484,
      "grad_norm": 0.07611919194459915,
      "learning_rate": 7.151515151515152e-05,
      "loss": 0.0143,
      "step": 1880
    },
    {
      "epoch": 28.560606060606062,
      "grad_norm": 0.07448887079954147,
      "learning_rate": 7.143939393939394e-05,
      "loss": 0.0125,
      "step": 1885
    },
    {
      "epoch": 28.636363636363637,
      "grad_norm": 0.035387080162763596,
      "learning_rate": 7.136363636363636e-05,
      "loss": 0.0109,
      "step": 1890
    },
    {
      "epoch": 28.71212121212121,
      "grad_norm": 0.17645961046218872,
      "learning_rate": 7.12878787878788e-05,
      "loss": 0.0155,
      "step": 1895
    },
    {
      "epoch": 28.78787878787879,
      "grad_norm": 0.051544226706027985,
      "learning_rate": 7.121212121212121e-05,
      "loss": 0.013,
      "step": 1900
    },
    {
      "epoch": 28.863636363636363,
      "grad_norm": 0.05301607772707939,
      "learning_rate": 7.113636363636363e-05,
      "loss": 0.0118,
      "step": 1905
    },
    {
      "epoch": 28.939393939393938,
      "grad_norm": 0.07802437245845795,
      "learning_rate": 7.106060606060607e-05,
      "loss": 0.0129,
      "step": 1910
    },
    {
      "epoch": 29.0,
      "eval_loss": 0.01311896089464426,
      "eval_mean_accuracy": 0.9597477246233818,
      "eval_mean_iou": 0.9253863680801655,
      "eval_runtime": 109.8897,
      "eval_samples_per_second": 2.139,
      "eval_steps_per_second": 0.273,
      "step": 1914
    },
    {
      "epoch": 29.015151515151516,
      "grad_norm": 0.04973312094807625,
      "learning_rate": 7.098484848484849e-05,
      "loss": 0.0116,
      "step": 1915
    },
    {
      "epoch": 29.09090909090909,
      "grad_norm": 0.03684161975979805,
      "learning_rate": 7.090909090909092e-05,
      "loss": 0.0136,
      "step": 1920
    },
    {
      "epoch": 29.166666666666668,
      "grad_norm": 0.06615979969501495,
      "learning_rate": 7.083333333333334e-05,
      "loss": 0.0105,
      "step": 1925
    },
    {
      "epoch": 29.242424242424242,
      "grad_norm": 0.07462740689516068,
      "learning_rate": 7.075757575757576e-05,
      "loss": 0.0109,
      "step": 1930
    },
    {
      "epoch": 29.318181818181817,
      "grad_norm": 0.08642291277647018,
      "learning_rate": 7.068181818181818e-05,
      "loss": 0.0121,
      "step": 1935
    },
    {
      "epoch": 29.393939393939394,
      "grad_norm": 0.08559373021125793,
      "learning_rate": 7.060606060606061e-05,
      "loss": 0.0187,
      "step": 1940
    },
    {
      "epoch": 29.46969696969697,
      "grad_norm": 0.10868417471647263,
      "learning_rate": 7.053030303030303e-05,
      "loss": 0.0142,
      "step": 1945
    },
    {
      "epoch": 29.545454545454547,
      "grad_norm": 0.13082125782966614,
      "learning_rate": 7.045454545454546e-05,
      "loss": 0.0156,
      "step": 1950
    },
    {
      "epoch": 29.62121212121212,
      "grad_norm": 0.06175585836172104,
      "learning_rate": 7.037878787878788e-05,
      "loss": 0.0126,
      "step": 1955
    },
    {
      "epoch": 29.696969696969695,
      "grad_norm": 0.06169668212532997,
      "learning_rate": 7.03030303030303e-05,
      "loss": 0.0121,
      "step": 1960
    },
    {
      "epoch": 29.772727272727273,
      "grad_norm": 0.18244794011116028,
      "learning_rate": 7.022727272727274e-05,
      "loss": 0.0169,
      "step": 1965
    },
    {
      "epoch": 29.848484848484848,
      "grad_norm": 0.5402779579162598,
      "learning_rate": 7.015151515151515e-05,
      "loss": 0.0163,
      "step": 1970
    },
    {
      "epoch": 29.924242424242426,
      "grad_norm": 0.12343169003725052,
      "learning_rate": 7.007575757575759e-05,
      "loss": 0.0132,
      "step": 1975
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.11872578412294388,
      "learning_rate": 7e-05,
      "loss": 0.0106,
      "step": 1980
    },
    {
      "epoch": 30.0,
      "eval_loss": 0.014531918801367283,
      "eval_mean_accuracy": 0.9674407250386775,
      "eval_mean_iou": 0.9203758956042165,
      "eval_runtime": 109.9396,
      "eval_samples_per_second": 2.138,
      "eval_steps_per_second": 0.273,
      "step": 1980
    },
    {
      "epoch": 30.075757575757574,
      "grad_norm": 0.09509027749300003,
      "learning_rate": 6.992424242424243e-05,
      "loss": 0.0104,
      "step": 1985
    },
    {
      "epoch": 30.151515151515152,
      "grad_norm": 0.06573574244976044,
      "learning_rate": 6.984848484848485e-05,
      "loss": 0.0127,
      "step": 1990
    },
    {
      "epoch": 30.227272727272727,
      "grad_norm": 0.40525177121162415,
      "learning_rate": 6.977272727272728e-05,
      "loss": 0.0143,
      "step": 1995
    },
    {
      "epoch": 30.303030303030305,
      "grad_norm": 0.06698071956634521,
      "learning_rate": 6.96969696969697e-05,
      "loss": 0.0089,
      "step": 2000
    },
    {
      "epoch": 30.37878787878788,
      "grad_norm": 0.04077392816543579,
      "learning_rate": 6.962121212121212e-05,
      "loss": 0.015,
      "step": 2005
    },
    {
      "epoch": 30.454545454545453,
      "grad_norm": 0.08249492198228836,
      "learning_rate": 6.954545454545455e-05,
      "loss": 0.0134,
      "step": 2010
    },
    {
      "epoch": 30.53030303030303,
      "grad_norm": 0.13396184146404266,
      "learning_rate": 6.946969696969697e-05,
      "loss": 0.0152,
      "step": 2015
    },
    {
      "epoch": 30.606060606060606,
      "grad_norm": 0.18429727852344513,
      "learning_rate": 6.93939393939394e-05,
      "loss": 0.0131,
      "step": 2020
    },
    {
      "epoch": 30.681818181818183,
      "grad_norm": 0.05011655390262604,
      "learning_rate": 6.931818181818182e-05,
      "loss": 0.0124,
      "step": 2025
    },
    {
      "epoch": 30.757575757575758,
      "grad_norm": 0.0901452973484993,
      "learning_rate": 6.924242424242424e-05,
      "loss": 0.0154,
      "step": 2030
    },
    {
      "epoch": 30.833333333333332,
      "grad_norm": 0.12369824945926666,
      "learning_rate": 6.916666666666666e-05,
      "loss": 0.0098,
      "step": 2035
    },
    {
      "epoch": 30.90909090909091,
      "grad_norm": 0.10107390582561493,
      "learning_rate": 6.90909090909091e-05,
      "loss": 0.0123,
      "step": 2040
    },
    {
      "epoch": 30.984848484848484,
      "grad_norm": 0.08938383311033249,
      "learning_rate": 6.901515151515152e-05,
      "loss": 0.0115,
      "step": 2045
    },
    {
      "epoch": 31.0,
      "eval_loss": 0.012659462168812752,
      "eval_mean_accuracy": 0.9515835593546439,
      "eval_mean_iou": 0.9266121481501388,
      "eval_runtime": 109.5393,
      "eval_samples_per_second": 2.145,
      "eval_steps_per_second": 0.274,
      "step": 2046
    },
    {
      "epoch": 31.060606060606062,
      "grad_norm": 0.0789368599653244,
      "learning_rate": 6.893939393939395e-05,
      "loss": 0.0094,
      "step": 2050
    },
    {
      "epoch": 31.136363636363637,
      "grad_norm": 0.06564489006996155,
      "learning_rate": 6.886363636363637e-05,
      "loss": 0.013,
      "step": 2055
    },
    {
      "epoch": 31.21212121212121,
      "grad_norm": 0.04669715091586113,
      "learning_rate": 6.878787878787879e-05,
      "loss": 0.0104,
      "step": 2060
    },
    {
      "epoch": 31.28787878787879,
      "grad_norm": 0.03585536777973175,
      "learning_rate": 6.871212121212122e-05,
      "loss": 0.0082,
      "step": 2065
    },
    {
      "epoch": 31.363636363636363,
      "grad_norm": 0.10969308763742447,
      "learning_rate": 6.863636363636364e-05,
      "loss": 0.0115,
      "step": 2070
    },
    {
      "epoch": 31.439393939393938,
      "grad_norm": 0.23658400774002075,
      "learning_rate": 6.856060606060606e-05,
      "loss": 0.0122,
      "step": 2075
    },
    {
      "epoch": 31.515151515151516,
      "grad_norm": 0.11215761303901672,
      "learning_rate": 6.848484848484848e-05,
      "loss": 0.0108,
      "step": 2080
    },
    {
      "epoch": 31.59090909090909,
      "grad_norm": 0.17185278236865997,
      "learning_rate": 6.840909090909091e-05,
      "loss": 0.0139,
      "step": 2085
    },
    {
      "epoch": 31.666666666666668,
      "grad_norm": 0.09880587458610535,
      "learning_rate": 6.833333333333333e-05,
      "loss": 0.013,
      "step": 2090
    },
    {
      "epoch": 31.742424242424242,
      "grad_norm": 0.08652031421661377,
      "learning_rate": 6.825757575757576e-05,
      "loss": 0.0094,
      "step": 2095
    },
    {
      "epoch": 31.818181818181817,
      "grad_norm": 0.10556855797767639,
      "learning_rate": 6.818181818181818e-05,
      "loss": 0.0145,
      "step": 2100
    },
    {
      "epoch": 31.893939393939394,
      "grad_norm": 0.09701216220855713,
      "learning_rate": 6.81060606060606e-05,
      "loss": 0.0155,
      "step": 2105
    },
    {
      "epoch": 31.96969696969697,
      "grad_norm": 0.04061540961265564,
      "learning_rate": 6.803030303030304e-05,
      "loss": 0.0131,
      "step": 2110
    },
    {
      "epoch": 32.0,
      "eval_loss": 0.012345309369266033,
      "eval_mean_accuracy": 0.962387705671614,
      "eval_mean_iou": 0.9288493428415148,
      "eval_runtime": 113.706,
      "eval_samples_per_second": 2.067,
      "eval_steps_per_second": 0.264,
      "step": 2112
    },
    {
      "epoch": 32.04545454545455,
      "grad_norm": 0.05108783021569252,
      "learning_rate": 6.795454545454546e-05,
      "loss": 0.0133,
      "step": 2115
    },
    {
      "epoch": 32.121212121212125,
      "grad_norm": 0.050104714930057526,
      "learning_rate": 6.787878787878789e-05,
      "loss": 0.0106,
      "step": 2120
    },
    {
      "epoch": 32.196969696969695,
      "grad_norm": 0.36575835943222046,
      "learning_rate": 6.78030303030303e-05,
      "loss": 0.0147,
      "step": 2125
    },
    {
      "epoch": 32.27272727272727,
      "grad_norm": 0.03118722140789032,
      "learning_rate": 6.772727272727273e-05,
      "loss": 0.0109,
      "step": 2130
    },
    {
      "epoch": 32.34848484848485,
      "grad_norm": 0.09519505500793457,
      "learning_rate": 6.765151515151515e-05,
      "loss": 0.0122,
      "step": 2135
    },
    {
      "epoch": 32.42424242424242,
      "grad_norm": 0.1005990281701088,
      "learning_rate": 6.757575757575758e-05,
      "loss": 0.0167,
      "step": 2140
    },
    {
      "epoch": 32.5,
      "grad_norm": 0.07473775744438171,
      "learning_rate": 6.750000000000001e-05,
      "loss": 0.0148,
      "step": 2145
    },
    {
      "epoch": 32.57575757575758,
      "grad_norm": 0.04295356944203377,
      "learning_rate": 6.742424242424242e-05,
      "loss": 0.015,
      "step": 2150
    },
    {
      "epoch": 32.65151515151515,
      "grad_norm": 0.174886554479599,
      "learning_rate": 6.734848484848485e-05,
      "loss": 0.0124,
      "step": 2155
    },
    {
      "epoch": 32.72727272727273,
      "grad_norm": 0.05481899902224541,
      "learning_rate": 6.727272727272727e-05,
      "loss": 0.0103,
      "step": 2160
    },
    {
      "epoch": 32.803030303030305,
      "grad_norm": 0.18934907019138336,
      "learning_rate": 6.71969696969697e-05,
      "loss": 0.0131,
      "step": 2165
    },
    {
      "epoch": 32.878787878787875,
      "grad_norm": 0.10034554451704025,
      "learning_rate": 6.712121212121213e-05,
      "loss": 0.0144,
      "step": 2170
    },
    {
      "epoch": 32.95454545454545,
      "grad_norm": 0.10720604658126831,
      "learning_rate": 6.704545454545455e-05,
      "loss": 0.0233,
      "step": 2175
    },
    {
      "epoch": 33.0,
      "eval_loss": 0.015099034644663334,
      "eval_mean_accuracy": 0.9412814198156036,
      "eval_mean_iou": 0.9178111398774149,
      "eval_runtime": 109.2296,
      "eval_samples_per_second": 2.151,
      "eval_steps_per_second": 0.275,
      "step": 2178
    },
    {
      "epoch": 33.03030303030303,
      "grad_norm": 0.39493682980537415,
      "learning_rate": 6.696969696969696e-05,
      "loss": 0.0143,
      "step": 2180
    },
    {
      "epoch": 33.10606060606061,
      "grad_norm": 0.06298752874135971,
      "learning_rate": 6.68939393939394e-05,
      "loss": 0.0174,
      "step": 2185
    },
    {
      "epoch": 33.18181818181818,
      "grad_norm": 0.1056286096572876,
      "learning_rate": 6.681818181818183e-05,
      "loss": 0.014,
      "step": 2190
    },
    {
      "epoch": 33.25757575757576,
      "grad_norm": 0.10794030129909515,
      "learning_rate": 6.674242424242425e-05,
      "loss": 0.0146,
      "step": 2195
    },
    {
      "epoch": 33.333333333333336,
      "grad_norm": 0.12694840133190155,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.0109,
      "step": 2200
    },
    {
      "epoch": 33.40909090909091,
      "grad_norm": 0.044256243854761124,
      "learning_rate": 6.659090909090909e-05,
      "loss": 0.0134,
      "step": 2205
    },
    {
      "epoch": 33.484848484848484,
      "grad_norm": 0.03317928686738014,
      "learning_rate": 6.651515151515152e-05,
      "loss": 0.0144,
      "step": 2210
    },
    {
      "epoch": 33.56060606060606,
      "grad_norm": 0.044657494872808456,
      "learning_rate": 6.643939393939394e-05,
      "loss": 0.0113,
      "step": 2215
    },
    {
      "epoch": 33.63636363636363,
      "grad_norm": 0.08492383360862732,
      "learning_rate": 6.636363636363638e-05,
      "loss": 0.0129,
      "step": 2220
    },
    {
      "epoch": 33.71212121212121,
      "grad_norm": 0.07876002788543701,
      "learning_rate": 6.628787878787878e-05,
      "loss": 0.0101,
      "step": 2225
    },
    {
      "epoch": 33.78787878787879,
      "grad_norm": 0.07676819711923599,
      "learning_rate": 6.621212121212121e-05,
      "loss": 0.0191,
      "step": 2230
    },
    {
      "epoch": 33.86363636363637,
      "grad_norm": 0.046629514545202255,
      "learning_rate": 6.613636363636365e-05,
      "loss": 0.0109,
      "step": 2235
    },
    {
      "epoch": 33.93939393939394,
      "grad_norm": 0.06711985915899277,
      "learning_rate": 6.606060606060607e-05,
      "loss": 0.0099,
      "step": 2240
    },
    {
      "epoch": 34.0,
      "eval_loss": 0.012880802154541016,
      "eval_mean_accuracy": 0.9502101359284216,
      "eval_mean_iou": 0.9247825417016556,
      "eval_runtime": 114.263,
      "eval_samples_per_second": 2.057,
      "eval_steps_per_second": 0.263,
      "step": 2244
    },
    {
      "epoch": 34.015151515151516,
      "grad_norm": 0.09088897705078125,
      "learning_rate": 6.598484848484849e-05,
      "loss": 0.0172,
      "step": 2245
    },
    {
      "epoch": 34.09090909090909,
      "grad_norm": 0.051903631538152695,
      "learning_rate": 6.59090909090909e-05,
      "loss": 0.0114,
      "step": 2250
    },
    {
      "epoch": 34.166666666666664,
      "grad_norm": 0.042612262070178986,
      "learning_rate": 6.583333333333334e-05,
      "loss": 0.0161,
      "step": 2255
    },
    {
      "epoch": 34.24242424242424,
      "grad_norm": 0.10715638846158981,
      "learning_rate": 6.575757575757576e-05,
      "loss": 0.0122,
      "step": 2260
    },
    {
      "epoch": 34.31818181818182,
      "grad_norm": 0.034101929515600204,
      "learning_rate": 6.568181818181819e-05,
      "loss": 0.0105,
      "step": 2265
    },
    {
      "epoch": 34.39393939393939,
      "grad_norm": 0.13515859842300415,
      "learning_rate": 6.560606060606061e-05,
      "loss": 0.0114,
      "step": 2270
    },
    {
      "epoch": 34.46969696969697,
      "grad_norm": 0.03108268976211548,
      "learning_rate": 6.553030303030303e-05,
      "loss": 0.0115,
      "step": 2275
    },
    {
      "epoch": 34.54545454545455,
      "grad_norm": 0.05103018507361412,
      "learning_rate": 6.545454545454546e-05,
      "loss": 0.0083,
      "step": 2280
    },
    {
      "epoch": 34.621212121212125,
      "grad_norm": 0.06111261993646622,
      "learning_rate": 6.537878787878788e-05,
      "loss": 0.0139,
      "step": 2285
    },
    {
      "epoch": 34.696969696969695,
      "grad_norm": 0.08672675490379333,
      "learning_rate": 6.530303030303032e-05,
      "loss": 0.0112,
      "step": 2290
    },
    {
      "epoch": 34.77272727272727,
      "grad_norm": 0.04751618579030037,
      "learning_rate": 6.522727272727272e-05,
      "loss": 0.0099,
      "step": 2295
    },
    {
      "epoch": 34.84848484848485,
      "grad_norm": 0.14341579377651215,
      "learning_rate": 6.515151515151516e-05,
      "loss": 0.0098,
      "step": 2300
    },
    {
      "epoch": 34.92424242424242,
      "grad_norm": 0.4330891966819763,
      "learning_rate": 6.507575757575757e-05,
      "loss": 0.0152,
      "step": 2305
    },
    {
      "epoch": 35.0,
      "grad_norm": 0.11044011265039444,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.0114,
      "step": 2310
    },
    {
      "epoch": 35.0,
      "eval_loss": 0.013168140314519405,
      "eval_mean_accuracy": 0.9382885427569804,
      "eval_mean_iou": 0.9169272562810415,
      "eval_runtime": 108.4986,
      "eval_samples_per_second": 2.166,
      "eval_steps_per_second": 0.277,
      "step": 2310
    },
    {
      "epoch": 35.07575757575758,
      "grad_norm": 0.0602533221244812,
      "learning_rate": 6.492424242424243e-05,
      "loss": 0.0111,
      "step": 2315
    },
    {
      "epoch": 35.15151515151515,
      "grad_norm": 0.027574222534894943,
      "learning_rate": 6.484848484848485e-05,
      "loss": 0.0103,
      "step": 2320
    },
    {
      "epoch": 35.22727272727273,
      "grad_norm": 0.12410042434930801,
      "learning_rate": 6.477272727272728e-05,
      "loss": 0.012,
      "step": 2325
    },
    {
      "epoch": 35.303030303030305,
      "grad_norm": 0.030065948143601418,
      "learning_rate": 6.46969696969697e-05,
      "loss": 0.0091,
      "step": 2330
    },
    {
      "epoch": 35.378787878787875,
      "grad_norm": 0.10490231961011887,
      "learning_rate": 6.462121212121213e-05,
      "loss": 0.0128,
      "step": 2335
    },
    {
      "epoch": 35.45454545454545,
      "grad_norm": 0.3587149977684021,
      "learning_rate": 6.454545454545455e-05,
      "loss": 0.0125,
      "step": 2340
    },
    {
      "epoch": 35.53030303030303,
      "grad_norm": 0.17342440783977509,
      "learning_rate": 6.446969696969697e-05,
      "loss": 0.0109,
      "step": 2345
    },
    {
      "epoch": 35.60606060606061,
      "grad_norm": 0.1740293949842453,
      "learning_rate": 6.439393939393939e-05,
      "loss": 0.0137,
      "step": 2350
    },
    {
      "epoch": 35.68181818181818,
      "grad_norm": 0.03855850547552109,
      "learning_rate": 6.431818181818182e-05,
      "loss": 0.0129,
      "step": 2355
    },
    {
      "epoch": 35.75757575757576,
      "grad_norm": 0.05703097954392433,
      "learning_rate": 6.424242424242424e-05,
      "loss": 0.0088,
      "step": 2360
    },
    {
      "epoch": 35.833333333333336,
      "grad_norm": 0.08064791560173035,
      "learning_rate": 6.416666666666668e-05,
      "loss": 0.0107,
      "step": 2365
    },
    {
      "epoch": 35.90909090909091,
      "grad_norm": 0.08592565357685089,
      "learning_rate": 6.40909090909091e-05,
      "loss": 0.0126,
      "step": 2370
    },
    {
      "epoch": 35.984848484848484,
      "grad_norm": 0.04512941092252731,
      "learning_rate": 6.401515151515152e-05,
      "loss": 0.0097,
      "step": 2375
    },
    {
      "epoch": 36.0,
      "eval_loss": 0.012168590910732746,
      "eval_mean_accuracy": 0.9575022167917465,
      "eval_mean_iou": 0.9288715408847481,
      "eval_runtime": 109.2965,
      "eval_samples_per_second": 2.15,
      "eval_steps_per_second": 0.274,
      "step": 2376
    },
    {
      "epoch": 36.06060606060606,
      "grad_norm": 0.23136784136295319,
      "learning_rate": 6.393939393939395e-05,
      "loss": 0.0116,
      "step": 2380
    },
    {
      "epoch": 36.13636363636363,
      "grad_norm": 0.05725252628326416,
      "learning_rate": 6.386363636363637e-05,
      "loss": 0.0097,
      "step": 2385
    },
    {
      "epoch": 36.21212121212121,
      "grad_norm": 0.05111449211835861,
      "learning_rate": 6.37878787878788e-05,
      "loss": 0.0142,
      "step": 2390
    },
    {
      "epoch": 36.28787878787879,
      "grad_norm": 0.12967385351657867,
      "learning_rate": 6.371212121212121e-05,
      "loss": 0.0086,
      "step": 2395
    },
    {
      "epoch": 36.36363636363637,
      "grad_norm": 0.049787554889917374,
      "learning_rate": 6.363636363636364e-05,
      "loss": 0.0079,
      "step": 2400
    },
    {
      "epoch": 36.43939393939394,
      "grad_norm": 0.15063980221748352,
      "learning_rate": 6.356060606060606e-05,
      "loss": 0.0149,
      "step": 2405
    },
    {
      "epoch": 36.515151515151516,
      "grad_norm": 0.09079291671514511,
      "learning_rate": 6.34848484848485e-05,
      "loss": 0.0103,
      "step": 2410
    },
    {
      "epoch": 36.59090909090909,
      "grad_norm": 0.13014796376228333,
      "learning_rate": 6.340909090909091e-05,
      "loss": 0.0154,
      "step": 2415
    },
    {
      "epoch": 36.666666666666664,
      "grad_norm": 0.04497504234313965,
      "learning_rate": 6.333333333333333e-05,
      "loss": 0.0108,
      "step": 2420
    },
    {
      "epoch": 36.74242424242424,
      "grad_norm": 0.08013498783111572,
      "learning_rate": 6.325757575757577e-05,
      "loss": 0.0087,
      "step": 2425
    },
    {
      "epoch": 36.81818181818182,
      "grad_norm": 0.05519306659698486,
      "learning_rate": 6.318181818181818e-05,
      "loss": 0.0124,
      "step": 2430
    },
    {
      "epoch": 36.89393939393939,
      "grad_norm": 0.11748310923576355,
      "learning_rate": 6.310606060606062e-05,
      "loss": 0.0153,
      "step": 2435
    },
    {
      "epoch": 36.96969696969697,
      "grad_norm": 0.0485047809779644,
      "learning_rate": 6.303030303030302e-05,
      "loss": 0.0106,
      "step": 2440
    },
    {
      "epoch": 37.0,
      "eval_loss": 0.012661498039960861,
      "eval_mean_accuracy": 0.9412131273448384,
      "eval_mean_iou": 0.9206736173927214,
      "eval_runtime": 109.77,
      "eval_samples_per_second": 2.141,
      "eval_steps_per_second": 0.273,
      "step": 2442
    },
    {
      "epoch": 37.04545454545455,
      "grad_norm": 0.044076643884181976,
      "learning_rate": 6.295454545454546e-05,
      "loss": 0.0106,
      "step": 2445
    },
    {
      "epoch": 37.121212121212125,
      "grad_norm": 0.06143128499388695,
      "learning_rate": 6.287878787878788e-05,
      "loss": 0.0094,
      "step": 2450
    },
    {
      "epoch": 37.196969696969695,
      "grad_norm": 0.031067537143826485,
      "learning_rate": 6.280303030303031e-05,
      "loss": 0.0107,
      "step": 2455
    },
    {
      "epoch": 37.27272727272727,
      "grad_norm": 0.06414441019296646,
      "learning_rate": 6.272727272727273e-05,
      "loss": 0.0118,
      "step": 2460
    },
    {
      "epoch": 37.34848484848485,
      "grad_norm": 0.04670654237270355,
      "learning_rate": 6.265151515151515e-05,
      "loss": 0.01,
      "step": 2465
