{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.930486462689853,
  "eval_steps": 500,
  "global_step": 28000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.017609509134932862,
      "grad_norm": 48.91790771484375,
      "learning_rate": 5.989540412044374e-05,
      "loss": 9.243912963867187,
      "step": 100
    },
    {
      "epoch": 0.035219018269865725,
      "grad_norm": 10.053231239318848,
      "learning_rate": 5.978975171685156e-05,
      "loss": 3.7935104370117188,
      "step": 200
    },
    {
      "epoch": 0.05282852740479859,
      "grad_norm": 10.990548133850098,
      "learning_rate": 5.968409931325938e-05,
      "loss": 3.084054260253906,
      "step": 300
    },
    {
      "epoch": 0.07043803653973145,
      "grad_norm": 25.284229278564453,
      "learning_rate": 5.9578446909667196e-05,
      "loss": 2.5839889526367186,
      "step": 400
    },
    {
      "epoch": 0.08804754567466432,
      "grad_norm": 19.381221771240234,
      "learning_rate": 5.9472794506075014e-05,
      "loss": 2.5452081298828126,
      "step": 500
    },
    {
      "epoch": 0.08804754567466432,
      "eval_loss": 0.46036240458488464,
      "eval_mIoU": 24.719834678855246,
      "eval_mean_accuracy": 30.302184775266984,
      "eval_overall_accuracy": 93.80432756432208,
      "eval_runtime": 595.2093,
      "eval_samples_per_second": 1.94,
      "eval_steps_per_second": 0.971,
      "step": 500
    },
    {
      "epoch": 0.10565705480959718,
      "grad_norm": 30.767061233520508,
      "learning_rate": 5.936714210248283e-05,
      "loss": 2.543914794921875,
      "step": 600
    },
    {
      "epoch": 0.12326656394453005,
      "grad_norm": 12.711847305297852,
      "learning_rate": 5.926148969889066e-05,
      "loss": 2.555495300292969,
      "step": 700
    },
    {
      "epoch": 0.1408760730794629,
      "grad_norm": 9.562529563903809,
      "learning_rate": 5.915583729529847e-05,
      "loss": 2.357043914794922,
      "step": 800
    },
    {
      "epoch": 0.15848558221439576,
      "grad_norm": 7.863616943359375,
      "learning_rate": 5.905018489170629e-05,
      "loss": 2.3811045837402345,
      "step": 900
    },
    {
      "epoch": 0.17609509134932863,
      "grad_norm": 19.924156188964844,
      "learning_rate": 5.8944532488114105e-05,
      "loss": 2.2813540649414064,
      "step": 1000
    },
    {
      "epoch": 0.17609509134932863,
      "eval_loss": 0.4134902060031891,
      "eval_mIoU": 28.85459800222985,
      "eval_mean_accuracy": 36.7015683094026,
      "eval_overall_accuracy": 94.40936266085802,
      "eval_runtime": 581.5599,
      "eval_samples_per_second": 1.986,
      "eval_steps_per_second": 0.994,
      "step": 1000
    },
    {
      "epoch": 0.1937046004842615,
      "grad_norm": 9.245338439941406,
      "learning_rate": 5.883888008452192e-05,
      "loss": 2.224588165283203,
      "step": 1100
    },
    {
      "epoch": 0.21131410961919436,
      "grad_norm": 28.38551902770996,
      "learning_rate": 5.873322768092974e-05,
      "loss": 2.2024305725097655,
      "step": 1200
    },
    {
      "epoch": 0.22892361875412723,
      "grad_norm": 21.654741287231445,
      "learning_rate": 5.862757527733756e-05,
      "loss": 2.333185577392578,
      "step": 1300
    },
    {
      "epoch": 0.2465331278890601,
      "grad_norm": 22.906660079956055,
      "learning_rate": 5.8521922873745385e-05,
      "loss": 2.1109620666503908,
      "step": 1400
    },
    {
      "epoch": 0.26414263702399293,
      "grad_norm": 19.337547302246094,
      "learning_rate": 5.8416270470153196e-05,
      "loss": 2.1600161743164064,
      "step": 1500
    },
    {
      "epoch": 0.26414263702399293,
      "eval_loss": 0.41332295536994934,
      "eval_mIoU": 30.867547066481954,
      "eval_mean_accuracy": 38.46061833204942,
      "eval_overall_accuracy": 94.4356427873884,
      "eval_runtime": 582.6409,
      "eval_samples_per_second": 1.982,
      "eval_steps_per_second": 0.992,
      "step": 1500
    },
    {
      "epoch": 0.2817521461589258,
      "grad_norm": 16.490278244018555,
      "learning_rate": 5.831061806656102e-05,
      "loss": 2.0159757995605467,
      "step": 1600
    },
    {
      "epoch": 0.29936165529385866,
      "grad_norm": 17.120445251464844,
      "learning_rate": 5.820496566296883e-05,
      "loss": 2.0717059326171876,
      "step": 1700
    },
    {
      "epoch": 0.31697116442879153,
      "grad_norm": 12.258465766906738,
      "learning_rate": 5.809931325937665e-05,
      "loss": 2.131772918701172,
      "step": 1800
    },
    {
      "epoch": 0.3345806735637244,
      "grad_norm": 13.74216079711914,
      "learning_rate": 5.799366085578447e-05,
      "loss": 1.8676652526855468,
      "step": 1900
    },
    {
      "epoch": 0.35219018269865726,
      "grad_norm": 25.953657150268555,
      "learning_rate": 5.788800845219229e-05,
      "loss": 1.9515042114257812,
      "step": 2000
    },
    {
      "epoch": 0.35219018269865726,
      "eval_loss": 0.4337688982486725,
      "eval_mIoU": 32.19576589621381,
      "eval_mean_accuracy": 40.59080396049274,
      "eval_overall_accuracy": 94.54845279842228,
      "eval_runtime": 584.2063,
      "eval_samples_per_second": 1.977,
      "eval_steps_per_second": 0.989,
      "step": 2000
    },
    {
      "epoch": 0.3697996918335901,
      "grad_norm": 19.75416374206543,
      "learning_rate": 5.778235604860011e-05,
      "loss": 1.784864959716797,
      "step": 2100
    },
    {
      "epoch": 0.387409200968523,
      "grad_norm": 22.587085723876953,
      "learning_rate": 5.767670364500792e-05,
      "loss": 1.9940145874023438,
      "step": 2200
    },
    {
      "epoch": 0.40501871010345586,
      "grad_norm": 35.613121032714844,
      "learning_rate": 5.757105124141575e-05,
      "loss": 1.9241729736328126,
      "step": 2300
    },
    {
      "epoch": 0.4226282192383887,
      "grad_norm": 38.178104400634766,
      "learning_rate": 5.746539883782356e-05,
      "loss": 1.7019007873535157,
      "step": 2400
    },
    {
      "epoch": 0.4402377283733216,
      "grad_norm": 35.151058197021484,
      "learning_rate": 5.735974643423138e-05,
      "loss": 1.8641450500488281,
      "step": 2500
    },
    {
      "epoch": 0.4402377283733216,
      "eval_loss": 0.39942070841789246,
      "eval_mIoU": 33.685957807527856,
      "eval_mean_accuracy": 44.48807007726972,
      "eval_overall_accuracy": 94.60921745795709,
      "eval_runtime": 588.2746,
      "eval_samples_per_second": 1.963,
      "eval_steps_per_second": 0.983,
      "step": 2500
    },
    {
      "epoch": 0.45784723750825446,
      "grad_norm": 13.256592750549316,
      "learning_rate": 5.7254094030639196e-05,
      "loss": 1.7548785400390625,
      "step": 2600
    },
    {
      "epoch": 0.4754567466431873,
      "grad_norm": 12.395865440368652,
      "learning_rate": 5.7148441627047014e-05,
      "loss": 1.7571292114257813,
      "step": 2700
    },
    {
      "epoch": 0.4930662557781202,
      "grad_norm": 10.11026668548584,
      "learning_rate": 5.704278922345484e-05,
      "loss": 1.8158778381347656,
      "step": 2800
    },
    {
      "epoch": 0.510675764913053,
      "grad_norm": 37.68540954589844,
      "learning_rate": 5.693713681986265e-05,
      "loss": 1.95743408203125,
      "step": 2900
    },
    {
      "epoch": 0.5282852740479859,
      "grad_norm": 6.611830234527588,
      "learning_rate": 5.6831484416270475e-05,
      "loss": 1.9499478149414062,
      "step": 3000
    },
    {
      "epoch": 0.5282852740479859,
      "eval_loss": 0.3753623068332672,
      "eval_mIoU": 33.48202504810993,
      "eval_mean_accuracy": 42.84434388722273,
      "eval_overall_accuracy": 94.69698059610474,
      "eval_runtime": 592.7509,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 0.975,
      "step": 3000
    },
    {
      "epoch": 0.5458947831829187,
      "grad_norm": 11.240772247314453,
      "learning_rate": 5.6725832012678287e-05,
      "loss": 1.9529837036132813,
      "step": 3100
    },
    {
      "epoch": 0.5635042923178516,
      "grad_norm": 5.816749095916748,
      "learning_rate": 5.662017960908611e-05,
      "loss": 1.7379286193847656,
      "step": 3200
    },
    {
      "epoch": 0.5811138014527845,
      "grad_norm": 14.43251895904541,
      "learning_rate": 5.651452720549393e-05,
      "loss": 1.878860626220703,
      "step": 3300
    },
    {
      "epoch": 0.5987233105877173,
      "grad_norm": 17.320858001708984,
      "learning_rate": 5.640887480190174e-05,
      "loss": 1.7022770690917968,
      "step": 3400
    },
    {
      "epoch": 0.6163328197226502,
      "grad_norm": 8.217686653137207,
      "learning_rate": 5.6303222398309566e-05,
      "loss": 1.7487423706054688,
      "step": 3500
    },
    {
      "epoch": 0.6163328197226502,
      "eval_loss": 0.3824031949043274,
      "eval_mIoU": 35.34702178902146,
      "eval_mean_accuracy": 43.66437157324913,
      "eval_overall_accuracy": 95.11653123995958,
