{
  "best_metric": 1.8335492610931396,
  "best_model_checkpoint": "miner_id_24/checkpoint-500",
  "epoch": 0.7215007215007215,
  "eval_steps": 100,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001443001443001443,
      "grad_norm": 0.21467295289039612,
      "learning_rate": 5e-06,
      "loss": 2.0025,
      "step": 1
    },
    {
      "epoch": 0.001443001443001443,
      "eval_loss": 2.499330997467041,
      "eval_runtime": 155.4624,
      "eval_samples_per_second": 7.507,
      "eval_steps_per_second": 1.878,
      "step": 1
    },
    {
      "epoch": 0.002886002886002886,
      "grad_norm": 0.31256306171417236,
      "learning_rate": 1e-05,
      "loss": 1.821,
      "step": 2
    },
    {
      "epoch": 0.004329004329004329,
      "grad_norm": 0.3174012303352356,
      "learning_rate": 1.5e-05,
      "loss": 1.9903,
      "step": 3
    },
    {
      "epoch": 0.005772005772005772,
      "grad_norm": 0.3661087155342102,
      "learning_rate": 2e-05,
      "loss": 2.0412,
      "step": 4
    },
    {
      "epoch": 0.007215007215007215,
      "grad_norm": 0.34538447856903076,
      "learning_rate": 2.5e-05,
      "loss": 1.9811,
      "step": 5
    },
    {
      "epoch": 0.008658008658008658,
      "grad_norm": 0.3496937155723572,
      "learning_rate": 3e-05,
      "loss": 2.1532,
      "step": 6
    },
    {
      "epoch": 0.010101010101010102,
      "grad_norm": 0.3897201418876648,
      "learning_rate": 3.5e-05,
      "loss": 2.0584,
      "step": 7
    },
    {
      "epoch": 0.011544011544011544,
      "grad_norm": 0.359721839427948,
      "learning_rate": 4e-05,
      "loss": 2.0975,
      "step": 8
    },
    {
      "epoch": 0.012987012987012988,
      "grad_norm": 0.368932843208313,
      "learning_rate": 4.5e-05,
      "loss": 2.0563,
      "step": 9
    },
    {
      "epoch": 0.01443001443001443,
      "grad_norm": 0.46016350388526917,
      "learning_rate": 5e-05,
      "loss": 1.9935,
      "step": 10
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 0.49919214844703674,
      "learning_rate": 5.500000000000001e-05,
      "loss": 2.0946,
      "step": 11
    },
    {
      "epoch": 0.017316017316017316,
      "grad_norm": 0.4046800136566162,
      "learning_rate": 6e-05,
      "loss": 2.2127,
      "step": 12
    },
    {
      "epoch": 0.01875901875901876,
      "grad_norm": 0.40110981464385986,
      "learning_rate": 6.500000000000001e-05,
      "loss": 2.0404,
      "step": 13
    },
    {
      "epoch": 0.020202020202020204,
      "grad_norm": 0.4125487506389618,
      "learning_rate": 7e-05,
      "loss": 2.0546,
      "step": 14
    },
    {
      "epoch": 0.021645021645021644,
      "grad_norm": 0.3947140872478485,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.164,
      "step": 15
    },
    {
      "epoch": 0.023088023088023088,
      "grad_norm": 0.4024007022380829,
      "learning_rate": 8e-05,
      "loss": 2.225,
      "step": 16
    },
    {
      "epoch": 0.024531024531024532,
      "grad_norm": 0.4380939304828644,
      "learning_rate": 8.5e-05,
      "loss": 2.128,
      "step": 17
    },
    {
      "epoch": 0.025974025974025976,
      "grad_norm": 0.409138947725296,
      "learning_rate": 9e-05,
      "loss": 2.0402,
      "step": 18
    },
    {
      "epoch": 0.027417027417027416,
      "grad_norm": 0.5016234517097473,
      "learning_rate": 9.5e-05,
      "loss": 2.0481,
      "step": 19
    },
    {
      "epoch": 0.02886002886002886,
      "grad_norm": 0.4870470464229584,
      "learning_rate": 0.0001,
      "loss": 2.1388,
      "step": 20
    },
    {
      "epoch": 0.030303030303030304,
      "grad_norm": 0.46186015009880066,
      "learning_rate": 9.999892908320647e-05,
      "loss": 1.8981,
      "step": 21
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 0.4449887275695801,
      "learning_rate": 9.999571637870036e-05,
      "loss": 1.959,
      "step": 22
    },
    {
      "epoch": 0.03318903318903319,
      "grad_norm": 0.4438563585281372,
      "learning_rate": 9.999036202410325e-05,
      "loss": 2.0846,
      "step": 23
    },
    {
      "epoch": 0.03463203463203463,
      "grad_norm": 0.48415109515190125,
      "learning_rate": 9.998286624877786e-05,
      "loss": 1.6974,
      "step": 24
    },
    {
      "epoch": 0.03607503607503607,
      "grad_norm": 0.44993868470191956,
      "learning_rate": 9.997322937381829e-05,
      "loss": 2.21,
      "step": 25
    },
    {
      "epoch": 0.03751803751803752,
      "grad_norm": 0.49077942967414856,
      "learning_rate": 9.996145181203615e-05,
      "loss": 1.9039,
      "step": 26
    },
    {
      "epoch": 0.03896103896103896,
      "grad_norm": 0.4759511947631836,
      "learning_rate": 9.994753406794301e-05,
      "loss": 2.018,
      "step": 27
    },
    {
      "epoch": 0.04040404040404041,
      "grad_norm": 0.4197351634502411,
      "learning_rate": 9.99314767377287e-05,
      "loss": 1.866,
      "step": 28
    },
    {
      "epoch": 0.04184704184704185,
      "grad_norm": 0.43352821469306946,
      "learning_rate": 9.991328050923581e-05,
      "loss": 1.8513,
      "step": 29
    },
    {
      "epoch": 0.04329004329004329,
      "grad_norm": 0.4750743508338928,
      "learning_rate": 9.989294616193017e-05,
      "loss": 2.0907,
      "step": 30
    },
    {
      "epoch": 0.044733044733044736,
      "grad_norm": 0.48469865322113037,
      "learning_rate": 9.98704745668676e-05,
      "loss": 1.833,
      "step": 31
    },
    {
      "epoch": 0.046176046176046176,
      "grad_norm": 0.47763270139694214,
      "learning_rate": 9.98458666866564e-05,
      "loss": 2.1126,
      "step": 32
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.4813288748264313,
      "learning_rate": 9.981912357541627e-05,
      "loss": 2.1389,
      "step": 33
    },
    {
      "epoch": 0.049062049062049064,
      "grad_norm": 0.5742578506469727,
      "learning_rate": 9.97902463787331e-05,
      "loss": 1.9601,
      "step": 34
    },
    {
      "epoch": 0.050505050505050504,
      "grad_norm": 0.5102812051773071,
      "learning_rate": 9.975923633360985e-05,
      "loss": 2.0476,
      "step": 35
    },
    {
      "epoch": 0.05194805194805195,
      "grad_norm": 0.5013228058815002,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.8781,
      "step": 36
    },
    {
      "epoch": 0.05339105339105339,
      "grad_norm": 0.5211206078529358,
      "learning_rate": 9.969082310281891e-05,
      "loss": 2.0521,
      "step": 37
    },
    {
      "epoch": 0.05483405483405483,
      "grad_norm": 0.6096839904785156,
      "learning_rate": 9.965342284774632e-05,
      "loss": 1.7361,
      "step": 38
    },
    {
      "epoch": 0.05627705627705628,
      "grad_norm": 0.6038843393325806,
      "learning_rate": 9.961389560529836e-05,
      "loss": 1.8981,
      "step": 39
    },
    {
      "epoch": 0.05772005772005772,
      "grad_norm": 0.5440136790275574,
      "learning_rate": 9.957224306869053e-05,
      "loss": 1.8275,
      "step": 40
    },
    {
      "epoch": 0.05916305916305916,
      "grad_norm": 0.8856310844421387,
      "learning_rate": 9.952846702217886e-05,
      "loss": 2.063,
      "step": 41
    },
    {
      "epoch": 0.06060606060606061,
      "grad_norm": 0.8176852464675903,
      "learning_rate": 9.948256934098352e-05,
      "loss": 1.6802,
      "step": 42
    },
    {
      "epoch": 0.06204906204906205,
      "grad_norm": 0.8885481953620911,
      "learning_rate": 9.943455199120837e-05,
      "loss": 1.8003,
      "step": 43
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 0.746778666973114,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.985,
      "step": 44
    },
    {
      "epoch": 0.06493506493506493,
      "grad_norm": 0.7613317966461182,
      "learning_rate": 9.933216660424395e-05,
      "loss": 1.8953,
      "step": 45
    },
    {
      "epoch": 0.06637806637806638,
      "grad_norm": 0.9400774836540222,
      "learning_rate": 9.927780295290389e-05,
      "loss": 1.919,
      "step": 46
    },
    {
      "epoch": 0.06782106782106782,
      "grad_norm": 0.8416843414306641,
      "learning_rate": 9.922132840449459e-05,
      "loss": 1.8115,
      "step": 47
    },
    {
      "epoch": 0.06926406926406926,
      "grad_norm": 0.8849237561225891,
      "learning_rate": 9.916274537819775e-05,
      "loss": 1.9293,
      "step": 48
    },
    {
      "epoch": 0.0707070707070707,
      "grad_norm": 0.9754048585891724,
      "learning_rate": 9.91020563835152e-05,
      "loss": 1.7725,
      "step": 49
    },
    {
      "epoch": 0.07215007215007214,
      "grad_norm": 1.8871734142303467,
      "learning_rate": 9.903926402016153e-05,
      "loss": 1.8307,
      "step": 50
    },
    {
      "epoch": 0.0735930735930736,
      "grad_norm": 0.7267559766769409,
      "learning_rate": 9.897437097795257e-05,
      "loss": 2.0867,
      "step": 51
    },
    {
      "epoch": 0.07503607503607504,
      "grad_norm": 0.8035733699798584,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.7587,
      "step": 52
    },
    {
      "epoch": 0.07647907647907648,
      "grad_norm": 0.6840614080429077,
      "learning_rate": 9.883829406604363e-05,
      "loss": 1.9629,
      "step": 53
    },
    {
      "epoch": 0.07792207792207792,
      "grad_norm": 0.49338486790657043,
      "learning_rate": 9.876711602542563e-05,
      "loss": 2.0719,
      "step": 54
    },
    {
      "epoch": 0.07936507936507936,
      "grad_norm": 0.3963480591773987,
      "learning_rate": 9.869384896386668e-05,
      "loss": 2.1635,
      "step": 55
    },
    {
      "epoch": 0.08080808080808081,
      "grad_norm": 0.3595295250415802,
      "learning_rate": 9.861849601988383e-05,
      "loss": 1.8126,
      "step": 56
    },
    {
      "epoch": 0.08225108225108226,
      "grad_norm": 0.3469742238521576,
      "learning_rate": 9.854106042134641e-05,
      "loss": 2.0279,
      "step": 57
    },
    {
      "epoch": 0.0836940836940837,
      "grad_norm": 0.33953097462654114,
      "learning_rate": 9.846154548533773e-05,
      "loss": 1.9088,
      "step": 58
    },
    {
      "epoch": 0.08513708513708514,
      "grad_norm": 0.305743932723999,
      "learning_rate": 9.837995461801299e-05,
      "loss": 2.0343,
      "step": 59
    },
    {
      "epoch": 0.08658008658008658,
      "grad_norm": 0.3304813802242279,
      "learning_rate": 9.829629131445342e-05,
      "loss": 2.0815,
      "step": 60
    },
    {
      "epoch": 0.08802308802308802,
      "grad_norm": 0.34086668491363525,
      "learning_rate": 9.821055915851647e-05,
      "loss": 2.1491,
      "step": 61
    },
    {
      "epoch": 0.08946608946608947,
      "grad_norm": 0.38692232966423035,
      "learning_rate": 9.812276182268236e-05,
      "loss": 1.9965,
      "step": 62
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.3496730625629425,
      "learning_rate": 9.803290306789676e-05,