    },
    {
      "epoch": 37.42424242424242,
      "grad_norm": 0.12874381244182587,
      "learning_rate": 6.257575757575758e-05,
      "loss": 0.0132,
      "step": 2470
    },
    {
      "epoch": 37.5,
      "grad_norm": 0.08384066820144653,
      "learning_rate": 6.25e-05,
      "loss": 0.0115,
      "step": 2475
    },
    {
      "epoch": 37.57575757575758,
      "grad_norm": 0.07006815075874329,
      "learning_rate": 6.242424242424243e-05,
      "loss": 0.0111,
      "step": 2480
    },
    {
      "epoch": 37.65151515151515,
      "grad_norm": 0.13485975563526154,
      "learning_rate": 6.234848484848485e-05,
      "loss": 0.0126,
      "step": 2485
    },
    {
      "epoch": 37.72727272727273,
      "grad_norm": 0.0457785427570343,
      "learning_rate": 6.227272727272727e-05,
      "loss": 0.0109,
      "step": 2490
    },
    {
      "epoch": 37.803030303030305,
      "grad_norm": 0.05054578185081482,
      "learning_rate": 6.219696969696969e-05,
      "loss": 0.0097,
      "step": 2495
    },
    {
      "epoch": 37.878787878787875,
      "grad_norm": 0.08781137317419052,
      "learning_rate": 6.212121212121213e-05,
      "loss": 0.0106,
      "step": 2500
    },
    {
      "epoch": 37.95454545454545,
      "grad_norm": 0.0430886372923851,
      "learning_rate": 6.204545454545455e-05,
      "loss": 0.0091,
      "step": 2505
    },
    {
      "epoch": 38.0,
      "eval_loss": 0.012076352722942829,
      "eval_mean_accuracy": 0.9493737201986838,
      "eval_mean_iou": 0.9252135582610722,
      "eval_runtime": 110.196,
      "eval_samples_per_second": 2.133,
      "eval_steps_per_second": 0.272,
      "step": 2508
    },
    {
      "epoch": 38.03030303030303,
      "grad_norm": 0.07825518399477005,
      "learning_rate": 6.196969696969698e-05,
      "loss": 0.017,
      "step": 2510
    },
    {
      "epoch": 38.10606060606061,
      "grad_norm": 0.04494224116206169,
      "learning_rate": 6.18939393939394e-05,
      "loss": 0.0079,
      "step": 2515
    },
    {
      "epoch": 38.18181818181818,
      "grad_norm": 0.16161903738975525,
      "learning_rate": 6.181818181818182e-05,
      "loss": 0.0103,
      "step": 2520
    },
    {
      "epoch": 38.25757575757576,
      "grad_norm": 0.2515649199485779,
      "learning_rate": 6.174242424242425e-05,
      "loss": 0.0097,
      "step": 2525
    },
    {
      "epoch": 38.333333333333336,
      "grad_norm": 0.06942258030176163,
      "learning_rate": 6.166666666666667e-05,
      "loss": 0.0092,
      "step": 2530
    },
    {
      "epoch": 38.40909090909091,
      "grad_norm": 0.029291298240423203,
      "learning_rate": 6.15909090909091e-05,
      "loss": 0.0115,
      "step": 2535
    },
    {
      "epoch": 38.484848484848484,
      "grad_norm": 0.09783357381820679,
      "learning_rate": 6.151515151515151e-05,
      "loss": 0.0111,
      "step": 2540
    },
    {
      "epoch": 38.56060606060606,
      "grad_norm": 0.03193418309092522,
      "learning_rate": 6.143939393939394e-05,
      "loss": 0.0093,
      "step": 2545
    },
    {
      "epoch": 38.63636363636363,
      "grad_norm": 0.08343544602394104,
      "learning_rate": 6.136363636363636e-05,
      "loss": 0.0106,
      "step": 2550
    },
    {
      "epoch": 38.71212121212121,
      "grad_norm": 0.13089774549007416,
      "learning_rate": 6.12878787878788e-05,
      "loss": 0.0113,
      "step": 2555
    },
    {
      "epoch": 38.78787878787879,
      "grad_norm": 0.15301284193992615,
      "learning_rate": 6.121212121212121e-05,
      "loss": 0.0128,
      "step": 2560
    },
    {
      "epoch": 38.86363636363637,
      "grad_norm": 0.0971105620265007,
      "learning_rate": 6.113636363636363e-05,
      "loss": 0.0105,
      "step": 2565
    },
    {
      "epoch": 38.93939393939394,
      "grad_norm": 0.050568465143442154,
      "learning_rate": 6.106060606060607e-05,
      "loss": 0.0124,
      "step": 2570
    },
    {
      "epoch": 39.0,
      "eval_loss": 0.011807882227003574,
      "eval_mean_accuracy": 0.9663675923528418,
      "eval_mean_iou": 0.9289225650519337,
      "eval_runtime": 127.5174,
      "eval_samples_per_second": 1.843,
      "eval_steps_per_second": 0.235,
      "step": 2574
    },
    {
      "epoch": 39.015151515151516,
      "grad_norm": 0.03226739168167114,
      "learning_rate": 6.098484848484849e-05,
      "loss": 0.014,
      "step": 2575
    },
    {
      "epoch": 39.09090909090909,
      "grad_norm": 0.06365498155355453,
      "learning_rate": 6.090909090909091e-05,
      "loss": 0.0078,
      "step": 2580
    },
    {
      "epoch": 39.166666666666664,
      "grad_norm": 0.07357432693243027,
      "learning_rate": 6.083333333333333e-05,
      "loss": 0.016,
      "step": 2585
    },
    {
      "epoch": 39.24242424242424,
      "grad_norm": 0.04025569185614586,
      "learning_rate": 6.075757575757576e-05,
      "loss": 0.0102,
      "step": 2590
    },
    {
      "epoch": 39.31818181818182,
      "grad_norm": 0.03255618363618851,
      "learning_rate": 6.0681818181818185e-05,
      "loss": 0.0118,
      "step": 2595
    },
    {
      "epoch": 39.39393939393939,
      "grad_norm": 0.11152070760726929,
      "learning_rate": 6.060606060606061e-05,
      "loss": 0.0108,
      "step": 2600
    },
    {
      "epoch": 39.46969696969697,
      "grad_norm": 0.05964520201086998,
      "learning_rate": 6.053030303030304e-05,
      "loss": 0.0111,
      "step": 2605
    },
    {
      "epoch": 39.54545454545455,
      "grad_norm": 0.0815061554312706,
      "learning_rate": 6.045454545454545e-05,
      "loss": 0.0123,
      "step": 2610
    },
    {
      "epoch": 39.621212121212125,
      "grad_norm": 0.05999678373336792,
      "learning_rate": 6.037878787878788e-05,
      "loss": 0.0092,
      "step": 2615
    },
    {
      "epoch": 39.696969696969695,
      "grad_norm": 0.04912338778376579,
      "learning_rate": 6.03030303030303e-05,
      "loss": 0.0088,
      "step": 2620
    },
    {
      "epoch": 39.77272727272727,
      "grad_norm": 0.036409035325050354,
      "learning_rate": 6.022727272727273e-05,
      "loss": 0.0133,
      "step": 2625
    },
    {
      "epoch": 39.84848484848485,
      "grad_norm": 0.0336834192276001,
      "learning_rate": 6.0151515151515156e-05,
      "loss": 0.0105,
      "step": 2630
    },
    {
      "epoch": 39.92424242424242,
      "grad_norm": 0.07044974714517593,
      "learning_rate": 6.0075757575757575e-05,
      "loss": 0.0097,
      "step": 2635
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.1458202302455902,
      "learning_rate": 6e-05,
      "loss": 0.0095,
      "step": 2640
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.01185819786041975,
      "eval_mean_accuracy": 0.9601634931506068,
      "eval_mean_iou": 0.929759355213382,
      "eval_runtime": 131.0559,
      "eval_samples_per_second": 1.793,
      "eval_steps_per_second": 0.229,
      "step": 2640
    },
    {
      "epoch": 40.07575757575758,
      "grad_norm": 0.03044210560619831,
      "learning_rate": 5.992424242424243e-05,
      "loss": 0.0094,
      "step": 2645
    },
    {
      "epoch": 40.15151515151515,
      "grad_norm": 0.039066560566425323,
      "learning_rate": 5.9848484848484854e-05,
      "loss": 0.0085,
      "step": 2650
    },
    {
      "epoch": 40.22727272727273,
      "grad_norm": 0.03763739392161369,
      "learning_rate": 5.977272727272728e-05,
      "loss": 0.0101,
      "step": 2655
    },
    {
      "epoch": 40.303030303030305,
      "grad_norm": 0.04523979499936104,
      "learning_rate": 5.969696969696969e-05,
      "loss": 0.0092,
      "step": 2660
    },
    {
      "epoch": 40.378787878787875,
      "grad_norm": 0.15006108582019806,
      "learning_rate": 5.962121212121212e-05,
      "loss": 0.0093,
      "step": 2665
    },
    {
      "epoch": 40.45454545454545,
      "grad_norm": 0.023997178301215172,
      "learning_rate": 5.9545454545454546e-05,
      "loss": 0.0095,
      "step": 2670
    },
    {
      "epoch": 40.53030303030303,
      "grad_norm": 0.05292406678199768,
      "learning_rate": 5.946969696969697e-05,
      "loss": 0.0138,
      "step": 2675
    },
    {
      "epoch": 40.60606060606061,
      "grad_norm": 0.06288266181945801,
      "learning_rate": 5.93939393939394e-05,
      "loss": 0.0105,
      "step": 2680
    },
    {
      "epoch": 40.68181818181818,
      "grad_norm": 0.052500661462545395,
      "learning_rate": 5.931818181818182e-05,
      "loss": 0.0107,
      "step": 2685
    },
    {
      "epoch": 40.75757575757576,
      "grad_norm": 0.0771554633975029,
      "learning_rate": 5.9242424242424244e-05,
      "loss": 0.0107,
      "step": 2690
    },
    {
      "epoch": 40.833333333333336,
      "grad_norm": 0.03165804222226143,
      "learning_rate": 5.916666666666667e-05,
      "loss": 0.0101,
      "step": 2695
    },
    {
      "epoch": 40.90909090909091,
      "grad_norm": 0.24504762887954712,
      "learning_rate": 5.90909090909091e-05,
      "loss": 0.0151,
      "step": 2700
    },
    {
      "epoch": 40.984848484848484,
      "grad_norm": 0.18077823519706726,
      "learning_rate": 5.901515151515152e-05,
      "loss": 0.0141,
      "step": 2705
    },
    {
      "epoch": 41.0,
      "eval_loss": 0.011890443973243237,
      "eval_mean_accuracy": 0.9532994616154152,
      "eval_mean_iou": 0.924595987031528,
      "eval_runtime": 128.3751,
      "eval_samples_per_second": 1.831,
      "eval_steps_per_second": 0.234,
      "step": 2706
    },
    {
      "epoch": 41.06060606060606,
      "grad_norm": 0.026771852746605873,
      "learning_rate": 5.8939393939393936e-05,
      "loss": 0.01,
      "step": 2710
    },
    {
      "epoch": 41.13636363636363,
      "grad_norm": 0.11302294582128525,
      "learning_rate": 5.886363636363636e-05,
      "loss": 0.0116,
      "step": 2715
    },
    {
      "epoch": 41.21212121212121,
      "grad_norm": 0.13912703096866608,
      "learning_rate": 5.878787878787879e-05,
      "loss": 0.0104,
      "step": 2720
    },
    {
      "epoch": 41.28787878787879,
      "grad_norm": 0.09919160604476929,
      "learning_rate": 5.871212121212122e-05,
      "loss": 0.0122,
      "step": 2725
    },
    {
      "epoch": 41.36363636363637,
      "grad_norm": 0.101755291223526,
      "learning_rate": 5.8636363636363634e-05,
      "loss": 0.0128,
      "step": 2730
    },
    {
      "epoch": 41.43939393939394,
      "grad_norm": 0.026371806859970093,
      "learning_rate": 5.856060606060606e-05,
      "loss": 0.009,
      "step": 2735
    },
    {
      "epoch": 41.515151515151516,
      "grad_norm": 0.07032539695501328,
      "learning_rate": 5.848484848484849e-05,
      "loss": 0.0072,
      "step": 2740
    },
    {
      "epoch": 41.59090909090909,
      "grad_norm": 0.04797264188528061,
      "learning_rate": 5.840909090909091e-05,
      "loss": 0.0153,
      "step": 2745
    },
    {
      "epoch": 41.666666666666664,
      "grad_norm": 0.035804979503154755,
      "learning_rate": 5.833333333333334e-05,
      "loss": 0.0093,
      "step": 2750
    },
    {
      "epoch": 41.74242424242424,
      "grad_norm": 0.04653352126479149,
      "learning_rate": 5.825757575757575e-05,
      "loss": 0.0108,
      "step": 2755
    },
    {
      "epoch": 41.81818181818182,
      "grad_norm": 0.09822138398885727,
      "learning_rate": 5.818181818181818e-05,
      "loss": 0.017,
      "step": 2760
    },
    {
      "epoch": 41.89393939393939,
      "grad_norm": 0.10393213480710983,
      "learning_rate": 5.810606060606061e-05,
      "loss": 0.0119,
      "step": 2765
    },
    {
      "epoch": 41.96969696969697,
      "grad_norm": 0.04201122745871544,
      "learning_rate": 5.803030303030304e-05,
      "loss": 0.01,
      "step": 2770
    },
    {
      "epoch": 42.0,
      "eval_loss": 0.011807852424681187,
      "eval_mean_accuracy": 0.9605352524545471,
      "eval_mean_iou": 0.9301918007440554,
      "eval_runtime": 134.0335,
      "eval_samples_per_second": 1.753,
      "eval_steps_per_second": 0.224,
      "step": 2772
    },
    {
      "epoch": 42.04545454545455,
      "grad_norm": 0.10560622066259384,
      "learning_rate": 5.7954545454545464e-05,
      "loss": 0.0099,
      "step": 2775
    },
    {
      "epoch": 42.121212121212125,
      "grad_norm": 0.9810836315155029,
      "learning_rate": 5.787878787878788e-05,
      "loss": 0.0126,
      "step": 2780
    },
    {
      "epoch": 42.196969696969695,
      "grad_norm": 0.05452556163072586,
      "learning_rate": 5.78030303030303e-05,
      "loss": 0.0101,
      "step": 2785
    },
    {
      "epoch": 42.27272727272727,
      "grad_norm": 0.10924031585454941,
      "learning_rate": 5.772727272727273e-05,
      "loss": 0.0122,
      "step": 2790
    },
    {
      "epoch": 42.34848484848485,
      "grad_norm": 0.03945128992199898,
      "learning_rate": 5.7651515151515156e-05,
      "loss": 0.0101,
      "step": 2795
    },
    {
      "epoch": 42.42424242424242,
      "grad_norm": 0.07899191975593567,
      "learning_rate": 5.757575757575758e-05,
      "loss": 0.0102,
      "step": 2800
    },
    {
      "epoch": 42.5,
      "grad_norm": 0.12503716349601746,
      "learning_rate": 5.7499999999999995e-05,
      "loss": 0.0083,
      "step": 2805
    },
    {
      "epoch": 42.57575757575758,
      "grad_norm": 3.1894569396972656,
      "learning_rate": 5.742424242424243e-05,
      "loss": 0.0367,
      "step": 2810
    },
    {
      "epoch": 42.65151515151515,
      "grad_norm": 0.0746053084731102,
      "learning_rate": 5.7348484848484854e-05,
      "loss": 0.0101,
      "step": 2815
    },
    {
      "epoch": 42.72727272727273,
      "grad_norm": 0.06412798911333084,
      "learning_rate": 5.727272727272728e-05,
      "loss": 0.0123,
      "step": 2820
    },
    {
      "epoch": 42.803030303030305,
      "grad_norm": 0.05188719928264618,
      "learning_rate": 5.719696969696971e-05,
      "loss": 0.0119,
      "step": 2825
    },
    {
      "epoch": 42.878787878787875,
      "grad_norm": 0.03282471001148224,
      "learning_rate": 5.712121212121212e-05,
      "loss": 0.0112,
      "step": 2830
    },
    {
      "epoch": 42.95454545454545,
      "grad_norm": 0.05583253130316734,
      "learning_rate": 5.7045454545454546e-05,
      "loss": 0.0104,
      "step": 2835
    },
    {
      "epoch": 43.0,
      "eval_loss": 0.011778323911130428,
      "eval_mean_accuracy": 0.9629572913433008,
      "eval_mean_iou": 0.9290905644352141,
      "eval_runtime": 130.6664,
      "eval_samples_per_second": 1.798,
      "eval_steps_per_second": 0.23,
      "step": 2838
    },
    {
      "epoch": 43.03030303030303,
      "grad_norm": 0.04248515143990517,
      "learning_rate": 5.696969696969697e-05,
      "loss": 0.0097,
      "step": 2840
    },
    {
      "epoch": 43.10606060606061,
      "grad_norm": 0.0590214841067791,
      "learning_rate": 5.68939393939394e-05,
      "loss": 0.0086,
      "step": 2845
    },
    {
      "epoch": 43.18181818181818,
      "grad_norm": 0.04899401590228081,
      "learning_rate": 5.6818181818181825e-05,
      "loss": 0.0108,
      "step": 2850
    },
    {
      "epoch": 43.25757575757576,
      "grad_norm": 0.056327465921640396,
      "learning_rate": 5.6742424242424244e-05,
      "loss": 0.0093,
      "step": 2855
    },
    {
      "epoch": 43.333333333333336,
      "grad_norm": 0.058888841420412064,
      "learning_rate": 5.666666666666667e-05,
      "loss": 0.0097,
      "step": 2860
    },
    {
      "epoch": 43.40909090909091,
      "grad_norm": 0.04053435102105141,
      "learning_rate": 5.65909090909091e-05,
      "loss": 0.0102,
      "step": 2865
    },
    {
      "epoch": 43.484848484848484,
      "grad_norm": 0.02590072900056839,
      "learning_rate": 5.651515151515152e-05,
      "loss": 0.0105,
      "step": 2870
    },
    {
      "epoch": 43.56060606060606,
      "grad_norm": 0.0485180988907814,
      "learning_rate": 5.643939393939395e-05,
      "loss": 0.0115,
      "step": 2875
    },
    {
      "epoch": 43.63636363636363,
      "grad_norm": 0.07296877354383469,
      "learning_rate": 5.636363636363636e-05,
      "loss": 0.0102,
      "step": 2880
    },
    {
      "epoch": 43.71212121212121,
      "grad_norm": 0.07502567023038864,
      "learning_rate": 5.628787878787879e-05,
      "loss": 0.0113,
      "step": 2885
    },
    {
      "epoch": 43.78787878787879,
      "grad_norm": 0.05460130795836449,
      "learning_rate": 5.6212121212121215e-05,
      "loss": 0.0117,
      "step": 2890
    },
    {
      "epoch": 43.86363636363637,
      "grad_norm": 0.05533159151673317,
      "learning_rate": 5.613636363636364e-05,
      "loss": 0.011,
      "step": 2895
    },
    {
      "epoch": 43.93939393939394,
      "grad_norm": 0.06939251720905304,
      "learning_rate": 5.606060606060606e-05,
      "loss": 0.0134,
      "step": 2900
    },
    {
      "epoch": 44.0,
      "eval_loss": 0.011285110376775265,
      "eval_mean_accuracy": 0.9619925689516045,
      "eval_mean_iou": 0.9309522451912047,
      "eval_runtime": 132.3218,
      "eval_samples_per_second": 1.776,
      "eval_steps_per_second": 0.227,
      "step": 2904
    },
    {
      "epoch": 44.015151515151516,
      "grad_norm": 0.07195951789617538,
      "learning_rate": 5.598484848484849e-05,
      "loss": 0.0103,
      "step": 2905
    },
    {
      "epoch": 44.09090909090909,
      "grad_norm": 0.021849235519766808,
      "learning_rate": 5.5909090909090913e-05,
      "loss": 0.0071,
      "step": 2910
    },
    {
      "epoch": 44.166666666666664,
      "grad_norm": 0.14573627710342407,
      "learning_rate": 5.583333333333334e-05,
      "loss": 0.0119,
      "step": 2915
    },
    {
      "epoch": 44.24242424242424,
      "grad_norm": 0.036952074617147446,
      "learning_rate": 5.5757575757575766e-05,
      "loss": 0.0078,
      "step": 2920
    },
    {
      "epoch": 44.31818181818182,
      "grad_norm": 0.0779380202293396,
      "learning_rate": 5.568181818181818e-05,
      "loss": 0.0105,
      "step": 2925
    },
    {
      "epoch": 44.39393939393939,
      "grad_norm": 0.04914235323667526,
      "learning_rate": 5.5606060606060605e-05,
      "loss": 0.0093,
      "step": 2930
    },
    {
      "epoch": 44.46969696969697,
      "grad_norm": 0.0895208790898323,
      "learning_rate": 5.553030303030303e-05,
      "loss": 0.0116,
      "step": 2935
    },
    {
      "epoch": 44.54545454545455,
      "grad_norm": 0.06198227033019066,
      "learning_rate": 5.545454545454546e-05,
      "loss": 0.0121,
      "step": 2940
    },
    {
      "epoch": 44.621212121212125,
      "grad_norm": 0.07971720397472382,
      "learning_rate": 5.5378787878787884e-05,
      "loss": 0.0154,
      "step": 2945
    },
    {
      "epoch": 44.696969696969695,
      "grad_norm": 0.04046093672513962,
      "learning_rate": 5.5303030303030304e-05,
      "loss": 0.0096,
      "step": 2950
    },
    {
      "epoch": 44.77272727272727,
      "grad_norm": 0.05711892619729042,
      "learning_rate": 5.522727272727273e-05,
      "loss": 0.0142,
      "step": 2955
    },
    {
      "epoch": 44.84848484848485,
      "grad_norm": 0.046940967440605164,
      "learning_rate": 5.5151515151515156e-05,
      "loss": 0.0091,
      "step": 2960
    },
    {
      "epoch": 44.92424242424242,
      "grad_norm": 0.03442925587296486,
      "learning_rate": 5.507575757575758e-05,
      "loss": 0.0103,
      "step": 2965
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.05698355659842491,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.0091,
      "step": 2970
    },
    {
      "epoch": 45.0,
      "eval_loss": 0.011407570913434029,
      "eval_mean_accuracy": 0.9561268306335386,
      "eval_mean_iou": 0.9304397496752612,
      "eval_runtime": 130.7831,
      "eval_samples_per_second": 1.797,
      "eval_steps_per_second": 0.229,
      "step": 2970
    },
    {
      "epoch": 45.07575757575758,
      "grad_norm": 0.14581868052482605,
      "learning_rate": 5.492424242424242e-05,
      "loss": 0.011,
      "step": 2975
    },
    {
      "epoch": 45.15151515151515,
      "grad_norm": 0.024033572524785995,
      "learning_rate": 5.484848484848485e-05,
      "loss": 0.0073,
      "step": 2980
    },
    {
      "epoch": 45.22727272727273,
      "grad_norm": 0.0981559008359909,
      "learning_rate": 5.4772727272727274e-05,
      "loss": 0.011,
      "step": 2985
    },
    {
      "epoch": 45.303030303030305,
      "grad_norm": 0.0462665893137455,
      "learning_rate": 5.46969696969697e-05,
      "loss": 0.0118,
      "step": 2990
    },
    {
      "epoch": 45.378787878787875,
      "grad_norm": 0.03924930840730667,
      "learning_rate": 5.462121212121213e-05,
      "loss": 0.0085,
      "step": 2995
    },
    {
      "epoch": 45.45454545454545,
      "grad_norm": 0.08342628180980682,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 0.0132,
      "step": 3000
    },
    {
      "epoch": 45.53030303030303,
      "grad_norm": 0.16838541626930237,
      "learning_rate": 5.446969696969697e-05,
      "loss": 0.0104,
      "step": 3005
    },
    {
      "epoch": 45.60606060606061,
      "grad_norm": 0.03201046213507652,
      "learning_rate": 5.43939393939394e-05,
      "loss": 0.0084,
      "step": 3010
    },
    {
      "epoch": 45.68181818181818,
      "grad_norm": 0.026701470836997032,
      "learning_rate": 5.4318181818181825e-05,
      "loss": 0.0095,
      "step": 3015
    },
    {
      "epoch": 45.75757575757576,
      "grad_norm": 0.0687679573893547,
      "learning_rate": 5.424242424242425e-05,
      "loss": 0.0117,
      "step": 3020
    },
    {
      "epoch": 45.833333333333336,
      "grad_norm": 0.06110064685344696,
      "learning_rate": 5.4166666666666664e-05,
      "loss": 0.0075,
      "step": 3025
    },
    {
      "epoch": 45.90909090909091,
      "grad_norm": 0.06796802580356598,
      "learning_rate": 5.409090909090909e-05,
      "loss": 0.0124,
      "step": 3030
    },
    {
      "epoch": 45.984848484848484,
      "grad_norm": 0.06645475327968597,
      "learning_rate": 5.401515151515152e-05,
      "loss": 0.0125,
      "step": 3035
    },
    {
      "epoch": 46.0,
      "eval_loss": 0.011490856297314167,
      "eval_mean_accuracy": 0.9480319279192205,
      "eval_mean_iou": 0.9256079041559514,
      "eval_runtime": 129.7631,
      "eval_samples_per_second": 1.811,
      "eval_steps_per_second": 0.231,
      "step": 3036
    },
    {
      "epoch": 46.06060606060606,
      "grad_norm": 0.08423695713281631,
      "learning_rate": 5.393939393939394e-05,
      "loss": 0.0135,
      "step": 3040
    },
    {
      "epoch": 46.13636363636363,
      "grad_norm": 0.09289564192295074,
      "learning_rate": 5.386363636363637e-05,
      "loss": 0.01,
      "step": 3045
    },
    {
      "epoch": 46.21212121212121,
      "grad_norm": 0.05097874999046326,
      "learning_rate": 5.378787878787879e-05,
      "loss": 0.0112,
      "step": 3050
    },
    {
      "epoch": 46.28787878787879,
      "grad_norm": 0.08499103784561157,
      "learning_rate": 5.3712121212121215e-05,
      "loss": 0.0108,