      "eval_runtime": 594.2993,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 0.973,
      "step": 3500
    },
    {
      "epoch": 0.6339423288575831,
      "grad_norm": 8.632674217224121,
      "learning_rate": 5.619756999471738e-05,
      "loss": 1.7975865173339844,
      "step": 3600
    },
    {
      "epoch": 0.6515518379925159,
      "grad_norm": 12.074219703674316,
      "learning_rate": 5.60919175911252e-05,
      "loss": 1.7007231140136718,
      "step": 3700
    },
    {
      "epoch": 0.6691613471274488,
      "grad_norm": 28.738109588623047,
      "learning_rate": 5.5986265187533014e-05,
      "loss": 1.7046678161621094,
      "step": 3800
    },
    {
      "epoch": 0.6867708562623817,
      "grad_norm": 17.928030014038086,
      "learning_rate": 5.588061278394084e-05,
      "loss": 1.876953125,
      "step": 3900
    },
    {
      "epoch": 0.7043803653973145,
      "grad_norm": 8.591980934143066,
      "learning_rate": 5.577496038034866e-05,
      "loss": 1.8151817321777344,
      "step": 4000
    },
    {
      "epoch": 0.7043803653973145,
      "eval_loss": 0.385084331035614,
      "eval_mIoU": 35.32089535040148,
      "eval_mean_accuracy": 44.71839496509271,
      "eval_overall_accuracy": 94.84546578807749,
      "eval_runtime": 594.7804,
      "eval_samples_per_second": 1.942,
      "eval_steps_per_second": 0.972,
      "step": 4000
    },
    {
      "epoch": 0.7219898745322474,
      "grad_norm": 28.210174560546875,
      "learning_rate": 5.566930797675647e-05,
      "loss": 1.6734780883789062,
      "step": 4100
    },
    {
      "epoch": 0.7395993836671803,
      "grad_norm": 48.83910369873047,
      "learning_rate": 5.556365557316429e-05,
      "loss": 1.6707070922851563,
      "step": 4200
    },
    {
      "epoch": 0.7572088928021131,
      "grad_norm": 40.12311935424805,
      "learning_rate": 5.5458003169572105e-05,
      "loss": 1.7087435913085938,
      "step": 4300
    },
    {
      "epoch": 0.774818401937046,
      "grad_norm": 6.389533042907715,
      "learning_rate": 5.535235076597993e-05,
      "loss": 1.7182331848144532,
      "step": 4400
    },
    {
      "epoch": 0.7924279110719789,
      "grad_norm": 9.220948219299316,
      "learning_rate": 5.524669836238774e-05,
      "loss": 1.7504454040527344,
      "step": 4500
    },
    {
      "epoch": 0.7924279110719789,
      "eval_loss": 0.3685193955898285,
      "eval_mIoU": 35.59346252124521,
      "eval_mean_accuracy": 46.61321173585393,
      "eval_overall_accuracy": 95.03028737518179,
      "eval_runtime": 596.9732,
      "eval_samples_per_second": 1.935,
      "eval_steps_per_second": 0.968,
      "step": 4500
    },
    {
      "epoch": 0.8100374202069117,
      "grad_norm": 10.630125045776367,
      "learning_rate": 5.5141045958795566e-05,
      "loss": 1.6900584411621093,
      "step": 4600
    },
    {
      "epoch": 0.8276469293418446,
      "grad_norm": 21.725378036499023,
      "learning_rate": 5.5035393555203384e-05,
      "loss": 1.5272064208984375,
      "step": 4700
    },
    {
      "epoch": 0.8452564384767774,
      "grad_norm": 34.975982666015625,
      "learning_rate": 5.49297411516112e-05,
      "loss": 1.5466600036621094,
      "step": 4800
    },
    {
      "epoch": 0.8628659476117103,
      "grad_norm": 10.243609428405762,
      "learning_rate": 5.482408874801902e-05,
      "loss": 1.7738760375976563,
      "step": 4900
    },
    {
      "epoch": 0.8804754567466432,
      "grad_norm": 21.549283981323242,
      "learning_rate": 5.471843634442683e-05,
      "loss": 1.770701904296875,
      "step": 5000
    },
    {
      "epoch": 0.8804754567466432,
      "eval_loss": 0.37507033348083496,
      "eval_mIoU": 33.710236086968045,
      "eval_mean_accuracy": 42.77628595940222,
      "eval_overall_accuracy": 94.85613901377756,
      "eval_runtime": 600.8517,
      "eval_samples_per_second": 1.922,
      "eval_steps_per_second": 0.962,
      "step": 5000
    },
    {
      "epoch": 0.898084965881576,
      "grad_norm": 4.752756595611572,
      "learning_rate": 5.461278394083466e-05,
      "loss": 1.8512596130371093,
      "step": 5100
    },
    {
      "epoch": 0.9156944750165089,
      "grad_norm": 16.470531463623047,
      "learning_rate": 5.4507131537242475e-05,
      "loss": 1.6076771545410156,
      "step": 5200
    },
    {
      "epoch": 0.9333039841514418,
      "grad_norm": 44.30925369262695,
      "learning_rate": 5.440147913365029e-05,
      "loss": 1.6375923156738281,
      "step": 5300
    },
    {
      "epoch": 0.9509134932863746,
      "grad_norm": 13.600090980529785,
      "learning_rate": 5.429582673005811e-05,
      "loss": 1.5879168701171875,
      "step": 5400
    },
    {
      "epoch": 0.9685230024213075,
      "grad_norm": 14.923215866088867,
      "learning_rate": 5.419017432646593e-05,
      "loss": 1.786272735595703,
      "step": 5500
    },
    {
      "epoch": 0.9685230024213075,
      "eval_loss": 0.3556274175643921,
      "eval_mIoU": 35.76118355021881,
      "eval_mean_accuracy": 45.60329420506218,
      "eval_overall_accuracy": 95.1163135875355,
      "eval_runtime": 610.4692,
      "eval_samples_per_second": 1.892,
      "eval_steps_per_second": 0.947,
      "step": 5500
    },
    {
      "epoch": 0.9861325115562404,
      "grad_norm": 13.075687408447266,
      "learning_rate": 5.408452192287375e-05,
      "loss": 1.6160453796386718,
      "step": 5600
    },
    {
      "epoch": 1.003697996918336,
      "grad_norm": 11.696391105651855,
      "learning_rate": 5.397886951928156e-05,
      "loss": 1.6222117614746094,
      "step": 5700
    },
    {
      "epoch": 1.0213075060532688,
      "grad_norm": 8.68483829498291,
      "learning_rate": 5.3873217115689384e-05,
      "loss": 1.2888230895996093,
      "step": 5800
    },
    {
      "epoch": 1.0389170151882017,
      "grad_norm": 11.437637329101562,
      "learning_rate": 5.37675647120972e-05,
      "loss": 1.4501405334472657,
      "step": 5900
    },
    {
      "epoch": 1.0565265243231345,
      "grad_norm": 23.864805221557617,
      "learning_rate": 5.366191230850502e-05,
      "loss": 1.4908659362792969,
      "step": 6000
    },
    {
      "epoch": 1.0565265243231345,
      "eval_loss": 0.37875378131866455,
      "eval_mIoU": 34.90903154363175,
      "eval_mean_accuracy": 45.42566443381862,
      "eval_overall_accuracy": 95.06103152320499,
      "eval_runtime": 616.9187,
      "eval_samples_per_second": 1.872,
      "eval_steps_per_second": 0.937,
      "step": 6000
    },
    {
      "epoch": 1.0741360334580674,
      "grad_norm": 5.937008857727051,
      "learning_rate": 5.355625990491284e-05,
      "loss": 1.3125173950195312,
      "step": 6100
    },
    {
      "epoch": 1.0917455425930003,
      "grad_norm": 17.99025535583496,
      "learning_rate": 5.3450607501320657e-05,
      "loss": 1.5096702575683594,
      "step": 6200
    },
    {
      "epoch": 1.1093550517279331,
      "grad_norm": 10.604606628417969,
      "learning_rate": 5.3344955097728475e-05,
      "loss": 1.3583757019042968,
      "step": 6300
    },
    {
      "epoch": 1.126964560862866,
      "grad_norm": 11.066084861755371,
      "learning_rate": 5.323930269413629e-05,
      "loss": 1.4467390441894532,
      "step": 6400
    },
    {
      "epoch": 1.1445740699977989,
      "grad_norm": 3.401655435562134,
      "learning_rate": 5.313365029054411e-05,
      "loss": 1.393478240966797,
      "step": 6500
    },
    {
      "epoch": 1.1445740699977989,
      "eval_loss": 0.37657374143600464,
      "eval_mIoU": 38.11100404001698,
      "eval_mean_accuracy": 48.92294650624167,
      "eval_overall_accuracy": 95.20334020837561,
      "eval_runtime": 638.676,
      "eval_samples_per_second": 1.808,
      "eval_steps_per_second": 0.905,
      "step": 6500
    },
    {
      "epoch": 1.1621835791327317,
      "grad_norm": 22.327383041381836,
      "learning_rate": 5.302799788695193e-05,
      "loss": 1.3139019775390626,
      "step": 6600
    },
    {
      "epoch": 1.1797930882676646,
      "grad_norm": 25.93901252746582,
      "learning_rate": 5.292234548335975e-05,
      "loss": 1.4768368530273437,
      "step": 6700
    },
    {
      "epoch": 1.1974025974025975,
      "grad_norm": 11.60429573059082,
      "learning_rate": 5.2816693079767565e-05,
      "loss": 1.3625910949707032,
      "step": 6800
    },
    {
      "epoch": 1.2150121065375303,
      "grad_norm": 4.614072799682617,
      "learning_rate": 5.2711040676175384e-05,
      "loss": 1.436598358154297,
      "step": 6900
    },
    {
      "epoch": 1.2326216156724632,
      "grad_norm": 32.9177360534668,
      "learning_rate": 5.26053882725832e-05,
      "loss": 1.3963890075683594,
      "step": 7000
    },
    {
      "epoch": 1.2326216156724632,
      "eval_loss": 0.36373773217201233,
      "eval_mIoU": 38.23142578696723,
      "eval_mean_accuracy": 48.00428340236774,
      "eval_overall_accuracy": 95.36992324895157,
      "eval_runtime": 623.8877,
      "eval_samples_per_second": 1.851,
      "eval_steps_per_second": 0.926,