      "loss": 2.0397,
      "step": 63
    },
    {
      "epoch": 0.09235209235209235,
      "grad_norm": 0.384067565202713,
      "learning_rate": 9.794098674340965e-05,
      "loss": 2.0058,
      "step": 64
    },
    {
      "epoch": 0.09379509379509379,
      "grad_norm": 0.3466632068157196,
      "learning_rate": 9.784701678661045e-05,
      "loss": 2.066,
      "step": 65
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.31923907995224,
      "learning_rate": 9.775099722285935e-05,
      "loss": 1.8307,
      "step": 66
    },
    {
      "epoch": 0.09668109668109669,
      "grad_norm": 0.39498981833457947,
      "learning_rate": 9.765293216531486e-05,
      "loss": 2.0947,
      "step": 67
    },
    {
      "epoch": 0.09812409812409813,
      "grad_norm": 0.36418530344963074,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.1766,
      "step": 68
    },
    {
      "epoch": 0.09956709956709957,
      "grad_norm": 0.3260754346847534,
      "learning_rate": 9.74506824594107e-05,
      "loss": 1.9282,
      "step": 69
    },
    {
      "epoch": 0.10101010101010101,
      "grad_norm": 0.40958088636398315,
      "learning_rate": 9.73465064747553e-05,
      "loss": 1.7713,
      "step": 70
    },
    {
      "epoch": 0.10245310245310245,
      "grad_norm": 0.4162747859954834,
      "learning_rate": 9.724030232334391e-05,
      "loss": 2.0031,
      "step": 71
    },
    {
      "epoch": 0.1038961038961039,
      "grad_norm": 0.41455021500587463,
      "learning_rate": 9.713207455460894e-05,
      "loss": 2.1278,
      "step": 72
    },
    {
      "epoch": 0.10533910533910534,
      "grad_norm": 0.40532439947128296,
      "learning_rate": 9.702182780466775e-05,
      "loss": 2.0676,
      "step": 73
    },
    {
      "epoch": 0.10678210678210678,
      "grad_norm": 0.3800431787967682,
      "learning_rate": 9.690956679612421e-05,
      "loss": 1.9015,
      "step": 74
    },
    {
      "epoch": 0.10822510822510822,
      "grad_norm": 0.4204113185405731,
      "learning_rate": 9.67952963378663e-05,
      "loss": 1.9664,
      "step": 75
    },
    {
      "epoch": 0.10966810966810966,
      "grad_norm": 0.4439535140991211,
      "learning_rate": 9.667902132486009e-05,
      "loss": 1.9959,
      "step": 76
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 0.4069725573062897,
      "learning_rate": 9.656074673794018e-05,
      "loss": 1.9064,
      "step": 77
    },
    {
      "epoch": 0.11255411255411256,
      "grad_norm": 0.42558419704437256,
      "learning_rate": 9.644047764359622e-05,
      "loss": 1.9374,
      "step": 78
    },
    {
      "epoch": 0.113997113997114,
      "grad_norm": 0.46812930703163147,
      "learning_rate": 9.631821919375591e-05,
      "loss": 1.8003,
      "step": 79
    },
    {
      "epoch": 0.11544011544011544,
      "grad_norm": 0.4569174349308014,
      "learning_rate": 9.619397662556435e-05,
      "loss": 1.7006,
      "step": 80
    },
    {
      "epoch": 0.11688311688311688,
      "grad_norm": 0.4715164601802826,
      "learning_rate": 9.606775526115963e-05,
      "loss": 1.9712,
      "step": 81
    },
    {
      "epoch": 0.11832611832611832,
      "grad_norm": 0.4455204904079437,
      "learning_rate": 9.593956050744492e-05,
      "loss": 1.8968,
      "step": 82
    },
    {
      "epoch": 0.11976911976911978,
      "grad_norm": 0.5169462561607361,
      "learning_rate": 9.580939785585681e-05,
      "loss": 1.963,
      "step": 83
    },
    {
      "epoch": 0.12121212121212122,
      "grad_norm": 0.4874756932258606,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.943,
      "step": 84
    },
    {
      "epoch": 0.12265512265512266,
      "grad_norm": 0.4800645112991333,
      "learning_rate": 9.554319124605879e-05,
      "loss": 1.8902,
      "step": 85
    },
    {
      "epoch": 0.1240981240981241,
      "grad_norm": 0.48956722021102905,
      "learning_rate": 9.540715869125407e-05,
      "loss": 1.7206,
      "step": 86
    },
    {
      "epoch": 0.12554112554112554,
      "grad_norm": 0.5349124670028687,
      "learning_rate": 9.526918104489777e-05,
      "loss": 1.8946,
      "step": 87
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.5947434306144714,
      "learning_rate": 9.512926421749304e-05,
      "loss": 1.8111,
      "step": 88
    },
    {
      "epoch": 0.12842712842712842,
      "grad_norm": 0.5332505702972412,
      "learning_rate": 9.498741420261108e-05,
      "loss": 1.8397,
      "step": 89
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 0.5390715599060059,
      "learning_rate": 9.484363707663442e-05,
      "loss": 1.847,
      "step": 90
    },
    {
      "epoch": 0.13131313131313133,
      "grad_norm": 0.5333109498023987,
      "learning_rate": 9.469793899849661e-05,
      "loss": 1.912,
      "step": 91
    },
    {
      "epoch": 0.13275613275613277,
      "grad_norm": 0.5515038967132568,
      "learning_rate": 9.45503262094184e-05,
      "loss": 1.8217,
      "step": 92
    },
    {
      "epoch": 0.1341991341991342,
      "grad_norm": 0.615770161151886,
      "learning_rate": 9.440080503264037e-05,
      "loss": 1.854,
      "step": 93
    },
    {
      "epoch": 0.13564213564213565,
      "grad_norm": 0.6496592164039612,
      "learning_rate": 9.42493818731521e-05,
      "loss": 1.7574,
      "step": 94
    },
    {
      "epoch": 0.1370851370851371,
      "grad_norm": 0.7257400155067444,
      "learning_rate": 9.409606321741775e-05,
      "loss": 1.8909,
      "step": 95
    },
    {
      "epoch": 0.13852813852813853,
      "grad_norm": 0.7065343856811523,
      "learning_rate": 9.394085563309827e-05,
      "loss": 1.9357,
      "step": 96
    },
    {
      "epoch": 0.13997113997113997,
      "grad_norm": 0.7580714225769043,
      "learning_rate": 9.378376576876999e-05,
      "loss": 1.9127,
      "step": 97
    },
    {
      "epoch": 0.1414141414141414,
      "grad_norm": 0.8716478943824768,
      "learning_rate": 9.362480035363986e-05,
      "loss": 1.7791,
      "step": 98
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 1.0269742012023926,
      "learning_rate": 9.34639661972572e-05,
      "loss": 1.8904,
      "step": 99
    },
    {
      "epoch": 0.1443001443001443,
      "grad_norm": 1.5355191230773926,
      "learning_rate": 9.330127018922194e-05,
      "loss": 1.8833,
      "step": 100
    },
    {
      "epoch": 0.1443001443001443,
      "eval_loss": 1.993816614151001,
      "eval_runtime": 157.0351,
      "eval_samples_per_second": 7.431,
      "eval_steps_per_second": 1.859,
      "step": 100
    },
    {
      "epoch": 0.14574314574314573,
      "grad_norm": 0.742597222328186,
      "learning_rate": 9.31367192988896e-05,
      "loss": 2.0576,
      "step": 101
    },
    {
      "epoch": 0.1471861471861472,
      "grad_norm": 0.8394694924354553,
      "learning_rate": 9.297032057507264e-05,
      "loss": 1.7602,
      "step": 102
    },
    {
      "epoch": 0.14862914862914864,
      "grad_norm": 0.838948667049408,
      "learning_rate": 9.280208114573859e-05,
      "loss": 1.8816,
      "step": 103
    },
    {
      "epoch": 0.15007215007215008,
      "grad_norm": 0.7070848941802979,
      "learning_rate": 9.263200821770461e-05,
      "loss": 2.0543,
      "step": 104
    },
    {
      "epoch": 0.15151515151515152,
      "grad_norm": 0.44883328676223755,
      "learning_rate": 9.246010907632895e-05,
      "loss": 1.9549,
      "step": 105
    },
    {
      "epoch": 0.15295815295815296,
      "grad_norm": 0.35736629366874695,
      "learning_rate": 9.228639108519868e-05,
      "loss": 2.0421,
      "step": 106
    },
    {
      "epoch": 0.1544011544011544,
      "grad_norm": 0.3133707642555237,
      "learning_rate": 9.211086168581433e-05,
      "loss": 1.8866,
      "step": 107
    },
    {
      "epoch": 0.15584415584415584,
      "grad_norm": 0.31356483697891235,
      "learning_rate": 9.193352839727121e-05,
      "loss": 1.921,
      "step": 108
    },
    {
      "epoch": 0.15728715728715728,
      "grad_norm": 0.3296065926551819,
      "learning_rate": 9.175439881593716e-05,
      "loss": 2.036,
      "step": 109
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 0.28739431500434875,
      "learning_rate": 9.157348061512727e-05,
      "loss": 1.9744,
      "step": 110
    },
    {
      "epoch": 0.16017316017316016,
      "grad_norm": 0.3230794072151184,
      "learning_rate": 9.139078154477512e-05,
      "loss": 2.0474,
      "step": 111
    },
    {
      "epoch": 0.16161616161616163,
      "grad_norm": 0.31099987030029297,
      "learning_rate": 9.120630943110077e-05,
      "loss": 2.1352,
      "step": 112
    },
    {
      "epoch": 0.16305916305916307,
      "grad_norm": 0.31235870718955994,
      "learning_rate": 9.102007217627568e-05,
      "loss": 1.9676,
      "step": 113
    },
    {
      "epoch": 0.1645021645021645,
      "grad_norm": 0.39123886823654175,
      "learning_rate": 9.083207775808396e-05,
      "loss": 1.9695,
      "step": 114
    },
    {
      "epoch": 0.16594516594516595,
      "grad_norm": 0.3258761167526245,
      "learning_rate": 9.064233422958077e-05,
      "loss": 1.747,
      "step": 115
    },
    {
      "epoch": 0.1673881673881674,
      "grad_norm": 0.35148924589157104,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.983,
      "step": 116
    },
    {
      "epoch": 0.16883116883116883,
      "grad_norm": 0.3492424488067627,
      "learning_rate": 9.025763242814291e-05,
      "loss": 1.9037,
      "step": 117
    },
    {
      "epoch": 0.17027417027417027,
      "grad_norm": 0.4515904486179352,
      "learning_rate": 9.006269063455304e-05,
      "loss": 1.9855,
      "step": 118
    },
    {
      "epoch": 0.1717171717171717,
      "grad_norm": 0.3563421964645386,
      "learning_rate": 8.986603268863536e-05,
      "loss": 1.8676,
      "step": 119
    },
    {
      "epoch": 0.17316017316017315,
      "grad_norm": 0.4804840385913849,
      "learning_rate": 8.966766701456177e-05,
      "loss": 2.0058,
      "step": 120
    },
    {
      "epoch": 0.1746031746031746,
      "grad_norm": 0.4358946979045868,
      "learning_rate": 8.94676021096575e-05,
      "loss": 2.0668,
      "step": 121
    },
    {
      "epoch": 0.17604617604617603,
      "grad_norm": 0.4688352644443512,
      "learning_rate": 8.926584654403724e-05,
      "loss": 2.0926,
      "step": 122
    },
    {
      "epoch": 0.1774891774891775,
      "grad_norm": 0.38327309489250183,
      "learning_rate": 8.906240896023794e-05,
      "loss": 1.9646,
      "step": 123
    },
    {
      "epoch": 0.17893217893217894,
      "grad_norm": 0.3731294572353363,
      "learning_rate": 8.885729807284856e-05,
      "loss": 1.9799,
      "step": 124
    },
    {
      "epoch": 0.18037518037518038,
      "grad_norm": 0.4308861494064331,
      "learning_rate": 8.865052266813685e-05,
      "loss": 2.1767,
      "step": 125