      "step": 3055
    },
    {
      "epoch": 46.36363636363637,
      "grad_norm": 0.028906868770718575,
      "learning_rate": 5.363636363636364e-05,
      "loss": 0.0091,
      "step": 3060
    },
    {
      "epoch": 46.43939393939394,
      "grad_norm": 0.062237221747636795,
      "learning_rate": 5.356060606060607e-05,
      "loss": 0.0089,
      "step": 3065
    },
    {
      "epoch": 46.515151515151516,
      "grad_norm": 0.0699303075671196,
      "learning_rate": 5.348484848484848e-05,
      "loss": 0.0098,
      "step": 3070
    },
    {
      "epoch": 46.59090909090909,
      "grad_norm": 0.0665397047996521,
      "learning_rate": 5.340909090909091e-05,
      "loss": 0.0102,
      "step": 3075
    },
    {
      "epoch": 46.666666666666664,
      "grad_norm": 0.050889141857624054,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.011,
      "step": 3080
    },
    {
      "epoch": 46.74242424242424,
      "grad_norm": 0.07601216435432434,
      "learning_rate": 5.325757575757576e-05,
      "loss": 0.0071,
      "step": 3085
    },
    {
      "epoch": 46.81818181818182,
      "grad_norm": 0.06293993443250656,
      "learning_rate": 5.3181818181818186e-05,
      "loss": 0.0088,
      "step": 3090
    },
    {
      "epoch": 46.89393939393939,
      "grad_norm": 0.0674889013171196,
      "learning_rate": 5.3106060606060605e-05,
      "loss": 0.0099,
      "step": 3095
    },
    {
      "epoch": 46.96969696969697,
      "grad_norm": 0.05907618999481201,
      "learning_rate": 5.303030303030303e-05,
      "loss": 0.0129,
      "step": 3100
    },
    {
      "epoch": 47.0,
      "eval_loss": 0.011220579035580158,
      "eval_mean_accuracy": 0.9632626871718033,
      "eval_mean_iou": 0.9314998352683258,
      "eval_runtime": 152.4159,
      "eval_samples_per_second": 1.542,
      "eval_steps_per_second": 0.197,
      "step": 3102
    },
    {
      "epoch": 47.04545454545455,
      "grad_norm": 0.1426934450864792,
      "learning_rate": 5.295454545454546e-05,
      "loss": 0.0127,
      "step": 3105
    },
    {
      "epoch": 47.121212121212125,
      "grad_norm": 0.09816721826791763,
      "learning_rate": 5.2878787878787884e-05,
      "loss": 0.0093,
      "step": 3110
    },
    {
      "epoch": 47.196969696969695,
      "grad_norm": 0.06219260394573212,
      "learning_rate": 5.280303030303031e-05,
      "loss": 0.0093,
      "step": 3115
    },
    {
      "epoch": 47.27272727272727,
      "grad_norm": 0.03978537395596504,
      "learning_rate": 5.272727272727272e-05,
      "loss": 0.0096,
      "step": 3120
    },
    {
      "epoch": 47.34848484848485,
      "grad_norm": 0.05753485858440399,
      "learning_rate": 5.265151515151515e-05,
      "loss": 0.0088,
      "step": 3125
    },
    {
      "epoch": 47.42424242424242,
      "grad_norm": 0.026985613629221916,
      "learning_rate": 5.2575757575757576e-05,
      "loss": 0.0116,
      "step": 3130
    },
    {
      "epoch": 47.5,
      "grad_norm": 0.20751309394836426,
      "learning_rate": 5.25e-05,
      "loss": 0.0122,
      "step": 3135
    },
    {
      "epoch": 47.57575757575758,
      "grad_norm": 0.034716032445430756,
      "learning_rate": 5.242424242424243e-05,
      "loss": 0.0082,
      "step": 3140
    },
    {
      "epoch": 47.65151515151515,
      "grad_norm": 0.04076504334807396,
      "learning_rate": 5.234848484848485e-05,
      "loss": 0.0108,
      "step": 3145
    },
    {
      "epoch": 47.72727272727273,
      "grad_norm": 0.022971780970692635,
      "learning_rate": 5.2272727272727274e-05,
      "loss": 0.009,
      "step": 3150
    },
    {
      "epoch": 47.803030303030305,
      "grad_norm": 0.02886657416820526,
      "learning_rate": 5.21969696969697e-05,
      "loss": 0.0105,
      "step": 3155
    },
    {
      "epoch": 47.878787878787875,
      "grad_norm": 0.04496586322784424,
      "learning_rate": 5.212121212121213e-05,
      "loss": 0.0085,
      "step": 3160
    },
    {
      "epoch": 47.95454545454545,
      "grad_norm": 0.09009754657745361,
      "learning_rate": 5.204545454545455e-05,
      "loss": 0.0108,
      "step": 3165
    },
    {
      "epoch": 48.0,
      "eval_loss": 0.011064937338232994,
      "eval_mean_accuracy": 0.9630187045319115,
      "eval_mean_iou": 0.9319738545809358,
      "eval_runtime": 130.8298,
      "eval_samples_per_second": 1.796,
      "eval_steps_per_second": 0.229,
      "step": 3168
    },
    {
      "epoch": 48.03030303030303,
      "grad_norm": 0.04920899495482445,
      "learning_rate": 5.1969696969696966e-05,
      "loss": 0.0101,
      "step": 3170
    },
    {
      "epoch": 48.10606060606061,
      "grad_norm": 0.034927595406770706,
      "learning_rate": 5.189393939393939e-05,
      "loss": 0.0095,
      "step": 3175
    },
    {
      "epoch": 48.18181818181818,
      "grad_norm": 0.0440865233540535,
      "learning_rate": 5.181818181818182e-05,
      "loss": 0.008,
      "step": 3180
    },
    {
      "epoch": 48.25757575757576,
      "grad_norm": 0.1620052605867386,
      "learning_rate": 5.1742424242424245e-05,
      "loss": 0.013,
      "step": 3185
    },
    {
      "epoch": 48.333333333333336,
      "grad_norm": 0.1449420005083084,
      "learning_rate": 5.166666666666667e-05,
      "loss": 0.0098,
      "step": 3190
    },
    {
      "epoch": 48.40909090909091,
      "grad_norm": 0.04209984093904495,
      "learning_rate": 5.159090909090909e-05,
      "loss": 0.0097,
      "step": 3195
    },
    {
      "epoch": 48.484848484848484,
      "grad_norm": 0.07320595532655716,
      "learning_rate": 5.151515151515152e-05,
      "loss": 0.0084,
      "step": 3200
    },
    {
      "epoch": 48.56060606060606,
      "grad_norm": 0.020307034254074097,
      "learning_rate": 5.143939393939394e-05,
      "loss": 0.0092,
      "step": 3205
    },
    {
      "epoch": 48.63636363636363,
      "grad_norm": 0.1068815141916275,
      "learning_rate": 5.136363636363637e-05,
      "loss": 0.0131,
      "step": 3210
    },
    {
      "epoch": 48.71212121212121,
      "grad_norm": 0.047849494963884354,
      "learning_rate": 5.1287878787878796e-05,
      "loss": 0.0128,
      "step": 3215
    },
    {
      "epoch": 48.78787878787879,
      "grad_norm": 0.06492839753627777,
      "learning_rate": 5.121212121212121e-05,
      "loss": 0.0103,
      "step": 3220
    },
    {
      "epoch": 48.86363636363637,
      "grad_norm": 0.07225367426872253,
      "learning_rate": 5.1136363636363635e-05,
      "loss": 0.0093,
      "step": 3225
    },
    {
      "epoch": 48.93939393939394,
      "grad_norm": 0.12909355759620667,
      "learning_rate": 5.106060606060606e-05,
      "loss": 0.0077,
      "step": 3230
    },
    {
      "epoch": 49.0,
      "eval_loss": 0.011090602725744247,
      "eval_mean_accuracy": 0.9584174549795046,
      "eval_mean_iou": 0.9314926902013566,
      "eval_runtime": 133.6755,
      "eval_samples_per_second": 1.758,
      "eval_steps_per_second": 0.224,
      "step": 3234
    },
    {
      "epoch": 49.015151515151516,
      "grad_norm": 0.05509858950972557,
      "learning_rate": 5.098484848484849e-05,
      "loss": 0.01,
      "step": 3235
    },
    {
      "epoch": 49.09090909090909,
      "grad_norm": 0.032800354063510895,
      "learning_rate": 5.090909090909091e-05,
      "loss": 0.0096,
      "step": 3240
    },
    {
      "epoch": 49.166666666666664,
      "grad_norm": 0.021322358399629593,
      "learning_rate": 5.0833333333333333e-05,
      "loss": 0.0082,
      "step": 3245
    },
    {
      "epoch": 49.24242424242424,
      "grad_norm": 0.046150654554367065,
      "learning_rate": 5.075757575757576e-05,
      "loss": 0.0107,
      "step": 3250
    },
    {
      "epoch": 49.31818181818182,
      "grad_norm": 0.05532953515648842,
      "learning_rate": 5.0681818181818186e-05,
      "loss": 0.0113,
      "step": 3255
    },
    {
      "epoch": 49.39393939393939,
      "grad_norm": 0.15334443747997284,
      "learning_rate": 5.060606060606061e-05,
      "loss": 0.0095,
      "step": 3260
    },
    {
      "epoch": 49.46969696969697,
      "grad_norm": 0.09281843155622482,
      "learning_rate": 5.0530303030303025e-05,
      "loss": 0.0122,
      "step": 3265
    },
    {
      "epoch": 49.54545454545455,
      "grad_norm": 0.03623240068554878,
      "learning_rate": 5.045454545454545e-05,
      "loss": 0.0075,
      "step": 3270
    },
    {
      "epoch": 49.621212121212125,
      "grad_norm": 0.045548029243946075,
      "learning_rate": 5.037878787878788e-05,
      "loss": 0.0104,
      "step": 3275
    },
    {
      "epoch": 49.696969696969695,
      "grad_norm": 0.03415241092443466,
      "learning_rate": 5.030303030303031e-05,
      "loss": 0.0104,
      "step": 3280
    },
    {
      "epoch": 49.77272727272727,
      "grad_norm": 0.12308188527822495,
      "learning_rate": 5.022727272727274e-05,
      "loss": 0.0083,
      "step": 3285
    },
    {
      "epoch": 49.84848484848485,
      "grad_norm": 0.08811760693788528,
      "learning_rate": 5.015151515151515e-05,
      "loss": 0.0105,
      "step": 3290
    },
    {
      "epoch": 49.92424242424242,
      "grad_norm": 0.04444101080298424,
      "learning_rate": 5.0075757575757576e-05,
      "loss": 0.0093,
      "step": 3295
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.26643773913383484,
      "learning_rate": 5e-05,
      "loss": 0.0107,
      "step": 3300
    },
    {
      "epoch": 50.0,
      "eval_loss": 0.010924181900918484,
      "eval_mean_accuracy": 0.9634208284402207,
      "eval_mean_iou": 0.9326889608867445,
      "eval_runtime": 134.5939,
      "eval_samples_per_second": 1.746,
      "eval_steps_per_second": 0.223,
      "step": 3300
    },
    {
      "epoch": 50.07575757575758,
      "grad_norm": 0.06223462522029877,
      "learning_rate": 4.992424242424243e-05,
      "loss": 0.0082,
      "step": 3305
    },
    {
      "epoch": 50.15151515151515,
      "grad_norm": 0.06327224522829056,
      "learning_rate": 4.984848484848485e-05,
      "loss": 0.0091,
      "step": 3310
    },
    {
      "epoch": 50.22727272727273,
      "grad_norm": 0.06920851767063141,
      "learning_rate": 4.9772727272727275e-05,
      "loss": 0.0092,
      "step": 3315
    },
    {
      "epoch": 50.303030303030305,
      "grad_norm": 0.02621302753686905,
      "learning_rate": 4.9696969696969694e-05,
      "loss": 0.0098,
      "step": 3320
    },
    {
      "epoch": 50.378787878787875,
      "grad_norm": 0.13362650573253632,
      "learning_rate": 4.962121212121213e-05,
      "loss": 0.0096,
      "step": 3325
    },
    {
      "epoch": 50.45454545454545,
      "grad_norm": 0.036353107541799545,
      "learning_rate": 4.9545454545454553e-05,
      "loss": 0.0076,
      "step": 3330
    },
    {
      "epoch": 50.53030303030303,
      "grad_norm": 0.030746368691325188,
      "learning_rate": 4.946969696969697e-05,
      "loss": 0.0077,
      "step": 3335
    },
    {
      "epoch": 50.60606060606061,
      "grad_norm": 0.019784174859523773,
      "learning_rate": 4.93939393939394e-05,
      "loss": 0.0138,
      "step": 3340
    },
    {
      "epoch": 50.68181818181818,
      "grad_norm": 0.03525077551603317,
      "learning_rate": 4.931818181818182e-05,
      "loss": 0.0079,
      "step": 3345
    },
    {
      "epoch": 50.75757575757576,
      "grad_norm": 0.04255997762084007,
      "learning_rate": 4.9242424242424245e-05,
      "loss": 0.0118,
      "step": 3350
    },
    {
      "epoch": 50.833333333333336,
      "grad_norm": 0.052387967705726624,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 0.0162,
      "step": 3355
    },
    {
      "epoch": 50.90909090909091,
      "grad_norm": 0.07029189169406891,
      "learning_rate": 4.909090909090909e-05,
      "loss": 0.0111,
      "step": 3360
    },
    {
      "epoch": 50.984848484848484,
      "grad_norm": 0.04237540066242218,
      "learning_rate": 4.901515151515152e-05,
      "loss": 0.0108,
      "step": 3365
    },
    {
      "epoch": 51.0,
      "eval_loss": 0.011673275381326675,
      "eval_mean_accuracy": 0.944879533220683,
      "eval_mean_iou": 0.9227064720209003,
      "eval_runtime": 129.208,
      "eval_samples_per_second": 1.819,
      "eval_steps_per_second": 0.232,
      "step": 3366
    },
    {
      "epoch": 51.06060606060606,
      "grad_norm": 0.043555933982133865,
      "learning_rate": 4.8939393939393944e-05,
      "loss": 0.0133,
      "step": 3370
    },
    {
      "epoch": 51.13636363636363,
      "grad_norm": 0.01770366169512272,
      "learning_rate": 4.886363636363637e-05,
      "loss": 0.0114,
      "step": 3375
    },
    {
      "epoch": 51.21212121212121,
      "grad_norm": 0.14457736909389496,
      "learning_rate": 4.878787878787879e-05,
      "loss": 0.0085,
      "step": 3380
    },
    {
      "epoch": 51.28787878787879,
      "grad_norm": 0.03227056935429573,
      "learning_rate": 4.8712121212121216e-05,
      "loss": 0.0072,
      "step": 3385
    },
    {
      "epoch": 51.36363636363637,
      "grad_norm": 0.05647565796971321,
      "learning_rate": 4.863636363636364e-05,
      "loss": 0.0123,
      "step": 3390
    },
    {
      "epoch": 51.43939393939394,
      "grad_norm": 0.02439090609550476,
      "learning_rate": 4.856060606060606e-05,
      "loss": 0.0079,
      "step": 3395
    },
    {
      "epoch": 51.515151515151516,
      "grad_norm": 0.062035221606492996,
      "learning_rate": 4.848484848484849e-05,
      "loss": 0.0087,
      "step": 3400
    },
    {
      "epoch": 51.59090909090909,
      "grad_norm": 0.0548180416226387,
      "learning_rate": 4.840909090909091e-05,
      "loss": 0.0105,
      "step": 3405
    },
    {
      "epoch": 51.666666666666664,
      "grad_norm": 0.05994485691189766,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 0.0102,
      "step": 3410
    },
    {
      "epoch": 51.74242424242424,
      "grad_norm": 0.031134238466620445,
      "learning_rate": 4.825757575757576e-05,
      "loss": 0.0134,
      "step": 3415
    },
    {
      "epoch": 51.81818181818182,
      "grad_norm": 0.025048216804862022,
      "learning_rate": 4.8181818181818186e-05,
      "loss": 0.0076,
      "step": 3420
    },
    {
      "epoch": 51.89393939393939,
      "grad_norm": 0.06273403763771057,
      "learning_rate": 4.810606060606061e-05,
      "loss": 0.0082,
      "step": 3425
    },
    {
      "epoch": 51.96969696969697,
      "grad_norm": 0.0426410473883152,
      "learning_rate": 4.803030303030303e-05,
      "loss": 0.0095,
      "step": 3430
    },
    {
      "epoch": 52.0,
      "eval_loss": 0.011105363257229328,
      "eval_mean_accuracy": 0.9644564674261143,
      "eval_mean_iou": 0.9319910891202007,
      "eval_runtime": 112.4443,
      "eval_samples_per_second": 2.09,
      "eval_steps_per_second": 0.267,
      "step": 3432
    },
    {
      "epoch": 52.04545454545455,
      "grad_norm": 0.10171551257371902,
      "learning_rate": 4.795454545454546e-05,
      "loss": 0.0079,
      "step": 3435
    },
    {
      "epoch": 52.121212121212125,
      "grad_norm": 0.042261794209480286,
      "learning_rate": 4.787878787878788e-05,
      "loss": 0.0127,
      "step": 3440
    },
    {
      "epoch": 52.196969696969695,
      "grad_norm": 0.04945836961269379,
      "learning_rate": 4.7803030303030304e-05,
      "loss": 0.0122,
      "step": 3445
    },
    {
      "epoch": 52.27272727272727,
      "grad_norm": 0.09596319496631622,
      "learning_rate": 4.772727272727273e-05,
      "loss": 0.0101,
      "step": 3450
    },
    {
      "epoch": 52.34848484848485,
      "grad_norm": 0.034693412482738495,
      "learning_rate": 4.765151515151515e-05,
      "loss": 0.0099,
      "step": 3455
    },
    {
      "epoch": 52.42424242424242,
      "grad_norm": 0.03737581893801689,
      "learning_rate": 4.7575757575757576e-05,
      "loss": 0.0071,
      "step": 3460
    },
    {
      "epoch": 52.5,
      "grad_norm": 0.03867292031645775,
      "learning_rate": 4.75e-05,
      "loss": 0.0106,
      "step": 3465
    },
    {
      "epoch": 52.57575757575758,
      "grad_norm": 0.05379658192396164,
      "learning_rate": 4.742424242424243e-05,
      "loss": 0.0109,
      "step": 3470
    },
    {
      "epoch": 52.65151515151515,
      "grad_norm": 0.06304959207773209,
      "learning_rate": 4.7348484848484855e-05,
      "loss": 0.0095,
      "step": 3475
    },
    {
      "epoch": 52.72727272727273,
      "grad_norm": 0.036937154829502106,
      "learning_rate": 4.7272727272727275e-05,
      "loss": 0.0077,
      "step": 3480
    },
    {
      "epoch": 52.803030303030305,
      "grad_norm": 0.030664615333080292,
      "learning_rate": 4.71969696969697e-05,
      "loss": 0.0104,
      "step": 3485
    },
    {
      "epoch": 52.878787878787875,
      "grad_norm": 0.05126954987645149,
      "learning_rate": 4.712121212121212e-05,
      "loss": 0.012,
      "step": 3490
    },
    {
      "epoch": 52.95454545454545,
      "grad_norm": 0.11736458539962769,
      "learning_rate": 4.704545454545455e-05,
      "loss": 0.0084,
      "step": 3495
    },
    {
      "epoch": 53.0,
      "eval_loss": 0.011148341931402683,
      "eval_mean_accuracy": 0.9540419344767453,
      "eval_mean_iou": 0.9298485245323042,
      "eval_runtime": 112.0124,
      "eval_samples_per_second": 2.098,
      "eval_steps_per_second": 0.268,
      "step": 3498
    },
    {
      "epoch": 53.03030303030303,
      "grad_norm": 0.03938106819987297,
      "learning_rate": 4.696969696969697e-05,
      "loss": 0.0075,
      "step": 3500
    },
    {
      "epoch": 53.10606060606061,
      "grad_norm": 0.025069458410143852,
      "learning_rate": 4.689393939393939e-05,
      "loss": 0.0079,
      "step": 3505
    },
    {
      "epoch": 53.18181818181818,
      "grad_norm": 0.1726539433002472,
      "learning_rate": 4.681818181818182e-05,
      "loss": 0.0106,
      "step": 3510
    },
    {
      "epoch": 53.25757575757576,
      "grad_norm": 0.026888813823461533,
      "learning_rate": 4.6742424242424245e-05,
      "loss": 0.0072,
      "step": 3515
    },
    {
      "epoch": 53.333333333333336,
      "grad_norm": 0.07678517699241638,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.0111,
      "step": 3520
    },
    {
      "epoch": 53.40909090909091,
      "grad_norm": 0.047660522162914276,
      "learning_rate": 4.659090909090909e-05,
      "loss": 0.0088,
      "step": 3525
    },
    {
      "epoch": 53.484848484848484,
      "grad_norm": 0.0497659407556057,
      "learning_rate": 4.651515151515152e-05,
      "loss": 0.0101,
      "step": 3530
    },
    {
      "epoch": 53.56060606060606,
      "grad_norm": 0.12443435937166214,
      "learning_rate": 4.6439393939393944e-05,
      "loss": 0.0126,
      "step": 3535
    },
    {
      "epoch": 53.63636363636363,
      "grad_norm": 0.05502001568675041,
      "learning_rate": 4.636363636363636e-05,
      "loss": 0.0098,
      "step": 3540
    },
    {
      "epoch": 53.71212121212121,
      "grad_norm": 0.07882898300886154,
      "learning_rate": 4.628787878787879e-05,
      "loss": 0.0103,
      "step": 3545
    },
    {
      "epoch": 53.78787878787879,
      "grad_norm": 0.09978947788476944,
      "learning_rate": 4.621212121212121e-05,
      "loss": 0.0106,
      "step": 3550
    },
    {
      "epoch": 53.86363636363637,
      "grad_norm": 0.07483825832605362,
      "learning_rate": 4.6136363636363635e-05,
      "loss": 0.0101,
      "step": 3555
    },
    {
      "epoch": 53.93939393939394,
      "grad_norm": 0.034889768809080124,
      "learning_rate": 4.606060606060607e-05,
      "loss": 0.0067,
      "step": 3560
    },
    {
      "epoch": 54.0,
      "eval_loss": 0.011029655113816261,
      "eval_mean_accuracy": 0.9662136909568274,
      "eval_mean_iou": 0.9316393430036829,
      "eval_runtime": 114.0109,
      "eval_samples_per_second": 2.061,
      "eval_steps_per_second": 0.263,
      "step": 3564
    },
    {
      "epoch": 54.015151515151516,
      "grad_norm": 0.03548719361424446,
      "learning_rate": 4.598484848484849e-05,
      "loss": 0.0087,
      "step": 3565
    },
    {
      "epoch": 54.09090909090909,
      "grad_norm": 0.04339136928319931,
      "learning_rate": 4.5909090909090914e-05,
      "loss": 0.0079,
      "step": 3570
    },
    {
      "epoch": 54.166666666666664,
      "grad_norm": 0.0416102297604084,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.009,
      "step": 3575
    },
    {
      "epoch": 54.24242424242424,
      "grad_norm": 0.0798560157418251,
      "learning_rate": 4.575757575757576e-05,
      "loss": 0.0105,
      "step": 3580
    },
    {
      "epoch": 54.31818181818182,
      "grad_norm": 0.04902668669819832,
      "learning_rate": 4.5681818181818186e-05,
      "loss": 0.0108,
      "step": 3585
    },
    {
      "epoch": 54.39393939393939,
      "grad_norm": 0.13727474212646484,
      "learning_rate": 4.5606060606060606e-05,
      "loss": 0.0092,
      "step": 3590
    },
    {
      "epoch": 54.46969696969697,
      "grad_norm": 0.054456207901239395,
      "learning_rate": 4.553030303030303e-05,
      "loss": 0.0115,
      "step": 3595
    },
    {
      "epoch": 54.54545454545455,
      "grad_norm": 0.08882977813482285,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.0117,
      "step": 3600
    },
    {
      "epoch": 54.621212121212125,
      "grad_norm": 0.1582881361246109,
      "learning_rate": 4.5378787878787885e-05,
      "loss": 0.0108,
      "step": 3605
    },
    {
      "epoch": 54.696969696969695,
      "grad_norm": 0.030232032760977745,
      "learning_rate": 4.5303030303030304e-05,
      "loss": 0.0112,
      "step": 3610
    },
    {
      "epoch": 54.77272727272727,
      "grad_norm": 0.039854396134614944,
      "learning_rate": 4.522727272727273e-05,
      "loss": 0.0085,
      "step": 3615
    },
    {
      "epoch": 54.84848484848485,
      "grad_norm": 0.08767293393611908,
      "learning_rate": 4.515151515151516e-05,
      "loss": 0.0096,
      "step": 3620
    },
    {
      "epoch": 54.92424242424242,
      "grad_norm": 0.01435124408453703,
      "learning_rate": 4.5075757575757577e-05,
      "loss": 0.0083,
      "step": 3625
    },
    {
      "epoch": 55.0,
      "grad_norm": 0.029528262093663216,
      "learning_rate": 4.5e-05,
      "loss": 0.0087,
      "step": 3630
    },
    {
      "epoch": 55.0,
      "eval_loss": 0.010832401923835278,
      "eval_mean_accuracy": 0.9664124280587,
      "eval_mean_iou": 0.9334025763664177,
      "eval_runtime": 114.2132,
      "eval_samples_per_second": 2.058,
      "eval_steps_per_second": 0.263,
      "step": 3630
    },
    {
      "epoch": 55.07575757575758,
      "grad_norm": 0.03741980716586113,
      "learning_rate": 4.492424242424242e-05,
      "loss": 0.0101,
      "step": 3635
    },
    {
      "epoch": 55.15151515151515,
      "grad_norm": 0.042487241327762604,
      "learning_rate": 4.484848484848485e-05,
      "loss": 0.0119,
      "step": 3640
    },
    {
      "epoch": 55.22727272727273,
      "grad_norm": 0.039345212280750275,
      "learning_rate": 4.4772727272727275e-05,
      "loss": 0.01,
