{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.055498458376156,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0041109969167523125,
      "grad_norm": 0.5151441097259521,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.1041,
      "step": 1
    },
    {
      "epoch": 0.008221993833504625,
      "grad_norm": 0.5600175261497498,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.1677,
      "step": 2
    },
    {
      "epoch": 0.012332990750256937,
      "grad_norm": 0.5339025855064392,
      "learning_rate": 6e-06,
      "loss": 1.2228,
      "step": 3
    },
    {
      "epoch": 0.01644398766700925,
      "grad_norm": 0.5728273391723633,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2463,
      "step": 4
    },
    {
      "epoch": 0.020554984583761562,
      "grad_norm": 0.5520905256271362,
      "learning_rate": 1e-05,
      "loss": 1.2652,
      "step": 5
    },
    {
      "epoch": 0.024665981500513873,
      "grad_norm": 0.5096163749694824,
      "learning_rate": 1.2e-05,
      "loss": 1.183,
      "step": 6
    },
    {
      "epoch": 0.02877697841726619,
      "grad_norm": 0.5273175835609436,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 1.1689,
      "step": 7
    },
    {
      "epoch": 0.0328879753340185,
      "grad_norm": 0.5087783336639404,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.1003,
      "step": 8
    },
    {
      "epoch": 0.03699897225077081,
      "grad_norm": 0.5192142724990845,
      "learning_rate": 1.8e-05,
      "loss": 1.1486,
      "step": 9
    },
    {
      "epoch": 0.041109969167523124,
      "grad_norm": 0.4961031377315521,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 10
    },
    {
      "epoch": 0.045220966084275435,
      "grad_norm": 0.5031614899635315,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 1.1498,
      "step": 11
    },
    {
      "epoch": 0.04933196300102775,
      "grad_norm": 0.5319791436195374,
      "learning_rate": 2.4e-05,
      "loss": 1.1691,
      "step": 12
    },
    {
      "epoch": 0.05344295991778006,
      "grad_norm": 0.5037567019462585,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 1.1247,
      "step": 13
    },
    {
      "epoch": 0.05755395683453238,
      "grad_norm": 0.55753493309021,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.1406,
      "step": 14
    },
    {
      "epoch": 0.06166495375128469,
      "grad_norm": 0.5214470624923706,
      "learning_rate": 3e-05,
      "loss": 1.0488,
      "step": 15
    },
    {
      "epoch": 0.065775950668037,
      "grad_norm": 0.5083463191986084,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 1.1067,
      "step": 16
    },
    {
      "epoch": 0.0698869475847893,
      "grad_norm": 0.6208635568618774,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 1.0384,
      "step": 17
    },
    {
      "epoch": 0.07399794450154162,
      "grad_norm": 0.6004786491394043,
      "learning_rate": 3.6e-05,
      "loss": 0.9681,
      "step": 18
    },
    {
      "epoch": 0.07810894141829394,
      "grad_norm": 0.6306799650192261,
      "learning_rate": 3.8e-05,
      "loss": 1.0048,
      "step": 19
    },
    {
      "epoch": 0.08221993833504625,
      "grad_norm": 0.6294493079185486,
      "learning_rate": 4e-05,
      "loss": 0.973,
      "step": 20
    },
    {
      "epoch": 0.08633093525179857,
      "grad_norm": 0.501514732837677,
      "learning_rate": 4.2e-05,
      "loss": 0.8689,
      "step": 21
    },
    {
      "epoch": 0.09044193216855087,
      "grad_norm": 0.6042078137397766,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.9267,
      "step": 22
    },
    {
      "epoch": 0.09455292908530319,
      "grad_norm": 0.5602875351905823,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.9484,
      "step": 23
    },
    {
      "epoch": 0.0986639260020555,
      "grad_norm": 0.4414026141166687,
      "learning_rate": 4.8e-05,
      "loss": 0.8629,
      "step": 24
    },
    {
      "epoch": 0.10277492291880781,
      "grad_norm": 0.40880197286605835,
      "learning_rate": 5e-05,
      "loss": 0.9374,
      "step": 25
    },
    {
      "epoch": 0.10688591983556012,
      "grad_norm": 0.4516941010951996,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.8808,
      "step": 26
    },
    {
      "epoch": 0.11099691675231244,
      "grad_norm": 0.44685766100883484,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.7999,