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.41706499457359314,
      "learning_rate": 8.844209160367299e-05,
      "loss": 1.9402,
      "step": 126
    },
    {
      "epoch": 0.18326118326118326,
      "grad_norm": 0.4007774293422699,
      "learning_rate": 8.823201380795001e-05,
      "loss": 1.9648,
      "step": 127
    },
    {
      "epoch": 0.1847041847041847,
      "grad_norm": 0.38914182782173157,
      "learning_rate": 8.802029828000156e-05,
      "loss": 1.9749,
      "step": 128
    },
    {
      "epoch": 0.18614718614718614,
      "grad_norm": 0.4113757312297821,
      "learning_rate": 8.780695408901613e-05,
      "loss": 1.8051,
      "step": 129
    },
    {
      "epoch": 0.18759018759018758,
      "grad_norm": 0.44275519251823425,
      "learning_rate": 8.759199037394887e-05,
      "loss": 1.8252,
      "step": 130
    },
    {
      "epoch": 0.18903318903318903,
      "grad_norm": 0.43470901250839233,
      "learning_rate": 8.737541634312985e-05,
      "loss": 1.8967,
      "step": 131
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.48708173632621765,
      "learning_rate": 8.715724127386972e-05,
      "loss": 1.9375,
      "step": 132
    },
    {
      "epoch": 0.1919191919191919,
      "grad_norm": 0.4494212865829468,
      "learning_rate": 8.693747451206232e-05,
      "loss": 1.8855,
      "step": 133
    },
    {
      "epoch": 0.19336219336219337,
      "grad_norm": 0.42587026953697205,
      "learning_rate": 8.671612547178428e-05,
      "loss": 1.7339,
      "step": 134
    },
    {
      "epoch": 0.19480519480519481,
      "grad_norm": 0.5141189694404602,
      "learning_rate": 8.649320363489179e-05,
      "loss": 1.8482,
      "step": 135
    },
    {
      "epoch": 0.19624819624819625,
      "grad_norm": 0.559985876083374,
      "learning_rate": 8.626871855061438e-05,
      "loss": 1.9328,
      "step": 136
    },
    {
      "epoch": 0.1976911976911977,
      "grad_norm": 0.5099179744720459,
      "learning_rate": 8.604267983514594e-05,
      "loss": 1.8337,
      "step": 137
    },
    {
      "epoch": 0.19913419913419914,
      "grad_norm": 0.4592433273792267,
      "learning_rate": 8.581509717123273e-05,
      "loss": 1.7073,
      "step": 138
    },
    {
      "epoch": 0.20057720057720058,
      "grad_norm": 0.5081384181976318,
      "learning_rate": 8.558598030775857e-05,
      "loss": 1.8063,
      "step": 139
    },
    {
      "epoch": 0.20202020202020202,
      "grad_norm": 0.5941727757453918,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.7908,
      "step": 140
    },
    {
      "epoch": 0.20346320346320346,
      "grad_norm": 0.5501158833503723,
      "learning_rate": 8.51231833058426e-05,
      "loss": 1.8062,
      "step": 141
    },
    {
      "epoch": 0.2049062049062049,
      "grad_norm": 0.5739471316337585,
      "learning_rate": 8.488952299208401e-05,
      "loss": 1.9205,
      "step": 142
    },
    {
      "epoch": 0.20634920634920634,
      "grad_norm": 0.6430457830429077,
      "learning_rate": 8.46543681272818e-05,
      "loss": 2.075,
      "step": 143
    },
    {
      "epoch": 0.2077922077922078,
      "grad_norm": 0.6881362795829773,
      "learning_rate": 8.44177287846877e-05,
      "loss": 1.9087,
      "step": 144
    },
    {
      "epoch": 0.20923520923520925,
      "grad_norm": 0.6772449612617493,
      "learning_rate": 8.417961510114356e-05,
      "loss": 1.5338,
      "step": 145
    },
    {
      "epoch": 0.2106782106782107,
      "grad_norm": 0.727997362613678,
      "learning_rate": 8.39400372766471e-05,
      "loss": 1.8159,
      "step": 146
    },
    {
      "epoch": 0.21212121212121213,
      "grad_norm": 0.7379339337348938,
      "learning_rate": 8.36990055739149e-05,
      "loss": 1.6428,
      "step": 147
    },
    {
      "epoch": 0.21356421356421357,
      "grad_norm": 0.8291444182395935,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.8209,
      "step": 148
    },
    {
      "epoch": 0.215007215007215,
      "grad_norm": 0.8952499628067017,
      "learning_rate": 8.321262189556409e-05,
      "loss": 1.6022,
      "step": 149
    },
    {
      "epoch": 0.21645021645021645,
      "grad_norm": 1.9928690195083618,
      "learning_rate": 8.296729075500344e-05,
      "loss": 1.62,
      "step": 150
    },
    {
      "epoch": 0.2178932178932179,
      "grad_norm": 0.5600587725639343,
      "learning_rate": 8.272054740543052e-05,
      "loss": 1.9747,
      "step": 151
    },
    {
      "epoch": 0.21933621933621933,
      "grad_norm": 0.8041567802429199,
      "learning_rate": 8.247240241650918e-05,
      "loss": 2.0701,
      "step": 152
    },
    {
      "epoch": 0.22077922077922077,
      "grad_norm": 0.8225897550582886,
      "learning_rate": 8.222286641794488e-05,
      "loss": 2.0347,
      "step": 153
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.7567336559295654,
      "learning_rate": 8.197195009902924e-05,
      "loss": 1.7514,
      "step": 154
    },
    {
      "epoch": 0.22366522366522368,
      "grad_norm": 0.6722519993782043,
      "learning_rate": 8.171966420818228e-05,
      "loss": 1.9292,
      "step": 155
    },
    {
      "epoch": 0.22510822510822512,
      "grad_norm": 0.4068787395954132,
      "learning_rate": 8.146601955249188e-05,
      "loss": 1.9831,
      "step": 156
    },
    {
      "epoch": 0.22655122655122656,
      "grad_norm": 0.32383856177330017,
      "learning_rate": 8.121102699725089e-05,
      "loss": 1.915,
      "step": 157
    },
    {
      "epoch": 0.227994227994228,
      "grad_norm": 0.31018707156181335,
      "learning_rate": 8.095469746549172e-05,
      "loss": 2.0141,
      "step": 158
    },
    {
      "epoch": 0.22943722943722944,
      "grad_norm": 0.28106921911239624,
      "learning_rate": 8.069704193751832e-05,
      "loss": 1.9433,
      "step": 159
    },
    {
      "epoch": 0.23088023088023088,
      "grad_norm": 0.30933740735054016,
      "learning_rate": 8.043807145043604e-05,
      "loss": 2.11,
      "step": 160
    },
    {
      "epoch": 0.23232323232323232,
      "grad_norm": 0.3066120445728302,
      "learning_rate": 8.017779709767858e-05,
      "loss": 2.0969,
      "step": 161
    },
    {
      "epoch": 0.23376623376623376,
      "grad_norm": 0.3253624737262726,
      "learning_rate": 7.991623002853296e-05,
      "loss": 1.7843,
      "step": 162
    },
    {
      "epoch": 0.2352092352092352,
      "grad_norm": 0.3151222765445709,
      "learning_rate": 7.965338144766186e-05,
      "loss": 1.9197,
      "step": 163
    },
    {
      "epoch": 0.23665223665223664,
      "grad_norm": 0.38550567626953125,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.0804,
      "step": 164
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.3623102307319641,
      "learning_rate": 7.912388484339012e-05,
      "loss": 1.8096,
      "step": 165
    },
    {
      "epoch": 0.23953823953823955,
      "grad_norm": 0.39857810735702515,
      "learning_rate": 7.88572595018617e-05,
      "loss": 2.001,
      "step": 166
    },
    {
      "epoch": 0.240981240981241,
      "grad_norm": 0.39852413535118103,
      "learning_rate": 7.858939801138061e-05,
      "loss": 1.9444,
      "step": 167
    },
    {
      "epoch": 0.24242424242424243,
      "grad_norm": 0.3307431638240814,
      "learning_rate": 7.832031184624164e-05,
      "loss": 1.9379,
      "step": 168
    },
    {
      "epoch": 0.24386724386724387,
      "grad_norm": 0.3271358609199524,
      "learning_rate": 7.80500125332005e-05,
      "loss": 1.902,
      "step": 169
    },
    {
      "epoch": 0.2453102453102453,
      "grad_norm": 0.41468778252601624,
      "learning_rate": 7.777851165098012e-05,
      "loss": 1.9665,
      "step": 170
    },
    {
      "epoch": 0.24675324675324675,
      "grad_norm": 0.3892781138420105,
      "learning_rate": 7.750582082977467e-05,
      "loss": 1.8492,
      "step": 171
    },
    {
      "epoch": 0.2481962481962482,
      "grad_norm": 0.38811030983924866,
      "learning_rate": 7.723195175075136e-05,
      "loss": 1.9571,
      "step": 172
    },
    {
      "epoch": 0.24963924963924963,
      "grad_norm": 0.43744465708732605,
      "learning_rate": 7.695691614555003e-05,
      "loss": 2.0064,
      "step": 173
    },
    {
      "epoch": 0.2510822510822511,
      "grad_norm": 0.4178030788898468,
      "learning_rate": 7.668072579578058e-05,
      "loss": 2.0999,
      "step": 174
    },
    {
      "epoch": 0.25252525252525254,
      "grad_norm": 0.40274107456207275,
      "learning_rate": 7.64033925325184e-05,
      "loss": 1.8941,
      "step": 175
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.4556225836277008,
      "learning_rate": 7.612492823579745e-05,
      "loss": 2.0221,
      "step": 176
    },
    {
      "epoch": 0.2554112554112554,
      "grad_norm": 0.45292896032333374,
      "learning_rate": 7.584534483410137e-05,
      "loss": 1.8131,
      "step": 177
    },
    {
      "epoch": 0.25685425685425683,
      "grad_norm": 0.42167505621910095,
      "learning_rate": 7.55646543038526e-05,
      "loss": 2.1179,
      "step": 178
    },
    {
      "epoch": 0.2582972582972583,
      "grad_norm": 0.42378920316696167,
      "learning_rate": 7.528286866889924e-05,
      "loss": 2.1155,
      "step": 179
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 0.4680819511413574,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.0742,
      "step": 180
    },
    {
      "epoch": 0.2611832611832612,
      "grad_norm": 0.4424709975719452,
      "learning_rate": 7.471606041430723e-05,
      "loss": 2.0885,
      "step": 181
    },
    {
      "epoch": 0.26262626262626265,
      "grad_norm": 0.531936764717102,
      "learning_rate": 7.443106207484776e-05,
      "loss": 1.8219,
      "step": 182
    },
    {
      "epoch": 0.26406926406926406,
      "grad_norm": 0.39951208233833313,
      "learning_rate": 7.414501719000187e-05,
      "loss": 1.8678,
      "step": 183
    },
    {
      "epoch": 0.26551226551226553,
      "grad_norm": 0.44766274094581604,
      "learning_rate": 7.385793801298042e-05,
      "loss": 2.0432,
      "step": 184
    },
    {
      "epoch": 0.26695526695526695,
      "grad_norm": 0.4364171326160431,
      "learning_rate": 7.35698368412999e-05,
      "loss": 1.8403,
      "step": 185
    },
    {
      "epoch": 0.2683982683982684,
      "grad_norm": 0.3969670534133911,
      "learning_rate": 7.328072601625557e-05,
      "loss": 1.8147,
      "step": 186
    },
    {
      "epoch": 0.2698412698412698,
      "grad_norm": 0.4603928327560425,
      "learning_rate": 7.2990617922393e-05,
      "loss": 1.8538,
      "step": 187
    },
    {
      "epoch": 0.2712842712842713,
      "grad_norm": 0.4237711727619171,
      "learning_rate": 7.269952498697734e-05,
      "loss": 1.8051,
      "step": 188
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.5007184147834778,
      "learning_rate": 7.240745967946113e-05,