      "step": 3645
    },
    {
      "epoch": 55.303030303030305,
      "grad_norm": 0.03277987986803055,
      "learning_rate": 4.46969696969697e-05,
      "loss": 0.0066,
      "step": 3650
    },
    {
      "epoch": 55.378787878787875,
      "grad_norm": 0.06374022364616394,
      "learning_rate": 4.462121212121213e-05,
      "loss": 0.0105,
      "step": 3655
    },
    {
      "epoch": 55.45454545454545,
      "grad_norm": 0.11683417111635208,
      "learning_rate": 4.454545454545455e-05,
      "loss": 0.0095,
      "step": 3660
    },
    {
      "epoch": 55.53030303030303,
      "grad_norm": 0.030355941504240036,
      "learning_rate": 4.4469696969696973e-05,
      "loss": 0.0095,
      "step": 3665
    },
    {
      "epoch": 55.60606060606061,
      "grad_norm": 0.059898145496845245,
      "learning_rate": 4.43939393939394e-05,
      "loss": 0.0106,
      "step": 3670
    },
    {
      "epoch": 55.68181818181818,
      "grad_norm": 0.03135860711336136,
      "learning_rate": 4.431818181818182e-05,
      "loss": 0.0105,
      "step": 3675
    },
    {
      "epoch": 55.75757575757576,
      "grad_norm": 0.050088077783584595,
      "learning_rate": 4.4242424242424246e-05,
      "loss": 0.0082,
      "step": 3680
    },
    {
      "epoch": 55.833333333333336,
      "grad_norm": 0.05700050666928291,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 0.0086,
      "step": 3685
    },
    {
      "epoch": 55.90909090909091,
      "grad_norm": 0.05043916031718254,
      "learning_rate": 4.409090909090909e-05,
      "loss": 0.0065,
      "step": 3690
    },
    {
      "epoch": 55.984848484848484,
      "grad_norm": 0.04749040678143501,
      "learning_rate": 4.401515151515152e-05,
      "loss": 0.0092,
      "step": 3695
    },
    {
      "epoch": 56.0,
      "eval_loss": 0.010891544632613659,
      "eval_mean_accuracy": 0.9643856128592406,
      "eval_mean_iou": 0.9333505509212876,
      "eval_runtime": 115.4044,
      "eval_samples_per_second": 2.036,
      "eval_steps_per_second": 0.26,
      "step": 3696
    },
    {
      "epoch": 56.06060606060606,
      "grad_norm": 0.03183702379465103,
      "learning_rate": 4.3939393939393944e-05,
      "loss": 0.01,
      "step": 3700
    },
    {
      "epoch": 56.13636363636363,
      "grad_norm": 0.028371397405862808,
      "learning_rate": 4.386363636363637e-05,
      "loss": 0.0084,
      "step": 3705
    },
    {
      "epoch": 56.21212121212121,
      "grad_norm": 0.021252259612083435,
      "learning_rate": 4.378787878787879e-05,
      "loss": 0.0076,
      "step": 3710
    },
    {
      "epoch": 56.28787878787879,
      "grad_norm": 0.08984354138374329,
      "learning_rate": 4.3712121212121216e-05,
      "loss": 0.0101,
      "step": 3715
    },
    {
      "epoch": 56.36363636363637,
      "grad_norm": 0.0206893403083086,
      "learning_rate": 4.3636363636363636e-05,
      "loss": 0.0097,
      "step": 3720
    },
    {
      "epoch": 56.43939393939394,
      "grad_norm": 0.02762766368687153,
      "learning_rate": 4.356060606060606e-05,
      "loss": 0.0094,
      "step": 3725
    },
    {
      "epoch": 56.515151515151516,
      "grad_norm": 0.029712922871112823,
      "learning_rate": 4.348484848484849e-05,
      "loss": 0.0077,
      "step": 3730
    },
    {
      "epoch": 56.59090909090909,
      "grad_norm": 0.0653505027294159,
      "learning_rate": 4.340909090909091e-05,
      "loss": 0.0098,
      "step": 3735
    },
    {
      "epoch": 56.666666666666664,
      "grad_norm": 0.04493647441267967,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.0099,
      "step": 3740
    },
    {
      "epoch": 56.74242424242424,
      "grad_norm": 0.060800135135650635,
      "learning_rate": 4.325757575757576e-05,
      "loss": 0.0126,
      "step": 3745
    },
    {
      "epoch": 56.81818181818182,
      "grad_norm": 0.04947813227772713,
      "learning_rate": 4.318181818181819e-05,
      "loss": 0.0105,
      "step": 3750
    },
    {
      "epoch": 56.89393939393939,
      "grad_norm": 0.02774951606988907,
      "learning_rate": 4.3106060606060606e-05,
      "loss": 0.007,
      "step": 3755
    },
    {
      "epoch": 56.96969696969697,
      "grad_norm": 0.06900814175605774,
      "learning_rate": 4.303030303030303e-05,
      "loss": 0.0088,
      "step": 3760
    },
    {
      "epoch": 57.0,
      "eval_loss": 0.011131164617836475,
      "eval_mean_accuracy": 0.9516373501330276,
      "eval_mean_iou": 0.9278086015722135,
      "eval_runtime": 116.5976,
      "eval_samples_per_second": 2.015,
      "eval_steps_per_second": 0.257,
      "step": 3762
    },
    {
      "epoch": 57.04545454545455,
      "grad_norm": 0.16135533154010773,
      "learning_rate": 4.295454545454546e-05,
      "loss": 0.0095,
      "step": 3765
    },
    {
      "epoch": 57.121212121212125,
      "grad_norm": 0.027069350704550743,
      "learning_rate": 4.287878787878788e-05,
      "loss": 0.0081,
      "step": 3770
    },
    {
      "epoch": 57.196969696969695,
      "grad_norm": 0.049404773861169815,
      "learning_rate": 4.2803030303030305e-05,
      "loss": 0.0098,
      "step": 3775
    },
    {
      "epoch": 57.27272727272727,
      "grad_norm": 0.02999548427760601,
      "learning_rate": 4.2727272727272724e-05,
      "loss": 0.009,
      "step": 3780
    },
    {
      "epoch": 57.34848484848485,
      "grad_norm": 0.10875272750854492,
      "learning_rate": 4.265151515151515e-05,
      "loss": 0.0109,
      "step": 3785
    },
    {
      "epoch": 57.42424242424242,
      "grad_norm": 0.06954370439052582,
      "learning_rate": 4.257575757575758e-05,
      "loss": 0.0091,
      "step": 3790
    },
    {
      "epoch": 57.5,
      "grad_norm": 0.03223840147256851,
      "learning_rate": 4.25e-05,
      "loss": 0.0099,
      "step": 3795
    },
    {
      "epoch": 57.57575757575758,
      "grad_norm": 0.07136336714029312,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.0095,
      "step": 3800
    },
    {
      "epoch": 57.65151515151515,
      "grad_norm": 0.021398568525910378,
      "learning_rate": 4.234848484848485e-05,
      "loss": 0.0077,
      "step": 3805
    },
    {
      "epoch": 57.72727272727273,
      "grad_norm": 0.07841000705957413,
      "learning_rate": 4.2272727272727275e-05,
      "loss": 0.0108,
      "step": 3810
    },
    {
      "epoch": 57.803030303030305,
      "grad_norm": 0.06136574223637581,
      "learning_rate": 4.21969696969697e-05,
      "loss": 0.0098,
      "step": 3815
    },
    {
      "epoch": 57.878787878787875,
      "grad_norm": 0.04557708650827408,
      "learning_rate": 4.212121212121212e-05,
      "loss": 0.0091,
      "step": 3820
    },
    {
      "epoch": 57.95454545454545,
      "grad_norm": 0.028318628668785095,
      "learning_rate": 4.204545454545455e-05,
      "loss": 0.0094,
      "step": 3825
    },
    {
      "epoch": 58.0,
      "eval_loss": 0.010866796597838402,
      "eval_mean_accuracy": 0.9655244247930653,
      "eval_mean_iou": 0.9327643953844954,
      "eval_runtime": 112.573,
      "eval_samples_per_second": 2.088,
      "eval_steps_per_second": 0.266,
      "step": 3828
    },
    {
      "epoch": 58.03030303030303,
      "grad_norm": 0.03608550503849983,
      "learning_rate": 4.196969696969697e-05,
      "loss": 0.009,
      "step": 3830
    },
    {
      "epoch": 58.10606060606061,
      "grad_norm": 0.03775801882147789,
      "learning_rate": 4.189393939393939e-05,
      "loss": 0.0101,
      "step": 3835
    },
    {
      "epoch": 58.18181818181818,
      "grad_norm": 0.014146799221634865,
      "learning_rate": 4.181818181818182e-05,
      "loss": 0.0075,
      "step": 3840
    },
    {
      "epoch": 58.25757575757576,
      "grad_norm": 0.16111214458942413,
      "learning_rate": 4.1742424242424246e-05,
      "loss": 0.0107,
      "step": 3845
    },
    {
      "epoch": 58.333333333333336,
      "grad_norm": 0.05974853038787842,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.0103,
      "step": 3850
    },
    {
      "epoch": 58.40909090909091,
      "grad_norm": 0.040087584406137466,
      "learning_rate": 4.159090909090909e-05,
      "loss": 0.0093,
      "step": 3855
    },
    {
      "epoch": 58.484848484848484,
      "grad_norm": 0.029749684035778046,
      "learning_rate": 4.151515151515152e-05,
      "loss": 0.0101,
      "step": 3860
    },
    {
      "epoch": 58.56060606060606,
      "grad_norm": 0.05692281946539879,
      "learning_rate": 4.143939393939394e-05,
      "loss": 0.009,
      "step": 3865
    },
    {
      "epoch": 58.63636363636363,
      "grad_norm": 0.04727243632078171,
      "learning_rate": 4.1363636363636364e-05,
      "loss": 0.0091,
      "step": 3870
    },
    {
      "epoch": 58.71212121212121,
      "grad_norm": 0.06694398075342178,
      "learning_rate": 4.128787878787879e-05,
      "loss": 0.0092,
      "step": 3875
    },
    {
      "epoch": 58.78787878787879,
      "grad_norm": 0.07738799601793289,
      "learning_rate": 4.1212121212121216e-05,
      "loss": 0.0077,
      "step": 3880
    },
    {
      "epoch": 58.86363636363637,
      "grad_norm": 0.042070720344781876,
      "learning_rate": 4.113636363636364e-05,
      "loss": 0.0097,
      "step": 3885
    },
    {
      "epoch": 58.93939393939394,
      "grad_norm": 0.050908736884593964,
      "learning_rate": 4.106060606060606e-05,
      "loss": 0.0089,
      "step": 3890
    },
    {
      "epoch": 59.0,
      "eval_loss": 0.01058129407465458,
      "eval_mean_accuracy": 0.9640679734824218,
      "eval_mean_iou": 0.9338439984688467,
      "eval_runtime": 108.5354,
      "eval_samples_per_second": 2.165,
      "eval_steps_per_second": 0.276,
      "step": 3894
    },
    {
      "epoch": 59.015151515151516,
      "grad_norm": 0.04071848839521408,
      "learning_rate": 4.098484848484849e-05,
      "loss": 0.0097,
      "step": 3895
    },
    {
      "epoch": 59.09090909090909,
      "grad_norm": 0.06917822360992432,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.0099,
      "step": 3900
    },
    {
      "epoch": 59.166666666666664,
      "grad_norm": 0.043478477746248245,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 0.0089,
      "step": 3905
    },
    {
      "epoch": 59.24242424242424,
      "grad_norm": 0.07832542806863785,
      "learning_rate": 4.075757575757576e-05,
      "loss": 0.0098,
      "step": 3910
    },
    {
      "epoch": 59.31818181818182,
      "grad_norm": 0.06101706624031067,
      "learning_rate": 4.068181818181818e-05,
      "loss": 0.0093,
      "step": 3915
    },
    {
      "epoch": 59.39393939393939,
      "grad_norm": 0.030086534097790718,
      "learning_rate": 4.0606060606060606e-05,
      "loss": 0.0092,
      "step": 3920
    },
    {
      "epoch": 59.46969696969697,
      "grad_norm": 0.028206603601574898,
      "learning_rate": 4.053030303030303e-05,
      "loss": 0.0081,
      "step": 3925
    },
    {
      "epoch": 59.54545454545455,
      "grad_norm": 0.06041272357106209,
      "learning_rate": 4.045454545454546e-05,
      "loss": 0.0093,
      "step": 3930
    },
    {
      "epoch": 59.621212121212125,
      "grad_norm": 0.09844840317964554,
      "learning_rate": 4.0378787878787885e-05,
      "loss": 0.008,
      "step": 3935
    },
    {
      "epoch": 59.696969696969695,
      "grad_norm": 0.04461321234703064,
      "learning_rate": 4.0303030303030305e-05,
      "loss": 0.01,
      "step": 3940
    },
    {
      "epoch": 59.77272727272727,
      "grad_norm": 0.034242309629917145,
      "learning_rate": 4.022727272727273e-05,
      "loss": 0.0077,
      "step": 3945
    },
    {
      "epoch": 59.84848484848485,
      "grad_norm": 0.21248526871204376,
      "learning_rate": 4.015151515151515e-05,
      "loss": 0.012,
      "step": 3950
    },
    {
      "epoch": 59.92424242424242,
      "grad_norm": 0.02297009713947773,
      "learning_rate": 4.007575757575758e-05,
      "loss": 0.0092,
      "step": 3955
    },
    {
      "epoch": 60.0,
      "grad_norm": 0.02758110500872135,
      "learning_rate": 4e-05,
      "loss": 0.0091,
      "step": 3960
    },
    {
      "epoch": 60.0,
      "eval_loss": 0.010756264440715313,
      "eval_mean_accuracy": 0.967835577034532,
      "eval_mean_iou": 0.9333732288022739,
      "eval_runtime": 108.7949,
      "eval_samples_per_second": 2.16,
      "eval_steps_per_second": 0.276,
      "step": 3960
    },
    {
      "epoch": 60.07575757575758,
      "grad_norm": 0.020747516304254532,
      "learning_rate": 3.992424242424242e-05,
      "loss": 0.0082,
      "step": 3965
    },
    {
      "epoch": 60.15151515151515,
      "grad_norm": 0.04350237920880318,
      "learning_rate": 3.984848484848485e-05,
      "loss": 0.0077,
      "step": 3970
    },
    {
      "epoch": 60.22727272727273,
      "grad_norm": 0.023618577048182487,
      "learning_rate": 3.9772727272727275e-05,
      "loss": 0.0125,
      "step": 3975
    },
    {
      "epoch": 60.303030303030305,
      "grad_norm": 0.06801124662160873,
      "learning_rate": 3.96969696969697e-05,
      "loss": 0.0084,
      "step": 3980
    },
    {
      "epoch": 60.378787878787875,
      "grad_norm": 0.047551725059747696,
      "learning_rate": 3.962121212121213e-05,
      "loss": 0.0105,
      "step": 3985
    },
    {
      "epoch": 60.45454545454545,
      "grad_norm": 0.1878127157688141,
      "learning_rate": 3.954545454545455e-05,
      "loss": 0.0098,
      "step": 3990
    },
    {
      "epoch": 60.53030303030303,
      "grad_norm": 0.041793301701545715,
      "learning_rate": 3.9469696969696974e-05,
      "loss": 0.0086,
      "step": 3995
    },
    {
      "epoch": 60.60606060606061,
      "grad_norm": 0.06554850935935974,
      "learning_rate": 3.939393939393939e-05,
      "loss": 0.0124,
      "step": 4000
    },
    {
      "epoch": 60.68181818181818,
      "grad_norm": 0.040125273168087006,
      "learning_rate": 3.931818181818182e-05,
      "loss": 0.0076,
      "step": 4005
    },
    {
      "epoch": 60.75757575757576,
      "grad_norm": 0.02520720846951008,
      "learning_rate": 3.924242424242424e-05,
      "loss": 0.0106,
      "step": 4010
    },
    {
      "epoch": 60.833333333333336,
      "grad_norm": 0.0420609712600708,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.0076,
      "step": 4015
    },
    {
      "epoch": 60.90909090909091,
      "grad_norm": 0.059919003397226334,
      "learning_rate": 3.909090909090909e-05,
      "loss": 0.0088,
      "step": 4020
    },
    {
      "epoch": 60.984848484848484,
      "grad_norm": 0.041421256959438324,
      "learning_rate": 3.901515151515152e-05,
      "loss": 0.0078,
      "step": 4025
    },
    {
      "epoch": 61.0,
      "eval_loss": 0.011201247572898865,
      "eval_mean_accuracy": 0.9692754170486914,
      "eval_mean_iou": 0.9311023382603762,
      "eval_runtime": 108.4252,
      "eval_samples_per_second": 2.167,
      "eval_steps_per_second": 0.277,
      "step": 4026
    },
    {
      "epoch": 61.06060606060606,
      "grad_norm": 0.06394296884536743,
      "learning_rate": 3.8939393939393944e-05,
      "loss": 0.0079,
      "step": 4030
    },
    {
      "epoch": 61.13636363636363,
      "grad_norm": 0.11667873710393906,
      "learning_rate": 3.8863636363636364e-05,
      "loss": 0.0117,
      "step": 4035
    },
    {
      "epoch": 61.21212121212121,
      "grad_norm": 0.0770072415471077,
      "learning_rate": 3.878787878787879e-05,
      "loss": 0.0111,
      "step": 4040
    },
    {
      "epoch": 61.28787878787879,
      "grad_norm": 0.044575486332178116,
      "learning_rate": 3.8712121212121217e-05,
      "loss": 0.0089,
      "step": 4045
    },
    {
      "epoch": 61.36363636363637,
      "grad_norm": 0.11559037119150162,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.0107,
      "step": 4050
    },
    {
      "epoch": 61.43939393939394,
      "grad_norm": 0.05656173452734947,
      "learning_rate": 3.856060606060606e-05,
      "loss": 0.0088,
      "step": 4055
    },
    {
      "epoch": 61.515151515151516,
      "grad_norm": 0.06268004328012466,
      "learning_rate": 3.848484848484848e-05,
      "loss": 0.0068,
      "step": 4060
    },
    {
      "epoch": 61.59090909090909,
      "grad_norm": 0.03743515536189079,
      "learning_rate": 3.840909090909091e-05,
      "loss": 0.0093,
      "step": 4065
    },
    {
      "epoch": 61.666666666666664,
      "grad_norm": 0.035856474190950394,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 0.008,
      "step": 4070
    },
    {
      "epoch": 61.74242424242424,
      "grad_norm": 0.01652700826525688,
      "learning_rate": 3.825757575757576e-05,
      "loss": 0.0077,
      "step": 4075
    },
    {
      "epoch": 61.81818181818182,
      "grad_norm": 0.036741528660058975,
      "learning_rate": 3.818181818181819e-05,
      "loss": 0.0103,
      "step": 4080
    },
    {
      "epoch": 61.89393939393939,
      "grad_norm": 0.03326910734176636,
      "learning_rate": 3.810606060606061e-05,
      "loss": 0.009,
      "step": 4085
    },
    {
      "epoch": 61.96969696969697,
      "grad_norm": 0.01467684656381607,
      "learning_rate": 3.803030303030303e-05,
      "loss": 0.0077,
      "step": 4090
    },
    {
      "epoch": 62.0,
      "eval_loss": 0.01053342130035162,
      "eval_mean_accuracy": 0.9651359640637462,
      "eval_mean_iou": 0.9339295319202497,
      "eval_runtime": 108.0145,
      "eval_samples_per_second": 2.176,
      "eval_steps_per_second": 0.278,
      "step": 4092
    },
    {
      "epoch": 62.04545454545455,
      "grad_norm": 0.0530879832804203,
      "learning_rate": 3.795454545454545e-05,
      "loss": 0.0102,
      "step": 4095
    },
    {
      "epoch": 62.121212121212125,
      "grad_norm": 0.046043481677770615,
      "learning_rate": 3.787878787878788e-05,
      "loss": 0.0086,
      "step": 4100
    },
    {
      "epoch": 62.196969696969695,
      "grad_norm": 0.035981521010398865,
      "learning_rate": 3.7803030303030305e-05,
      "loss": 0.0088,
      "step": 4105
    },
    {
      "epoch": 62.27272727272727,
      "grad_norm": 0.02888447232544422,
      "learning_rate": 3.7727272727272725e-05,
      "loss": 0.0103,
      "step": 4110
    },
    {
      "epoch": 62.34848484848485,
      "grad_norm": 0.06069885566830635,
      "learning_rate": 3.765151515151516e-05,
      "loss": 0.0101,
      "step": 4115
    },
    {
      "epoch": 62.42424242424242,
      "grad_norm": 0.05272584408521652,
      "learning_rate": 3.757575757575758e-05,
      "loss": 0.0097,
      "step": 4120
    },
    {
      "epoch": 62.5,
      "grad_norm": 0.030612776055932045,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0092,
      "step": 4125
    },
    {
      "epoch": 62.57575757575758,
      "grad_norm": 0.07009588181972504,
      "learning_rate": 3.742424242424243e-05,
      "loss": 0.011,
      "step": 4130
    },
    {
      "epoch": 62.65151515151515,
      "grad_norm": 0.0507914163172245,
      "learning_rate": 3.734848484848485e-05,
      "loss": 0.0067,
      "step": 4135
    },
    {
      "epoch": 62.72727272727273,
      "grad_norm": 0.06143215671181679,
      "learning_rate": 3.7272727272727276e-05,
      "loss": 0.0079,
      "step": 4140
    },
    {
      "epoch": 62.803030303030305,
      "grad_norm": 0.038194965571165085,
      "learning_rate": 3.7196969696969695e-05,
      "loss": 0.0084,
      "step": 4145
    },
    {
      "epoch": 62.878787878787875,
      "grad_norm": 0.054870542138814926,
      "learning_rate": 3.712121212121212e-05,
      "loss": 0.0093,
      "step": 4150
    },
    {
      "epoch": 62.95454545454545,
      "grad_norm": 0.10727784037590027,
      "learning_rate": 3.704545454545455e-05,
      "loss": 0.011,
      "step": 4155
    },
    {
      "epoch": 63.0,
      "eval_loss": 0.010348420590162277,
      "eval_mean_accuracy": 0.9636346792508489,
      "eval_mean_iou": 0.9344185248472444,
      "eval_runtime": 109.43,
      "eval_samples_per_second": 2.147,
      "eval_steps_per_second": 0.274,
      "step": 4158
    },
    {
      "epoch": 63.03030303030303,
      "grad_norm": 0.047154661267995834,
      "learning_rate": 3.6969696969696974e-05,
      "loss": 0.0111,
      "step": 4160
    },
    {
      "epoch": 63.10606060606061,
      "grad_norm": 0.06417280435562134,
      "learning_rate": 3.68939393939394e-05,
      "loss": 0.0083,
      "step": 4165
    },
    {
      "epoch": 63.18181818181818,
      "grad_norm": 0.06893038004636765,
      "learning_rate": 3.681818181818182e-05,
      "loss": 0.0066,
      "step": 4170
    },
    {
      "epoch": 63.25757575757576,
      "grad_norm": 0.011962244287133217,
      "learning_rate": 3.6742424242424246e-05,
      "loss": 0.0085,
      "step": 4175
    },
    {
      "epoch": 63.333333333333336,
      "grad_norm": 0.0482768639922142,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.0078,
      "step": 4180
    },
    {
      "epoch": 63.40909090909091,
      "grad_norm": 0.0487326942384243,
      "learning_rate": 3.659090909090909e-05,
      "loss": 0.0089,
      "step": 4185
    },
    {
      "epoch": 63.484848484848484,
      "grad_norm": 0.07458975911140442,
      "learning_rate": 3.651515151515152e-05,
      "loss": 0.0123,
      "step": 4190
    },
    {
      "epoch": 63.56060606060606,
      "grad_norm": 0.07712392508983612,
      "learning_rate": 3.643939393939394e-05,
      "loss": 0.0092,
      "step": 4195
    },
    {
      "epoch": 63.63636363636363,
      "grad_norm": 0.07650120556354523,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.0087,
      "step": 4200
    },
    {
      "epoch": 63.71212121212121,
      "grad_norm": 0.09635181725025177,
      "learning_rate": 3.628787878787879e-05,
      "loss": 0.0114,
      "step": 4205
    },
    {
      "epoch": 63.78787878787879,
      "grad_norm": 0.05205988511443138,
      "learning_rate": 3.621212121212122e-05,
      "loss": 0.0117,
      "step": 4210
    },
    {
      "epoch": 63.86363636363637,
      "grad_norm": 0.041871070861816406,
      "learning_rate": 3.613636363636364e-05,
      "loss": 0.0079,
      "step": 4215
    },
    {
      "epoch": 63.93939393939394,
      "grad_norm": 0.1929158717393875,
      "learning_rate": 3.606060606060606e-05,
      "loss": 0.0154,
      "step": 4220
    },
    {
      "epoch": 64.0,
      "eval_loss": 0.01114614587277174,
      "eval_mean_accuracy": 0.9494857835742524,
      "eval_mean_iou": 0.925588950919848,
      "eval_runtime": 114.6437,
      "eval_samples_per_second": 2.05,
      "eval_steps_per_second": 0.262,
      "step": 4224
    },
    {
      "epoch": 64.01515151515152,
      "grad_norm": 0.02728956937789917,
      "learning_rate": 3.598484848484849e-05,
      "loss": 0.0107,
      "step": 4225
    },
    {
      "epoch": 64.0909090909091,
      "grad_norm": 0.03523927554488182,
      "learning_rate": 3.590909090909091e-05,
      "loss": 0.0099,
      "step": 4230
    },
    {
      "epoch": 64.16666666666667,
      "grad_norm": 0.020485861226916313,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 0.0079,
      "step": 4235
    },