      "step": 7000
    },
    {
      "epoch": 1.250231124807396,
      "grad_norm": 9.769020080566406,
      "learning_rate": 5.249973586899103e-05,
      "loss": 1.2838912963867188,
      "step": 7100
    },
    {
      "epoch": 1.267840633942329,
      "grad_norm": 11.216667175292969,
      "learning_rate": 5.239408346539884e-05,
      "loss": 1.3618894958496093,
      "step": 7200
    },
    {
      "epoch": 1.2854501430772618,
      "grad_norm": 3.4321463108062744,
      "learning_rate": 5.2288431061806656e-05,
      "loss": 1.2915570068359374,
      "step": 7300
    },
    {
      "epoch": 1.3030596522121947,
      "grad_norm": 7.617103099822998,
      "learning_rate": 5.2182778658214474e-05,
      "loss": 1.3785501098632813,
      "step": 7400
    },
    {
      "epoch": 1.3206691613471275,
      "grad_norm": 5.255311965942383,
      "learning_rate": 5.207712625462229e-05,
      "loss": 1.4688572692871094,
      "step": 7500
    },
    {
      "epoch": 1.3206691613471275,
      "eval_loss": 0.3581092953681946,
      "eval_mIoU": 36.78684041274723,
      "eval_mean_accuracy": 46.12063614366271,
      "eval_overall_accuracy": 95.14375926096203,
      "eval_runtime": 632.8318,
      "eval_samples_per_second": 1.825,
      "eval_steps_per_second": 0.913,
      "step": 7500
    },
    {
      "epoch": 1.3382786704820604,
      "grad_norm": 59.6718635559082,
      "learning_rate": 5.197147385103011e-05,
      "loss": 1.5980494689941407,
      "step": 7600
    },
    {
      "epoch": 1.355888179616993,
      "grad_norm": 40.1579475402832,
      "learning_rate": 5.186582144743793e-05,
      "loss": 1.398350830078125,
      "step": 7700
    },
    {
      "epoch": 1.3734976887519261,
      "grad_norm": 44.09864807128906,
      "learning_rate": 5.1760169043845754e-05,
      "loss": 1.4520651245117187,
      "step": 7800
    },
    {
      "epoch": 1.3911071978868588,
      "grad_norm": 7.595520496368408,
      "learning_rate": 5.1654516640253565e-05,
      "loss": 1.414281005859375,
      "step": 7900
    },
    {
      "epoch": 1.4087167070217919,
      "grad_norm": 22.861289978027344,
      "learning_rate": 5.154886423666139e-05,
      "loss": 1.3919259643554687,
      "step": 8000
    },
    {
      "epoch": 1.4087167070217919,
      "eval_loss": 0.3777351677417755,
      "eval_mIoU": 37.04354406013157,
      "eval_mean_accuracy": 49.01216587877698,
      "eval_overall_accuracy": 95.28077327843867,
      "eval_runtime": 633.5258,
      "eval_samples_per_second": 1.823,
      "eval_steps_per_second": 0.912,
      "step": 8000
    },
    {
      "epoch": 1.4263262161567245,
      "grad_norm": 17.126745223999023,
      "learning_rate": 5.14432118330692e-05,
      "loss": 1.3992294311523437,
      "step": 8100
    },
    {
      "epoch": 1.4439357252916576,
      "grad_norm": 25.60263442993164,
      "learning_rate": 5.133755942947702e-05,
      "loss": 1.5045437622070312,
      "step": 8200
    },
    {
      "epoch": 1.4615452344265902,
      "grad_norm": 39.46979904174805,
      "learning_rate": 5.123190702588484e-05,
      "loss": 1.386280517578125,
      "step": 8300
    },
    {
      "epoch": 1.4791547435615233,
      "grad_norm": 26.430803298950195,
      "learning_rate": 5.1126254622292656e-05,
      "loss": 1.395216827392578,
      "step": 8400
    },
    {
      "epoch": 1.496764252696456,
      "grad_norm": 3.6278374195098877,
      "learning_rate": 5.102060221870048e-05,
      "loss": 1.3238540649414063,
      "step": 8500
    },
    {
      "epoch": 1.496764252696456,
      "eval_loss": 0.3692784607410431,
      "eval_mIoU": 38.03621070175311,
      "eval_mean_accuracy": 49.83983428993074,
      "eval_overall_accuracy": 95.30611112520292,
      "eval_runtime": 632.3728,
      "eval_samples_per_second": 1.826,
      "eval_steps_per_second": 0.914,
      "step": 8500
    },
    {
      "epoch": 1.514373761831389,
      "grad_norm": 10.106583595275879,
      "learning_rate": 5.091494981510829e-05,
      "loss": 1.341611328125,
      "step": 8600
    },
    {
      "epoch": 1.5319832709663217,
      "grad_norm": 8.109859466552734,
      "learning_rate": 5.080929741151612e-05,
      "loss": 1.286796875,
      "step": 8700
    },
    {
      "epoch": 1.5495927801012548,
      "grad_norm": 40.51174545288086,
      "learning_rate": 5.070364500792393e-05,
      "loss": 1.4733267211914063,
      "step": 8800
    },
    {
      "epoch": 1.5672022892361874,
      "grad_norm": 8.529606819152832,
      "learning_rate": 5.0597992604331754e-05,
      "loss": 1.3715184020996094,
      "step": 8900
    },
    {
      "epoch": 1.5848117983711205,
      "grad_norm": 10.91648006439209,
      "learning_rate": 5.0492340200739565e-05,
      "loss": 1.4365826416015626,
      "step": 9000
    },
    {
      "epoch": 1.5848117983711205,
      "eval_loss": 0.3527390956878662,
      "eval_mIoU": 38.72966019649981,
      "eval_mean_accuracy": 50.33583101361378,
      "eval_overall_accuracy": 95.30509519370604,
      "eval_runtime": 622.2524,
      "eval_samples_per_second": 1.856,
      "eval_steps_per_second": 0.929,
      "step": 9000
    },
    {
      "epoch": 1.6024213075060532,
      "grad_norm": 17.285547256469727,
      "learning_rate": 5.038668779714738e-05,
      "loss": 1.298778076171875,
      "step": 9100
    },
    {
      "epoch": 1.6200308166409862,
      "grad_norm": 3.4417006969451904,
      "learning_rate": 5.028103539355521e-05,
      "loss": 1.3848799133300782,
      "step": 9200
    },
    {
      "epoch": 1.637640325775919,
      "grad_norm": 11.60331916809082,
      "learning_rate": 5.017538298996302e-05,
      "loss": 1.396776123046875,
      "step": 9300
    },
    {
      "epoch": 1.655249834910852,
      "grad_norm": 42.88685607910156,
      "learning_rate": 5.0069730586370845e-05,
      "loss": 1.5503982543945312,
      "step": 9400
    },
    {
      "epoch": 1.6728593440457846,
      "grad_norm": 10.325550079345703,
      "learning_rate": 4.9964078182778656e-05,
      "loss": 1.3557667541503906,
      "step": 9500
    },
    {
      "epoch": 1.6728593440457846,
      "eval_loss": 0.3433685302734375,
      "eval_mIoU": 37.68474411400127,
      "eval_mean_accuracy": 48.22518380563681,
      "eval_overall_accuracy": 95.38149978175308,
      "eval_runtime": 643.0868,
      "eval_samples_per_second": 1.796,
      "eval_steps_per_second": 0.899,
      "step": 9500
    },
    {
      "epoch": 1.6904688531807177,
      "grad_norm": 7.645025253295898,
      "learning_rate": 4.985842577918648e-05,
      "loss": 1.3318984985351563,
      "step": 9600
    },
    {
      "epoch": 1.7080783623156504,
      "grad_norm": 2.253293514251709,
      "learning_rate": 4.97527733755943e-05,
      "loss": 1.4566912841796875,
      "step": 9700
    },
    {
      "epoch": 1.7256878714505834,
      "grad_norm": 2.5841846466064453,
      "learning_rate": 4.964712097200211e-05,
      "loss": 1.3018017578125,
      "step": 9800
    },
    {
      "epoch": 1.743297380585516,
      "grad_norm": 5.959505081176758,
      "learning_rate": 4.9541468568409935e-05,
      "loss": 1.3685440063476562,
      "step": 9900
    },
    {
      "epoch": 1.7609068897204492,
      "grad_norm": 21.39006996154785,
      "learning_rate": 4.943581616481775e-05,
      "loss": 1.267444839477539,
      "step": 10000
    },
    {
      "epoch": 1.7609068897204492,
      "eval_loss": 0.3590288758277893,
      "eval_mIoU": 39.250987764136376,
      "eval_mean_accuracy": 51.270012242428145,
      "eval_overall_accuracy": 95.32280232483174,
      "eval_runtime": 643.0953,
      "eval_samples_per_second": 1.796,
      "eval_steps_per_second": 0.899,
      "step": 10000
    },
    {
      "epoch": 1.7785163988553818,
      "grad_norm": 64.21682739257812,
      "learning_rate": 4.933016376122557e-05,
      "loss": 1.2918376159667968,
      "step": 10100
    },
    {
      "epoch": 1.796125907990315,
      "grad_norm": 32.8283805847168,
      "learning_rate": 4.922451135763338e-05,
      "loss": 1.4452650451660156,
      "step": 10200
    },
    {
      "epoch": 1.8137354171252476,
      "grad_norm": 11.545291900634766,
      "learning_rate": 4.911885895404121e-05,
      "loss": 1.2878465270996093,
      "step": 10300
    },
    {
      "epoch": 1.8313449262601806,
      "grad_norm": 18.071176528930664,
      "learning_rate": 4.9013206550449026e-05,
      "loss": 1.3438943481445313,
      "step": 10400
    },
    {
      "epoch": 1.8489544353951133,
      "grad_norm": 7.273808479309082,
      "learning_rate": 4.8907554146856844e-05,
      "loss": 1.2591412353515625,
      "step": 10500
    },
    {
      "epoch": 1.8489544353951133,
      "eval_loss": 0.39277809858322144,
      "eval_mIoU": 36.1147332767591,
      "eval_mean_accuracy": 45.97273320824087,
      "eval_overall_accuracy": 95.28447403020157,
      "eval_runtime": 612.8948,
      "eval_samples_per_second": 1.884,
      "eval_steps_per_second": 0.943,
      "step": 10500
    },
    {
      "epoch": 1.8665639445300464,