      "loss": 1.8868,
      "step": 189
    },
    {
      "epoch": 0.2741702741702742,
      "grad_norm": 0.5073036551475525,
      "learning_rate": 7.211443451095007e-05,
      "loss": 1.7303,
      "step": 190
    },
    {
      "epoch": 0.2756132756132756,
      "grad_norm": 0.5766152143478394,
      "learning_rate": 7.18204620336671e-05,
      "loss": 1.9035,
      "step": 191
    },
    {
      "epoch": 0.27705627705627706,
      "grad_norm": 0.5440102815628052,
      "learning_rate": 7.152555484041476e-05,
      "loss": 1.8722,
      "step": 192
    },
    {
      "epoch": 0.2784992784992785,
      "grad_norm": 0.5761400461196899,
      "learning_rate": 7.122972556403567e-05,
      "loss": 1.8134,
      "step": 193
    },
    {
      "epoch": 0.27994227994227994,
      "grad_norm": 0.6331032514572144,
      "learning_rate": 7.09329868768714e-05,
      "loss": 1.8894,
      "step": 194
    },
    {
      "epoch": 0.2813852813852814,
      "grad_norm": 0.6492578983306885,
      "learning_rate": 7.063535149021973e-05,
      "loss": 1.8969,
      "step": 195
    },
    {
      "epoch": 0.2828282828282828,
      "grad_norm": 0.6990649700164795,
      "learning_rate": 7.033683215379002e-05,
      "loss": 2.0232,
      "step": 196
    },
    {
      "epoch": 0.2842712842712843,
      "grad_norm": 0.778406023979187,
      "learning_rate": 7.003744165515705e-05,
      "loss": 1.8087,
      "step": 197
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.7730764150619507,
      "learning_rate": 6.973719281921335e-05,
      "loss": 1.8531,
      "step": 198
    },
    {
      "epoch": 0.28715728715728717,
      "grad_norm": 0.8641883730888367,
      "learning_rate": 6.943609850761979e-05,
      "loss": 1.9818,
      "step": 199
    },
    {
      "epoch": 0.2886002886002886,
      "grad_norm": 1.346280813217163,
      "learning_rate": 6.91341716182545e-05,
      "loss": 1.7613,
      "step": 200
    },
    {
      "epoch": 0.2886002886002886,
      "eval_loss": 1.930011510848999,
      "eval_runtime": 156.8582,
      "eval_samples_per_second": 7.44,
      "eval_steps_per_second": 1.862,
      "step": 200
    },
    {
      "epoch": 0.29004329004329005,
      "grad_norm": 0.33210301399230957,
      "learning_rate": 6.883142508466054e-05,
      "loss": 1.8346,
      "step": 201
    },
    {
      "epoch": 0.29148629148629146,
      "grad_norm": 0.48873820900917053,
      "learning_rate": 6.852787187549182e-05,
      "loss": 1.7156,
      "step": 202
    },
    {
      "epoch": 0.29292929292929293,
      "grad_norm": 0.5536623001098633,
      "learning_rate": 6.82235249939575e-05,
      "loss": 1.9248,
      "step": 203
    },
    {
      "epoch": 0.2943722943722944,
      "grad_norm": 0.5415230393409729,
      "learning_rate": 6.7918397477265e-05,
      "loss": 1.8952,
      "step": 204
    },
    {
      "epoch": 0.2958152958152958,
      "grad_norm": 0.4394938349723816,
      "learning_rate": 6.761250239606169e-05,
      "loss": 1.8813,
      "step": 205
    },
    {
      "epoch": 0.2972582972582973,
      "grad_norm": 0.37631645798683167,
      "learning_rate": 6.730585285387465e-05,
      "loss": 1.9335,
      "step": 206
    },
    {
      "epoch": 0.2987012987012987,
      "grad_norm": 0.3685065507888794,
      "learning_rate": 6.699846198654971e-05,
      "loss": 1.878,
      "step": 207
    },
    {
      "epoch": 0.30014430014430016,
      "grad_norm": 0.2967827618122101,
      "learning_rate": 6.669034296168855e-05,
      "loss": 1.9316,
      "step": 208
    },
    {
      "epoch": 0.30158730158730157,
      "grad_norm": 0.3100146949291229,
      "learning_rate": 6.638150897808468e-05,
      "loss": 1.9145,
      "step": 209
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 0.3605823516845703,
      "learning_rate": 6.607197326515808e-05,
      "loss": 1.7016,
      "step": 210
    },
    {
      "epoch": 0.30447330447330445,
      "grad_norm": 0.364137202501297,
      "learning_rate": 6.57617490823885e-05,
      "loss": 1.9892,
      "step": 211
    },
    {
      "epoch": 0.3059163059163059,
      "grad_norm": 0.29972338676452637,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0432,
      "step": 212
    },
    {
      "epoch": 0.30735930735930733,
      "grad_norm": 0.3284769058227539,
      "learning_rate": 6.513928849212873e-05,
      "loss": 2.0474,
      "step": 213
    },
    {
      "epoch": 0.3088023088023088,
      "grad_norm": 0.3190697431564331,
      "learning_rate": 6.482707874877854e-05,
      "loss": 2.0264,
      "step": 214
    },
    {
      "epoch": 0.31024531024531027,
      "grad_norm": 0.32829707860946655,
      "learning_rate": 6.451423386272312e-05,
      "loss": 1.845,
      "step": 215
    },
    {
      "epoch": 0.3116883116883117,
      "grad_norm": 0.38648197054862976,
      "learning_rate": 6.420076723519614e-05,
      "loss": 2.068,
      "step": 216
    },
    {
      "epoch": 0.31313131313131315,
      "grad_norm": 0.3774462640285492,
      "learning_rate": 6.388669229406462e-05,
      "loss": 1.7653,
      "step": 217
    },
    {
      "epoch": 0.31457431457431456,
      "grad_norm": 0.35081520676612854,
      "learning_rate": 6.357202249325371e-05,
      "loss": 1.8903,
      "step": 218
    },
    {
      "epoch": 0.31601731601731603,
      "grad_norm": 0.377341628074646,
      "learning_rate": 6.32567713121704e-05,
      "loss": 1.8658,
      "step": 219
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 0.33360525965690613,
      "learning_rate": 6.294095225512603e-05,
      "loss": 1.894,
      "step": 220
    },
    {
      "epoch": 0.3189033189033189,
      "grad_norm": 0.3708435893058777,
      "learning_rate": 6.26245788507579e-05,
      "loss": 1.8171,
      "step": 221
    },
    {
      "epoch": 0.3203463203463203,
      "grad_norm": 0.4073418378829956,
      "learning_rate": 6.230766465144967e-05,
      "loss": 2.0069,
      "step": 222
    },
    {
      "epoch": 0.3217893217893218,
      "grad_norm": 0.3824368417263031,
      "learning_rate": 6.199022323275083e-05,
      "loss": 1.8745,
      "step": 223
    },
    {
      "epoch": 0.32323232323232326,
      "grad_norm": 0.37881967425346375,
      "learning_rate": 6.167226819279528e-05,
      "loss": 2.0335,
      "step": 224
    },
    {
      "epoch": 0.3246753246753247,
      "grad_norm": 0.3804931938648224,
      "learning_rate": 6.135381315171867e-05,
      "loss": 1.9527,
      "step": 225
    },
    {
      "epoch": 0.32611832611832614,
      "grad_norm": 0.35501721501350403,
      "learning_rate": 6.103487175107507e-05,
      "loss": 1.9177,
      "step": 226
    },
    {
      "epoch": 0.32756132756132755,
      "grad_norm": 0.4466770887374878,
      "learning_rate": 6.071545765325254e-05,
      "loss": 2.1138,
      "step": 227
    },
    {
      "epoch": 0.329004329004329,
      "grad_norm": 0.42904266715049744,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 1.9386,
      "step": 228
    },
    {
      "epoch": 0.33044733044733043,
      "grad_norm": 0.4379250705242157,
      "learning_rate": 6.007526611628086e-05,
      "loss": 1.9915,
      "step": 229
    },
    {
      "epoch": 0.3318903318903319,
      "grad_norm": 0.5087229013442993,
      "learning_rate": 5.9754516100806423e-05,
      "loss": 1.8386,
      "step": 230
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.4807024896144867,
      "learning_rate": 5.9433348234327765e-05,
      "loss": 1.9612,
      "step": 231
    },
    {
      "epoch": 0.3347763347763348,
      "grad_norm": 0.387539267539978,
      "learning_rate": 5.911177627460739e-05,
      "loss": 1.9333,
      "step": 232
    },
    {
      "epoch": 0.3362193362193362,
      "grad_norm": 0.4602753221988678,
      "learning_rate": 5.8789813996717736e-05,
      "loss": 1.8672,
      "step": 233
    },
    {
      "epoch": 0.33766233766233766,
      "grad_norm": 0.40583816170692444,
      "learning_rate": 5.8467475192451226e-05,
      "loss": 1.9556,
      "step": 234
    },
    {
      "epoch": 0.33910533910533913,
      "grad_norm": 0.5055876970291138,
      "learning_rate": 5.814477366972945e-05,
      "loss": 1.9021,
      "step": 235
    },
    {
      "epoch": 0.34054834054834054,
      "grad_norm": 0.43246597051620483,
      "learning_rate": 5.782172325201155e-05,
      "loss": 1.6321,
      "step": 236
    },
    {
      "epoch": 0.341991341991342,
      "grad_norm": 0.44223451614379883,
      "learning_rate": 5.749833777770225e-05,
      "loss": 1.7132,
      "step": 237
    },
    {
      "epoch": 0.3434343434343434,
      "grad_norm": 0.47704342007637024,
      "learning_rate": 5.717463109955896e-05,
      "loss": 1.8639,
      "step": 238
    },
    {
      "epoch": 0.3448773448773449,
      "grad_norm": 0.618209719657898,
      "learning_rate": 5.685061708409841e-05,
      "loss": 1.8064,
      "step": 239
    },
    {
      "epoch": 0.3463203463203463,
      "grad_norm": 0.47115299105644226,
      "learning_rate": 5.6526309611002594e-05,
      "loss": 1.8133,
      "step": 240
    },
    {
      "epoch": 0.3477633477633478,
      "grad_norm": 0.4711276888847351,
      "learning_rate": 5.6201722572524275e-05,
      "loss": 1.8172,
      "step": 241
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 0.5899491310119629,
      "learning_rate": 5.587686987289189e-05,
      "loss": 1.9716,
      "step": 242
    },
    {
      "epoch": 0.35064935064935066,
      "grad_norm": 0.6002910137176514,
      "learning_rate": 5.5551765427713884e-05,
      "loss": 1.9386,
      "step": 243
    },
    {
      "epoch": 0.35209235209235207,
      "grad_norm": 0.6195904612541199,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.9987,
      "step": 244
    },
    {
      "epoch": 0.35353535353535354,
      "grad_norm": 0.6184617877006531,
      "learning_rate": 5.490085701647805e-05,
      "loss": 1.822,
      "step": 245
    },
    {
      "epoch": 0.354978354978355,
      "grad_norm": 0.5911903977394104,
      "learning_rate": 5.457508093317013e-05,
      "loss": 1.8101,
      "step": 246
    },
    {
      "epoch": 0.3564213564213564,
      "grad_norm": 0.7217023372650146,
      "learning_rate": 5.4249108868622086e-05,
      "loss": 1.9276,
      "step": 247
    },
    {
      "epoch": 0.3578643578643579,
      "grad_norm": 0.6757352352142334,
      "learning_rate": 5.392295478639225e-05,
      "loss": 1.8208,
      "step": 248
    },
    {
      "epoch": 0.3593073593073593,
      "grad_norm": 0.8589720726013184,
      "learning_rate": 5.359663265783598e-05,
      "loss": 1.7009,
      "step": 249
    },
    {
      "epoch": 0.36075036075036077,
      "grad_norm": 1.2887048721313477,
      "learning_rate": 5.327015646150716e-05,
      "loss": 1.8405,
      "step": 250
    },
    {
      "epoch": 0.3621933621933622,
      "grad_norm": 0.20886729657649994,
      "learning_rate": 5.294354018255945e-05,