    {
      "epoch": 64.24242424242425,
      "grad_norm": 0.06435656547546387,
      "learning_rate": 3.575757575757576e-05,
      "loss": 0.0094,
      "step": 4240
    },
    {
      "epoch": 64.31818181818181,
      "grad_norm": 0.17279578745365143,
      "learning_rate": 3.568181818181818e-05,
      "loss": 0.0118,
      "step": 4245
    },
    {
      "epoch": 64.39393939393939,
      "grad_norm": 0.05728883296251297,
      "learning_rate": 3.560606060606061e-05,
      "loss": 0.0122,
      "step": 4250
    },
    {
      "epoch": 64.46969696969697,
      "grad_norm": 0.0455138236284256,
      "learning_rate": 3.553030303030303e-05,
      "loss": 0.0098,
      "step": 4255
    },
    {
      "epoch": 64.54545454545455,
      "grad_norm": 0.03397032991051674,
      "learning_rate": 3.545454545454546e-05,
      "loss": 0.0079,
      "step": 4260
    },
    {
      "epoch": 64.62121212121212,
      "grad_norm": 0.04753456264734268,
      "learning_rate": 3.537878787878788e-05,
      "loss": 0.012,
      "step": 4265
    },
    {
      "epoch": 64.6969696969697,
      "grad_norm": 0.036434393376111984,
      "learning_rate": 3.5303030303030305e-05,
      "loss": 0.0084,
      "step": 4270
    },
    {
      "epoch": 64.77272727272727,
      "grad_norm": 0.02974160760641098,
      "learning_rate": 3.522727272727273e-05,
      "loss": 0.0088,
      "step": 4275
    },
    {
      "epoch": 64.84848484848484,
      "grad_norm": 0.020718812942504883,
      "learning_rate": 3.515151515151515e-05,
      "loss": 0.0084,
      "step": 4280
    },
    {
      "epoch": 64.92424242424242,
      "grad_norm": 0.04110213369131088,
      "learning_rate": 3.507575757575758e-05,
      "loss": 0.0115,
      "step": 4285
    },
    {
      "epoch": 65.0,
      "grad_norm": 0.043455030769109726,
      "learning_rate": 3.5e-05,
      "loss": 0.0085,
      "step": 4290
    },
    {
      "epoch": 65.0,
      "eval_loss": 0.010995491407811642,
      "eval_mean_accuracy": 0.9520561376460893,
      "eval_mean_iou": 0.9275973044083908,
      "eval_runtime": 108.8047,
      "eval_samples_per_second": 2.16,
      "eval_steps_per_second": 0.276,
      "step": 4290
    },
    {
      "epoch": 65.07575757575758,
      "grad_norm": 0.09342098981142044,
      "learning_rate": 3.492424242424242e-05,
      "loss": 0.0123,
      "step": 4295
    },
    {
      "epoch": 65.15151515151516,
      "grad_norm": 0.020615678280591965,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.0073,
      "step": 4300
    },
    {
      "epoch": 65.22727272727273,
      "grad_norm": 0.04690653830766678,
      "learning_rate": 3.4772727272727276e-05,
      "loss": 0.0112,
      "step": 4305
    },
    {
      "epoch": 65.3030303030303,
      "grad_norm": 0.04761623963713646,
      "learning_rate": 3.46969696969697e-05,
      "loss": 0.0114,
      "step": 4310
    },
    {
      "epoch": 65.37878787878788,
      "grad_norm": 0.04484567418694496,
      "learning_rate": 3.462121212121212e-05,
      "loss": 0.009,
      "step": 4315
    },
    {
      "epoch": 65.45454545454545,
      "grad_norm": 0.044064294546842575,
      "learning_rate": 3.454545454545455e-05,
      "loss": 0.0104,
      "step": 4320
    },
    {
      "epoch": 65.53030303030303,
      "grad_norm": 0.05786487087607384,
      "learning_rate": 3.4469696969696974e-05,
      "loss": 0.0086,
      "step": 4325
    },
    {
      "epoch": 65.60606060606061,
      "grad_norm": 0.024088772013783455,
      "learning_rate": 3.4393939393939394e-05,
      "loss": 0.0078,
      "step": 4330
    },
    {
      "epoch": 65.68181818181819,
      "grad_norm": 0.07135248184204102,
      "learning_rate": 3.431818181818182e-05,
      "loss": 0.0093,
      "step": 4335
    },
    {
      "epoch": 65.75757575757575,
      "grad_norm": 0.04373057186603546,
      "learning_rate": 3.424242424242424e-05,
      "loss": 0.0093,
      "step": 4340
    },
    {
      "epoch": 65.83333333333333,
      "grad_norm": 0.07045969367027283,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 0.0103,
      "step": 4345
    },
    {
      "epoch": 65.9090909090909,
      "grad_norm": 0.027760986238718033,
      "learning_rate": 3.409090909090909e-05,
      "loss": 0.0058,
      "step": 4350
    },
    {
      "epoch": 65.98484848484848,
      "grad_norm": 0.07308410853147507,
      "learning_rate": 3.401515151515152e-05,
      "loss": 0.0088,
      "step": 4355
    },
    {
      "epoch": 66.0,
      "eval_loss": 0.01095026545226574,
      "eval_mean_accuracy": 0.9619027323789294,
      "eval_mean_iou": 0.9312023131233754,
      "eval_runtime": 109.896,
      "eval_samples_per_second": 2.138,
      "eval_steps_per_second": 0.273,
      "step": 4356
    },
    {
      "epoch": 66.06060606060606,
      "grad_norm": 0.0892305001616478,
      "learning_rate": 3.3939393939393945e-05,
      "loss": 0.0093,
      "step": 4360
    },
    {
      "epoch": 66.13636363636364,
      "grad_norm": 0.02317708544433117,
      "learning_rate": 3.3863636363636364e-05,
      "loss": 0.0136,
      "step": 4365
    },
    {
      "epoch": 66.21212121212122,
      "grad_norm": 0.02812225930392742,
      "learning_rate": 3.378787878787879e-05,
      "loss": 0.0067,
      "step": 4370
    },
    {
      "epoch": 66.28787878787878,
      "grad_norm": 0.037195827811956406,
      "learning_rate": 3.371212121212121e-05,
      "loss": 0.0065,
      "step": 4375
    },
    {
      "epoch": 66.36363636363636,
      "grad_norm": 0.04202352091670036,
      "learning_rate": 3.3636363636363636e-05,
      "loss": 0.007,
      "step": 4380
    },
    {
      "epoch": 66.43939393939394,
      "grad_norm": 0.08171532303094864,
      "learning_rate": 3.356060606060606e-05,
      "loss": 0.0132,
      "step": 4385
    },
    {
      "epoch": 66.51515151515152,
      "grad_norm": 0.04003671184182167,
      "learning_rate": 3.348484848484848e-05,
      "loss": 0.0068,
      "step": 4390
    },
    {
      "epoch": 66.5909090909091,
      "grad_norm": 0.01925681158900261,
      "learning_rate": 3.3409090909090915e-05,
      "loss": 0.0088,
      "step": 4395
    },
    {
      "epoch": 66.66666666666667,
      "grad_norm": 0.08983631432056427,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0091,
      "step": 4400
    },
    {
      "epoch": 66.74242424242425,
      "grad_norm": 0.05802718177437782,
      "learning_rate": 3.325757575757576e-05,
      "loss": 0.0087,
      "step": 4405
    },
    {
      "epoch": 66.81818181818181,
      "grad_norm": 0.06412912160158157,
      "learning_rate": 3.318181818181819e-05,
      "loss": 0.0101,
      "step": 4410
    },
    {
      "epoch": 66.89393939393939,
      "grad_norm": 0.08606050908565521,
      "learning_rate": 3.310606060606061e-05,
      "loss": 0.0099,
      "step": 4415
    },
    {
      "epoch": 66.96969696969697,
      "grad_norm": 0.031664662063121796,
      "learning_rate": 3.303030303030303e-05,
      "loss": 0.0082,
      "step": 4420
    },
    {
      "epoch": 67.0,
      "eval_loss": 0.010841302573680878,
      "eval_mean_accuracy": 0.9618311010164877,
      "eval_mean_iou": 0.9304414638316839,
      "eval_runtime": 108.0753,
      "eval_samples_per_second": 2.174,
      "eval_steps_per_second": 0.278,
      "step": 4422
    },
    {
      "epoch": 67.04545454545455,
      "grad_norm": 0.016359848901629448,
      "learning_rate": 3.295454545454545e-05,
      "loss": 0.0081,
      "step": 4425
    },
    {
      "epoch": 67.12121212121212,
      "grad_norm": 0.03398958593606949,
      "learning_rate": 3.287878787878788e-05,
      "loss": 0.0099,
      "step": 4430
    },
    {
      "epoch": 67.1969696969697,
      "grad_norm": 0.03569827228784561,
      "learning_rate": 3.2803030303030305e-05,
      "loss": 0.0111,
      "step": 4435
    },
    {
      "epoch": 67.27272727272727,
      "grad_norm": 0.0373200848698616,
      "learning_rate": 3.272727272727273e-05,
      "loss": 0.0081,
      "step": 4440
    },
    {
      "epoch": 67.34848484848484,
      "grad_norm": 0.03528263047337532,
      "learning_rate": 3.265151515151516e-05,
      "loss": 0.0116,
      "step": 4445
    },
    {
      "epoch": 67.42424242424242,
      "grad_norm": 0.03887326642870903,
      "learning_rate": 3.257575757575758e-05,
      "loss": 0.0092,
      "step": 4450
    },
    {
      "epoch": 67.5,
      "grad_norm": 0.1109720841050148,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.0077,
      "step": 4455
    },
    {
      "epoch": 67.57575757575758,
      "grad_norm": 0.020121807232499123,
      "learning_rate": 3.2424242424242423e-05,
      "loss": 0.0104,
      "step": 4460
    },
    {
      "epoch": 67.65151515151516,
      "grad_norm": 0.0855100005865097,
      "learning_rate": 3.234848484848485e-05,
      "loss": 0.0104,
      "step": 4465
    },
    {
      "epoch": 67.72727272727273,
      "grad_norm": 0.05511549115180969,
      "learning_rate": 3.2272727272727276e-05,
      "loss": 0.0078,
      "step": 4470
    },
    {
      "epoch": 67.8030303030303,
      "grad_norm": 0.039750467985868454,
      "learning_rate": 3.2196969696969696e-05,
      "loss": 0.0072,
      "step": 4475
    },
    {
      "epoch": 67.87878787878788,
      "grad_norm": 0.04774370416998863,
      "learning_rate": 3.212121212121212e-05,
      "loss": 0.0092,
      "step": 4480
    },
    {
      "epoch": 67.95454545454545,
      "grad_norm": 0.041000884026288986,
      "learning_rate": 3.204545454545455e-05,
      "loss": 0.007,
      "step": 4485
    },
    {
      "epoch": 68.0,
      "eval_loss": 0.010789004154503345,
      "eval_mean_accuracy": 0.954017106143361,
      "eval_mean_iou": 0.9297507732141983,
      "eval_runtime": 107.8059,
      "eval_samples_per_second": 2.18,
      "eval_steps_per_second": 0.278,
      "step": 4488
    },
    {
      "epoch": 68.03030303030303,
      "grad_norm": 0.03352880850434303,
      "learning_rate": 3.1969696969696974e-05,
      "loss": 0.0092,
      "step": 4490
    },
    {
      "epoch": 68.10606060606061,
      "grad_norm": 0.03404291719198227,
      "learning_rate": 3.18939393939394e-05,
      "loss": 0.011,
      "step": 4495
    },
    {
      "epoch": 68.18181818181819,
      "grad_norm": 0.11808786541223526,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.011,
      "step": 4500
    },
    {
      "epoch": 68.25757575757575,
      "grad_norm": 0.05828849598765373,
      "learning_rate": 3.174242424242425e-05,
      "loss": 0.0072,
      "step": 4505
    },
    {
      "epoch": 68.33333333333333,
      "grad_norm": 0.07018955051898956,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.0088,
      "step": 4510
    },
    {
      "epoch": 68.4090909090909,
      "grad_norm": 0.05007249861955643,
      "learning_rate": 3.159090909090909e-05,
      "loss": 0.0097,
      "step": 4515
    },
    {
      "epoch": 68.48484848484848,
      "grad_norm": 0.016938215121626854,
      "learning_rate": 3.151515151515151e-05,
      "loss": 0.0079,
      "step": 4520
    },
    {
      "epoch": 68.56060606060606,
      "grad_norm": 0.07808490842580795,
      "learning_rate": 3.143939393939394e-05,
      "loss": 0.0108,
      "step": 4525
    },
    {
      "epoch": 68.63636363636364,
      "grad_norm": 0.03241580352187157,
      "learning_rate": 3.1363636363636365e-05,
      "loss": 0.0081,
      "step": 4530
    },
    {
      "epoch": 68.71212121212122,
      "grad_norm": 0.05491955578327179,
      "learning_rate": 3.128787878787879e-05,
      "loss": 0.0103,
      "step": 4535
    },
    {
      "epoch": 68.78787878787878,
      "grad_norm": 0.07096435129642487,
      "learning_rate": 3.121212121212122e-05,
      "loss": 0.0098,
      "step": 4540
    },
    {
      "epoch": 68.86363636363636,
      "grad_norm": 0.05701403319835663,
      "learning_rate": 3.113636363636364e-05,
      "loss": 0.0091,
      "step": 4545
    },
    {
      "epoch": 68.93939393939394,
      "grad_norm": 0.01829889975488186,
      "learning_rate": 3.106060606060606e-05,
      "loss": 0.0064,
      "step": 4550
    },
    {
      "epoch": 69.0,
      "eval_loss": 0.01066575013101101,
      "eval_mean_accuracy": 0.9564576227967804,
      "eval_mean_iou": 0.9316168659043537,
      "eval_runtime": 109.2388,
      "eval_samples_per_second": 2.151,
      "eval_steps_per_second": 0.275,
      "step": 4554
    },
    {
      "epoch": 69.01515151515152,
      "grad_norm": 0.03056393750011921,
      "learning_rate": 3.098484848484849e-05,
      "loss": 0.0104,
      "step": 4555
    },
    {
      "epoch": 69.0909090909091,
      "grad_norm": 0.04383070394396782,
      "learning_rate": 3.090909090909091e-05,
      "loss": 0.0084,
      "step": 4560
    },
    {
      "epoch": 69.16666666666667,
      "grad_norm": 0.030420484021306038,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 0.0084,
      "step": 4565
    },
    {
      "epoch": 69.24242424242425,
      "grad_norm": 0.044958751648664474,
      "learning_rate": 3.0757575757575755e-05,
      "loss": 0.0091,
      "step": 4570
    },
    {
      "epoch": 69.31818181818181,
      "grad_norm": 0.0623924694955349,
      "learning_rate": 3.068181818181818e-05,
      "loss": 0.0067,
      "step": 4575
    },
    {
      "epoch": 69.39393939393939,
      "grad_norm": 0.060913193970918655,
      "learning_rate": 3.060606060606061e-05,
      "loss": 0.009,
      "step": 4580
    },
    {
      "epoch": 69.46969696969697,
      "grad_norm": 0.05079806596040726,
      "learning_rate": 3.0530303030303034e-05,
      "loss": 0.0084,
      "step": 4585
    },
    {
      "epoch": 69.54545454545455,
      "grad_norm": 0.04185197129845619,
      "learning_rate": 3.0454545454545456e-05,
      "loss": 0.0072,
      "step": 4590
    },
    {
      "epoch": 69.62121212121212,
      "grad_norm": 0.03326252102851868,
      "learning_rate": 3.037878787878788e-05,
      "loss": 0.01,
      "step": 4595
    },
    {
      "epoch": 69.6969696969697,
      "grad_norm": 0.060958389192819595,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 0.0098,
      "step": 4600
    },
    {
      "epoch": 69.77272727272727,
      "grad_norm": 0.06861080974340439,
      "learning_rate": 3.0227272727272725e-05,
      "loss": 0.0109,
      "step": 4605
    },
    {
      "epoch": 69.84848484848484,
      "grad_norm": 0.03341318294405937,
      "learning_rate": 3.015151515151515e-05,
      "loss": 0.0099,
      "step": 4610
    },
    {
      "epoch": 69.92424242424242,
      "grad_norm": 0.014403872191905975,
      "learning_rate": 3.0075757575757578e-05,
      "loss": 0.0085,
      "step": 4615
    },
    {
      "epoch": 70.0,
      "grad_norm": 0.04575015604496002,
      "learning_rate": 3e-05,
      "loss": 0.0109,
      "step": 4620
    },
    {
      "epoch": 70.0,
      "eval_loss": 0.010572757571935654,
      "eval_mean_accuracy": 0.9613365853132964,
      "eval_mean_iou": 0.9321565584821054,
      "eval_runtime": 105.735,
      "eval_samples_per_second": 2.223,
      "eval_steps_per_second": 0.284,
      "step": 4620
    },
    {
      "epoch": 70.07575757575758,
      "grad_norm": 0.021903809159994125,
      "learning_rate": 2.9924242424242427e-05,
      "loss": 0.0087,
      "step": 4625
    },
    {
      "epoch": 70.15151515151516,
      "grad_norm": 0.04126216098666191,
      "learning_rate": 2.9848484848484847e-05,
      "loss": 0.0074,
      "step": 4630
    },
    {
      "epoch": 70.22727272727273,
      "grad_norm": 0.05117930471897125,
      "learning_rate": 2.9772727272727273e-05,
      "loss": 0.0098,
      "step": 4635
    },
    {
      "epoch": 70.3030303030303,
      "grad_norm": 0.0580342672765255,
      "learning_rate": 2.96969696969697e-05,
      "loss": 0.0098,
      "step": 4640
    },
    {
      "epoch": 70.37878787878788,
      "grad_norm": 0.04827859625220299,
      "learning_rate": 2.9621212121212122e-05,
      "loss": 0.0095,
      "step": 4645
    },
    {
      "epoch": 70.45454545454545,
      "grad_norm": 0.03709656372666359,
      "learning_rate": 2.954545454545455e-05,
      "loss": 0.0112,
      "step": 4650
    },
    {
      "epoch": 70.53030303030303,
      "grad_norm": 0.022457921877503395,
      "learning_rate": 2.9469696969696968e-05,
      "loss": 0.0091,
      "step": 4655
    },
    {
      "epoch": 70.60606060606061,
      "grad_norm": 0.019703667610883713,
      "learning_rate": 2.9393939393939394e-05,
      "loss": 0.0095,
      "step": 4660
    },
    {
      "epoch": 70.68181818181819,
      "grad_norm": 0.02515508607029915,
      "learning_rate": 2.9318181818181817e-05,
      "loss": 0.008,
      "step": 4665
    },
    {
      "epoch": 70.75757575757575,
      "grad_norm": 0.04397687688469887,
      "learning_rate": 2.9242424242424243e-05,
      "loss": 0.0085,
      "step": 4670
    },
    {
      "epoch": 70.83333333333333,
      "grad_norm": 0.02078959532082081,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.0069,
      "step": 4675
    },
    {
      "epoch": 70.9090909090909,
      "grad_norm": 0.02069249004125595,
      "learning_rate": 2.909090909090909e-05,
      "loss": 0.0064,
      "step": 4680
    },
    {
      "epoch": 70.98484848484848,
      "grad_norm": 0.13526493310928345,
      "learning_rate": 2.901515151515152e-05,
      "loss": 0.0109,
      "step": 4685
    },
    {
      "epoch": 71.0,
      "eval_loss": 0.010431384667754173,
      "eval_mean_accuracy": 0.95973108244374,
      "eval_mean_iou": 0.9332468541568577,
      "eval_runtime": 105.9866,
      "eval_samples_per_second": 2.217,
      "eval_steps_per_second": 0.283,
      "step": 4686
    },
    {
      "epoch": 71.06060606060606,
      "grad_norm": 0.13918663561344147,
      "learning_rate": 2.893939393939394e-05,
      "loss": 0.0092,
      "step": 4690
    },
    {
      "epoch": 71.13636363636364,
      "grad_norm": 0.10762596875429153,
      "learning_rate": 2.8863636363636365e-05,
      "loss": 0.0106,
      "step": 4695
    },
    {
      "epoch": 71.21212121212122,
      "grad_norm": 0.028599543496966362,
      "learning_rate": 2.878787878787879e-05,
      "loss": 0.0058,
      "step": 4700
    },
    {
      "epoch": 71.28787878787878,
      "grad_norm": 0.09230702370405197,
      "learning_rate": 2.8712121212121214e-05,
      "loss": 0.0084,
      "step": 4705
    },
    {
      "epoch": 71.36363636363636,
      "grad_norm": 0.041058190166950226,
      "learning_rate": 2.863636363636364e-05,
      "loss": 0.0103,
      "step": 4710
    },
    {
      "epoch": 71.43939393939394,
      "grad_norm": 0.11795422434806824,
      "learning_rate": 2.856060606060606e-05,
      "loss": 0.0087,
      "step": 4715
    },
    {
      "epoch": 71.51515151515152,
      "grad_norm": 0.05824560299515724,
      "learning_rate": 2.8484848484848486e-05,
      "loss": 0.0085,
      "step": 4720
    },
    {
      "epoch": 71.5909090909091,
      "grad_norm": 0.045908164232969284,
      "learning_rate": 2.8409090909090912e-05,
      "loss": 0.009,
      "step": 4725
    },
    {
      "epoch": 71.66666666666667,
      "grad_norm": 0.03186006844043732,
      "learning_rate": 2.8333333333333335e-05,
      "loss": 0.0085,
      "step": 4730
    },
    {
      "epoch": 71.74242424242425,
      "grad_norm": 0.029093284159898758,
      "learning_rate": 2.825757575757576e-05,
      "loss": 0.0095,
      "step": 4735
    },
    {
      "epoch": 71.81818181818181,
      "grad_norm": 0.03256039693951607,
      "learning_rate": 2.818181818181818e-05,
      "loss": 0.0087,
      "step": 4740
    },
    {
      "epoch": 71.89393939393939,
      "grad_norm": 0.04024430736899376,
      "learning_rate": 2.8106060606060607e-05,
      "loss": 0.0091,
      "step": 4745
    },
    {
      "epoch": 71.96969696969697,
      "grad_norm": 0.0841880515217781,
      "learning_rate": 2.803030303030303e-05,
      "loss": 0.0098,
      "step": 4750
    },
    {
      "epoch": 72.0,
      "eval_loss": 0.010481276549398899,
      "eval_mean_accuracy": 0.9595870865871287,
      "eval_mean_iou": 0.9331010417171366,
      "eval_runtime": 106.4568,
      "eval_samples_per_second": 2.207,
      "eval_steps_per_second": 0.282,
      "step": 4752
    },
    {
      "epoch": 72.04545454545455,
      "grad_norm": 0.12360089272260666,
      "learning_rate": 2.7954545454545457e-05,
      "loss": 0.0093,
      "step": 4755
    },
    {
      "epoch": 72.12121212121212,
      "grad_norm": 0.04862156882882118,
      "learning_rate": 2.7878787878787883e-05,
      "loss": 0.0103,
      "step": 4760
    },
    {
      "epoch": 72.1969696969697,
      "grad_norm": 0.03480615094304085,
      "learning_rate": 2.7803030303030303e-05,
      "loss": 0.0101,
      "step": 4765
    },
    {
      "epoch": 72.27272727272727,
      "grad_norm": 0.023709965869784355,
      "learning_rate": 2.772727272727273e-05,
      "loss": 0.0074,
      "step": 4770
    },
    {
      "epoch": 72.34848484848484,
      "grad_norm": 0.027628546580672264,
      "learning_rate": 2.7651515151515152e-05,
      "loss": 0.0079,
      "step": 4775
    },
    {
      "epoch": 72.42424242424242,
      "grad_norm": 0.029094085097312927,
      "learning_rate": 2.7575757575757578e-05,
      "loss": 0.0093,
      "step": 4780
    },
    {
      "epoch": 72.5,
      "grad_norm": 0.028976084664463997,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.0074,
      "step": 4785
    },
    {
      "epoch": 72.57575757575758,
      "grad_norm": 0.027473242953419685,
      "learning_rate": 2.7424242424242424e-05,
      "loss": 0.0087,
      "step": 4790
    },
    {
      "epoch": 72.65151515151516,
      "grad_norm": 0.015115257352590561,
      "learning_rate": 2.734848484848485e-05,
      "loss": 0.006,
      "step": 4795
    },
    {
      "epoch": 72.72727272727273,
      "grad_norm": 0.02262251451611519,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.0062,
      "step": 4800
    },
    {
      "epoch": 72.8030303030303,
      "grad_norm": 0.025969522073864937,
      "learning_rate": 2.71969696969697e-05,
      "loss": 0.0104,
      "step": 4805
    },
    {
      "epoch": 72.87878787878788,
      "grad_norm": 0.047238316386938095,
      "learning_rate": 2.7121212121212126e-05,
      "loss": 0.0127,
      "step": 4810
    },
    {
      "epoch": 72.95454545454545,
      "grad_norm": 0.041452668607234955,
      "learning_rate": 2.7045454545454545e-05,
      "loss": 0.0096,
      "step": 4815
    },
    {
      "epoch": 73.0,
      "eval_loss": 0.01050542015582323,
      "eval_mean_accuracy": 0.9616987270861475,
      "eval_mean_iou": 0.9322744602052287,
      "eval_runtime": 105.5734,
      "eval_samples_per_second": 2.226,
      "eval_steps_per_second": 0.284,
      "step": 4818
    },
    {
      "epoch": 73.03030303030303,
      "grad_norm": 0.03450143337249756,
      "learning_rate": 2.696969696969697e-05,
      "loss": 0.0093,
      "step": 4820
    },
    {
      "epoch": 73.10606060606061,
      "grad_norm": 0.04642605036497116,
      "learning_rate": 2.6893939393939394e-05,
      "loss": 0.0113,
      "step": 4825
    },
    {
      "epoch": 73.18181818181819,