      "grad_norm": 27.94432830810547,
      "learning_rate": 4.880190174326466e-05,
      "loss": 1.218763427734375,
      "step": 10600
    },
    {
      "epoch": 1.884173453664979,
      "grad_norm": 8.407777786254883,
      "learning_rate": 4.8696249339672474e-05,
      "loss": 1.3651194763183594,
      "step": 10700
    },
    {
      "epoch": 1.901782962799912,
      "grad_norm": 20.770652770996094,
      "learning_rate": 4.85905969360803e-05,
      "loss": 1.3555903625488281,
      "step": 10800
    },
    {
      "epoch": 1.9193924719348447,
      "grad_norm": 13.678817749023438,
      "learning_rate": 4.848494453248811e-05,
      "loss": 1.3331718444824219,
      "step": 10900
    },
    {
      "epoch": 1.9370019810697778,
      "grad_norm": 7.2792181968688965,
      "learning_rate": 4.8379292128895935e-05,
      "loss": 1.1297405242919922,
      "step": 11000
    },
    {
      "epoch": 1.9370019810697778,
      "eval_loss": 0.36189690232276917,
      "eval_mIoU": 38.97650392834806,
      "eval_mean_accuracy": 50.780002728102545,
      "eval_overall_accuracy": 95.34987445517535,
      "eval_runtime": 612.1048,
      "eval_samples_per_second": 1.887,
      "eval_steps_per_second": 0.944,
      "step": 11000
    },
    {
      "epoch": 1.9546114902047105,
      "grad_norm": 31.04173469543457,
      "learning_rate": 4.827363972530375e-05,
      "loss": 1.4935745239257812,
      "step": 11100
    },
    {
      "epoch": 1.9722209993396436,
      "grad_norm": 20.730045318603516,
      "learning_rate": 4.816798732171157e-05,
      "loss": 1.4070867919921874,
      "step": 11200
    },
    {
      "epoch": 1.9898305084745762,
      "grad_norm": 18.578184127807617,
      "learning_rate": 4.806233491811939e-05,
      "loss": 1.2718235778808593,
      "step": 11300
    },
    {
      "epoch": 2.007395993836672,
      "grad_norm": 7.441476821899414,
      "learning_rate": 4.79566825145272e-05,
      "loss": 1.2224794769287108,
      "step": 11400
    },
    {
      "epoch": 2.0250055029716045,
      "grad_norm": 7.709177494049072,
      "learning_rate": 4.7851030110935026e-05,
      "loss": 1.1453880310058593,
      "step": 11500
    },
    {
      "epoch": 2.0250055029716045,
      "eval_loss": 0.3579792082309723,
      "eval_mIoU": 39.1309522428395,
      "eval_mean_accuracy": 49.93403955617436,
      "eval_overall_accuracy": 95.3825044838381,
      "eval_runtime": 615.2836,
      "eval_samples_per_second": 1.877,
      "eval_steps_per_second": 0.939,
      "step": 11500
    },
    {
      "epoch": 2.0426150121065376,
      "grad_norm": 42.05501174926758,
      "learning_rate": 4.7745377707342844e-05,
      "loss": 1.2940916442871093,
      "step": 11600
    },
    {
      "epoch": 2.0602245212414703,
      "grad_norm": 13.565258026123047,
      "learning_rate": 4.763972530375066e-05,
      "loss": 1.138370361328125,
      "step": 11700
    },
    {
      "epoch": 2.0778340303764034,
      "grad_norm": 35.50266647338867,
      "learning_rate": 4.753407290015848e-05,
      "loss": 1.0196060180664062,
      "step": 11800
    },
    {
      "epoch": 2.095443539511336,
      "grad_norm": 19.02262306213379,
      "learning_rate": 4.74284204965663e-05,
      "loss": 1.1524028778076172,
      "step": 11900
    },
    {
      "epoch": 2.113053048646269,
      "grad_norm": 7.430933475494385,
      "learning_rate": 4.732276809297412e-05,
      "loss": 1.2204737854003906,
      "step": 12000
    },
    {
      "epoch": 2.113053048646269,
      "eval_loss": 0.38723915815353394,
      "eval_mIoU": 39.744055139390696,
      "eval_mean_accuracy": 52.037140691988846,
      "eval_overall_accuracy": 95.38114407361844,
      "eval_runtime": 614.3063,
      "eval_samples_per_second": 1.88,
      "eval_steps_per_second": 0.941,
      "step": 12000
    },
    {
      "epoch": 2.1306625577812017,
      "grad_norm": 12.622140884399414,
      "learning_rate": 4.7217115689381935e-05,
      "loss": 1.0923126220703125,
      "step": 12100
    },
    {
      "epoch": 2.148272066916135,
      "grad_norm": 26.150724411010742,
      "learning_rate": 4.711146328578975e-05,
      "loss": 1.1749604797363282,
      "step": 12200
    },
    {
      "epoch": 2.1658815760510675,
      "grad_norm": 12.140750885009766,
      "learning_rate": 4.700581088219757e-05,
      "loss": 1.1458908081054688,
      "step": 12300
    },
    {
      "epoch": 2.1834910851860005,
      "grad_norm": 5.676041126251221,
      "learning_rate": 4.690015847860539e-05,
      "loss": 1.1439741516113282,
      "step": 12400
    },
    {
      "epoch": 2.201100594320933,
      "grad_norm": 5.724509239196777,
      "learning_rate": 4.679450607501321e-05,
      "loss": 1.1061095428466796,
      "step": 12500
    },
    {
      "epoch": 2.201100594320933,
      "eval_loss": 0.3751870095729828,
      "eval_mIoU": 39.40716115137035,
      "eval_mean_accuracy": 50.12508859035146,
      "eval_overall_accuracy": 95.35122363598316,
      "eval_runtime": 613.3143,
      "eval_samples_per_second": 1.883,
      "eval_steps_per_second": 0.942,
      "step": 12500
    },
    {
      "epoch": 2.2187101034558663,
      "grad_norm": 21.661672592163086,
      "learning_rate": 4.6688853671421026e-05,
      "loss": 1.1017567443847656,
      "step": 12600
    },
    {
      "epoch": 2.236319612590799,
      "grad_norm": 3.0723938941955566,
      "learning_rate": 4.6583201267828844e-05,
      "loss": 1.097019500732422,
      "step": 12700
    },
    {
      "epoch": 2.253929121725732,
      "grad_norm": 20.003185272216797,
      "learning_rate": 4.647754886423666e-05,
      "loss": 1.0861209106445313,
      "step": 12800
    },
    {
      "epoch": 2.2715386308606647,
      "grad_norm": 17.646045684814453,
      "learning_rate": 4.637189646064448e-05,
      "loss": 1.1462979888916016,
      "step": 12900
    },
    {
      "epoch": 2.2891481399955977,
      "grad_norm": 6.879117488861084,
      "learning_rate": 4.62662440570523e-05,
      "loss": 1.2343074035644532,
      "step": 13000
    },
    {
      "epoch": 2.2891481399955977,
      "eval_loss": 0.35330045223236084,
      "eval_mIoU": 39.403422282692816,
      "eval_mean_accuracy": 50.84406228069122,
      "eval_overall_accuracy": 95.43958259351326,
      "eval_runtime": 614.1751,
      "eval_samples_per_second": 1.881,
      "eval_steps_per_second": 0.941,
      "step": 13000
    },
    {
      "epoch": 2.3067576491305304,
      "grad_norm": 83.66311645507812,
      "learning_rate": 4.616059165346012e-05,
      "loss": 1.0837049102783203,
      "step": 13100
    },
    {
      "epoch": 2.3243671582654635,
      "grad_norm": 27.105422973632812,
      "learning_rate": 4.6054939249867935e-05,
      "loss": 1.2298228454589843,
      "step": 13200
    },
    {
      "epoch": 2.341976667400396,
      "grad_norm": 7.361494541168213,
      "learning_rate": 4.594928684627575e-05,
      "loss": 1.1874884796142577,
      "step": 13300
    },
    {
      "epoch": 2.359586176535329,
      "grad_norm": 5.626405239105225,
      "learning_rate": 4.584363444268357e-05,
      "loss": 1.1711844635009765,
      "step": 13400
    },
    {
      "epoch": 2.377195685670262,
      "grad_norm": 5.629500865936279,
      "learning_rate": 4.5737982039091396e-05,
      "loss": 1.1417552947998046,
      "step": 13500
    },
    {
      "epoch": 2.377195685670262,
      "eval_loss": 0.3440675735473633,
      "eval_mIoU": 41.19118866588967,
      "eval_mean_accuracy": 52.981668891642265,
      "eval_overall_accuracy": 95.56694691315359,
      "eval_runtime": 615.9929,
      "eval_samples_per_second": 1.875,
      "eval_steps_per_second": 0.938,
      "step": 13500
    },
    {
      "epoch": 2.394805194805195,
      "grad_norm": 7.499139785766602,
      "learning_rate": 4.563232963549921e-05,
      "loss": 1.1351351165771484,
      "step": 13600
    },
    {
      "epoch": 2.4124147039401276,
      "grad_norm": 6.671785831451416,
      "learning_rate": 4.552667723190703e-05,
      "loss": 1.158738479614258,
      "step": 13700
    },
    {
      "epoch": 2.4300242130750607,
      "grad_norm": 5.020116806030273,
      "learning_rate": 4.5421024828314844e-05,
      "loss": 1.0873625183105469,
      "step": 13800
    },
    {
      "epoch": 2.4476337222099933,
      "grad_norm": 11.06270694732666,
      "learning_rate": 4.531537242472266e-05,
      "loss": 1.1327620697021485,
      "step": 13900
    },
    {
      "epoch": 2.4652432313449264,
      "grad_norm": 39.09502410888672,
      "learning_rate": 4.520972002113048e-05,
      "loss": 1.163224563598633,
      "step": 14000
    },
    {
      "epoch": 2.4652432313449264,
      "eval_loss": 0.37537702918052673,
      "eval_mIoU": 37.81560249206644,
      "eval_mean_accuracy": 47.31413884036508,
      "eval_overall_accuracy": 95.4164120232388,
      "eval_runtime": 621.4201,
      "eval_samples_per_second": 1.859,
      "eval_steps_per_second": 0.93,
      "step": 14000
    },
    {
      "epoch": 2.482852740479859,