      "loss": 1.9762,
      "step": 251
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.2683543562889099,
      "learning_rate": 5.26167978121472e-05,
      "loss": 1.8028,
      "step": 252
    },
    {
      "epoch": 0.36507936507936506,
      "grad_norm": 0.28197985887527466,
      "learning_rate": 5.228994334682604e-05,
      "loss": 1.7985,
      "step": 253
    },
    {
      "epoch": 0.3665223665223665,
      "grad_norm": 0.3346744179725647,
      "learning_rate": 5.196299078795344e-05,
      "loss": 1.9574,
      "step": 254
    },
    {
      "epoch": 0.36796536796536794,
      "grad_norm": 0.33492788672447205,
      "learning_rate": 5.1635954141088813e-05,
      "loss": 1.7003,
      "step": 255
    },
    {
      "epoch": 0.3694083694083694,
      "grad_norm": 0.41365864872932434,
      "learning_rate": 5.1308847415393666e-05,
      "loss": 2.0219,
      "step": 256
    },
    {
      "epoch": 0.3708513708513709,
      "grad_norm": 0.4230329990386963,
      "learning_rate": 5.0981684623031415e-05,
      "loss": 1.8089,
      "step": 257
    },
    {
      "epoch": 0.3722943722943723,
      "grad_norm": 0.40661102533340454,
      "learning_rate": 5.0654479778567223e-05,
      "loss": 1.8797,
      "step": 258
    },
    {
      "epoch": 0.37373737373737376,
      "grad_norm": 0.3464805781841278,
      "learning_rate": 5.0327246898367597e-05,
      "loss": 1.7252,
      "step": 259
    },
    {
      "epoch": 0.37518037518037517,
      "grad_norm": 0.31361910700798035,
      "learning_rate": 5e-05,
      "loss": 2.0597,
      "step": 260
    },
    {
      "epoch": 0.37662337662337664,
      "grad_norm": 0.43410736322402954,
      "learning_rate": 4.9672753101632415e-05,
      "loss": 1.8767,
      "step": 261
    },
    {
      "epoch": 0.37806637806637805,
      "grad_norm": 0.4131929576396942,
      "learning_rate": 4.934552022143279e-05,
      "loss": 1.929,
      "step": 262
    },
    {
      "epoch": 0.3795093795093795,
      "grad_norm": 0.3646806478500366,
      "learning_rate": 4.901831537696859e-05,
      "loss": 2.0679,
      "step": 263
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.2950616478919983,
      "learning_rate": 4.869115258460635e-05,
      "loss": 1.9789,
      "step": 264
    },
    {
      "epoch": 0.3823953823953824,
      "grad_norm": 0.387099027633667,
      "learning_rate": 4.83640458589112e-05,
      "loss": 1.9296,
      "step": 265
    },
    {
      "epoch": 0.3838383838383838,
      "grad_norm": 0.351723849773407,
      "learning_rate": 4.8037009212046586e-05,
      "loss": 1.8302,
      "step": 266
    },
    {
      "epoch": 0.3852813852813853,
      "grad_norm": 0.40654313564300537,
      "learning_rate": 4.7710056653173976e-05,
      "loss": 1.8677,
      "step": 267
    },
    {
      "epoch": 0.38672438672438675,
      "grad_norm": 0.39013293385505676,
      "learning_rate": 4.738320218785281e-05,
      "loss": 2.0443,
      "step": 268
    },
    {
      "epoch": 0.38816738816738816,
      "grad_norm": 0.3763997256755829,
      "learning_rate": 4.7056459817440544e-05,
      "loss": 1.9559,
      "step": 269
    },
    {
      "epoch": 0.38961038961038963,
      "grad_norm": 0.39660537242889404,
      "learning_rate": 4.6729843538492847e-05,
      "loss": 1.8978,
      "step": 270
    },
    {
      "epoch": 0.39105339105339104,
      "grad_norm": 0.40315741300582886,
      "learning_rate": 4.640336734216403e-05,
      "loss": 1.8238,
      "step": 271
    },
    {
      "epoch": 0.3924963924963925,
      "grad_norm": 0.4773310720920563,
      "learning_rate": 4.607704521360776e-05,
      "loss": 1.8576,
      "step": 272
    },
    {
      "epoch": 0.3939393939393939,
      "grad_norm": 0.35388076305389404,
      "learning_rate": 4.575089113137792e-05,
      "loss": 1.9199,
      "step": 273
    },
    {
      "epoch": 0.3953823953823954,
      "grad_norm": 0.38064417243003845,
      "learning_rate": 4.542491906682989e-05,
      "loss": 2.0109,
      "step": 274
    },
    {
      "epoch": 0.3968253968253968,
      "grad_norm": 0.4209693670272827,
      "learning_rate": 4.509914298352197e-05,
      "loss": 1.9023,
      "step": 275
    },
    {
      "epoch": 0.39826839826839827,
      "grad_norm": 0.432314932346344,
      "learning_rate": 4.477357683661734e-05,
      "loss": 2.014,
      "step": 276
    },
    {
      "epoch": 0.3997113997113997,
      "grad_norm": 0.4496338665485382,
      "learning_rate": 4.444823457228612e-05,
      "loss": 1.7333,
      "step": 277
    },
    {
      "epoch": 0.40115440115440115,
      "grad_norm": 0.4074253439903259,
      "learning_rate": 4.412313012710813e-05,
      "loss": 1.9451,
      "step": 278
    },
    {
      "epoch": 0.4025974025974026,
      "grad_norm": 0.43842747807502747,
      "learning_rate": 4.379827742747575e-05,
      "loss": 1.7648,
      "step": 279
    },
    {
      "epoch": 0.40404040404040403,
      "grad_norm": 0.41590142250061035,
      "learning_rate": 4.347369038899744e-05,
      "loss": 1.991,
      "step": 280
    },
    {
      "epoch": 0.4054834054834055,
      "grad_norm": 0.41823965311050415,
      "learning_rate": 4.3149382915901606e-05,
      "loss": 1.8753,
      "step": 281
    },
    {
      "epoch": 0.4069264069264069,
      "grad_norm": 0.4189210534095764,
      "learning_rate": 4.282536890044104e-05,
      "loss": 1.9947,
      "step": 282
    },
    {
      "epoch": 0.4083694083694084,
      "grad_norm": 0.43361666798591614,
      "learning_rate": 4.250166222229774e-05,
      "loss": 1.949,
      "step": 283
    },
    {
      "epoch": 0.4098124098124098,
      "grad_norm": 0.4770269989967346,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 1.7797,
      "step": 284
    },
    {
      "epoch": 0.41125541125541126,
      "grad_norm": 0.44815993309020996,
      "learning_rate": 4.185522633027057e-05,
      "loss": 1.6703,
      "step": 285
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 0.5173628926277161,
      "learning_rate": 4.153252480754877e-05,
      "loss": 1.8344,
      "step": 286
    },
    {
      "epoch": 0.41414141414141414,
      "grad_norm": 0.5571794509887695,
      "learning_rate": 4.1210186003282275e-05,
      "loss": 2.0112,
      "step": 287
    },
    {
      "epoch": 0.4155844155844156,
      "grad_norm": 0.5537445545196533,
      "learning_rate": 4.088822372539263e-05,
      "loss": 1.7687,
      "step": 288
    },
    {
      "epoch": 0.417027417027417,
      "grad_norm": 0.5494181513786316,
      "learning_rate": 4.0566651765672246e-05,
      "loss": 1.8711,
      "step": 289
    },
    {
      "epoch": 0.4184704184704185,
      "grad_norm": 0.49445641040802,
      "learning_rate": 4.0245483899193595e-05,
      "loss": 1.759,
      "step": 290
    },
    {
      "epoch": 0.4199134199134199,
      "grad_norm": 0.569502592086792,
      "learning_rate": 3.992473388371915e-05,
      "loss": 1.9987,
      "step": 291
    },
    {
      "epoch": 0.4213564213564214,
      "grad_norm": 0.6796965599060059,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.871,
      "step": 292
    },
    {
      "epoch": 0.4227994227994228,
      "grad_norm": 0.5192450284957886,
      "learning_rate": 3.928454234674747e-05,
      "loss": 1.5954,
      "step": 293
    },
    {
      "epoch": 0.42424242424242425,
      "grad_norm": 0.5506890416145325,
      "learning_rate": 3.896512824892495e-05,
      "loss": 1.9684,
      "step": 294
    },
    {
      "epoch": 0.42568542568542567,
      "grad_norm": 0.5987568497657776,
      "learning_rate": 3.864618684828134e-05,
      "loss": 2.034,
      "step": 295
    },
    {
      "epoch": 0.42712842712842713,
      "grad_norm": 0.582402229309082,
      "learning_rate": 3.832773180720475e-05,
      "loss": 1.8364,
      "step": 296
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.6683496832847595,
      "learning_rate": 3.800977676724919e-05,
      "loss": 1.8847,
      "step": 297
    },
    {
      "epoch": 0.43001443001443,
      "grad_norm": 0.6951612830162048,
      "learning_rate": 3.769233534855035e-05,
      "loss": 1.7696,
      "step": 298
    },
    {
      "epoch": 0.4314574314574315,
      "grad_norm": 0.9686669707298279,
      "learning_rate": 3.73754211492421e-05,
      "loss": 1.6039,
      "step": 299
    },
    {
      "epoch": 0.4329004329004329,
      "grad_norm": 1.6390401124954224,
      "learning_rate": 3.705904774487396e-05,
      "loss": 2.0773,
      "step": 300
    },
    {
      "epoch": 0.4329004329004329,
      "eval_loss": 1.8472508192062378,
      "eval_runtime": 156.8612,
      "eval_samples_per_second": 7.44,
      "eval_steps_per_second": 1.862,
      "step": 300
    },
    {
      "epoch": 0.43434343434343436,
      "grad_norm": 0.2445702999830246,
      "learning_rate": 3.6743228687829595e-05,
      "loss": 1.9193,
      "step": 301
    },
    {
      "epoch": 0.4357864357864358,
      "grad_norm": 0.23023849725723267,
      "learning_rate": 3.642797750674629e-05,
      "loss": 1.685,
      "step": 302
    },
    {
      "epoch": 0.43722943722943725,
      "grad_norm": 0.21904835104942322,
      "learning_rate": 3.6113307705935396e-05,
      "loss": 1.8372,
      "step": 303
    },
    {
      "epoch": 0.43867243867243866,
      "grad_norm": 0.2425857037305832,
      "learning_rate": 3.579923276480387e-05,
      "loss": 1.7388,
      "step": 304
    },
    {
      "epoch": 0.4401154401154401,
      "grad_norm": 0.25920021533966064,
      "learning_rate": 3.5485766137276894e-05,
      "loss": 1.8383,
      "step": 305
    },
    {
      "epoch": 0.44155844155844154,
      "grad_norm": 0.2727745771408081,
      "learning_rate": 3.5172921251221455e-05,
      "loss": 1.8585,
      "step": 306
    },
    {
      "epoch": 0.443001443001443,
      "grad_norm": 0.2757931351661682,
      "learning_rate": 3.486071150787128e-05,
      "loss": 1.6702,
      "step": 307
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.31464436650276184,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.9148,
      "step": 308
    },
    {
      "epoch": 0.4458874458874459,
      "grad_norm": 0.3319891393184662,
      "learning_rate": 3.423825091761153e-05,
      "loss": 2.0834,
      "step": 309
    },
    {
      "epoch": 0.44733044733044736,
      "grad_norm": 0.3357025384902954,
      "learning_rate": 3.392802673484193e-05,
      "loss": 1.9062,
      "step": 310
    },
    {
      "epoch": 0.44877344877344877,
      "grad_norm": 0.30716291069984436,
      "learning_rate": 3.361849102191533e-05,
      "loss": 2.0304,
      "step": 311
    },
    {
      "epoch": 0.45021645021645024,
      "grad_norm": 0.31525135040283203,
      "learning_rate": 3.330965703831146e-05,
      "loss": 1.8541,
      "step": 312
    },
    {
      "epoch": 0.45165945165945165,
      "grad_norm": 0.2891955077648163,
      "learning_rate": 3.300153801345028e-05,