      "grad_norm": 0.046312347054481506,
      "learning_rate": 2.681818181818182e-05,
      "loss": 0.0092,
      "step": 4830
    },
    {
      "epoch": 73.25757575757575,
      "grad_norm": 0.06011101230978966,
      "learning_rate": 2.674242424242424e-05,
      "loss": 0.0079,
      "step": 4835
    },
    {
      "epoch": 73.33333333333333,
      "grad_norm": 0.03597854822874069,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.0104,
      "step": 4840
    },
    {
      "epoch": 73.4090909090909,
      "grad_norm": 0.0697672963142395,
      "learning_rate": 2.6590909090909093e-05,
      "loss": 0.0076,
      "step": 4845
    },
    {
      "epoch": 73.48484848484848,
      "grad_norm": 0.05983448401093483,
      "learning_rate": 2.6515151515151516e-05,
      "loss": 0.0088,
      "step": 4850
    },
    {
      "epoch": 73.56060606060606,
      "grad_norm": 0.053877897560596466,
      "learning_rate": 2.6439393939393942e-05,
      "loss": 0.0093,
      "step": 4855
    },
    {
      "epoch": 73.63636363636364,
      "grad_norm": 0.04413070157170296,
      "learning_rate": 2.636363636363636e-05,
      "loss": 0.0113,
      "step": 4860
    },
    {
      "epoch": 73.71212121212122,
      "grad_norm": 0.04516996443271637,
      "learning_rate": 2.6287878787878788e-05,
      "loss": 0.0083,
      "step": 4865
    },
    {
      "epoch": 73.78787878787878,
      "grad_norm": 0.06201249733567238,
      "learning_rate": 2.6212121212121214e-05,
      "loss": 0.0065,
      "step": 4870
    },
    {
      "epoch": 73.86363636363636,
      "grad_norm": 0.03763002157211304,
      "learning_rate": 2.6136363636363637e-05,
      "loss": 0.0094,
      "step": 4875
    },
    {
      "epoch": 73.93939393939394,
      "grad_norm": 0.041292596608400345,
      "learning_rate": 2.6060606060606063e-05,
      "loss": 0.0078,
      "step": 4880
    },
    {
      "epoch": 74.0,
      "eval_loss": 0.010468543507158756,
      "eval_mean_accuracy": 0.9679491248736601,
      "eval_mean_iou": 0.9339771975612798,
      "eval_runtime": 106.63,
      "eval_samples_per_second": 2.204,
      "eval_steps_per_second": 0.281,
      "step": 4884
    },
    {
      "epoch": 74.01515151515152,
      "grad_norm": 0.05016251280903816,
      "learning_rate": 2.5984848484848483e-05,
      "loss": 0.0089,
      "step": 4885
    },
    {
      "epoch": 74.0909090909091,
      "grad_norm": 0.11201584339141846,
      "learning_rate": 2.590909090909091e-05,
      "loss": 0.0092,
      "step": 4890
    },
    {
      "epoch": 74.16666666666667,
      "grad_norm": 0.04583180323243141,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 0.0068,
      "step": 4895
    },
    {
      "epoch": 74.24242424242425,
      "grad_norm": 0.026928547769784927,
      "learning_rate": 2.575757575757576e-05,
      "loss": 0.0084,
      "step": 4900
    },
    {
      "epoch": 74.31818181818181,
      "grad_norm": 0.02214164473116398,
      "learning_rate": 2.5681818181818185e-05,
      "loss": 0.0064,
      "step": 4905
    },
    {
      "epoch": 74.39393939393939,
      "grad_norm": 0.01238306239247322,
      "learning_rate": 2.5606060606060604e-05,
      "loss": 0.0105,
      "step": 4910
    },
    {
      "epoch": 74.46969696969697,
      "grad_norm": 0.08929905295372009,
      "learning_rate": 2.553030303030303e-05,
      "loss": 0.0086,
      "step": 4915
    },
    {
      "epoch": 74.54545454545455,
      "grad_norm": 0.04563877359032631,
      "learning_rate": 2.5454545454545454e-05,
      "loss": 0.0071,
      "step": 4920
    },
    {
      "epoch": 74.62121212121212,
      "grad_norm": 0.0775129497051239,
      "learning_rate": 2.537878787878788e-05,
      "loss": 0.0124,
      "step": 4925
    },
    {
      "epoch": 74.6969696969697,
      "grad_norm": 0.03705916181206703,
      "learning_rate": 2.5303030303030306e-05,
      "loss": 0.0077,
      "step": 4930
    },
    {
      "epoch": 74.77272727272727,
      "grad_norm": 0.09436459839344025,
      "learning_rate": 2.5227272727272726e-05,
      "loss": 0.0075,
      "step": 4935
    },
    {
      "epoch": 74.84848484848484,
      "grad_norm": 0.02611716091632843,
      "learning_rate": 2.5151515151515155e-05,
      "loss": 0.0073,
      "step": 4940
    },
    {
      "epoch": 74.92424242424242,
      "grad_norm": 0.13315583765506744,
      "learning_rate": 2.5075757575757575e-05,
      "loss": 0.0133,
      "step": 4945
    },
    {
      "epoch": 75.0,
      "grad_norm": 0.09374118596315384,
      "learning_rate": 2.5e-05,
      "loss": 0.0072,
      "step": 4950
    },
    {
      "epoch": 75.0,
      "eval_loss": 0.010879242792725563,
      "eval_mean_accuracy": 0.9690582856922338,
      "eval_mean_iou": 0.931420747773667,
      "eval_runtime": 106.9048,
      "eval_samples_per_second": 2.198,
      "eval_steps_per_second": 0.281,
      "step": 4950
    },
    {
      "epoch": 75.07575757575758,
      "grad_norm": 0.0337580181658268,
      "learning_rate": 2.4924242424242424e-05,
      "loss": 0.0095,
      "step": 4955
    },
    {
      "epoch": 75.15151515151516,
      "grad_norm": 0.029575346037745476,
      "learning_rate": 2.4848484848484847e-05,
      "loss": 0.0094,
      "step": 4960
    },
    {
      "epoch": 75.22727272727273,
      "grad_norm": 0.07215619832277298,
      "learning_rate": 2.4772727272727277e-05,
      "loss": 0.0072,
      "step": 4965
    },
    {
      "epoch": 75.3030303030303,
      "grad_norm": 0.02778995782136917,
      "learning_rate": 2.46969696969697e-05,
      "loss": 0.007,
      "step": 4970
    },
    {
      "epoch": 75.37878787878788,
      "grad_norm": 0.06139788776636124,
      "learning_rate": 2.4621212121212123e-05,
      "loss": 0.0129,
      "step": 4975
    },
    {
      "epoch": 75.45454545454545,
      "grad_norm": 0.08555522561073303,
      "learning_rate": 2.4545454545454545e-05,
      "loss": 0.0078,
      "step": 4980
    },
    {
      "epoch": 75.53030303030303,
      "grad_norm": 0.03697648644447327,
      "learning_rate": 2.4469696969696972e-05,
      "loss": 0.009,
      "step": 4985
    },
    {
      "epoch": 75.60606060606061,
      "grad_norm": 0.08047065883874893,
      "learning_rate": 2.4393939393939395e-05,
      "loss": 0.0075,
      "step": 4990
    },
    {
      "epoch": 75.68181818181819,
      "grad_norm": 0.03715177997946739,
      "learning_rate": 2.431818181818182e-05,
      "loss": 0.009,
      "step": 4995
    },
    {
      "epoch": 75.75757575757575,
      "grad_norm": 0.04096238687634468,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 0.0088,
      "step": 5000
    },
    {
      "epoch": 75.83333333333333,
      "grad_norm": 0.0307797621935606,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.0078,
      "step": 5005
    },
    {
      "epoch": 75.9090909090909,
      "grad_norm": 0.07030709087848663,
      "learning_rate": 2.4090909090909093e-05,
      "loss": 0.0089,
      "step": 5010
    },
    {
      "epoch": 75.98484848484848,
      "grad_norm": 0.078398697078228,
      "learning_rate": 2.4015151515151516e-05,
      "loss": 0.0111,
      "step": 5015
    },
    {
      "epoch": 76.0,
      "eval_loss": 0.010334459133446217,
      "eval_mean_accuracy": 0.9650514028903405,
      "eval_mean_iou": 0.933976846618951,
      "eval_runtime": 107.8342,
      "eval_samples_per_second": 2.179,
      "eval_steps_per_second": 0.278,
      "step": 5016
    },
    {
      "epoch": 76.06060606060606,
      "grad_norm": 0.04548473283648491,
      "learning_rate": 2.393939393939394e-05,
      "loss": 0.0103,
      "step": 5020
    },
    {
      "epoch": 76.13636363636364,
      "grad_norm": 0.02619260363280773,
      "learning_rate": 2.3863636363636365e-05,
      "loss": 0.0089,
      "step": 5025
    },
    {
      "epoch": 76.21212121212122,
      "grad_norm": 0.027637366205453873,
      "learning_rate": 2.3787878787878788e-05,
      "loss": 0.0101,
      "step": 5030
    },
    {
      "epoch": 76.28787878787878,
      "grad_norm": 0.06628289073705673,
      "learning_rate": 2.3712121212121214e-05,
      "loss": 0.0088,
      "step": 5035
    },
    {
      "epoch": 76.36363636363636,
      "grad_norm": 0.07368367910385132,
      "learning_rate": 2.3636363636363637e-05,
      "loss": 0.0079,
      "step": 5040
    },
    {
      "epoch": 76.43939393939394,
      "grad_norm": 0.17397284507751465,
      "learning_rate": 2.356060606060606e-05,
      "loss": 0.0098,
      "step": 5045
    },
    {
      "epoch": 76.51515151515152,
      "grad_norm": 0.025283606722950935,
      "learning_rate": 2.3484848484848487e-05,
      "loss": 0.008,
      "step": 5050
    },
    {
      "epoch": 76.5909090909091,
      "grad_norm": 0.06463988870382309,
      "learning_rate": 2.340909090909091e-05,
      "loss": 0.008,
      "step": 5055
    },
    {
      "epoch": 76.66666666666667,
      "grad_norm": 0.037584226578474045,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.0076,
      "step": 5060
    },
    {
      "epoch": 76.74242424242425,
      "grad_norm": 0.05195797234773636,
      "learning_rate": 2.325757575757576e-05,
      "loss": 0.0118,
      "step": 5065
    },
    {
      "epoch": 76.81818181818181,
      "grad_norm": 0.05388859659433365,
      "learning_rate": 2.318181818181818e-05,
      "loss": 0.0075,
      "step": 5070
    },
    {
      "epoch": 76.89393939393939,
      "grad_norm": 0.024480002000927925,
      "learning_rate": 2.3106060606060605e-05,
      "loss": 0.0059,
      "step": 5075
    },
    {
      "epoch": 76.96969696969697,
      "grad_norm": 0.04549216851592064,
      "learning_rate": 2.3030303030303034e-05,
      "loss": 0.0094,
      "step": 5080
    },
    {
      "epoch": 77.0,
      "eval_loss": 0.010321916081011295,
      "eval_mean_accuracy": 0.9622106599500384,
      "eval_mean_iou": 0.9345596941208864,
      "eval_runtime": 106.8565,
      "eval_samples_per_second": 2.199,
      "eval_steps_per_second": 0.281,
      "step": 5082
    },
    {
      "epoch": 77.04545454545455,
      "grad_norm": 0.07242515683174133,
      "learning_rate": 2.2954545454545457e-05,
      "loss": 0.0078,
      "step": 5085
    },
    {
      "epoch": 77.12121212121212,
      "grad_norm": 0.04360293224453926,
      "learning_rate": 2.287878787878788e-05,
      "loss": 0.0052,
      "step": 5090
    },
    {
      "epoch": 77.1969696969697,
      "grad_norm": 0.058590881526470184,
      "learning_rate": 2.2803030303030303e-05,
      "loss": 0.0094,
      "step": 5095
    },
    {
      "epoch": 77.27272727272727,
      "grad_norm": 0.04384826868772507,
      "learning_rate": 2.272727272727273e-05,
      "loss": 0.0074,
      "step": 5100
    },
    {
      "epoch": 77.34848484848484,
      "grad_norm": 0.051105745136737823,
      "learning_rate": 2.2651515151515152e-05,
      "loss": 0.0118,
      "step": 5105
    },
    {
      "epoch": 77.42424242424242,
      "grad_norm": 0.03958512470126152,
      "learning_rate": 2.257575757575758e-05,
      "loss": 0.0096,
      "step": 5110
    },
    {
      "epoch": 77.5,
      "grad_norm": 0.022392839193344116,
      "learning_rate": 2.25e-05,
      "loss": 0.0098,
      "step": 5115
    },
    {
      "epoch": 77.57575757575758,
      "grad_norm": 0.05451478436589241,
      "learning_rate": 2.2424242424242424e-05,
      "loss": 0.0092,
      "step": 5120
    },
    {
      "epoch": 77.65151515151516,
      "grad_norm": 0.048957522958517075,
      "learning_rate": 2.234848484848485e-05,
      "loss": 0.0087,
      "step": 5125
    },
    {
      "epoch": 77.72727272727273,
      "grad_norm": 0.04050649702548981,
      "learning_rate": 2.2272727272727274e-05,
      "loss": 0.0087,
      "step": 5130
    },
    {
      "epoch": 77.8030303030303,
      "grad_norm": 0.02958143688738346,
      "learning_rate": 2.21969696969697e-05,
      "loss": 0.0075,
      "step": 5135
    },
    {
      "epoch": 77.87878787878788,
      "grad_norm": 0.11438674479722977,
      "learning_rate": 2.2121212121212123e-05,
      "loss": 0.0097,
      "step": 5140
    },
    {
      "epoch": 77.95454545454545,
      "grad_norm": 0.04339442774653435,
      "learning_rate": 2.2045454545454546e-05,
      "loss": 0.0099,
      "step": 5145
    },
    {
      "epoch": 78.0,
      "eval_loss": 0.010354283265769482,
      "eval_mean_accuracy": 0.9629959526939507,
      "eval_mean_iou": 0.9342123058117338,
      "eval_runtime": 106.2039,
      "eval_samples_per_second": 2.213,
      "eval_steps_per_second": 0.282,
      "step": 5148
    },
    {
      "epoch": 78.03030303030303,
      "grad_norm": 0.03370332345366478,
      "learning_rate": 2.1969696969696972e-05,
      "loss": 0.0082,
      "step": 5150
    },
    {
      "epoch": 78.10606060606061,
      "grad_norm": 0.01700020395219326,
      "learning_rate": 2.1893939393939395e-05,
      "loss": 0.0059,
      "step": 5155
    },
    {
      "epoch": 78.18181818181819,
      "grad_norm": 0.06151702627539635,
      "learning_rate": 2.1818181818181818e-05,
      "loss": 0.0076,
      "step": 5160
    },
    {
      "epoch": 78.25757575757575,
      "grad_norm": 0.037882000207901,
      "learning_rate": 2.1742424242424244e-05,
      "loss": 0.0088,
      "step": 5165
    },
    {
      "epoch": 78.33333333333333,
      "grad_norm": 0.029390713199973106,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 0.01,
      "step": 5170
    },
    {
      "epoch": 78.4090909090909,
      "grad_norm": 0.026748094707727432,
      "learning_rate": 2.1590909090909093e-05,
      "loss": 0.0091,
      "step": 5175
    },
    {
      "epoch": 78.48484848484848,
      "grad_norm": 0.024010296911001205,
      "learning_rate": 2.1515151515151516e-05,
      "loss": 0.0082,
      "step": 5180
    },
    {
      "epoch": 78.56060606060606,
      "grad_norm": 0.023141875863075256,
      "learning_rate": 2.143939393939394e-05,
      "loss": 0.0083,
      "step": 5185
    },
    {
      "epoch": 78.63636363636364,
      "grad_norm": 0.030583282932639122,
      "learning_rate": 2.1363636363636362e-05,
      "loss": 0.0085,
      "step": 5190
    },
    {
      "epoch": 78.71212121212122,
      "grad_norm": 0.05524233728647232,
      "learning_rate": 2.128787878787879e-05,
      "loss": 0.0092,
      "step": 5195
    },
    {
      "epoch": 78.78787878787878,
      "grad_norm": 0.10357318073511124,
      "learning_rate": 2.1212121212121215e-05,
      "loss": 0.0092,
      "step": 5200
    },
    {
      "epoch": 78.86363636363636,
      "grad_norm": 0.07604549080133438,
      "learning_rate": 2.1136363636363638e-05,
      "loss": 0.0086,
      "step": 5205
    },
    {
      "epoch": 78.93939393939394,
      "grad_norm": 0.05063147470355034,
      "learning_rate": 2.106060606060606e-05,
      "loss": 0.0097,
      "step": 5210
    },
    {
      "epoch": 79.0,
      "eval_loss": 0.0101555734872818,
      "eval_mean_accuracy": 0.9619379393989145,
      "eval_mean_iou": 0.9352963602802227,
      "eval_runtime": 105.5398,
      "eval_samples_per_second": 2.227,
      "eval_steps_per_second": 0.284,
      "step": 5214
    },
    {
      "epoch": 79.01515151515152,
      "grad_norm": 0.04026450216770172,
      "learning_rate": 2.0984848484848483e-05,
      "loss": 0.009,
      "step": 5215
    },
    {
      "epoch": 79.0909090909091,
      "grad_norm": 0.048702992498874664,
      "learning_rate": 2.090909090909091e-05,
      "loss": 0.0091,
      "step": 5220
    },
    {
      "epoch": 79.16666666666667,
      "grad_norm": 0.033094290643930435,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0064,
      "step": 5225
    },
    {
      "epoch": 79.24242424242425,
      "grad_norm": 0.059840645641088486,
      "learning_rate": 2.075757575757576e-05,
      "loss": 0.007,
      "step": 5230
    },
    {
      "epoch": 79.31818181818181,
      "grad_norm": 0.039399925619363785,
      "learning_rate": 2.0681818181818182e-05,
      "loss": 0.0092,
      "step": 5235
    },
    {
      "epoch": 79.39393939393939,
      "grad_norm": 0.04825712367892265,
      "learning_rate": 2.0606060606060608e-05,
      "loss": 0.0079,
      "step": 5240
    },
    {
      "epoch": 79.46969696969697,
      "grad_norm": 0.03731779009103775,
      "learning_rate": 2.053030303030303e-05,
      "loss": 0.0088,
      "step": 5245
    },
    {
      "epoch": 79.54545454545455,
      "grad_norm": 0.026014264672994614,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 0.0074,
      "step": 5250
    },
    {
      "epoch": 79.62121212121212,
      "grad_norm": 0.0447966642677784,
      "learning_rate": 2.037878787878788e-05,
      "loss": 0.0099,
      "step": 5255
    },
    {
      "epoch": 79.6969696969697,
      "grad_norm": 0.03869086876511574,
      "learning_rate": 2.0303030303030303e-05,
      "loss": 0.0069,
      "step": 5260
    },
    {
      "epoch": 79.77272727272727,
      "grad_norm": 0.03913033381104469,
      "learning_rate": 2.022727272727273e-05,
      "loss": 0.0078,
      "step": 5265
    },
    {
      "epoch": 79.84848484848484,
      "grad_norm": 0.029129832983016968,
      "learning_rate": 2.0151515151515152e-05,
      "loss": 0.0103,
      "step": 5270
    },
    {
      "epoch": 79.92424242424242,
      "grad_norm": 0.03608020767569542,
      "learning_rate": 2.0075757575757575e-05,
      "loss": 0.0092,
      "step": 5275
    },
    {
      "epoch": 80.0,
      "grad_norm": 0.23000779747962952,
      "learning_rate": 2e-05,
      "loss": 0.0105,
      "step": 5280
    },
    {
      "epoch": 80.0,
      "eval_loss": 0.010239748284220695,
      "eval_mean_accuracy": 0.9660301298494031,
      "eval_mean_iou": 0.9354270795397134,
      "eval_runtime": 105.837,
      "eval_samples_per_second": 2.22,
      "eval_steps_per_second": 0.283,
      "step": 5280
    },
    {
      "epoch": 80.07575757575758,
      "grad_norm": 0.0508938729763031,
      "learning_rate": 1.9924242424242425e-05,
      "loss": 0.0085,
      "step": 5285
    },
    {
      "epoch": 80.15151515151516,
      "grad_norm": 0.030208123847842216,
      "learning_rate": 1.984848484848485e-05,
      "loss": 0.0067,
      "step": 5290
    },
    {
      "epoch": 80.22727272727273,
      "grad_norm": 0.03110942803323269,
      "learning_rate": 1.9772727272727274e-05,
      "loss": 0.008,
      "step": 5295
    },
    {
      "epoch": 80.3030303030303,
      "grad_norm": 0.026498429477214813,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.0076,
      "step": 5300
    },
    {
      "epoch": 80.37878787878788,
      "grad_norm": 0.025276949629187584,
      "learning_rate": 1.962121212121212e-05,
      "loss": 0.0072,
      "step": 5305
    },
    {
      "epoch": 80.45454545454545,
      "grad_norm": 0.03470022603869438,
      "learning_rate": 1.9545454545454546e-05,
      "loss": 0.0107,
      "step": 5310
    },
    {
      "epoch": 80.53030303030303,
      "grad_norm": 0.07964533567428589,
      "learning_rate": 1.9469696969696972e-05,
      "loss": 0.0085,
      "step": 5315
    },
    {
      "epoch": 80.60606060606061,
      "grad_norm": 0.01591130532324314,
      "learning_rate": 1.9393939393939395e-05,
      "loss": 0.0092,
      "step": 5320
    },
    {
      "epoch": 80.68181818181819,
      "grad_norm": 0.03381310775876045,
      "learning_rate": 1.9318181818181818e-05,
      "loss": 0.0088,
      "step": 5325
    },
    {
      "epoch": 80.75757575757575,
      "grad_norm": 0.038283053785562515,
      "learning_rate": 1.924242424242424e-05,
      "loss": 0.0105,
      "step": 5330
    },
    {
      "epoch": 80.83333333333333,
      "grad_norm": 0.11912844330072403,
      "learning_rate": 1.9166666666666667e-05,
      "loss": 0.0074,
      "step": 5335
    },
    {
      "epoch": 80.9090909090909,
      "grad_norm": 0.029994424432516098,
      "learning_rate": 1.9090909090909094e-05,
      "loss": 0.0063,
      "step": 5340
    },
    {
      "epoch": 80.98484848484848,
      "grad_norm": 0.035760145634412766,
      "learning_rate": 1.9015151515151516e-05,
      "loss": 0.0097,
      "step": 5345
    },
    {
      "epoch": 81.0,
      "eval_loss": 0.010220495983958244,
      "eval_mean_accuracy": 0.9672068648884197,
      "eval_mean_iou": 0.9357393126688426,
      "eval_runtime": 106.3903,
      "eval_samples_per_second": 2.209,
      "eval_steps_per_second": 0.282,
      "step": 5346
    },
    {
      "epoch": 81.06060606060606,
      "grad_norm": 0.02253742516040802,
      "learning_rate": 1.893939393939394e-05,
      "loss": 0.0096,
      "step": 5350
    },
    {
      "epoch": 81.13636363636364,
      "grad_norm": 0.07096920907497406,
      "learning_rate": 1.8863636363636362e-05,
      "loss": 0.0074,
      "step": 5355
    },
    {
      "epoch": 81.21212121212122,
      "grad_norm": 0.07337644696235657,
      "learning_rate": 1.878787878787879e-05,
      "loss": 0.0077,
      "step": 5360
    },
    {
      "epoch": 81.28787878787878,
      "grad_norm": 0.020210614427924156,
      "learning_rate": 1.8712121212121215e-05,
      "loss": 0.0082,
      "step": 5365
    },
    {
      "epoch": 81.36363636363636,
      "grad_norm": 0.05023518204689026,
      "learning_rate": 1.8636363636363638e-05,
      "loss": 0.01,
      "step": 5370
    },
    {
      "epoch": 81.43939393939394,
      "grad_norm": 0.01918661594390869,
      "learning_rate": 1.856060606060606e-05,
      "loss": 0.0092,
      "step": 5375
    },
    {
      "epoch": 81.51515151515152,
      "grad_norm": 0.03883567452430725,
      "learning_rate": 1.8484848484848487e-05,
      "loss": 0.008,
      "step": 5380
    },
    {
      "epoch": 81.5909090909091,
      "grad_norm": 0.02644503302872181,
      "learning_rate": 1.840909090909091e-05,
      "loss": 0.008,
      "step": 5385
    },
    {
      "epoch": 81.66666666666667,
      "grad_norm": 0.03461665287613869,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.0072,
      "step": 5390
    },
    {
      "epoch": 81.74242424242425,
      "grad_norm": 0.030406402423977852,
      "learning_rate": 1.825757575757576e-05,
      "loss": 0.0071,
      "step": 5395
    },
    {
      "epoch": 81.81818181818181,
      "grad_norm": 0.05844994634389877,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.01,
      "step": 5400
    },
    {
      "epoch": 81.89393939393939,
      "grad_norm": 0.10142076760530472,
      "learning_rate": 1.810606060606061e-05,
      "loss": 0.0109,
      "step": 5405
    },
    {
      "epoch": 81.96969696969697,
      "grad_norm": 0.029981547966599464,
      "learning_rate": 1.803030303030303e-05,
      "loss": 0.0095,
      "step": 5410
    },
    {
      "epoch": 82.0,
      "eval_loss": 0.01014484278857708,
      "eval_mean_accuracy": 0.9602110800612472,
      "eval_mean_iou": 0.9346652621241677,
      "eval_runtime": 106.7084,
      "eval_samples_per_second": 2.202,
      "eval_steps_per_second": 0.281,
      "step": 5412
    },
    {
      "epoch": 82.04545454545455,
      "grad_norm": 0.025870341807603836,
      "learning_rate": 1.7954545454545454e-05,
      "loss": 0.0054,
      "step": 5415
    },
    {
      "epoch": 82.12121212121212,
      "grad_norm": 0.03955446928739548,