      "grad_norm": 3.848895311355591,
      "learning_rate": 4.51040676175383e-05,
      "loss": 1.0415728759765626,
      "step": 14100
    },
    {
      "epoch": 2.500462249614792,
      "grad_norm": 4.380424976348877,
      "learning_rate": 4.499841521394612e-05,
      "loss": 1.1635037994384765,
      "step": 14200
    },
    {
      "epoch": 2.5180717587497248,
      "grad_norm": 37.84204864501953,
      "learning_rate": 4.4892762810353935e-05,
      "loss": 1.2334834289550782,
      "step": 14300
    },
    {
      "epoch": 2.535681267884658,
      "grad_norm": 7.818023204803467,
      "learning_rate": 4.478711040676176e-05,
      "loss": 1.0910034942626954,
      "step": 14400
    },
    {
      "epoch": 2.5532907770195905,
      "grad_norm": 19.556774139404297,
      "learning_rate": 4.468145800316957e-05,
      "loss": 1.1642417907714844,
      "step": 14500
    },
    {
      "epoch": 2.5532907770195905,
      "eval_loss": 0.3476269245147705,
      "eval_mIoU": 39.42816513930243,
      "eval_mean_accuracy": 50.40911446506301,
      "eval_overall_accuracy": 95.54313032142011,
      "eval_runtime": 630.7714,
      "eval_samples_per_second": 1.831,
      "eval_steps_per_second": 0.916,
      "step": 14500
    },
    {
      "epoch": 2.5709002861545236,
      "grad_norm": 38.74016189575195,
      "learning_rate": 4.457580559957739e-05,
      "loss": 1.0831634521484375,
      "step": 14600
    },
    {
      "epoch": 2.5885097952894562,
      "grad_norm": 11.159551620483398,
      "learning_rate": 4.447015319598521e-05,
      "loss": 1.1328179168701171,
      "step": 14700
    },
    {
      "epoch": 2.6061193044243893,
      "grad_norm": 7.955896854400635,
      "learning_rate": 4.4364500792393026e-05,
      "loss": 1.124356231689453,
      "step": 14800
    },
    {
      "epoch": 2.623728813559322,
      "grad_norm": 6.735036849975586,
      "learning_rate": 4.425884838880085e-05,
      "loss": 1.060822982788086,
      "step": 14900
    },
    {
      "epoch": 2.641338322694255,
      "grad_norm": 6.450618743896484,
      "learning_rate": 4.415319598520866e-05,
      "loss": 1.0960148620605468,
      "step": 15000
    },
    {
      "epoch": 2.641338322694255,
      "eval_loss": 0.3715696632862091,
      "eval_mIoU": 40.637422462526686,
      "eval_mean_accuracy": 51.30163306486706,
      "eval_overall_accuracy": 95.44940700778713,
      "eval_runtime": 632.007,
      "eval_samples_per_second": 1.828,
      "eval_steps_per_second": 0.915,
      "step": 15000
    },
    {
      "epoch": 2.6589478318291877,
      "grad_norm": 5.860879421234131,
      "learning_rate": 4.404754358161649e-05,
      "loss": 1.030592041015625,
      "step": 15100
    },
    {
      "epoch": 2.676557340964121,
      "grad_norm": 5.151623725891113,
      "learning_rate": 4.39418911780243e-05,
      "loss": 0.9981183624267578,
      "step": 15200
    },
    {
      "epoch": 2.6941668500990534,
      "grad_norm": 51.52006149291992,
      "learning_rate": 4.383623877443212e-05,
      "loss": 1.1631646728515626,
      "step": 15300
    },
    {
      "epoch": 2.711776359233986,
      "grad_norm": 1.7929599285125732,
      "learning_rate": 4.3730586370839935e-05,
      "loss": 1.034713134765625,
      "step": 15400
    },
    {
      "epoch": 2.729385868368919,
      "grad_norm": 79.84156799316406,
      "learning_rate": 4.362493396724775e-05,
      "loss": 1.1282410430908203,
      "step": 15500
    },
    {
      "epoch": 2.729385868368919,
      "eval_loss": 0.34207868576049805,
      "eval_mIoU": 41.725837022101224,
      "eval_mean_accuracy": 51.67503102984655,
      "eval_overall_accuracy": 95.68064702021611,
      "eval_runtime": 646.8847,
      "eval_samples_per_second": 1.785,
      "eval_steps_per_second": 0.894,
      "step": 15500
    },
    {
      "epoch": 2.7469953775038523,
      "grad_norm": 3.692952871322632,
      "learning_rate": 4.351928156365558e-05,
      "loss": 1.072099609375,
      "step": 15600
    },
    {
      "epoch": 2.764604886638785,
      "grad_norm": 57.948753356933594,
      "learning_rate": 4.341362916006339e-05,
      "loss": 1.1195967864990235,
      "step": 15700
    },
    {
      "epoch": 2.7822143957737175,
      "grad_norm": 8.935181617736816,
      "learning_rate": 4.3307976756471214e-05,
      "loss": 1.0987003326416016,
      "step": 15800
    },
    {
      "epoch": 2.7998239049086506,
      "grad_norm": 3.0621280670166016,
      "learning_rate": 4.3202324352879025e-05,
      "loss": 1.1626800537109374,
      "step": 15900
    },
    {
      "epoch": 2.8174334140435837,
      "grad_norm": 11.499585151672363,
      "learning_rate": 4.309667194928685e-05,
      "loss": 1.2044084930419923,
      "step": 16000
    },
    {
      "epoch": 2.8174334140435837,
      "eval_loss": 0.4090178608894348,
      "eval_mIoU": 38.788988881844546,
      "eval_mean_accuracy": 49.14422837445873,
      "eval_overall_accuracy": 95.46250347451213,
      "eval_runtime": 686.1121,
      "eval_samples_per_second": 1.683,
      "eval_steps_per_second": 0.842,
      "step": 16000
    },
    {
      "epoch": 2.8350429231785164,
      "grad_norm": 7.883459091186523,
      "learning_rate": 4.299101954569467e-05,
      "loss": 1.1145830535888672,
      "step": 16100
    },
    {
      "epoch": 2.852652432313449,
      "grad_norm": 26.763925552368164,
      "learning_rate": 4.288536714210248e-05,
      "loss": 1.1529721069335936,
      "step": 16200
    },
    {
      "epoch": 2.870261941448382,
      "grad_norm": 11.193148612976074,
      "learning_rate": 4.2779714738510305e-05,
      "loss": 1.1887353515625,
      "step": 16300
    },
    {
      "epoch": 2.887871450583315,
      "grad_norm": 47.4949836730957,
      "learning_rate": 4.2674062334918116e-05,
      "loss": 1.2108176422119141,
      "step": 16400
    },
    {
      "epoch": 2.905480959718248,
      "grad_norm": 18.99823570251465,
      "learning_rate": 4.256840993132594e-05,
      "loss": 1.0943442535400392,
      "step": 16500
    },
    {
      "epoch": 2.905480959718248,
      "eval_loss": 0.35304567217826843,
      "eval_mIoU": 41.13725855455199,
      "eval_mean_accuracy": 52.20547744211784,
      "eval_overall_accuracy": 95.69274142707064,
      "eval_runtime": 591.0741,
      "eval_samples_per_second": 1.954,
      "eval_steps_per_second": 0.978,
      "step": 16500
    },
    {
      "epoch": 2.9230904688531805,
      "grad_norm": 6.136702537536621,
      "learning_rate": 4.246275752773375e-05,
      "loss": 1.0625466156005858,
      "step": 16600
    },
    {
      "epoch": 2.9406999779881136,
      "grad_norm": 14.154541015625,
      "learning_rate": 4.235710512414158e-05,
      "loss": 1.0732533264160156,
      "step": 16700
    },
    {
      "epoch": 2.9583094871230466,
      "grad_norm": 3.0931496620178223,
      "learning_rate": 4.2251452720549396e-05,
      "loss": 1.0319291687011718,
      "step": 16800
    },
    {
      "epoch": 2.9759189962579793,
      "grad_norm": 80.85027313232422,
      "learning_rate": 4.2145800316957214e-05,
      "loss": 1.0832841491699219,
      "step": 16900
    },
    {
      "epoch": 2.993528505392912,
      "grad_norm": 3.672075033187866,
      "learning_rate": 4.204014791336503e-05,
      "loss": 1.2637503051757812,
      "step": 17000
    },
    {
      "epoch": 2.993528505392912,
      "eval_loss": 0.33654382824897766,
      "eval_mIoU": 41.44872246605363,
      "eval_mean_accuracy": 54.77046429937636,
      "eval_overall_accuracy": 95.56333863890016,
      "eval_runtime": 595.7763,
      "eval_samples_per_second": 1.939,
      "eval_steps_per_second": 0.97,
      "step": 17000
    },
    {
      "epoch": 3.0110939907550076,
      "grad_norm": 7.291171550750732,
      "learning_rate": 4.193449550977284e-05,
      "loss": 1.0205813598632814,
      "step": 17100
    },
    {
      "epoch": 3.0287034998899407,
      "grad_norm": 5.288455009460449,
      "learning_rate": 4.182884310618067e-05,
      "loss": 1.025837631225586,
      "step": 17200
    },
    {
      "epoch": 3.0463130090248733,
      "grad_norm": 2.849029302597046,
      "learning_rate": 4.172319070258848e-05,
      "loss": 0.8235783386230469,
      "step": 17300
    },
    {
      "epoch": 3.0639225181598064,
      "grad_norm": 4.3493499755859375,
      "learning_rate": 4.1617538298996305e-05,
      "loss": 0.9443639373779297,
      "step": 17400
    },
    {
      "epoch": 3.081532027294739,
      "grad_norm": 17.059249877929688,
      "learning_rate": 4.151188589540412e-05,
      "loss": 0.9543634033203126,
      "step": 17500
    },
    {
      "epoch": 3.081532027294739,
      "eval_loss": 0.3980509042739868,
      "eval_mIoU": 39.32169521997824,
      "eval_mean_accuracy": 48.75387080815462,
      "eval_overall_accuracy": 95.51444445853625,
      "eval_runtime": 595.8295,
      "eval_samples_per_second": 1.938,
      "eval_steps_per_second": 0.97,
      "step": 17500
    },
    {
      "epoch": 3.099141536429672,
      "grad_norm": 20.158220291137695,