      "loss": 1.9576,
      "step": 313
    },
    {
      "epoch": 0.4531024531024531,
      "grad_norm": 0.3366871774196625,
      "learning_rate": 3.2694147146125345e-05,
      "loss": 1.9217,
      "step": 314
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.3520633578300476,
      "learning_rate": 3.2387497603938326e-05,
      "loss": 1.9279,
      "step": 315
    },
    {
      "epoch": 0.455988455988456,
      "grad_norm": 0.4141858220100403,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 1.7844,
      "step": 316
    },
    {
      "epoch": 0.4574314574314574,
      "grad_norm": 0.3475481867790222,
      "learning_rate": 3.177647500604252e-05,
      "loss": 1.833,
      "step": 317
    },
    {
      "epoch": 0.4588744588744589,
      "grad_norm": 0.366415411233902,
      "learning_rate": 3.147212812450819e-05,
      "loss": 2.0163,
      "step": 318
    },
    {
      "epoch": 0.4603174603174603,
      "grad_norm": 0.3866393566131592,
      "learning_rate": 3.116857491533947e-05,
      "loss": 1.7986,
      "step": 319
    },
    {
      "epoch": 0.46176046176046176,
      "grad_norm": 0.3545255661010742,
      "learning_rate": 3.086582838174551e-05,
      "loss": 2.0761,
      "step": 320
    },
    {
      "epoch": 0.46320346320346323,
      "grad_norm": 0.42348945140838623,
      "learning_rate": 3.056390149238022e-05,
      "loss": 1.9795,
      "step": 321
    },
    {
      "epoch": 0.46464646464646464,
      "grad_norm": 0.36564403772354126,
      "learning_rate": 3.0262807180786647e-05,
      "loss": 1.8851,
      "step": 322
    },
    {
      "epoch": 0.4660894660894661,
      "grad_norm": 0.3915037512779236,
      "learning_rate": 2.996255834484296e-05,
      "loss": 1.9392,
      "step": 323
    },
    {
      "epoch": 0.4675324675324675,
      "grad_norm": 0.38861963152885437,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.9099,
      "step": 324
    },
    {
      "epoch": 0.468975468975469,
      "grad_norm": 0.38369008898735046,
      "learning_rate": 2.936464850978027e-05,
      "loss": 1.7798,
      "step": 325
    },
    {
      "epoch": 0.4704184704184704,
      "grad_norm": 0.42101967334747314,
      "learning_rate": 2.9067013123128613e-05,
      "loss": 1.8606,
      "step": 326
    },
    {
      "epoch": 0.47186147186147187,
      "grad_norm": 0.41875502467155457,
      "learning_rate": 2.8770274435964355e-05,
      "loss": 1.6667,
      "step": 327
    },
    {
      "epoch": 0.4733044733044733,
      "grad_norm": 0.41931435465812683,
      "learning_rate": 2.8474445159585235e-05,
      "loss": 1.8602,
      "step": 328
    },
    {
      "epoch": 0.47474747474747475,
      "grad_norm": 0.46239984035491943,
      "learning_rate": 2.8179537966332887e-05,
      "loss": 1.9873,
      "step": 329
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.4297107756137848,
      "learning_rate": 2.7885565489049946e-05,
      "loss": 1.8147,
      "step": 330
    },
    {
      "epoch": 0.47763347763347763,
      "grad_norm": 0.4730760455131531,
      "learning_rate": 2.759254032053888e-05,
      "loss": 2.121,
      "step": 331
    },
    {
      "epoch": 0.4790764790764791,
      "grad_norm": 0.4633348882198334,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 1.862,
      "step": 332
    },
    {
      "epoch": 0.4805194805194805,
      "grad_norm": 0.451509565114975,
      "learning_rate": 2.700938207760701e-05,
      "loss": 1.717,
      "step": 333
    },
    {
      "epoch": 0.481962481962482,
      "grad_norm": 0.5135688781738281,
      "learning_rate": 2.671927398374443e-05,
      "loss": 1.7786,
      "step": 334
    },
    {
      "epoch": 0.4834054834054834,
      "grad_norm": 0.5174142718315125,
      "learning_rate": 2.6430163158700115e-05,
      "loss": 1.8283,
      "step": 335
    },
    {
      "epoch": 0.48484848484848486,
      "grad_norm": 0.4473469853401184,
      "learning_rate": 2.6142061987019577e-05,
      "loss": 1.9845,
      "step": 336
    },
    {
      "epoch": 0.4862914862914863,
      "grad_norm": 0.4503369629383087,
      "learning_rate": 2.5854982809998153e-05,
      "loss": 1.759,
      "step": 337
    },
    {
      "epoch": 0.48773448773448774,
      "grad_norm": 0.6512835621833801,
      "learning_rate": 2.556893792515227e-05,
      "loss": 1.7662,
      "step": 338
    },
    {
      "epoch": 0.48917748917748916,
      "grad_norm": 0.5281562209129333,
      "learning_rate": 2.5283939585692783e-05,
      "loss": 1.791,
      "step": 339
    },
    {
      "epoch": 0.4906204906204906,
      "grad_norm": 0.6262437701225281,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.7711,
      "step": 340
    },
    {
      "epoch": 0.49206349206349204,
      "grad_norm": 0.5214657783508301,
      "learning_rate": 2.471713133110078e-05,
      "loss": 1.7583,
      "step": 341
    },
    {
      "epoch": 0.4935064935064935,
      "grad_norm": 0.5274417996406555,
      "learning_rate": 2.4435345696147403e-05,
      "loss": 1.8388,
      "step": 342
    },
    {
      "epoch": 0.494949494949495,
      "grad_norm": 0.5707380175590515,
      "learning_rate": 2.4154655165898627e-05,
      "loss": 2.0271,
      "step": 343
    },
    {
      "epoch": 0.4963924963924964,
      "grad_norm": 0.6870908141136169,
      "learning_rate": 2.3875071764202563e-05,
      "loss": 2.0183,
      "step": 344
    },
    {
      "epoch": 0.49783549783549785,
      "grad_norm": 0.6021436452865601,
      "learning_rate": 2.3596607467481603e-05,
      "loss": 1.8119,
      "step": 345
    },
    {
      "epoch": 0.49927849927849927,
      "grad_norm": 0.7010958790779114,
      "learning_rate": 2.3319274204219428e-05,
      "loss": 1.7931,
      "step": 346
    },
    {
      "epoch": 0.5007215007215007,
      "grad_norm": 0.7699779868125916,
      "learning_rate": 2.3043083854449988e-05,
      "loss": 1.6772,
      "step": 347
    },
    {
      "epoch": 0.5021645021645021,
      "grad_norm": 0.7229447364807129,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 1.8659,
      "step": 348
    },
    {
      "epoch": 0.5036075036075036,
      "grad_norm": 0.8113691210746765,
      "learning_rate": 2.2494179170225333e-05,
      "loss": 1.7651,
      "step": 349
    },
    {
      "epoch": 0.5050505050505051,
      "grad_norm": 1.5266594886779785,
      "learning_rate": 2.2221488349019903e-05,
      "loss": 1.9331,
      "step": 350
    },
    {
      "epoch": 0.5064935064935064,
      "grad_norm": 0.2073773890733719,
      "learning_rate": 2.194998746679952e-05,
      "loss": 1.8606,
      "step": 351
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 0.21894040703773499,
      "learning_rate": 2.167968815375837e-05,
      "loss": 1.6546,
      "step": 352
    },
    {
      "epoch": 0.5093795093795094,
      "grad_norm": 0.24405646324157715,
      "learning_rate": 2.1410601988619394e-05,
      "loss": 1.6878,
      "step": 353
    },
    {
      "epoch": 0.5108225108225108,
      "grad_norm": 0.23255270719528198,
      "learning_rate": 2.1142740498138324e-05,
      "loss": 1.7686,
      "step": 354
    },
    {
      "epoch": 0.5122655122655123,
      "grad_norm": 0.2648371458053589,
      "learning_rate": 2.08761151566099e-05,
      "loss": 1.7577,
      "step": 355
    },
    {
      "epoch": 0.5137085137085137,
      "grad_norm": 0.28337547183036804,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.8133,
      "step": 356
    },
    {
      "epoch": 0.5151515151515151,
      "grad_norm": 0.29984724521636963,
      "learning_rate": 2.034661855233815e-05,
      "loss": 1.6594,
      "step": 357
    },
    {
      "epoch": 0.5165945165945166,
      "grad_norm": 0.3129202425479889,
      "learning_rate": 2.008376997146705e-05,
      "loss": 1.9644,
      "step": 358
    },
    {
      "epoch": 0.5180375180375181,
      "grad_norm": 0.678475558757782,
      "learning_rate": 1.982220290232143e-05,
      "loss": 1.6171,
      "step": 359
    },
    {
      "epoch": 0.5194805194805194,
      "grad_norm": 0.2939353585243225,
      "learning_rate": 1.9561928549563968e-05,
      "loss": 1.9871,
      "step": 360
    },
    {
      "epoch": 0.5209235209235209,
      "grad_norm": 0.28600218892097473,
      "learning_rate": 1.9302958062481673e-05,
      "loss": 1.8726,
      "step": 361
    },
    {
      "epoch": 0.5223665223665224,
      "grad_norm": 0.30675143003463745,
      "learning_rate": 1.9045302534508297e-05,
      "loss": 1.8862,
      "step": 362
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 0.32877466082572937,
      "learning_rate": 1.8788973002749112e-05,
      "loss": 1.7572,
      "step": 363
    },
    {
      "epoch": 0.5252525252525253,
      "grad_norm": 0.3716675043106079,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 1.8445,
      "step": 364
    },
    {
      "epoch": 0.5266955266955267,
      "grad_norm": 0.3779400885105133,
      "learning_rate": 1.8280335791817733e-05,
      "loss": 1.9895,
      "step": 365
    },
    {
      "epoch": 0.5281385281385281,
      "grad_norm": 0.38026732206344604,
      "learning_rate": 1.8028049900970767e-05,
      "loss": 2.0318,
      "step": 366
    },
    {
      "epoch": 0.5295815295815296,
      "grad_norm": 0.37040016055107117,
      "learning_rate": 1.777713358205514e-05,
      "loss": 1.8038,
      "step": 367
    },
    {
      "epoch": 0.5310245310245311,
      "grad_norm": 0.3922972083091736,
      "learning_rate": 1.7527597583490822e-05,
      "loss": 1.7281,
      "step": 368
    },
    {
      "epoch": 0.5324675324675324,
      "grad_norm": 0.4298830032348633,
      "learning_rate": 1.7279452594569483e-05,
      "loss": 1.8546,
      "step": 369
    },
    {
      "epoch": 0.5339105339105339,
      "grad_norm": 0.39382922649383545,
      "learning_rate": 1.703270924499656e-05,
      "loss": 1.9061,
      "step": 370
    },
    {
      "epoch": 0.5353535353535354,
      "grad_norm": 0.4678158164024353,
      "learning_rate": 1.678737810443593e-05,
      "loss": 1.8942,
      "step": 371
    },
    {
      "epoch": 0.5367965367965368,
      "grad_norm": 0.3930918872356415,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.845,
      "step": 372
    },
    {
      "epoch": 0.5382395382395382,
      "grad_norm": 0.3660011291503906,
      "learning_rate": 1.6300994426085103e-05,
      "loss": 1.8379,
      "step": 373
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 0.444845050573349,
      "learning_rate": 1.605996272335291e-05,
      "loss": 2.0223,
      "step": 374
    },
    {
      "epoch": 0.5411255411255411,
      "grad_norm": 0.4140741527080536,
      "learning_rate": 1.5820384898856434e-05,
      "loss": 1.8988,
      "step": 375
    },
    {
      "epoch": 0.5425685425685426,
      "grad_norm": 0.4683399498462677,
      "learning_rate": 1.5582271215312294e-05,
      "loss": 1.7753,
      "step": 376
    },
    {
      "epoch": 0.5440115440115441,