      "learning_rate": 1.787878787878788e-05,
      "loss": 0.012,
      "step": 5420
    },
    {
      "epoch": 82.1969696969697,
      "grad_norm": 0.07262364774942398,
      "learning_rate": 1.7803030303030303e-05,
      "loss": 0.0081,
      "step": 5425
    },
    {
      "epoch": 82.27272727272727,
      "grad_norm": 0.041742660105228424,
      "learning_rate": 1.772727272727273e-05,
      "loss": 0.0096,
      "step": 5430
    },
    {
      "epoch": 82.34848484848484,
      "grad_norm": 0.020702151581645012,
      "learning_rate": 1.7651515151515153e-05,
      "loss": 0.0077,
      "step": 5435
    },
    {
      "epoch": 82.42424242424242,
      "grad_norm": 0.029360808432102203,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 0.0098,
      "step": 5440
    },
    {
      "epoch": 82.5,
      "grad_norm": 0.017996549606323242,
      "learning_rate": 1.75e-05,
      "loss": 0.0058,
      "step": 5445
    },
    {
      "epoch": 82.57575757575758,
      "grad_norm": 0.027227619662880898,
      "learning_rate": 1.7424242424242425e-05,
      "loss": 0.0072,
      "step": 5450
    },
    {
      "epoch": 82.65151515151516,
      "grad_norm": 0.04631071537733078,
      "learning_rate": 1.734848484848485e-05,
      "loss": 0.0101,
      "step": 5455
    },
    {
      "epoch": 82.72727272727273,
      "grad_norm": 0.06160657852888107,
      "learning_rate": 1.7272727272727274e-05,
      "loss": 0.0076,
      "step": 5460
    },
    {
      "epoch": 82.8030303030303,
      "grad_norm": 0.04007303714752197,
      "learning_rate": 1.7196969696969697e-05,
      "loss": 0.0082,
      "step": 5465
    },
    {
      "epoch": 82.87878787878788,
      "grad_norm": 0.04131500422954559,
      "learning_rate": 1.712121212121212e-05,
      "loss": 0.0089,
      "step": 5470
    },
    {
      "epoch": 82.95454545454545,
      "grad_norm": 0.10315566509962082,
      "learning_rate": 1.7045454545454546e-05,
      "loss": 0.0075,
      "step": 5475
    },
    {
      "epoch": 83.0,
      "eval_loss": 0.010236662812530994,
      "eval_mean_accuracy": 0.9597521924037878,
      "eval_mean_iou": 0.9347674157201992,
      "eval_runtime": 106.5113,
      "eval_samples_per_second": 2.206,
      "eval_steps_per_second": 0.282,
      "step": 5478
    },
    {
      "epoch": 83.03030303030303,
      "grad_norm": 0.023468932136893272,
      "learning_rate": 1.6969696969696972e-05,
      "loss": 0.0088,
      "step": 5480
    },
    {
      "epoch": 83.10606060606061,
      "grad_norm": 0.05640023574233055,
      "learning_rate": 1.6893939393939395e-05,
      "loss": 0.0087,
      "step": 5485
    },
    {
      "epoch": 83.18181818181819,
      "grad_norm": 0.05830927938222885,
      "learning_rate": 1.6818181818181818e-05,
      "loss": 0.008,
      "step": 5490
    },
    {
      "epoch": 83.25757575757575,
      "grad_norm": 0.016806235536932945,
      "learning_rate": 1.674242424242424e-05,
      "loss": 0.0084,
      "step": 5495
    },
    {
      "epoch": 83.33333333333333,
      "grad_norm": 0.057013098150491714,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0098,
      "step": 5500
    },
    {
      "epoch": 83.4090909090909,
      "grad_norm": 0.03845953568816185,
      "learning_rate": 1.6590909090909094e-05,
      "loss": 0.0084,
      "step": 5505
    },
    {
      "epoch": 83.48484848484848,
      "grad_norm": 0.05152120441198349,
      "learning_rate": 1.6515151515151517e-05,
      "loss": 0.0062,
      "step": 5510
    },
    {
      "epoch": 83.56060606060606,
      "grad_norm": 0.08059060573577881,
      "learning_rate": 1.643939393939394e-05,
      "loss": 0.0094,
      "step": 5515
    },
    {
      "epoch": 83.63636363636364,
      "grad_norm": 0.05539998039603233,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.0096,
      "step": 5520
    },
    {
      "epoch": 83.71212121212122,
      "grad_norm": 0.028666911646723747,
      "learning_rate": 1.628787878787879e-05,
      "loss": 0.0094,
      "step": 5525
    },
    {
      "epoch": 83.78787878787878,
      "grad_norm": 0.03674125671386719,
      "learning_rate": 1.6212121212121212e-05,
      "loss": 0.0095,
      "step": 5530
    },
    {
      "epoch": 83.86363636363636,
      "grad_norm": 0.05053194984793663,
      "learning_rate": 1.6136363636363638e-05,
      "loss": 0.007,
      "step": 5535
    },
    {
      "epoch": 83.93939393939394,
      "grad_norm": 0.041537247598171234,
      "learning_rate": 1.606060606060606e-05,
      "loss": 0.0071,
      "step": 5540
    },
    {
      "epoch": 84.0,
      "eval_loss": 0.010387294925749302,
      "eval_mean_accuracy": 0.9576568693033082,
      "eval_mean_iou": 0.9333989184635098,
      "eval_runtime": 107.1463,
      "eval_samples_per_second": 2.193,
      "eval_steps_per_second": 0.28,
      "step": 5544
    },
    {
      "epoch": 84.01515151515152,
      "grad_norm": 0.03256434574723244,
      "learning_rate": 1.5984848484848487e-05,
      "loss": 0.0071,
      "step": 5545
    },
    {
      "epoch": 84.0909090909091,
      "grad_norm": 0.030848747119307518,
      "learning_rate": 1.590909090909091e-05,
      "loss": 0.0085,
      "step": 5550
    },
    {
      "epoch": 84.16666666666667,
      "grad_norm": 0.021305767819285393,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.0064,
      "step": 5555
    },
    {
      "epoch": 84.24242424242425,
      "grad_norm": 0.02515929937362671,
      "learning_rate": 1.5757575757575756e-05,
      "loss": 0.0085,
      "step": 5560
    },
    {
      "epoch": 84.31818181818181,
      "grad_norm": 0.03549756854772568,
      "learning_rate": 1.5681818181818182e-05,
      "loss": 0.0094,
      "step": 5565
    },
    {
      "epoch": 84.39393939393939,
      "grad_norm": 0.1029946506023407,
      "learning_rate": 1.560606060606061e-05,
      "loss": 0.0089,
      "step": 5570
    },
    {
      "epoch": 84.46969696969697,
      "grad_norm": 0.07234491407871246,
      "learning_rate": 1.553030303030303e-05,
      "loss": 0.0081,
      "step": 5575
    },
    {
      "epoch": 84.54545454545455,
      "grad_norm": 0.030196094885468483,
      "learning_rate": 1.5454545454545454e-05,
      "loss": 0.0075,
      "step": 5580
    },
    {
      "epoch": 84.62121212121212,
      "grad_norm": 0.0997152179479599,
      "learning_rate": 1.5378787878787877e-05,
      "loss": 0.0118,
      "step": 5585
    },
    {
      "epoch": 84.6969696969697,
      "grad_norm": 0.02381710521876812,
      "learning_rate": 1.5303030303030304e-05,
      "loss": 0.0093,
      "step": 5590
    },
    {
      "epoch": 84.77272727272727,
      "grad_norm": 0.07447588443756104,
      "learning_rate": 1.5227272727272728e-05,
      "loss": 0.0092,
      "step": 5595
    },
    {
      "epoch": 84.84848484848484,
      "grad_norm": 0.047184038907289505,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 0.0078,
      "step": 5600
    },
    {
      "epoch": 84.92424242424242,
      "grad_norm": 0.021866420283913612,
      "learning_rate": 1.5075757575757576e-05,
      "loss": 0.0068,
      "step": 5605
    },
    {
      "epoch": 85.0,
      "grad_norm": 0.04825587570667267,
      "learning_rate": 1.5e-05,
      "loss": 0.0086,
      "step": 5610
    },
    {
      "epoch": 85.0,
      "eval_loss": 0.010119255632162094,
      "eval_mean_accuracy": 0.9619490271843834,
      "eval_mean_iou": 0.9356255638112433,
      "eval_runtime": 106.3434,
      "eval_samples_per_second": 2.21,
      "eval_steps_per_second": 0.282,
      "step": 5610
    },
    {
      "epoch": 85.07575757575758,
      "grad_norm": 0.029690535739064217,
      "learning_rate": 1.4924242424242423e-05,
      "loss": 0.0062,
      "step": 5615
    },
    {
      "epoch": 85.15151515151516,
      "grad_norm": 0.04636257514357567,
      "learning_rate": 1.484848484848485e-05,
      "loss": 0.0073,
      "step": 5620
    },
    {
      "epoch": 85.22727272727273,
      "grad_norm": 0.014345095492899418,
      "learning_rate": 1.4772727272727274e-05,
      "loss": 0.0074,
      "step": 5625
    },
    {
      "epoch": 85.3030303030303,
      "grad_norm": 0.022371189668774605,
      "learning_rate": 1.4696969696969697e-05,
      "loss": 0.0086,
      "step": 5630
    },
    {
      "epoch": 85.37878787878788,
      "grad_norm": 0.09568944573402405,
      "learning_rate": 1.4621212121212122e-05,
      "loss": 0.0097,
      "step": 5635
    },
    {
      "epoch": 85.45454545454545,
      "grad_norm": 0.057022806257009506,
      "learning_rate": 1.4545454545454545e-05,
      "loss": 0.0104,
      "step": 5640
    },
    {
      "epoch": 85.53030303030303,
      "grad_norm": 0.03688698261976242,
      "learning_rate": 1.446969696969697e-05,
      "loss": 0.0076,
      "step": 5645
    },
    {
      "epoch": 85.60606060606061,
      "grad_norm": 0.03131138160824776,
      "learning_rate": 1.4393939393939396e-05,
      "loss": 0.0073,
      "step": 5650
    },
    {
      "epoch": 85.68181818181819,
      "grad_norm": 0.07043741643428802,
      "learning_rate": 1.431818181818182e-05,
      "loss": 0.0107,
      "step": 5655
    },
    {
      "epoch": 85.75757575757575,
      "grad_norm": 0.03284035995602608,
      "learning_rate": 1.4242424242424243e-05,
      "loss": 0.0079,
      "step": 5660
    },
    {
      "epoch": 85.83333333333333,
      "grad_norm": 0.03744242712855339,
      "learning_rate": 1.4166666666666668e-05,
      "loss": 0.0087,
      "step": 5665
    },
    {
      "epoch": 85.9090909090909,
      "grad_norm": 0.026261677965521812,
      "learning_rate": 1.409090909090909e-05,
      "loss": 0.0101,
      "step": 5670
    },
    {
      "epoch": 85.98484848484848,
      "grad_norm": 0.42324909567832947,
      "learning_rate": 1.4015151515151515e-05,
      "loss": 0.0064,
      "step": 5675
    },
    {
      "epoch": 86.0,
      "eval_loss": 0.010133696720004082,
      "eval_mean_accuracy": 0.9649061427860722,
      "eval_mean_iou": 0.9358921934087867,
      "eval_runtime": 105.6902,
      "eval_samples_per_second": 2.223,
      "eval_steps_per_second": 0.284,
      "step": 5676
    },
    {
      "epoch": 86.06060606060606,
      "grad_norm": 0.04835553467273712,
      "learning_rate": 1.3939393939393942e-05,
      "loss": 0.0116,
      "step": 5680
    },
    {
      "epoch": 86.13636363636364,
      "grad_norm": 0.17864994704723358,
      "learning_rate": 1.3863636363636364e-05,
      "loss": 0.0085,
      "step": 5685
    },
    {
      "epoch": 86.21212121212122,
      "grad_norm": 0.06496840715408325,
      "learning_rate": 1.3787878787878789e-05,
      "loss": 0.0057,
      "step": 5690
    },
    {
      "epoch": 86.28787878787878,
      "grad_norm": 0.04384487122297287,
      "learning_rate": 1.3712121212121212e-05,
      "loss": 0.0082,
      "step": 5695
    },
    {
      "epoch": 86.36363636363636,
      "grad_norm": 0.02799801714718342,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 0.0071,
      "step": 5700
    },
    {
      "epoch": 86.43939393939394,
      "grad_norm": 0.0365811288356781,
      "learning_rate": 1.3560606060606063e-05,
      "loss": 0.0117,
      "step": 5705
    },
    {
      "epoch": 86.51515151515152,
      "grad_norm": 0.04523276910185814,
      "learning_rate": 1.3484848484848486e-05,
      "loss": 0.0116,
      "step": 5710
    },
    {
      "epoch": 86.5909090909091,
      "grad_norm": 0.06717886030673981,
      "learning_rate": 1.340909090909091e-05,
      "loss": 0.0103,
      "step": 5715
    },
    {
      "epoch": 86.66666666666667,
      "grad_norm": 0.018165141344070435,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0063,
      "step": 5720
    },
    {
      "epoch": 86.74242424242425,
      "grad_norm": 0.04554758220911026,
      "learning_rate": 1.3257575757575758e-05,
      "loss": 0.0083,
      "step": 5725
    },
    {
      "epoch": 86.81818181818181,
      "grad_norm": 0.05116431787610054,
      "learning_rate": 1.318181818181818e-05,
      "loss": 0.0093,
      "step": 5730
    },
    {
      "epoch": 86.89393939393939,
      "grad_norm": 0.032237567007541656,
      "learning_rate": 1.3106060606060607e-05,
      "loss": 0.0087,
      "step": 5735
    },
    {
      "epoch": 86.96969696969697,
      "grad_norm": 0.04752449318766594,
      "learning_rate": 1.3030303030303032e-05,
      "loss": 0.0084,
      "step": 5740
    },
    {
      "epoch": 87.0,
      "eval_loss": 0.010347366333007812,
      "eval_mean_accuracy": 0.9626808588188064,
      "eval_mean_iou": 0.9347012984674028,
      "eval_runtime": 106.7704,
      "eval_samples_per_second": 2.201,
      "eval_steps_per_second": 0.281,
      "step": 5742
    },
    {
      "epoch": 87.04545454545455,
      "grad_norm": 0.028220541775226593,
      "learning_rate": 1.2954545454545455e-05,
      "loss": 0.0094,
      "step": 5745
    },
    {
      "epoch": 87.12121212121212,
      "grad_norm": 0.07758155465126038,
      "learning_rate": 1.287878787878788e-05,
      "loss": 0.0074,
      "step": 5750
    },
    {
      "epoch": 87.1969696969697,
      "grad_norm": 0.04379045218229294,
      "learning_rate": 1.2803030303030302e-05,
      "loss": 0.0094,
      "step": 5755
    },
    {
      "epoch": 87.27272727272727,
      "grad_norm": 0.04330086708068848,
      "learning_rate": 1.2727272727272727e-05,
      "loss": 0.0103,
      "step": 5760
    },
    {
      "epoch": 87.34848484848484,
      "grad_norm": 0.08726771920919418,
      "learning_rate": 1.2651515151515153e-05,
      "loss": 0.009,
      "step": 5765
    },
    {
      "epoch": 87.42424242424242,
      "grad_norm": 0.03261416777968407,
      "learning_rate": 1.2575757575757578e-05,
      "loss": 0.0072,
      "step": 5770
    },
    {
      "epoch": 87.5,
      "grad_norm": 0.022511860355734825,
      "learning_rate": 1.25e-05,
      "loss": 0.006,
      "step": 5775
    },
    {
      "epoch": 87.57575757575758,
      "grad_norm": 0.04309185594320297,
      "learning_rate": 1.2424242424242424e-05,
      "loss": 0.0111,
      "step": 5780
    },
    {
      "epoch": 87.65151515151516,
      "grad_norm": 0.047899894416332245,
      "learning_rate": 1.234848484848485e-05,
      "loss": 0.0083,
      "step": 5785
    },
    {
      "epoch": 87.72727272727273,
      "grad_norm": 0.026579799130558968,
      "learning_rate": 1.2272727272727273e-05,
      "loss": 0.0076,
      "step": 5790
    },
    {
      "epoch": 87.8030303030303,
      "grad_norm": 0.019098084419965744,
      "learning_rate": 1.2196969696969697e-05,
      "loss": 0.0067,
      "step": 5795
    },
    {
      "epoch": 87.87878787878788,
      "grad_norm": 0.10605199635028839,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.0098,
      "step": 5800
    },
    {
      "epoch": 87.95454545454545,
      "grad_norm": 0.03990662470459938,
      "learning_rate": 1.2045454545454547e-05,
      "loss": 0.0079,
      "step": 5805
    },
    {
      "epoch": 88.0,
      "eval_loss": 0.01004178635776043,
      "eval_mean_accuracy": 0.9642326447128751,
      "eval_mean_iou": 0.9360301306864204,
      "eval_runtime": 107.2039,
      "eval_samples_per_second": 2.192,
      "eval_steps_per_second": 0.28,
      "step": 5808
    },
    {
      "epoch": 88.03030303030303,
      "grad_norm": 0.048288002610206604,
      "learning_rate": 1.196969696969697e-05,
      "loss": 0.0077,
      "step": 5810
    },
    {
      "epoch": 88.10606060606061,
      "grad_norm": 0.030538788065314293,
      "learning_rate": 1.1893939393939394e-05,
      "loss": 0.0077,
      "step": 5815
    },
    {
      "epoch": 88.18181818181819,
      "grad_norm": 0.05115768685936928,
      "learning_rate": 1.1818181818181819e-05,
      "loss": 0.0069,
      "step": 5820
    },
    {
      "epoch": 88.25757575757575,
      "grad_norm": 0.04398134723305702,
      "learning_rate": 1.1742424242424243e-05,
      "loss": 0.0085,
      "step": 5825
    },
    {
      "epoch": 88.33333333333333,
      "grad_norm": 0.038140542805194855,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.0075,
      "step": 5830
    },
    {
      "epoch": 88.4090909090909,
      "grad_norm": 0.024708544835448265,
      "learning_rate": 1.159090909090909e-05,
      "loss": 0.0078,
      "step": 5835
    },
    {
      "epoch": 88.48484848484848,
      "grad_norm": 0.0738573670387268,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 0.0098,
      "step": 5840
    },
    {
      "epoch": 88.56060606060606,
      "grad_norm": 0.03331197425723076,
      "learning_rate": 1.143939393939394e-05,
      "loss": 0.0095,
      "step": 5845
    },
    {
      "epoch": 88.63636363636364,
      "grad_norm": 0.04650338739156723,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 0.0096,
      "step": 5850
    },
    {
      "epoch": 88.71212121212122,
      "grad_norm": 0.07829892635345459,
      "learning_rate": 1.128787878787879e-05,
      "loss": 0.0114,
      "step": 5855
    },
    {
      "epoch": 88.78787878787878,
      "grad_norm": 0.025248106569051743,
      "learning_rate": 1.1212121212121212e-05,
      "loss": 0.0077,
      "step": 5860
    },
    {
      "epoch": 88.86363636363636,
      "grad_norm": 0.04843801632523537,
      "learning_rate": 1.1136363636363637e-05,
      "loss": 0.0078,
      "step": 5865
    },
    {
      "epoch": 88.93939393939394,
      "grad_norm": 0.019280917942523956,
      "learning_rate": 1.1060606060606061e-05,
      "loss": 0.0072,
      "step": 5870
    },
    {
      "epoch": 89.0,
      "eval_loss": 0.010203158482909203,
      "eval_mean_accuracy": 0.9602195705207235,
      "eval_mean_iou": 0.9345833029064393,
      "eval_runtime": 106.0707,
      "eval_samples_per_second": 2.216,
      "eval_steps_per_second": 0.283,
      "step": 5874
    },
    {
      "epoch": 89.01515151515152,
      "grad_norm": 0.01911124773323536,
      "learning_rate": 1.0984848484848486e-05,
      "loss": 0.0056,
      "step": 5875
    },
    {
      "epoch": 89.0909090909091,
      "grad_norm": 0.04301591217517853,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 0.0088,
      "step": 5880
    },
    {
      "epoch": 89.16666666666667,
      "grad_norm": 0.015321165323257446,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 0.0065,
      "step": 5885
    },
    {
      "epoch": 89.24242424242425,
      "grad_norm": 0.042060486972332,
      "learning_rate": 1.0757575757575758e-05,
      "loss": 0.01,
      "step": 5890
    },
    {
      "epoch": 89.31818181818181,
      "grad_norm": 0.036771778017282486,
      "learning_rate": 1.0681818181818181e-05,
      "loss": 0.0057,
      "step": 5895
    },
    {
      "epoch": 89.39393939393939,
      "grad_norm": 0.049904946237802505,
      "learning_rate": 1.0606060606060607e-05,
      "loss": 0.0079,
      "step": 5900
    },
    {
      "epoch": 89.46969696969697,
      "grad_norm": 0.0670136958360672,
      "learning_rate": 1.053030303030303e-05,
      "loss": 0.0077,
      "step": 5905
    },
    {
      "epoch": 89.54545454545455,
      "grad_norm": 0.03494875505566597,
      "learning_rate": 1.0454545454545455e-05,
      "loss": 0.0121,
      "step": 5910
    },
    {
      "epoch": 89.62121212121212,
      "grad_norm": 0.04414621740579605,
      "learning_rate": 1.037878787878788e-05,
      "loss": 0.0079,
      "step": 5915
    },
    {
      "epoch": 89.6969696969697,
      "grad_norm": 0.0474022701382637,
      "learning_rate": 1.0303030303030304e-05,
      "loss": 0.0082,
      "step": 5920
    },
    {
      "epoch": 89.77272727272727,
      "grad_norm": 0.02139611355960369,
      "learning_rate": 1.0227272727272729e-05,
      "loss": 0.0093,
      "step": 5925
    },
    {
      "epoch": 89.84848484848484,
      "grad_norm": 0.028572553768754005,
      "learning_rate": 1.0151515151515152e-05,
      "loss": 0.0062,
      "step": 5930
    },
    {
      "epoch": 89.92424242424242,
      "grad_norm": 0.040755320340394974,
      "learning_rate": 1.0075757575757576e-05,
      "loss": 0.0103,
      "step": 5935
    },
    {
      "epoch": 90.0,
      "grad_norm": 0.08273164182901382,
      "learning_rate": 1e-05,
      "loss": 0.0097,
      "step": 5940
    },
    {
      "epoch": 90.0,
      "eval_loss": 0.010044960305094719,
      "eval_mean_accuracy": 0.9663917665982252,
      "eval_mean_iou": 0.9361569281606473,
      "eval_runtime": 107.8665,
      "eval_samples_per_second": 2.179,
      "eval_steps_per_second": 0.278,
      "step": 5940
    },
    {
      "epoch": 90.07575757575758,
      "grad_norm": 0.04758373275399208,
      "learning_rate": 9.924242424242425e-06,
      "loss": 0.0076,
      "step": 5945
    },
    {
      "epoch": 90.15151515151516,
      "grad_norm": 0.05593794584274292,
      "learning_rate": 9.848484848484848e-06,
      "loss": 0.0075,
      "step": 5950
    },
    {
      "epoch": 90.22727272727273,
      "grad_norm": 0.045310258865356445,
      "learning_rate": 9.772727272727273e-06,
      "loss": 0.0067,
      "step": 5955
    },
    {
      "epoch": 90.3030303030303,
      "grad_norm": 0.05919276177883148,
      "learning_rate": 9.696969696969698e-06,
      "loss": 0.0093,
      "step": 5960
    },
    {
      "epoch": 90.37878787878788,
      "grad_norm": 0.03449501469731331,
      "learning_rate": 9.62121212121212e-06,
      "loss": 0.01,
      "step": 5965
    },
    {
      "epoch": 90.45454545454545,
      "grad_norm": 0.027477266266942024,
      "learning_rate": 9.545454545454547e-06,
      "loss": 0.009,
      "step": 5970
    },
    {
      "epoch": 90.53030303030303,
      "grad_norm": 0.06122530996799469,
      "learning_rate": 9.46969696969697e-06,
      "loss": 0.0095,
      "step": 5975
    },
    {
      "epoch": 90.60606060606061,
      "grad_norm": 0.03298007324337959,
      "learning_rate": 9.393939393939394e-06,
      "loss": 0.0051,
      "step": 5980
    },
    {
      "epoch": 90.68181818181819,
      "grad_norm": 0.04602937772870064,
      "learning_rate": 9.318181818181819e-06,
      "loss": 0.0076,
      "step": 5985
    },
    {
      "epoch": 90.75757575757575,
      "grad_norm": 0.04156597703695297,
      "learning_rate": 9.242424242424244e-06,
      "loss": 0.0107,
      "step": 5990
    },
    {
      "epoch": 90.83333333333333,
      "grad_norm": 0.055980145931243896,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.0075,
      "step": 5995
    },
    {
      "epoch": 90.9090909090909,
      "grad_norm": 0.06538745760917664,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.0126,
      "step": 6000
    },
    {
      "epoch": 90.98484848484848,
      "grad_norm": 0.03500562906265259,
      "learning_rate": 9.015151515151516e-06,
      "loss": 0.0059,
      "step": 6005
    },
    {
      "epoch": 91.0,
      "eval_loss": 0.010132785886526108,
      "eval_mean_accuracy": 0.9687225114477022,
      "eval_mean_iou": 0.9359031842548372,
      "eval_runtime": 106.0169,
      "eval_samples_per_second": 2.217,
      "eval_steps_per_second": 0.283,
      "step": 6006
    },
    {
      "epoch": 91.06060606060606,
      "grad_norm": 0.05061809718608856,
      "learning_rate": 8.93939393939394e-06,
      "loss": 0.0102,