      "learning_rate": 4.140623349181194e-05,
      "loss": 0.96615478515625,
      "step": 17600
    },
    {
      "epoch": 3.116751045564605,
      "grad_norm": 13.208799362182617,
      "learning_rate": 4.130058108821976e-05,
      "loss": 0.8107662963867187,
      "step": 17700
    },
    {
      "epoch": 3.134360554699538,
      "grad_norm": 1.974561095237732,
      "learning_rate": 4.119492868462757e-05,
      "loss": 0.8889852142333985,
      "step": 17800
    },
    {
      "epoch": 3.1519700638344705,
      "grad_norm": 19.499155044555664,
      "learning_rate": 4.1089276281035395e-05,
      "loss": 0.8983279418945312,
      "step": 17900
    },
    {
      "epoch": 3.1695795729694036,
      "grad_norm": 7.43156623840332,
      "learning_rate": 4.0983623877443214e-05,
      "loss": 0.843604736328125,
      "step": 18000
    },
    {
      "epoch": 3.1695795729694036,
      "eval_loss": 0.3665158748626709,
      "eval_mIoU": 40.75046102280888,
      "eval_mean_accuracy": 51.73736939649708,
      "eval_overall_accuracy": 95.71766147365818,
      "eval_runtime": 597.7525,
      "eval_samples_per_second": 1.932,
      "eval_steps_per_second": 0.967,
      "step": 18000
    },
    {
      "epoch": 3.1871890821043363,
      "grad_norm": 8.212807655334473,
      "learning_rate": 4.087797147385103e-05,
      "loss": 0.8693031311035156,
      "step": 18100
    },
    {
      "epoch": 3.2047985912392694,
      "grad_norm": 10.272090911865234,
      "learning_rate": 4.077231907025885e-05,
      "loss": 0.7922476196289062,
      "step": 18200
    },
    {
      "epoch": 3.222408100374202,
      "grad_norm": 9.316787719726562,
      "learning_rate": 4.066666666666667e-05,
      "loss": 0.9222467803955078,
      "step": 18300
    },
    {
      "epoch": 3.240017609509135,
      "grad_norm": 5.906984329223633,
      "learning_rate": 4.0561014263074486e-05,
      "loss": 0.933861312866211,
      "step": 18400
    },
    {
      "epoch": 3.2576271186440677,
      "grad_norm": 35.925132751464844,
      "learning_rate": 4.0455361859482304e-05,
      "loss": 1.147016830444336,
      "step": 18500
    },
    {
      "epoch": 3.2576271186440677,
      "eval_loss": 0.3637526333332062,
      "eval_mIoU": 40.01437254491868,
      "eval_mean_accuracy": 52.52752892592268,
      "eval_overall_accuracy": 95.57071008723535,
      "eval_runtime": 598.7356,
      "eval_samples_per_second": 1.929,
      "eval_steps_per_second": 0.965,
      "step": 18500
    },
    {
      "epoch": 3.275236627779001,
      "grad_norm": 44.39639663696289,
      "learning_rate": 4.034970945589012e-05,
      "loss": 0.9909713745117188,
      "step": 18600
    },
    {
      "epoch": 3.2928461369139335,
      "grad_norm": 7.367189407348633,
      "learning_rate": 4.024405705229794e-05,
      "loss": 0.899927978515625,
      "step": 18700
    },
    {
      "epoch": 3.3104556460488666,
      "grad_norm": 3.2386536598205566,
      "learning_rate": 4.013840464870576e-05,
      "loss": 0.900641860961914,
      "step": 18800
    },
    {
      "epoch": 3.328065155183799,
      "grad_norm": 5.339411735534668,
      "learning_rate": 4.003275224511358e-05,
      "loss": 0.8285768127441406,
      "step": 18900
    },
    {
      "epoch": 3.3456746643187323,
      "grad_norm": 7.77389669418335,
      "learning_rate": 3.9927099841521395e-05,
      "loss": 1.016840362548828,
      "step": 19000
    },
    {
      "epoch": 3.3456746643187323,
      "eval_loss": 0.35560256242752075,
      "eval_mIoU": 41.41963957275985,
      "eval_mean_accuracy": 52.9759184171816,
      "eval_overall_accuracy": 95.71307690112621,
      "eval_runtime": 598.8628,
      "eval_samples_per_second": 1.929,
      "eval_steps_per_second": 0.965,
      "step": 19000
    },
    {
      "epoch": 3.363284173453665,
      "grad_norm": 13.394330024719238,
      "learning_rate": 3.9821447437929213e-05,
      "loss": 0.8661341857910156,
      "step": 19100
    },
    {
      "epoch": 3.380893682588598,
      "grad_norm": 6.459191799163818,
      "learning_rate": 3.971579503433703e-05,
      "loss": 0.8622269439697265,
      "step": 19200
    },
    {
      "epoch": 3.3985031917235307,
      "grad_norm": 16.69118309020996,
      "learning_rate": 3.961014263074485e-05,
      "loss": 0.9276141357421875,
      "step": 19300
    },
    {
      "epoch": 3.4161127008584637,
      "grad_norm": 11.399658203125,
      "learning_rate": 3.9504490227152675e-05,
      "loss": 0.9722150421142578,
      "step": 19400
    },
    {
      "epoch": 3.4337222099933964,
      "grad_norm": 14.789872169494629,
      "learning_rate": 3.9398837823560486e-05,
      "loss": 0.930595703125,
      "step": 19500
    },
    {
      "epoch": 3.4337222099933964,
      "eval_loss": 0.3928566873073578,
      "eval_mIoU": 40.82311861205255,
      "eval_mean_accuracy": 51.636349706196874,
      "eval_overall_accuracy": 95.65218277307815,
      "eval_runtime": 599.9531,
      "eval_samples_per_second": 1.925,
      "eval_steps_per_second": 0.963,
      "step": 19500
    },
    {
      "epoch": 3.4513317191283295,
      "grad_norm": 47.58243942260742,
      "learning_rate": 3.9293185419968304e-05,
      "loss": 0.948566665649414,
      "step": 19600
    },
    {
      "epoch": 3.468941228263262,
      "grad_norm": 10.375826835632324,
      "learning_rate": 3.918753301637612e-05,
      "loss": 0.9331759643554688,
      "step": 19700
    },
    {
      "epoch": 3.486550737398195,
      "grad_norm": 27.57236099243164,
      "learning_rate": 3.908188061278394e-05,
      "loss": 0.9229088592529296,
      "step": 19800
    },
    {
      "epoch": 3.504160246533128,
      "grad_norm": 14.850144386291504,
      "learning_rate": 3.897622820919176e-05,
      "loss": 0.9040148162841797,
      "step": 19900
    },
    {
      "epoch": 3.5217697556680605,
      "grad_norm": 9.243043899536133,
      "learning_rate": 3.887057580559958e-05,
      "loss": 1.018120574951172,
      "step": 20000
    },
    {
      "epoch": 3.5217697556680605,
      "eval_loss": 0.3454851508140564,
      "eval_mIoU": 41.75683968881324,
      "eval_mean_accuracy": 54.63857579379155,
      "eval_overall_accuracy": 95.76613058775534,
      "eval_runtime": 613.735,
      "eval_samples_per_second": 1.882,
      "eval_steps_per_second": 0.942,
      "step": 20000
    },
    {
      "epoch": 3.5393792648029936,
      "grad_norm": 19.1420955657959,
      "learning_rate": 3.87649234020074e-05,
      "loss": 0.8436627197265625,
      "step": 20100
    },
    {
      "epoch": 3.5569887739379267,
      "grad_norm": 15.668157577514648,
      "learning_rate": 3.865927099841521e-05,
      "loss": 0.9943603515625,
      "step": 20200
    },
    {
      "epoch": 3.5745982830728593,
      "grad_norm": 8.847800254821777,
      "learning_rate": 3.855361859482303e-05,
      "loss": 0.959801254272461,
      "step": 20300
    },
    {
      "epoch": 3.592207792207792,
      "grad_norm": 14.564043045043945,
      "learning_rate": 3.844796619123085e-05,
      "loss": 0.8551026153564453,
      "step": 20400
    },
    {
      "epoch": 3.609817301342725,
      "grad_norm": 9.54574203491211,
      "learning_rate": 3.834231378763867e-05,
      "loss": 0.843503189086914,
      "step": 20500
    },
    {
      "epoch": 3.609817301342725,
      "eval_loss": 0.3464469313621521,
      "eval_mIoU": 41.94745733642343,
      "eval_mean_accuracy": 51.44057812491615,
      "eval_overall_accuracy": 95.820711474398,
      "eval_runtime": 619.4723,
      "eval_samples_per_second": 1.864,
      "eval_steps_per_second": 0.933,
      "step": 20500
    },
    {
      "epoch": 3.627426810477658,
      "grad_norm": 6.365943431854248,
      "learning_rate": 3.823666138404649e-05,
      "loss": 0.9280338287353516,
      "step": 20600
    },
    {
      "epoch": 3.645036319612591,
      "grad_norm": 49.77190017700195,
      "learning_rate": 3.8131008980454304e-05,
      "loss": 0.814898681640625,
      "step": 20700
    },
    {
      "epoch": 3.6626458287475234,
      "grad_norm": 22.503215789794922,
      "learning_rate": 3.802535657686213e-05,
      "loss": 0.9596369171142578,
      "step": 20800
    },
    {
      "epoch": 3.6802553378824565,
      "grad_norm": 3.472125291824341,
      "learning_rate": 3.791970417326994e-05,
      "loss": 0.7934391021728515,
      "step": 20900
    },
    {
      "epoch": 3.6978648470173896,
      "grad_norm": 7.981621265411377,
      "learning_rate": 3.7814051769677765e-05,
      "loss": 0.9479257202148438,
      "step": 21000
    },
    {
      "epoch": 3.6978648470173896,
      "eval_loss": 0.3813023269176483,
      "eval_mIoU": 40.63642183599385,
      "eval_mean_accuracy": 54.05427297237617,
      "eval_overall_accuracy": 95.5580978063278,
      "eval_runtime": 615.8201,
      "eval_samples_per_second": 1.876,
      "eval_steps_per_second": 0.939,
      "step": 21000
    },
    {
      "epoch": 3.7154743561523222,
      "grad_norm": 28.353187561035156,
      "learning_rate": 3.770839936608558e-05,