      "grad_norm": 0.39313170313835144,
      "learning_rate": 1.5345631872718214e-05,
      "loss": 1.9094,
      "step": 377
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.3654247224330902,
      "learning_rate": 1.5110477007916001e-05,
      "loss": 1.7358,
      "step": 378
    },
    {
      "epoch": 0.5468975468975469,
      "grad_norm": 0.44113779067993164,
      "learning_rate": 1.4876816694157419e-05,
      "loss": 1.9154,
      "step": 379
    },
    {
      "epoch": 0.5483405483405484,
      "grad_norm": 0.39124253392219543,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.9706,
      "step": 380
    },
    {
      "epoch": 0.5497835497835498,
      "grad_norm": 0.46143198013305664,
      "learning_rate": 1.4414019692241437e-05,
      "loss": 1.809,
      "step": 381
    },
    {
      "epoch": 0.5512265512265512,
      "grad_norm": 0.4628862142562866,
      "learning_rate": 1.4184902828767287e-05,
      "loss": 1.7279,
      "step": 382
    },
    {
      "epoch": 0.5526695526695526,
      "grad_norm": 0.49773210287094116,
      "learning_rate": 1.3957320164854059e-05,
      "loss": 1.8587,
      "step": 383
    },
    {
      "epoch": 0.5541125541125541,
      "grad_norm": 0.4421599507331848,
      "learning_rate": 1.373128144938563e-05,
      "loss": 1.7864,
      "step": 384
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.4303032457828522,
      "learning_rate": 1.3506796365108232e-05,
      "loss": 1.8105,
      "step": 385
    },
    {
      "epoch": 0.556998556998557,
      "grad_norm": 0.4077335000038147,
      "learning_rate": 1.3283874528215733e-05,
      "loss": 1.7978,
      "step": 386
    },
    {
      "epoch": 0.5584415584415584,
      "grad_norm": 0.5202664732933044,
      "learning_rate": 1.3062525487937699e-05,
      "loss": 1.6826,
      "step": 387
    },
    {
      "epoch": 0.5598845598845599,
      "grad_norm": 0.46949782967567444,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 1.8609,
      "step": 388
    },
    {
      "epoch": 0.5613275613275613,
      "grad_norm": 0.6220776438713074,
      "learning_rate": 1.2624583656870154e-05,
      "loss": 1.6619,
      "step": 389
    },
    {
      "epoch": 0.5627705627705628,
      "grad_norm": 0.5599220991134644,
      "learning_rate": 1.2408009626051137e-05,
      "loss": 1.8675,
      "step": 390
    },
    {
      "epoch": 0.5642135642135642,
      "grad_norm": 0.48732367157936096,
      "learning_rate": 1.2193045910983863e-05,
      "loss": 1.8726,
      "step": 391
    },
    {
      "epoch": 0.5656565656565656,
      "grad_norm": 0.5116375088691711,
      "learning_rate": 1.1979701719998453e-05,
      "loss": 1.8269,
      "step": 392
    },
    {
      "epoch": 0.5670995670995671,
      "grad_norm": 0.5755485892295837,
      "learning_rate": 1.1767986192049984e-05,
      "loss": 2.0062,
      "step": 393
    },
    {
      "epoch": 0.5685425685425686,
      "grad_norm": 0.5523481369018555,
      "learning_rate": 1.1557908396327028e-05,
      "loss": 1.8802,
      "step": 394
    },
    {
      "epoch": 0.56998556998557,
      "grad_norm": 0.5976821780204773,
      "learning_rate": 1.134947733186315e-05,
      "loss": 1.9027,
      "step": 395
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.672989547252655,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 1.6237,
      "step": 396
    },
    {
      "epoch": 0.5728715728715729,
      "grad_norm": 0.717589259147644,
      "learning_rate": 1.0937591039762085e-05,
      "loss": 1.8292,
      "step": 397
    },
    {
      "epoch": 0.5743145743145743,
      "grad_norm": 0.6616857051849365,
      "learning_rate": 1.0734153455962765e-05,
      "loss": 1.6972,
      "step": 398
    },
    {
      "epoch": 0.5757575757575758,
      "grad_norm": 0.7580000162124634,
      "learning_rate": 1.0532397890342505e-05,
      "loss": 1.4864,
      "step": 399
    },
    {
      "epoch": 0.5772005772005772,
      "grad_norm": 1.0676060914993286,
      "learning_rate": 1.0332332985438248e-05,
      "loss": 1.6269,
      "step": 400
    },
    {
      "epoch": 0.5772005772005772,
      "eval_loss": 1.8380851745605469,
      "eval_runtime": 156.9174,
      "eval_samples_per_second": 7.437,
      "eval_steps_per_second": 1.861,
      "step": 400
    },
    {
      "epoch": 0.5786435786435786,
      "grad_norm": 0.22474224865436554,
      "learning_rate": 1.013396731136465e-05,
      "loss": 1.7378,
      "step": 401
    },
    {
      "epoch": 0.5800865800865801,
      "grad_norm": 0.20946919918060303,
      "learning_rate": 9.937309365446973e-06,
      "loss": 1.8411,
      "step": 402
    },
    {
      "epoch": 0.5815295815295816,
      "grad_norm": 0.203902006149292,
      "learning_rate": 9.742367571857091e-06,
      "loss": 1.6986,
      "step": 403
    },
    {
      "epoch": 0.5829725829725829,
      "grad_norm": 0.2520330548286438,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.7644,
      "step": 404
    },
    {
      "epoch": 0.5844155844155844,
      "grad_norm": 0.23222126066684723,
      "learning_rate": 9.357665770419244e-06,
      "loss": 1.8559,
      "step": 405
    },
    {
      "epoch": 0.5858585858585859,
      "grad_norm": 0.25231245160102844,
      "learning_rate": 9.167922241916055e-06,
      "loss": 1.9885,
      "step": 406
    },
    {
      "epoch": 0.5873015873015873,
      "grad_norm": 0.24505528807640076,
      "learning_rate": 8.97992782372432e-06,
      "loss": 1.8296,
      "step": 407
    },
    {
      "epoch": 0.5887445887445888,
      "grad_norm": 0.32213860750198364,
      "learning_rate": 8.793690568899216e-06,
      "loss": 1.928,
      "step": 408
    },
    {
      "epoch": 0.5901875901875901,
      "grad_norm": 0.28556668758392334,
      "learning_rate": 8.609218455224893e-06,
      "loss": 1.9806,
      "step": 409
    },
    {
      "epoch": 0.5916305916305916,
      "grad_norm": 0.28748056292533875,
      "learning_rate": 8.426519384872733e-06,
      "loss": 2.0535,
      "step": 410
    },
    {
      "epoch": 0.5930735930735931,
      "grad_norm": 0.3097786605358124,
      "learning_rate": 8.245601184062852e-06,
      "loss": 1.8546,
      "step": 411
    },
    {
      "epoch": 0.5945165945165946,
      "grad_norm": 0.34151002764701843,
      "learning_rate": 8.066471602728803e-06,
      "loss": 2.0569,
      "step": 412
    },
    {
      "epoch": 0.5959595959595959,
      "grad_norm": 0.31291815638542175,
      "learning_rate": 7.889138314185678e-06,
      "loss": 1.7969,
      "step": 413
    },
    {
      "epoch": 0.5974025974025974,
      "grad_norm": 0.3612555265426636,
      "learning_rate": 7.71360891480134e-06,
      "loss": 1.7247,
      "step": 414
    },
    {
      "epoch": 0.5988455988455988,
      "grad_norm": 0.3746615946292877,
      "learning_rate": 7.539890923671062e-06,
      "loss": 1.9217,
      "step": 415
    },
    {
      "epoch": 0.6002886002886003,
      "grad_norm": 0.4344548285007477,
      "learning_rate": 7.367991782295391e-06,
      "loss": 2.1042,
      "step": 416
    },
    {
      "epoch": 0.6017316017316018,
      "grad_norm": 0.31650206446647644,
      "learning_rate": 7.197918854261432e-06,
      "loss": 1.814,
      "step": 417
    },
    {
      "epoch": 0.6031746031746031,
      "grad_norm": 0.37703970074653625,
      "learning_rate": 7.029679424927365e-06,
      "loss": 1.9636,
      "step": 418
    },
    {
      "epoch": 0.6046176046176046,
      "grad_norm": 0.37509685754776,
      "learning_rate": 6.863280701110408e-06,
      "loss": 1.7446,
      "step": 419
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 0.33785712718963623,
      "learning_rate": 6.698729810778065e-06,
      "loss": 1.8171,
      "step": 420
    },
    {
      "epoch": 0.6075036075036075,
      "grad_norm": 0.36930790543556213,
      "learning_rate": 6.536033802742813e-06,
      "loss": 1.9212,
      "step": 421
    },
    {
      "epoch": 0.6089466089466089,
      "grad_norm": 0.41896510124206543,
      "learning_rate": 6.375199646360142e-06,
      "loss": 1.9301,
      "step": 422
    },
    {
      "epoch": 0.6103896103896104,
      "grad_norm": 0.3452630639076233,
      "learning_rate": 6.216234231230012e-06,
      "loss": 2.0033,
      "step": 423
    },
    {
      "epoch": 0.6118326118326118,
      "grad_norm": 0.4596862196922302,
      "learning_rate": 6.059144366901736e-06,
      "loss": 1.9092,
      "step": 424
    },
    {
      "epoch": 0.6132756132756133,
      "grad_norm": 0.42444470524787903,
      "learning_rate": 5.903936782582253e-06,
      "loss": 1.9446,
      "step": 425
    },
    {
      "epoch": 0.6147186147186147,
      "grad_norm": 0.45510318875312805,
      "learning_rate": 5.750618126847912e-06,
      "loss": 1.9616,
      "step": 426
    },
    {
      "epoch": 0.6161616161616161,
      "grad_norm": 0.4246458113193512,
      "learning_rate": 5.599194967359639e-06,
      "loss": 2.0316,
      "step": 427
    },
    {
      "epoch": 0.6176046176046176,
      "grad_norm": 0.43000173568725586,
      "learning_rate": 5.449673790581611e-06,
      "loss": 1.8851,
      "step": 428
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 0.4248652160167694,
      "learning_rate": 5.302061001503394e-06,
      "loss": 2.0064,
      "step": 429
    },
    {
      "epoch": 0.6204906204906205,
      "grad_norm": 0.4813087284564972,
      "learning_rate": 5.156362923365588e-06,
      "loss": 1.9639,
      "step": 430
    },
    {
      "epoch": 0.6219336219336219,
      "grad_norm": 0.42517709732055664,
      "learning_rate": 5.012585797388936e-06,
      "loss": 1.9537,
      "step": 431
    },
    {
      "epoch": 0.6233766233766234,
      "grad_norm": 0.413674920797348,
      "learning_rate": 4.87073578250698e-06,
      "loss": 1.911,
      "step": 432
    },
    {
      "epoch": 0.6248196248196248,
      "grad_norm": 0.44890064001083374,
      "learning_rate": 4.730818955102234e-06,
      "loss": 1.8757,
      "step": 433
    },
    {
      "epoch": 0.6262626262626263,
      "grad_norm": 0.5599820613861084,
      "learning_rate": 4.592841308745932e-06,
      "loss": 1.7325,
      "step": 434
    },
    {
      "epoch": 0.6277056277056277,
      "grad_norm": 0.504962682723999,
      "learning_rate": 4.456808753941205e-06,
      "loss": 1.9556,
      "step": 435
    },
    {
      "epoch": 0.6291486291486291,
      "grad_norm": 0.4306437373161316,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.8258,
      "step": 436
    },
    {
      "epoch": 0.6305916305916306,
      "grad_norm": 0.481152206659317,
      "learning_rate": 4.190602144143207e-06,
      "loss": 1.9023,
      "step": 437
    },
    {
      "epoch": 0.6320346320346321,
      "grad_norm": 0.43929627537727356,
      "learning_rate": 4.06043949255509e-06,
      "loss": 1.6293,
      "step": 438
    },
    {
      "epoch": 0.6334776334776335,
      "grad_norm": 0.5508288145065308,