      "step": 27
    },
    {
      "epoch": 0.11510791366906475,
      "grad_norm": 0.3397183418273926,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.8174,
      "step": 28
    },
    {
      "epoch": 0.11921891058581706,
      "grad_norm": 0.3303641974925995,
      "learning_rate": 5.8e-05,
      "loss": 0.8282,
      "step": 29
    },
    {
      "epoch": 0.12332990750256938,
      "grad_norm": 0.35168400406837463,
      "learning_rate": 6e-05,
      "loss": 0.78,
      "step": 30
    },
    {
      "epoch": 0.12744090441932168,
      "grad_norm": 0.3936440050601959,
      "learning_rate": 6.2e-05,
      "loss": 0.7541,
      "step": 31
    },
    {
      "epoch": 0.131551901336074,
      "grad_norm": 0.25425174832344055,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.8346,
      "step": 32
    },
    {
      "epoch": 0.13566289825282632,
      "grad_norm": 0.24264051020145416,
      "learning_rate": 6.6e-05,
      "loss": 0.806,
      "step": 33
    },
    {
      "epoch": 0.1397738951695786,
      "grad_norm": 0.243570938706398,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.8501,
      "step": 34
    },
    {
      "epoch": 0.14388489208633093,
      "grad_norm": 0.2476535439491272,
      "learning_rate": 7e-05,
      "loss": 0.8286,
      "step": 35
    },
    {
      "epoch": 0.14799588900308325,
      "grad_norm": 0.2289520651102066,
      "learning_rate": 7.2e-05,
      "loss": 0.8261,
      "step": 36
    },
    {
      "epoch": 0.15210688591983557,
      "grad_norm": 0.24161428213119507,
      "learning_rate": 7.4e-05,
      "loss": 0.7589,
      "step": 37
    },
    {
      "epoch": 0.15621788283658788,
      "grad_norm": 0.23260338604450226,
      "learning_rate": 7.6e-05,
      "loss": 0.7867,
      "step": 38
    },
    {
      "epoch": 0.16032887975334018,
      "grad_norm": 0.23059183359146118,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.7848,
      "step": 39
    },
    {
      "epoch": 0.1644398766700925,
      "grad_norm": 0.21534432470798492,
      "learning_rate": 8e-05,
      "loss": 0.8778,
      "step": 40
    },
    {
      "epoch": 0.1685508735868448,
      "grad_norm": 0.20831508934497833,
      "learning_rate": 8.2e-05,
      "loss": 0.7785,
      "step": 41
    },
    {
      "epoch": 0.17266187050359713,
      "grad_norm": 0.2098928987979889,
      "learning_rate": 8.4e-05,
      "loss": 0.765,
      "step": 42
    },
    {
      "epoch": 0.17677286742034942,
      "grad_norm": 0.21017888188362122,
      "learning_rate": 8.6e-05,
      "loss": 0.8173,
      "step": 43
    },
    {
      "epoch": 0.18088386433710174,
      "grad_norm": 0.19125813245773315,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.836,
      "step": 44
    },
    {
      "epoch": 0.18499486125385406,
      "grad_norm": 0.20498064160346985,
      "learning_rate": 9e-05,
      "loss": 0.7212,
      "step": 45
    },
    {
      "epoch": 0.18910585817060638,
      "grad_norm": 0.20564904808998108,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.7611,
      "step": 46
    },
    {
      "epoch": 0.1932168550873587,
      "grad_norm": 0.21227578818798065,
      "learning_rate": 9.4e-05,
      "loss": 0.7702,
      "step": 47
    },
    {
      "epoch": 0.197327852004111,
      "grad_norm": 0.20378409326076508,
      "learning_rate": 9.6e-05,
      "loss": 0.8454,
      "step": 48
    },
    {
      "epoch": 0.2014388489208633,
      "grad_norm": 0.1814655214548111,
      "learning_rate": 9.8e-05,
      "loss": 0.7202,
      "step": 49
    },
    {
      "epoch": 0.20554984583761562,
      "grad_norm": 0.21666166186332703,
      "learning_rate": 0.0001,
      "loss": 0.7824,
      "step": 50
    },
    {
      "epoch": 0.20966084275436794,
      "grad_norm": 0.20601877570152283,
      "learning_rate": 0.00010200000000000001,
      "loss": 0.7651,
      "step": 51
    },
    {
      "epoch": 0.21377183967112023,
      "grad_norm": 0.20845171809196472,
      "learning_rate": 0.00010400000000000001,
      "loss": 0.7811,
      "step": 52
    },
    {
      "epoch": 0.21788283658787255,
      "grad_norm": 0.2032238095998764,
      "learning_rate": 0.00010600000000000002,
      "loss": 0.8483,
      "step": 53
    },
    {
      "epoch": 0.22199383350462487,
      "grad_norm": 0.241808220744133,
      "learning_rate": 0.00010800000000000001,