      "loss": 0.9741096496582031,
      "step": 21100
    },
    {
      "epoch": 3.733083865287255,
      "grad_norm": 15.872313499450684,
      "learning_rate": 3.7602746962493395e-05,
      "loss": 0.933817138671875,
      "step": 21200
    },
    {
      "epoch": 3.750693374422188,
      "grad_norm": 9.873473167419434,
      "learning_rate": 3.749709455890122e-05,
      "loss": 0.9153755187988282,
      "step": 21300
    },
    {
      "epoch": 3.768302883557121,
      "grad_norm": 3.8525328636169434,
      "learning_rate": 3.739144215530903e-05,
      "loss": 0.8332781219482421,
      "step": 21400
    },
    {
      "epoch": 3.7859123926920537,
      "grad_norm": 7.059874534606934,
      "learning_rate": 3.7285789751716856e-05,
      "loss": 0.8763603210449219,
      "step": 21500
    },
    {
      "epoch": 3.7859123926920537,
      "eval_loss": 0.3823384940624237,
      "eval_mIoU": 42.07849087378942,
      "eval_mean_accuracy": 53.026848381662504,
      "eval_overall_accuracy": 95.76488247165432,
      "eval_runtime": 611.8506,
      "eval_samples_per_second": 1.888,
      "eval_steps_per_second": 0.945,
      "step": 21500
    },
    {
      "epoch": 3.8035219018269864,
      "grad_norm": 3.2705864906311035,
      "learning_rate": 3.718013734812467e-05,
      "loss": 0.957579345703125,
      "step": 21600
    },
    {
      "epoch": 3.8211314109619194,
      "grad_norm": 2.298734664916992,
      "learning_rate": 3.707448494453249e-05,
      "loss": 0.939592514038086,
      "step": 21700
    },
    {
      "epoch": 3.8387409200968525,
      "grad_norm": 9.922309875488281,
      "learning_rate": 3.6968832540940304e-05,
      "loss": 0.9532173156738282,
      "step": 21800
    },
    {
      "epoch": 3.856350429231785,
      "grad_norm": 65.7906494140625,
      "learning_rate": 3.686318013734812e-05,
      "loss": 0.9110845184326172,
      "step": 21900
    },
    {
      "epoch": 3.873959938366718,
      "grad_norm": 5.08626127243042,
      "learning_rate": 3.675752773375595e-05,
      "loss": 0.9018997192382813,
      "step": 22000
    },
    {
      "epoch": 3.873959938366718,
      "eval_loss": 0.36784595251083374,
      "eval_mIoU": 40.651954450635344,
      "eval_mean_accuracy": 51.59310256241124,
      "eval_overall_accuracy": 95.59825715564546,
      "eval_runtime": 620.1319,
      "eval_samples_per_second": 1.863,
      "eval_steps_per_second": 0.932,
      "step": 22000
    },
    {
      "epoch": 3.891569447501651,
      "grad_norm": 3.0670690536499023,
      "learning_rate": 3.665187533016376e-05,
      "loss": 1.059849166870117,
      "step": 22100
    },
    {
      "epoch": 3.909178956636584,
      "grad_norm": 23.933759689331055,
      "learning_rate": 3.654622292657158e-05,
      "loss": 0.8858226013183593,
      "step": 22200
    },
    {
      "epoch": 3.9267884657715166,
      "grad_norm": 2.215423345565796,
      "learning_rate": 3.6440570522979395e-05,
      "loss": 0.9452084350585938,
      "step": 22300
    },
    {
      "epoch": 3.9443979749064493,
      "grad_norm": 309.3457336425781,
      "learning_rate": 3.633491811938722e-05,
      "loss": 0.9484812927246093,
      "step": 22400
    },
    {
      "epoch": 3.9620074840413824,
      "grad_norm": 2.219196319580078,
      "learning_rate": 3.622926571579504e-05,
      "loss": 1.0205068969726563,
      "step": 22500
    },
    {
      "epoch": 3.9620074840413824,
      "eval_loss": 0.3499414622783661,
      "eval_mIoU": 40.56537017455844,
      "eval_mean_accuracy": 52.12166106488223,
      "eval_overall_accuracy": 95.7546244039164,
      "eval_runtime": 612.0467,
      "eval_samples_per_second": 1.887,
      "eval_steps_per_second": 0.944,
      "step": 22500
    },
    {
      "epoch": 3.9796169931763155,
      "grad_norm": 24.519412994384766,
      "learning_rate": 3.6123613312202856e-05,
      "loss": 0.9474005889892578,
      "step": 22600
    },
    {
      "epoch": 3.997226502311248,
      "grad_norm": 6.02849817276001,
      "learning_rate": 3.6017960908610674e-05,
      "loss": 0.9866950225830078,
      "step": 22700
    },
    {
      "epoch": 4.014791987673344,
      "grad_norm": 10.79825210571289,
      "learning_rate": 3.5912308505018486e-05,
      "loss": 0.7545838928222657,
      "step": 22800
    },
    {
      "epoch": 4.032401496808276,
      "grad_norm": 8.403230667114258,
      "learning_rate": 3.580665610142631e-05,
      "loss": 0.8569242095947266,
      "step": 22900
    },
    {
      "epoch": 4.050011005943209,
      "grad_norm": 7.723455905914307,
      "learning_rate": 3.570100369783412e-05,
      "loss": 0.6824626922607422,
      "step": 23000
    },
    {
      "epoch": 4.050011005943209,
      "eval_loss": 0.39329513907432556,
      "eval_mIoU": 41.46868879356731,
      "eval_mean_accuracy": 51.606976880689324,
      "eval_overall_accuracy": 95.72947019106381,
      "eval_runtime": 613.8892,
      "eval_samples_per_second": 1.881,
      "eval_steps_per_second": 0.942,
      "step": 23000
    },
    {
      "epoch": 4.067620515078143,
      "grad_norm": 96.14293670654297,
      "learning_rate": 3.559535129424195e-05,
      "loss": 0.7759761810302734,
      "step": 23100
    },
    {
      "epoch": 4.085230024213075,
      "grad_norm": 19.32292938232422,
      "learning_rate": 3.5489698890649765e-05,
      "loss": 0.7245157623291015,
      "step": 23200
    },
    {
      "epoch": 4.102839533348008,
      "grad_norm": 4.930475234985352,
      "learning_rate": 3.538404648705758e-05,
      "loss": 0.7503673553466796,
      "step": 23300
    },
    {
      "epoch": 4.1204490424829405,
      "grad_norm": 8.997782707214355,
      "learning_rate": 3.52783940834654e-05,
      "loss": 0.7232845306396485,
      "step": 23400
    },
    {
      "epoch": 4.138058551617874,
      "grad_norm": 2.9045639038085938,
      "learning_rate": 3.517274167987321e-05,
      "loss": 0.7996186065673828,
      "step": 23500
    },
    {
      "epoch": 4.138058551617874,
      "eval_loss": 0.39504244923591614,
      "eval_mIoU": 41.049576590313016,
      "eval_mean_accuracy": 53.10629746092167,
      "eval_overall_accuracy": 95.63078711043188,
      "eval_runtime": 622.9242,
      "eval_samples_per_second": 1.854,
      "eval_steps_per_second": 0.928,
      "step": 23500
    },
    {
      "epoch": 4.155668060752807,
      "grad_norm": 21.821067810058594,
      "learning_rate": 3.506708927628104e-05,
      "loss": 0.7994131469726562,
      "step": 23600
    },
    {
      "epoch": 4.173277569887739,
      "grad_norm": 2.4144773483276367,
      "learning_rate": 3.496143687268885e-05,
      "loss": 0.7240392303466797,
      "step": 23700
    },
    {
      "epoch": 4.190887079022672,
      "grad_norm": 5.613311290740967,
      "learning_rate": 3.4855784469096674e-05,
      "loss": 0.7988972473144531,
      "step": 23800
    },
    {
      "epoch": 4.2084965881576055,
      "grad_norm": 2.1184654235839844,
      "learning_rate": 3.475013206550449e-05,
      "loss": 0.6988441467285156,
      "step": 23900
    },
    {
      "epoch": 4.226106097292538,
      "grad_norm": 3.8888204097747803,
      "learning_rate": 3.464447966191231e-05,
      "loss": 0.8608739471435547,
      "step": 24000
    },
    {
      "epoch": 4.226106097292538,
      "eval_loss": 0.4150463938713074,
      "eval_mIoU": 41.66218151688048,
      "eval_mean_accuracy": 52.747956293795106,
      "eval_overall_accuracy": 95.68294574688008,
      "eval_runtime": 615.4286,
      "eval_samples_per_second": 1.877,
      "eval_steps_per_second": 0.939,
      "step": 24000
    },
    {
      "epoch": 4.243715606427471,
      "grad_norm": 1.9162812232971191,
      "learning_rate": 3.453882725832013e-05,
      "loss": 0.7657085418701172,
      "step": 24100
    },
    {
      "epoch": 4.2613251155624035,
      "grad_norm": 11.44760513305664,
      "learning_rate": 3.4433174854727947e-05,
      "loss": 0.7565657043457031,
      "step": 24200
    },
    {
      "epoch": 4.278934624697337,
      "grad_norm": 11.871203422546387,
      "learning_rate": 3.4327522451135765e-05,
      "loss": 0.7799417114257813,
      "step": 24300
    },
    {
      "epoch": 4.29654413383227,
      "grad_norm": 13.05679988861084,
      "learning_rate": 3.422187004754358e-05,
      "loss": 0.7182006072998047,
      "step": 24400
    },
    {
      "epoch": 4.314153642967202,
      "grad_norm": 17.702442169189453,
      "learning_rate": 3.41162176439514e-05,
      "loss": 0.721866455078125,
      "step": 24500
    },
    {
      "epoch": 4.314153642967202,
      "eval_loss": 0.4177318215370178,
      "eval_mIoU": 41.28552847096313,
      "eval_mean_accuracy": 52.55204362802653,
      "eval_overall_accuracy": 95.7574832800663,
      "eval_runtime": 615.2573,
      "eval_samples_per_second": 1.877,
      "eval_steps_per_second": 0.939,
      "step": 24500
    },
    {
      "epoch": 4.331763152102135,
      "grad_norm": 4.1911492347717285,
      "learning_rate": 3.401056524035922e-05,