      "learning_rate": 3.932244738840379e-06,
      "loss": 1.8766,
      "step": 439
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 0.503813624382019,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 1.7276,
      "step": 440
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.5829843282699585,
      "learning_rate": 3.681780806244095e-06,
      "loss": 1.7256,
      "step": 441
    },
    {
      "epoch": 0.6378066378066378,
      "grad_norm": 0.5285375714302063,
      "learning_rate": 3.5595223564037884e-06,
      "loss": 1.7845,
      "step": 442
    },
    {
      "epoch": 0.6392496392496393,
      "grad_norm": 0.5721433758735657,
      "learning_rate": 3.4392532620598216e-06,
      "loss": 1.6747,
      "step": 443
    },
    {
      "epoch": 0.6406926406926406,
      "grad_norm": 0.5801364183425903,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 1.9831,
      "step": 444
    },
    {
      "epoch": 0.6421356421356421,
      "grad_norm": 0.6420133113861084,
      "learning_rate": 3.2047036621337236e-06,
      "loss": 1.842,
      "step": 445
    },
    {
      "epoch": 0.6435786435786436,
      "grad_norm": 0.657759428024292,
      "learning_rate": 3.0904332038757977e-06,
      "loss": 1.732,
      "step": 446
    },
    {
      "epoch": 0.645021645021645,
      "grad_norm": 0.6824421882629395,
      "learning_rate": 2.978172195332263e-06,
      "loss": 1.5208,
      "step": 447
    },
    {
      "epoch": 0.6464646464646465,
      "grad_norm": 0.7443853616714478,
      "learning_rate": 2.8679254453910785e-06,
      "loss": 1.7159,
      "step": 448
    },
    {
      "epoch": 0.6479076479076479,
      "grad_norm": 0.8423395156860352,
      "learning_rate": 2.759697676656098e-06,
      "loss": 1.6525,
      "step": 449
    },
    {
      "epoch": 0.6493506493506493,
      "grad_norm": 1.326777696609497,
      "learning_rate": 2.653493525244721e-06,
      "loss": 1.8648,
      "step": 450
    },
    {
      "epoch": 0.6507936507936508,
      "grad_norm": 0.20779266953468323,
      "learning_rate": 2.549317540589308e-06,
      "loss": 1.6705,
      "step": 451
    },
    {
      "epoch": 0.6522366522366523,
      "grad_norm": 0.21033185720443726,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.7687,
      "step": 452
    },
    {
      "epoch": 0.6536796536796536,
      "grad_norm": 0.2421332150697708,
      "learning_rate": 2.3470678346851518e-06,
      "loss": 1.689,
      "step": 453
    },
    {
      "epoch": 0.6551226551226551,
      "grad_norm": 0.26601821184158325,
      "learning_rate": 2.2490027771406687e-06,
      "loss": 1.8311,
      "step": 454
    },
    {
      "epoch": 0.6565656565656566,
      "grad_norm": 0.2268337607383728,
      "learning_rate": 2.152983213389559e-06,
      "loss": 1.8122,
      "step": 455
    },
    {
      "epoch": 0.658008658008658,
      "grad_norm": 0.27295178174972534,
      "learning_rate": 2.0590132565903476e-06,
      "loss": 1.8871,
      "step": 456
    },
    {
      "epoch": 0.6594516594516594,
      "grad_norm": 0.25046512484550476,
      "learning_rate": 1.9670969321032407e-06,
      "loss": 1.9112,
      "step": 457
    },
    {
      "epoch": 0.6608946608946609,
      "grad_norm": 0.2787816822528839,
      "learning_rate": 1.8772381773176417e-06,
      "loss": 1.8829,
      "step": 458
    },
    {
      "epoch": 0.6623376623376623,
      "grad_norm": 0.28614646196365356,
      "learning_rate": 1.7894408414835362e-06,
      "loss": 1.9613,
      "step": 459
    },
    {
      "epoch": 0.6637806637806638,
      "grad_norm": 0.29104024171829224,
      "learning_rate": 1.70370868554659e-06,
      "loss": 1.883,
      "step": 460
    },
    {
      "epoch": 0.6652236652236653,
      "grad_norm": 0.3024556040763855,
      "learning_rate": 1.620045381987012e-06,
      "loss": 1.7852,
      "step": 461
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.3180261552333832,
      "learning_rate": 1.5384545146622852e-06,
      "loss": 1.8743,
      "step": 462
    },
    {
      "epoch": 0.6681096681096681,
      "grad_norm": 0.34793442487716675,
      "learning_rate": 1.4589395786535953e-06,
      "loss": 1.7692,
      "step": 463
    },
    {
      "epoch": 0.6695526695526696,
      "grad_norm": 0.3252454698085785,
      "learning_rate": 1.3815039801161721e-06,
      "loss": 1.9941,
      "step": 464
    },
    {
      "epoch": 0.670995670995671,
      "grad_norm": 0.38873234391212463,
      "learning_rate": 1.3061510361333185e-06,
      "loss": 1.8367,
      "step": 465
    },
    {
      "epoch": 0.6724386724386724,
      "grad_norm": 0.3405764102935791,
      "learning_rate": 1.232883974574367e-06,
      "loss": 1.9568,
      "step": 466
    },
    {
      "epoch": 0.6738816738816739,
      "grad_norm": 0.4072246253490448,
      "learning_rate": 1.1617059339563807e-06,
      "loss": 1.9118,
      "step": 467
    },
    {
      "epoch": 0.6753246753246753,
      "grad_norm": 0.38311657309532166,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.9212,
      "step": 468
    },
    {
      "epoch": 0.6767676767676768,
      "grad_norm": 0.372346431016922,
      "learning_rate": 1.0256290220474307e-06,
      "loss": 1.8625,
      "step": 469
    },
    {
      "epoch": 0.6782106782106783,
      "grad_norm": 0.32480254769325256,
      "learning_rate": 9.607359798384785e-07,
      "loss": 2.0927,
      "step": 470
    },
    {
      "epoch": 0.6796536796536796,
      "grad_norm": 0.349486380815506,
      "learning_rate": 8.979436164848088e-07,
      "loss": 1.8977,
      "step": 471
    },
    {
      "epoch": 0.6810966810966811,
      "grad_norm": 0.5002724528312683,
      "learning_rate": 8.372546218022747e-07,
      "loss": 1.7969,
      "step": 472
    },
    {
      "epoch": 0.6825396825396826,
      "grad_norm": 0.43671897053718567,
      "learning_rate": 7.786715955054203e-07,
      "loss": 2.0032,
      "step": 473
    },
    {
      "epoch": 0.683982683982684,
      "grad_norm": 0.43542805314064026,
      "learning_rate": 7.221970470961125e-07,
      "loss": 1.9024,
      "step": 474
    },
    {
      "epoch": 0.6854256854256854,
      "grad_norm": 0.4874102771282196,
      "learning_rate": 6.678333957560512e-07,
      "loss": 1.6566,
      "step": 475
    },
    {
      "epoch": 0.6868686868686869,
      "grad_norm": 0.4170205891132355,
      "learning_rate": 6.15582970243117e-07,
      "loss": 1.8643,
      "step": 476
    },
    {
      "epoch": 0.6883116883116883,
      "grad_norm": 0.4337291717529297,
      "learning_rate": 5.654480087916303e-07,
      "loss": 2.0587,
      "step": 477
    },
    {
      "epoch": 0.6897546897546898,
      "grad_norm": 0.4315203130245209,
      "learning_rate": 5.174306590164879e-07,
      "loss": 2.0489,
      "step": 478
    },
    {
      "epoch": 0.6911976911976911,
      "grad_norm": 0.43969807028770447,
      "learning_rate": 4.715329778211375e-07,
      "loss": 1.9063,
      "step": 479
    },
    {
      "epoch": 0.6926406926406926,
      "grad_norm": 0.39184656739234924,
      "learning_rate": 4.277569313094809e-07,
      "loss": 1.7943,
      "step": 480
    },
    {
      "epoch": 0.6940836940836941,
      "grad_norm": 0.43797311186790466,
      "learning_rate": 3.8610439470164737e-07,
      "loss": 1.9579,
      "step": 481
    },
    {
      "epoch": 0.6955266955266955,
      "grad_norm": 0.45759716629981995,
      "learning_rate": 3.465771522536854e-07,
      "loss": 1.7868,
      "step": 482
    },
    {
      "epoch": 0.696969696969697,
      "grad_norm": 0.530158281326294,
      "learning_rate": 3.09176897181096e-07,
      "loss": 1.9113,
      "step": 483
    },
    {
      "epoch": 0.6984126984126984,
      "grad_norm": 0.4709303081035614,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 1.9702,
      "step": 484
    },
    {
      "epoch": 0.6998556998556998,
      "grad_norm": 0.5338622331619263,
      "learning_rate": 2.407636663901591e-07,
      "loss": 1.8965,
      "step": 485
    },
    {
      "epoch": 0.7012987012987013,
      "grad_norm": 0.48375779390335083,
      "learning_rate": 2.0975362126691712e-07,
      "loss": 1.6047,
      "step": 486
    },
    {
      "epoch": 0.7027417027417028,
      "grad_norm": 0.4987216293811798,
      "learning_rate": 1.8087642458373134e-07,
      "loss": 1.8085,
      "step": 487
    },
    {
      "epoch": 0.7041847041847041,
      "grad_norm": 0.5824238657951355,
      "learning_rate": 1.5413331334360182e-07,
      "loss": 1.6619,
      "step": 488
    },
    {
      "epoch": 0.7056277056277056,
      "grad_norm": 0.5118747353553772,
      "learning_rate": 1.2952543313240472e-07,
      "loss": 1.7263,
      "step": 489
    },
    {
      "epoch": 0.7070707070707071,
      "grad_norm": 0.6156334280967712,
      "learning_rate": 1.0705383806982606e-07,
      "loss": 1.8636,
      "step": 490
    },
    {
      "epoch": 0.7085137085137085,
      "grad_norm": 0.587309718132019,
      "learning_rate": 8.671949076420882e-08,
      "loss": 1.6529,
      "step": 491
    },
    {
      "epoch": 0.70995670995671,
      "grad_norm": 0.5975311398506165,
      "learning_rate": 6.852326227130834e-08,
      "loss": 1.875,
      "step": 492
    },
    {
      "epoch": 0.7113997113997114,
      "grad_norm": 0.5370944142341614,
      "learning_rate": 5.246593205699424e-08,
      "loss": 1.8962,
      "step": 493
    },
    {
      "epoch": 0.7128427128427128,
      "grad_norm": 0.6013877391815186,
      "learning_rate": 3.8548187963854956e-08,
      "loss": 1.9036,
      "step": 494
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.6089028120040894,
      "learning_rate": 2.6770626181715773e-08,
      "loss": 1.7561,
      "step": 495
    },
    {
      "epoch": 0.7157287157287158,
      "grad_norm": 0.6195142865180969,
      "learning_rate": 1.7133751222137007e-08,
      "loss": 1.8843,
      "step": 496
    },
    {
      "epoch": 0.7171717171717171,
      "grad_norm": 0.7459947466850281,
      "learning_rate": 9.637975896759077e-09,
      "loss": 1.5594,
      "step": 497
    },
    {
      "epoch": 0.7186147186147186,
      "grad_norm": 0.8303995132446289,
      "learning_rate": 4.2836212996499865e-09,
      "loss": 1.6858,
      "step": 498
    },
    {
      "epoch": 0.7200577200577201,
      "grad_norm": 0.9504781365394592,
      "learning_rate": 1.0709167935385455e-09,
      "loss": 1.4528,
      "step": 499
    },
    {
      "epoch": 0.7215007215007215,
      "grad_norm": 1.5718448162078857,
      "learning_rate": 0.0,
      "loss": 1.6021,
      "step": 500
    },
    {
      "epoch": 0.7215007215007215,
      "eval_loss": 1.8335492610931396,
      "eval_runtime": 156.8048,
      "eval_samples_per_second": 7.442,
      "eval_steps_per_second": 1.862,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4073840298371318e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}