      "step": 6010
    },
    {
      "epoch": 91.13636363636364,
      "grad_norm": 0.021474868059158325,
      "learning_rate": 8.863636363636365e-06,
      "loss": 0.0071,
      "step": 6015
    },
    {
      "epoch": 91.21212121212122,
      "grad_norm": 0.04241073131561279,
      "learning_rate": 8.787878787878788e-06,
      "loss": 0.007,
      "step": 6020
    },
    {
      "epoch": 91.28787878787878,
      "grad_norm": 0.07912750542163849,
      "learning_rate": 8.712121212121212e-06,
      "loss": 0.0098,
      "step": 6025
    },
    {
      "epoch": 91.36363636363636,
      "grad_norm": 0.09092885255813599,
      "learning_rate": 8.636363636363637e-06,
      "loss": 0.011,
      "step": 6030
    },
    {
      "epoch": 91.43939393939394,
      "grad_norm": 0.047319039702415466,
      "learning_rate": 8.56060606060606e-06,
      "loss": 0.0087,
      "step": 6035
    },
    {
      "epoch": 91.51515151515152,
      "grad_norm": 0.0530010461807251,
      "learning_rate": 8.484848484848486e-06,
      "loss": 0.0083,
      "step": 6040
    },
    {
      "epoch": 91.5909090909091,
      "grad_norm": 0.013299948535859585,
      "learning_rate": 8.409090909090909e-06,
      "loss": 0.0084,
      "step": 6045
    },
    {
      "epoch": 91.66666666666667,
      "grad_norm": 0.02709808573126793,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0068,
      "step": 6050
    },
    {
      "epoch": 91.74242424242425,
      "grad_norm": 0.029436960816383362,
      "learning_rate": 8.257575757575758e-06,
      "loss": 0.0072,
      "step": 6055
    },
    {
      "epoch": 91.81818181818181,
      "grad_norm": 0.049926042556762695,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.0088,
      "step": 6060
    },
    {
      "epoch": 91.89393939393939,
      "grad_norm": 0.009414332918822765,
      "learning_rate": 8.106060606060606e-06,
      "loss": 0.0078,
      "step": 6065
    },
    {
      "epoch": 91.96969696969697,
      "grad_norm": 0.03354932740330696,
      "learning_rate": 8.03030303030303e-06,
      "loss": 0.0077,
      "step": 6070
    },
    {
      "epoch": 92.0,
      "eval_loss": 0.010086090303957462,
      "eval_mean_accuracy": 0.9616878237516803,
      "eval_mean_iou": 0.9356321022689166,
      "eval_runtime": 106.1282,
      "eval_samples_per_second": 2.214,
      "eval_steps_per_second": 0.283,
      "step": 6072
    },
    {
      "epoch": 92.04545454545455,
      "grad_norm": 0.053115952759981155,
      "learning_rate": 7.954545454545455e-06,
      "loss": 0.0096,
      "step": 6075
    },
    {
      "epoch": 92.12121212121212,
      "grad_norm": 0.029202813282608986,
      "learning_rate": 7.878787878787878e-06,
      "loss": 0.009,
      "step": 6080
    },
    {
      "epoch": 92.1969696969697,
      "grad_norm": 0.038335345685482025,
      "learning_rate": 7.803030303030304e-06,
      "loss": 0.0097,
      "step": 6085
    },
    {
      "epoch": 92.27272727272727,
      "grad_norm": 0.04186587780714035,
      "learning_rate": 7.727272727272727e-06,
      "loss": 0.008,
      "step": 6090
    },
    {
      "epoch": 92.34848484848484,
      "grad_norm": 0.06838929653167725,
      "learning_rate": 7.651515151515152e-06,
      "loss": 0.0095,
      "step": 6095
    },
    {
      "epoch": 92.42424242424242,
      "grad_norm": 0.019481556490063667,
      "learning_rate": 7.5757575757575764e-06,
      "loss": 0.0069,
      "step": 6100
    },
    {
      "epoch": 92.5,
      "grad_norm": 0.030994949862360954,
      "learning_rate": 7.5e-06,
      "loss": 0.0079,
      "step": 6105
    },
    {
      "epoch": 92.57575757575758,
      "grad_norm": 0.029112888500094414,
      "learning_rate": 7.424242424242425e-06,
      "loss": 0.0084,
      "step": 6110
    },
    {
      "epoch": 92.65151515151516,
      "grad_norm": 0.042757272720336914,
      "learning_rate": 7.3484848484848486e-06,
      "loss": 0.0085,
      "step": 6115
    },
    {
      "epoch": 92.72727272727273,
      "grad_norm": 0.0160219743847847,
      "learning_rate": 7.272727272727272e-06,
      "loss": 0.0066,
      "step": 6120
    },
    {
      "epoch": 92.8030303030303,
      "grad_norm": 0.03635643050074577,
      "learning_rate": 7.196969696969698e-06,
      "loss": 0.008,
      "step": 6125
    },
    {
      "epoch": 92.87878787878788,
      "grad_norm": 0.038166750222444534,
      "learning_rate": 7.1212121212121215e-06,
      "loss": 0.0078,
      "step": 6130
    },
    {
      "epoch": 92.95454545454545,
      "grad_norm": 0.030236735939979553,
      "learning_rate": 7.045454545454545e-06,
      "loss": 0.0077,
      "step": 6135
    },
    {
      "epoch": 93.0,
      "eval_loss": 0.010099419392645359,
      "eval_mean_accuracy": 0.9645648111757396,
      "eval_mean_iou": 0.9361532613577457,
      "eval_runtime": 106.4611,
      "eval_samples_per_second": 2.207,
      "eval_steps_per_second": 0.282,
      "step": 6138
    },
    {
      "epoch": 93.03030303030303,
      "grad_norm": 0.02095762826502323,
      "learning_rate": 6.969696969696971e-06,
      "loss": 0.0087,
      "step": 6140
    },
    {
      "epoch": 93.10606060606061,
      "grad_norm": 0.017821820452809334,
      "learning_rate": 6.8939393939393945e-06,
      "loss": 0.0086,
      "step": 6145
    },
    {
      "epoch": 93.18181818181819,
      "grad_norm": 0.14403508603572845,
      "learning_rate": 6.818181818181818e-06,
      "loss": 0.0121,
      "step": 6150
    },
    {
      "epoch": 93.25757575757575,
      "grad_norm": 0.019874388352036476,
      "learning_rate": 6.742424242424243e-06,
      "loss": 0.0061,
      "step": 6155
    },
    {
      "epoch": 93.33333333333333,
      "grad_norm": 0.02095934934914112,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0094,
      "step": 6160
    },
    {
      "epoch": 93.4090909090909,
      "grad_norm": 0.018587643280625343,
      "learning_rate": 6.59090909090909e-06,
      "loss": 0.0066,
      "step": 6165
    },
    {
      "epoch": 93.48484848484848,
      "grad_norm": 0.01562019344419241,
      "learning_rate": 6.515151515151516e-06,
      "loss": 0.008,
      "step": 6170
    },
    {
      "epoch": 93.56060606060606,
      "grad_norm": 0.04144290089607239,
      "learning_rate": 6.43939393939394e-06,
      "loss": 0.0087,
      "step": 6175
    },
    {
      "epoch": 93.63636363636364,
      "grad_norm": 0.011395107954740524,
      "learning_rate": 6.363636363636363e-06,
      "loss": 0.0081,
      "step": 6180
    },
    {
      "epoch": 93.71212121212122,
      "grad_norm": 0.02482740767300129,
      "learning_rate": 6.287878787878789e-06,
      "loss": 0.0069,
      "step": 6185
    },
    {
      "epoch": 93.78787878787878,
      "grad_norm": 0.05444609746336937,
      "learning_rate": 6.212121212121212e-06,
      "loss": 0.0097,
      "step": 6190
    },
    {
      "epoch": 93.86363636363636,
      "grad_norm": 0.02874091826379299,
      "learning_rate": 6.136363636363636e-06,
      "loss": 0.0085,
      "step": 6195
    },
    {
      "epoch": 93.93939393939394,
      "grad_norm": 0.0396924652159214,
      "learning_rate": 6.060606060606061e-06,
      "loss": 0.0075,
      "step": 6200
    },
    {
      "epoch": 94.0,
      "eval_loss": 0.010059281252324581,
      "eval_mean_accuracy": 0.9638755673860402,
      "eval_mean_iou": 0.9360322464735353,
      "eval_runtime": 107.266,
      "eval_samples_per_second": 2.191,
      "eval_steps_per_second": 0.28,
      "step": 6204
    },
    {
      "epoch": 94.01515151515152,
      "grad_norm": 0.08742281049489975,
      "learning_rate": 5.984848484848485e-06,
      "loss": 0.0101,
      "step": 6205
    },
    {
      "epoch": 94.0909090909091,
      "grad_norm": 0.048075735569000244,
      "learning_rate": 5.909090909090909e-06,
      "loss": 0.0095,
      "step": 6210
    },
    {
      "epoch": 94.16666666666667,
      "grad_norm": 0.03732714802026749,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.0077,
      "step": 6215
    },
    {
      "epoch": 94.24242424242425,
      "grad_norm": 0.06750783324241638,
      "learning_rate": 5.7575757575757586e-06,
      "loss": 0.0074,
      "step": 6220
    },
    {
      "epoch": 94.31818181818181,
      "grad_norm": 0.03990792855620384,
      "learning_rate": 5.681818181818182e-06,
      "loss": 0.0071,
      "step": 6225
    },
    {
      "epoch": 94.39393939393939,
      "grad_norm": 0.030566111207008362,
      "learning_rate": 5.606060606060606e-06,
      "loss": 0.0066,
      "step": 6230
    },
    {
      "epoch": 94.46969696969697,
      "grad_norm": 0.04407535493373871,
      "learning_rate": 5.530303030303031e-06,
      "loss": 0.0101,
      "step": 6235
    },
    {
      "epoch": 94.54545454545455,
      "grad_norm": 0.03922566398978233,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.0094,
      "step": 6240
    },
    {
      "epoch": 94.62121212121212,
      "grad_norm": 0.03121493011713028,
      "learning_rate": 5.378787878787879e-06,
      "loss": 0.0111,
      "step": 6245
    },
    {
      "epoch": 94.6969696969697,
      "grad_norm": 0.05284569039940834,
      "learning_rate": 5.303030303030304e-06,
      "loss": 0.0067,
      "step": 6250
    },
    {
      "epoch": 94.77272727272727,
      "grad_norm": 0.023286443203687668,
      "learning_rate": 5.2272727272727274e-06,
      "loss": 0.0071,
      "step": 6255
    },
    {
      "epoch": 94.84848484848484,
      "grad_norm": 0.031161462888121605,
      "learning_rate": 5.151515151515152e-06,
      "loss": 0.0078,
      "step": 6260
    },
    {
      "epoch": 94.92424242424242,
      "grad_norm": 0.020004475489258766,
      "learning_rate": 5.075757575757576e-06,
      "loss": 0.007,
      "step": 6265
    },
    {
      "epoch": 95.0,
      "grad_norm": 0.08035217225551605,
      "learning_rate": 5e-06,
      "loss": 0.0093,
      "step": 6270
    },
    {
      "epoch": 95.0,
      "eval_loss": 0.010007022880017757,
      "eval_mean_accuracy": 0.9631529460569334,
      "eval_mean_iou": 0.93608926081122,
      "eval_runtime": 111.1944,
      "eval_samples_per_second": 2.113,
      "eval_steps_per_second": 0.27,
      "step": 6270
    },
    {
      "epoch": 95.07575757575758,
      "grad_norm": 0.028469273820519447,
      "learning_rate": 4.924242424242424e-06,
      "loss": 0.0073,
      "step": 6275
    },
    {
      "epoch": 95.15151515151516,
      "grad_norm": 0.02897004783153534,
      "learning_rate": 4.848484848484849e-06,
      "loss": 0.0099,
      "step": 6280
    },
    {
      "epoch": 95.22727272727273,
      "grad_norm": 0.06267004460096359,
      "learning_rate": 4.772727272727273e-06,
      "loss": 0.0083,
      "step": 6285
    },
    {
      "epoch": 95.3030303030303,
      "grad_norm": 0.07927574962377548,
      "learning_rate": 4.696969696969697e-06,
      "loss": 0.0085,
      "step": 6290
    },
    {
      "epoch": 95.37878787878788,
      "grad_norm": 0.03160750865936279,
      "learning_rate": 4.621212121212122e-06,
      "loss": 0.0062,
      "step": 6295
    },
    {
      "epoch": 95.45454545454545,
      "grad_norm": 0.060727473348379135,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.0087,
      "step": 6300
    },
    {
      "epoch": 95.53030303030303,
      "grad_norm": 0.06161715090274811,
      "learning_rate": 4.46969696969697e-06,
      "loss": 0.0089,
      "step": 6305
    },
    {
      "epoch": 95.60606060606061,
      "grad_norm": 0.038173578679561615,
      "learning_rate": 4.393939393939394e-06,
      "loss": 0.0079,
      "step": 6310
    },
    {
      "epoch": 95.68181818181819,
      "grad_norm": 0.050391267985105515,
      "learning_rate": 4.3181818181818185e-06,
      "loss": 0.0066,
      "step": 6315
    },
    {
      "epoch": 95.75757575757575,
      "grad_norm": 0.07662074267864227,
      "learning_rate": 4.242424242424243e-06,
      "loss": 0.0108,
      "step": 6320
    },
    {
      "epoch": 95.83333333333333,
      "grad_norm": 0.047124240547418594,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0097,
      "step": 6325
    },
    {
      "epoch": 95.9090909090909,
      "grad_norm": 0.03968175873160362,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 0.0057,
      "step": 6330
    },
    {
      "epoch": 95.98484848484848,
      "grad_norm": 0.029545683413743973,
      "learning_rate": 4.015151515151515e-06,
      "loss": 0.0087,
      "step": 6335
    },
    {
      "epoch": 96.0,
      "eval_loss": 0.00999161135405302,
      "eval_mean_accuracy": 0.9658166808231234,
      "eval_mean_iou": 0.9366139867645246,
      "eval_runtime": 111.1417,
      "eval_samples_per_second": 2.114,
      "eval_steps_per_second": 0.27,
      "step": 6336
    },
    {
      "epoch": 96.06060606060606,
      "grad_norm": 0.05038948729634285,
      "learning_rate": 3.939393939393939e-06,
      "loss": 0.0094,
      "step": 6340
    },
    {
      "epoch": 96.13636363636364,
      "grad_norm": 0.022214466705918312,
      "learning_rate": 3.863636363636364e-06,
      "loss": 0.008,
      "step": 6345
    },
    {
      "epoch": 96.21212121212122,
      "grad_norm": 0.06709983944892883,
      "learning_rate": 3.7878787878787882e-06,
      "loss": 0.0071,
      "step": 6350
    },
    {
      "epoch": 96.28787878787878,
      "grad_norm": 0.054274335503578186,
      "learning_rate": 3.7121212121212124e-06,
      "loss": 0.0072,
      "step": 6355
    },
    {
      "epoch": 96.36363636363636,
      "grad_norm": 0.05126475915312767,
      "learning_rate": 3.636363636363636e-06,
      "loss": 0.0084,
      "step": 6360
    },
    {
      "epoch": 96.43939393939394,
      "grad_norm": 0.038151271641254425,
      "learning_rate": 3.5606060606060608e-06,
      "loss": 0.0093,
      "step": 6365
    },
    {
      "epoch": 96.51515151515152,
      "grad_norm": 0.023309187963604927,
      "learning_rate": 3.4848484848484854e-06,
      "loss": 0.0078,
      "step": 6370
    },
    {
      "epoch": 96.5909090909091,
      "grad_norm": 0.0359482504427433,
      "learning_rate": 3.409090909090909e-06,
      "loss": 0.0078,
      "step": 6375
    },
    {
      "epoch": 96.66666666666667,
      "grad_norm": 0.03098476119339466,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0074,
      "step": 6380
    },
    {
      "epoch": 96.74242424242425,
      "grad_norm": 0.027104081586003304,
      "learning_rate": 3.257575757575758e-06,
      "loss": 0.009,
      "step": 6385
    },
    {
      "epoch": 96.81818181818181,
      "grad_norm": 0.08690879493951797,
      "learning_rate": 3.1818181818181817e-06,
      "loss": 0.0123,
      "step": 6390
    },
    {
      "epoch": 96.89393939393939,
      "grad_norm": 0.054520461708307266,
      "learning_rate": 3.106060606060606e-06,
      "loss": 0.0073,
      "step": 6395
    },
    {
      "epoch": 96.96969696969697,
      "grad_norm": 0.051536113023757935,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 0.0077,
      "step": 6400
    },
    {
      "epoch": 97.0,
      "eval_loss": 0.009941344149410725,
      "eval_mean_accuracy": 0.962475472796219,
      "eval_mean_iou": 0.9363337790342602,
      "eval_runtime": 106.2994,
      "eval_samples_per_second": 2.211,
      "eval_steps_per_second": 0.282,
      "step": 6402
    },
    {
      "epoch": 97.04545454545455,
      "grad_norm": 0.052524786442518234,
      "learning_rate": 2.9545454545454547e-06,
      "loss": 0.0077,
      "step": 6405
    },
    {
      "epoch": 97.12121212121212,
      "grad_norm": 0.017357109114527702,
      "learning_rate": 2.8787878787878793e-06,
      "loss": 0.0093,
      "step": 6410
    },
    {
      "epoch": 97.1969696969697,
      "grad_norm": 0.01797247678041458,
      "learning_rate": 2.803030303030303e-06,
      "loss": 0.0078,
      "step": 6415
    },
    {
      "epoch": 97.27272727272727,
      "grad_norm": 0.035160791128873825,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.008,
      "step": 6420
    },
    {
      "epoch": 97.34848484848484,
      "grad_norm": 0.017760787159204483,
      "learning_rate": 2.651515151515152e-06,
      "loss": 0.0101,
      "step": 6425
    },
    {
      "epoch": 97.42424242424242,
      "grad_norm": 0.014801735989749432,
      "learning_rate": 2.575757575757576e-06,
      "loss": 0.0075,
      "step": 6430
    },
    {
      "epoch": 97.5,
      "grad_norm": 0.05000997707247734,
      "learning_rate": 2.5e-06,
      "loss": 0.0071,
      "step": 6435
    },
    {
      "epoch": 97.57575757575758,
      "grad_norm": 0.029684199020266533,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 0.0064,
      "step": 6440
    },
    {
      "epoch": 97.65151515151516,
      "grad_norm": 0.07723288238048553,
      "learning_rate": 2.3484848484848486e-06,
      "loss": 0.0092,
      "step": 6445
    },
    {
      "epoch": 97.72727272727273,
      "grad_norm": 0.04107890650629997,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 0.0075,
      "step": 6450
    },
    {
      "epoch": 97.8030303030303,
      "grad_norm": 0.0177910253405571,
      "learning_rate": 2.196969696969697e-06,
      "loss": 0.0089,
      "step": 6455
    },
    {
      "epoch": 97.87878787878788,
      "grad_norm": 0.04167649894952774,
      "learning_rate": 2.1212121212121216e-06,
      "loss": 0.0084,
      "step": 6460
    },
    {
      "epoch": 97.95454545454545,
      "grad_norm": 0.06255833059549332,
      "learning_rate": 2.0454545454545457e-06,
      "loss": 0.0102,
      "step": 6465
    },
    {
      "epoch": 98.0,
      "eval_loss": 0.010090483352541924,
      "eval_mean_accuracy": 0.9589360580546592,
      "eval_mean_iou": 0.9347751680750506,
      "eval_runtime": 107.6025,
      "eval_samples_per_second": 2.184,
      "eval_steps_per_second": 0.279,
      "step": 6468
    },
    {
      "epoch": 98.03030303030303,
      "grad_norm": 0.022944649681448936,
      "learning_rate": 1.9696969696969695e-06,
      "loss": 0.0066,
      "step": 6470
    },
    {
      "epoch": 98.10606060606061,
      "grad_norm": 0.022588102146983147,
      "learning_rate": 1.8939393939393941e-06,
      "loss": 0.0092,
      "step": 6475
    },
    {
      "epoch": 98.18181818181819,
      "grad_norm": 0.07336974889039993,
      "learning_rate": 1.818181818181818e-06,
      "loss": 0.008,
      "step": 6480
    },
    {
      "epoch": 98.25757575757575,
      "grad_norm": 0.02980232983827591,
      "learning_rate": 1.7424242424242427e-06,
      "loss": 0.0065,
      "step": 6485
    },
    {
      "epoch": 98.33333333333333,
      "grad_norm": 0.042453814297914505,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0084,
      "step": 6490
    },
    {
      "epoch": 98.4090909090909,
      "grad_norm": 0.02873948961496353,
      "learning_rate": 1.5909090909090908e-06,
      "loss": 0.0064,
      "step": 6495
    },
    {
      "epoch": 98.48484848484848,
      "grad_norm": 0.08460472524166107,
      "learning_rate": 1.5151515151515152e-06,
      "loss": 0.0074,
      "step": 6500
    },
    {
      "epoch": 98.56060606060606,
      "grad_norm": 0.029787493869662285,
      "learning_rate": 1.4393939393939396e-06,
      "loss": 0.0083,
      "step": 6505
    },
    {
      "epoch": 98.63636363636364,
      "grad_norm": 0.02598843164741993,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 0.0069,
      "step": 6510
    },
    {
      "epoch": 98.71212121212122,
      "grad_norm": 0.04243359714746475,
      "learning_rate": 1.287878787878788e-06,
      "loss": 0.0077,
      "step": 6515
    },
    {
      "epoch": 98.78787878787878,
      "grad_norm": 0.020701134577393532,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 0.0097,
      "step": 6520
    },
    {
      "epoch": 98.86363636363636,
      "grad_norm": 0.06972921639680862,
      "learning_rate": 1.1363636363636364e-06,
      "loss": 0.0099,
      "step": 6525
    },
    {
      "epoch": 98.93939393939394,
      "grad_norm": 0.0856766402721405,
      "learning_rate": 1.0606060606060608e-06,
      "loss": 0.0101,
      "step": 6530
    },
    {
      "epoch": 99.0,
      "eval_loss": 0.009899638593196869,
      "eval_mean_accuracy": 0.9650524563362263,
      "eval_mean_iou": 0.9367337452205721,
      "eval_runtime": 106.413,
      "eval_samples_per_second": 2.208,
      "eval_steps_per_second": 0.282,
      "step": 6534
    },
    {
      "epoch": 99.01515151515152,
      "grad_norm": 0.03808484598994255,
      "learning_rate": 9.848484848484847e-07,
      "loss": 0.0107,
      "step": 6535
    },
    {
      "epoch": 99.0909090909091,
      "grad_norm": 0.025922929868102074,
      "learning_rate": 9.09090909090909e-07,
      "loss": 0.0067,
      "step": 6540
    },
    {
      "epoch": 99.16666666666667,
      "grad_norm": 0.0265145655721426,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0069,
      "step": 6545
    },
    {
      "epoch": 99.24242424242425,
      "grad_norm": 0.017802009359002113,
      "learning_rate": 7.575757575757576e-07,
      "loss": 0.0086,
      "step": 6550
    },
    {
      "epoch": 99.31818181818181,
      "grad_norm": 0.027775214985013008,
      "learning_rate": 6.818181818181818e-07,
      "loss": 0.0056,
      "step": 6555
    },
    {
      "epoch": 99.39393939393939,
      "grad_norm": 0.039006222039461136,
      "learning_rate": 6.060606060606061e-07,
      "loss": 0.0072,
      "step": 6560
    },
    {
      "epoch": 99.46969696969697,
      "grad_norm": 0.01686597801744938,
      "learning_rate": 5.303030303030304e-07,
      "loss": 0.008,
      "step": 6565
    },
    {
      "epoch": 99.54545454545455,
      "grad_norm": 0.039760857820510864,
      "learning_rate": 4.545454545454545e-07,
      "loss": 0.0084,
      "step": 6570
    },
    {
      "epoch": 99.62121212121212,
      "grad_norm": 0.014488129876554012,
      "learning_rate": 3.787878787878788e-07,
      "loss": 0.0082,
      "step": 6575
    },
    {
      "epoch": 99.6969696969697,
      "grad_norm": 0.02692151814699173,
      "learning_rate": 3.0303030303030305e-07,
      "loss": 0.011,
      "step": 6580
    },
    {
      "epoch": 99.77272727272727,
      "grad_norm": 0.03423842787742615,
      "learning_rate": 2.2727272727272726e-07,
      "loss": 0.0084,
      "step": 6585
    },
    {
      "epoch": 99.84848484848484,
      "grad_norm": 0.019152682274580002,
      "learning_rate": 1.5151515151515152e-07,
      "loss": 0.0092,
      "step": 6590
    },
    {
      "epoch": 99.92424242424242,
      "grad_norm": 0.0756276473402977,
      "learning_rate": 7.575757575757576e-08,
      "loss": 0.0098,
      "step": 6595
    },
    {
      "epoch": 100.0,
      "grad_norm": 0.021151289343833923,
      "learning_rate": 0.0,
      "loss": 0.0077,
      "step": 6600
    },
    {
      "epoch": 100.0,
      "eval_loss": 0.009974845685064793,
      "eval_mean_accuracy": 0.9624854642750108,
      "eval_mean_iou": 0.936197517466446,
      "eval_runtime": 106.4232,
      "eval_samples_per_second": 2.208,
      "eval_steps_per_second": 0.282,
      "step": 6600
    }
  ],
  "logging_steps": 5,
  "max_steps": 6600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}