      "loss": 0.803135986328125,
      "step": 24600
    },
    {
      "epoch": 4.349372661237068,
      "grad_norm": 4.093879222869873,
      "learning_rate": 3.390491283676704e-05,
      "loss": 0.8821774291992187,
      "step": 24700
    },
    {
      "epoch": 4.366982170372001,
      "grad_norm": 7.071560859680176,
      "learning_rate": 3.3799260433174856e-05,
      "loss": 0.8138728332519531,
      "step": 24800
    },
    {
      "epoch": 4.384591679506934,
      "grad_norm": 6.610219478607178,
      "learning_rate": 3.3693608029582674e-05,
      "loss": 0.7457758331298828,
      "step": 24900
    },
    {
      "epoch": 4.402201188641866,
      "grad_norm": 9.474441528320312,
      "learning_rate": 3.358795562599049e-05,
      "loss": 0.7298115539550781,
      "step": 25000
    },
    {
      "epoch": 4.402201188641866,
      "eval_loss": 0.38879069685935974,
      "eval_mIoU": 40.43565637669389,
      "eval_mean_accuracy": 51.68485905246275,
      "eval_overall_accuracy": 95.56142798749916,
      "eval_runtime": 615.8924,
      "eval_samples_per_second": 1.875,
      "eval_steps_per_second": 0.938,
      "step": 25000
    },
    {
      "epoch": 4.4198106977768,
      "grad_norm": 16.882156372070312,
      "learning_rate": 3.348230322239831e-05,
      "loss": 0.7813544464111328,
      "step": 25100
    },
    {
      "epoch": 4.437420206911733,
      "grad_norm": 6.276833534240723,
      "learning_rate": 3.337665081880613e-05,
      "loss": 0.8473509216308593,
      "step": 25200
    },
    {
      "epoch": 4.455029716046665,
      "grad_norm": 16.890676498413086,
      "learning_rate": 3.3270998415213946e-05,
      "loss": 0.7772435760498047,
      "step": 25300
    },
    {
      "epoch": 4.472639225181598,
      "grad_norm": 35.11220932006836,
      "learning_rate": 3.3165346011621765e-05,
      "loss": 0.7341469573974609,
      "step": 25400
    },
    {
      "epoch": 4.4902487343165305,
      "grad_norm": 4.116140365600586,
      "learning_rate": 3.305969360802958e-05,
      "loss": 0.8031134033203124,
      "step": 25500
    },
    {
      "epoch": 4.4902487343165305,
      "eval_loss": 0.4111625552177429,
      "eval_mIoU": 40.938853369665615,
      "eval_mean_accuracy": 52.64935812387781,
      "eval_overall_accuracy": 95.67260973381273,
      "eval_runtime": 617.4648,
      "eval_samples_per_second": 1.871,
      "eval_steps_per_second": 0.936,
      "step": 25500
    },
    {
      "epoch": 4.507858243451464,
      "grad_norm": 19.544837951660156,
      "learning_rate": 3.29540412044374e-05,
      "loss": 0.6685540008544922,
      "step": 25600
    },
    {
      "epoch": 4.525467752586397,
      "grad_norm": 26.763763427734375,
      "learning_rate": 3.284838880084522e-05,
      "loss": 0.7029299926757813,
      "step": 25700
    },
    {
      "epoch": 4.543077261721329,
      "grad_norm": 11.457063674926758,
      "learning_rate": 3.2742736397253044e-05,
      "loss": 0.7187101745605469,
      "step": 25800
    },
    {
      "epoch": 4.560686770856263,
      "grad_norm": 4.882247447967529,
      "learning_rate": 3.2637083993660855e-05,
      "loss": 0.8353469085693359,
      "step": 25900
    },
    {
      "epoch": 4.5782962799911955,
      "grad_norm": 7.698673248291016,
      "learning_rate": 3.2531431590068674e-05,
      "loss": 0.7884654998779297,
      "step": 26000
    },
    {
      "epoch": 4.5782962799911955,
      "eval_loss": 0.4020868241786957,
      "eval_mIoU": 41.061296108771664,
      "eval_mean_accuracy": 52.12832238084668,
      "eval_overall_accuracy": 95.63389468502689,
      "eval_runtime": 615.7111,
      "eval_samples_per_second": 1.876,
      "eval_steps_per_second": 0.939,
      "step": 26000
    },
    {
      "epoch": 4.595905789126128,
      "grad_norm": 27.729001998901367,
      "learning_rate": 3.242577918647649e-05,
      "loss": 0.7050986480712891,
      "step": 26100
    },
    {
      "epoch": 4.613515298261061,
      "grad_norm": 4.676613807678223,
      "learning_rate": 3.232012678288431e-05,
      "loss": 0.8271269989013672,
      "step": 26200
    },
    {
      "epoch": 4.631124807395993,
      "grad_norm": 9.141484260559082,
      "learning_rate": 3.221447437929213e-05,
      "loss": 0.8618730163574219,
      "step": 26300
    },
    {
      "epoch": 4.648734316530927,
      "grad_norm": 22.31208610534668,
      "learning_rate": 3.2108821975699946e-05,
      "loss": 0.9146401977539063,
      "step": 26400
    },
    {
      "epoch": 4.66634382566586,
      "grad_norm": 5.31740665435791,
      "learning_rate": 3.200316957210777e-05,
      "loss": 0.7881410217285156,
      "step": 26500
    },
    {
      "epoch": 4.66634382566586,
      "eval_loss": 0.3889371454715729,
      "eval_mIoU": 39.90579024702969,
      "eval_mean_accuracy": 49.79534666070949,
      "eval_overall_accuracy": 95.63201805213829,
      "eval_runtime": 601.1009,
      "eval_samples_per_second": 1.921,
      "eval_steps_per_second": 0.962,
      "step": 26500
    },
    {
      "epoch": 4.683953334800792,
      "grad_norm": 96.29251098632812,
      "learning_rate": 3.189751716851558e-05,
      "loss": 0.7128916931152344,
      "step": 26600
    },
    {
      "epoch": 4.701562843935726,
      "grad_norm": 10.139142036437988,
      "learning_rate": 3.17918647649234e-05,
      "loss": 0.7628507232666015,
      "step": 26700
    },
    {
      "epoch": 4.719172353070658,
      "grad_norm": 3.24714732170105,
      "learning_rate": 3.168621236133122e-05,
      "loss": 0.8892244720458984,
      "step": 26800
    },
    {
      "epoch": 4.736781862205591,
      "grad_norm": 5.197122097015381,
      "learning_rate": 3.158055995773904e-05,
      "loss": 0.7311017608642578,
      "step": 26900
    },
    {
      "epoch": 4.754391371340524,
      "grad_norm": 33.49533462524414,
      "learning_rate": 3.147490755414686e-05,
      "loss": 0.7185327911376953,
      "step": 27000
    },
    {
      "epoch": 4.754391371340524,
      "eval_loss": 0.423967570066452,
      "eval_mIoU": 41.53527504353472,
      "eval_mean_accuracy": 52.98828510157863,
      "eval_overall_accuracy": 95.73358940355705,
      "eval_runtime": 593.0491,
      "eval_samples_per_second": 1.948,
      "eval_steps_per_second": 0.975,
      "step": 27000
    },
    {
      "epoch": 4.772000880475456,
      "grad_norm": 42.050235748291016,
      "learning_rate": 3.1369255150554673e-05,
      "loss": 0.748544921875,
      "step": 27100
    },
    {
      "epoch": 4.78961038961039,
      "grad_norm": 12.394989013671875,
      "learning_rate": 3.12636027469625e-05,
      "loss": 0.7695165252685547,
      "step": 27200
    },
    {
      "epoch": 4.8072198987453225,
      "grad_norm": 8.161137580871582,
      "learning_rate": 3.115795034337031e-05,
      "loss": 0.7662505340576172,
      "step": 27300
    },
    {
      "epoch": 4.824829407880255,
      "grad_norm": 5.988201141357422,
      "learning_rate": 3.1052297939778135e-05,
      "loss": 0.6720768737792969,
      "step": 27400
    },
    {
      "epoch": 4.842438917015189,
      "grad_norm": 15.133417129516602,
      "learning_rate": 3.0946645536185946e-05,
      "loss": 0.851641845703125,
      "step": 27500
    },
    {
      "epoch": 4.842438917015189,
      "eval_loss": 0.36105746030807495,
      "eval_mIoU": 42.808893326574484,
      "eval_mean_accuracy": 53.29861054528663,
      "eval_overall_accuracy": 95.81305796965893,
      "eval_runtime": 592.1747,
      "eval_samples_per_second": 1.95,
      "eval_steps_per_second": 0.976,
      "step": 27500
    },
    {
      "epoch": 4.860048426150121,
      "grad_norm": 11.938694953918457,
      "learning_rate": 3.0840993132593764e-05,
      "loss": 0.7331785583496093,
      "step": 27600
    },
    {
      "epoch": 4.877657935285054,
      "grad_norm": 6.203324794769287,
      "learning_rate": 3.073534072900159e-05,
      "loss": 0.7468854522705078,
      "step": 27700
    },
    {
      "epoch": 4.895267444419987,
      "grad_norm": 18.217580795288086,
      "learning_rate": 3.06296883254094e-05,
      "loss": 0.7335443115234375,
      "step": 27800
    },
    {
      "epoch": 4.912876953554919,
      "grad_norm": 21.941572189331055,
      "learning_rate": 3.0524035921817226e-05,
      "loss": 0.8398058319091797,
      "step": 27900
    },
    {
      "epoch": 4.930486462689853,
      "grad_norm": 9.89223861694336,
      "learning_rate": 3.041838351822504e-05,
      "loss": 0.8591268157958984,
      "step": 28000
    },
    {
      "epoch": 4.930486462689853,
      "eval_loss": 0.37627097964286804,
      "eval_mIoU": 42.55151434436105,
      "eval_mean_accuracy": 53.672845497296365,
      "eval_overall_accuracy": 95.75246142102526,
      "eval_runtime": 598.4996,
      "eval_samples_per_second": 1.93,
      "eval_steps_per_second": 0.966,
      "step": 28000
    }
  ],
  "logging_steps": 100,
  "max_steps": 56790,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.9002130312914797e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}