      "loss": 0.7992,
      "step": 54
    },
    {
      "epoch": 0.2261048304213772,
      "grad_norm": 0.26223278045654297,
      "learning_rate": 0.00011000000000000002,
      "loss": 0.7839,
      "step": 55
    },
    {
      "epoch": 0.2302158273381295,
      "grad_norm": 0.20659419894218445,
      "learning_rate": 0.00011200000000000001,
      "loss": 0.7907,
      "step": 56
    },
    {
      "epoch": 0.2343268242548818,
      "grad_norm": 0.18811534345149994,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.7224,
      "step": 57
    },
    {
      "epoch": 0.23843782117163412,
      "grad_norm": 0.2020961493253708,
      "learning_rate": 0.000116,
      "loss": 0.7795,
      "step": 58
    },
    {
      "epoch": 0.24254881808838644,
      "grad_norm": 0.22986751794815063,
      "learning_rate": 0.000118,
      "loss": 0.7843,
      "step": 59
    },
    {
      "epoch": 0.24665981500513876,
      "grad_norm": 0.21348616480827332,
      "learning_rate": 0.00012,
      "loss": 0.817,
      "step": 60
    },
    {
      "epoch": 0.25077081192189105,
      "grad_norm": 0.21506637334823608,
      "learning_rate": 0.000122,
      "loss": 0.7852,
      "step": 61
    },
    {
      "epoch": 0.25488180883864336,
      "grad_norm": 0.2041584700345993,
      "learning_rate": 0.000124,
      "loss": 0.7452,
      "step": 62
    },
    {
      "epoch": 0.2589928057553957,
      "grad_norm": 0.18704313039779663,
      "learning_rate": 0.000126,
      "loss": 0.7696,
      "step": 63
    },
    {
      "epoch": 0.263103802672148,
      "grad_norm": 0.19755174219608307,
      "learning_rate": 0.00012800000000000002,
      "loss": 0.7934,
      "step": 64
    },
    {
      "epoch": 0.2672147995889003,
      "grad_norm": 0.2039138674736023,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.8554,
      "step": 65
    },
    {
      "epoch": 0.27132579650565264,
      "grad_norm": 0.20087020099163055,
      "learning_rate": 0.000132,
      "loss": 0.7663,
      "step": 66
    },
    {
      "epoch": 0.27543679342240496,
      "grad_norm": 0.19009239971637726,
      "learning_rate": 0.000134,
      "loss": 0.731,
      "step": 67
    },
    {
      "epoch": 0.2795477903391572,
      "grad_norm": 0.2003493756055832,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.8049,
      "step": 68
    },
    {
      "epoch": 0.28365878725590954,
      "grad_norm": 0.20522601902484894,
      "learning_rate": 0.000138,
      "loss": 0.803,
      "step": 69
    },
    {
      "epoch": 0.28776978417266186,
      "grad_norm": 0.20307479798793793,
      "learning_rate": 0.00014,
      "loss": 0.7891,
      "step": 70
    },
    {
      "epoch": 0.2918807810894142,
      "grad_norm": 0.19717766344547272,
      "learning_rate": 0.000142,
      "loss": 0.6747,
      "step": 71
    },
    {
      "epoch": 0.2959917780061665,
      "grad_norm": 0.2092077136039734,
      "learning_rate": 0.000144,
      "loss": 0.7276,
      "step": 72
    },
    {
      "epoch": 0.3001027749229188,
      "grad_norm": 0.18728697299957275,
      "learning_rate": 0.000146,
      "loss": 0.7613,
      "step": 73
    },
    {
      "epoch": 0.30421377183967113,
      "grad_norm": 0.20962484180927277,
      "learning_rate": 0.000148,
      "loss": 0.7826,
      "step": 74
    },
    {
      "epoch": 0.30832476875642345,
      "grad_norm": 0.20955528318881989,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.7594,
      "step": 75
    },
    {
      "epoch": 0.31243576567317577,
      "grad_norm": 0.21371231973171234,
      "learning_rate": 0.000152,
      "loss": 0.8138,
      "step": 76
    },
    {
      "epoch": 0.31654676258992803,
      "grad_norm": 0.19737087190151215,
      "learning_rate": 0.000154,
      "loss": 0.7475,
      "step": 77
    },
    {
      "epoch": 0.32065775950668035,
      "grad_norm": 0.21145501732826233,
      "learning_rate": 0.00015600000000000002,
      "loss": 0.7096,
      "step": 78
    },
    {
      "epoch": 0.32476875642343267,
      "grad_norm": 0.1865813434123993,
      "learning_rate": 0.00015800000000000002,
      "loss": 0.7511,
      "step": 79
    },
    {
      "epoch": 0.328879753340185,
      "grad_norm": 0.2682845890522003,
      "learning_rate": 0.00016,
      "loss": 0.717,
      "step": 80
    },
    {
      "epoch": 0.3329907502569373,
      "grad_norm": 0.20705395936965942,
      "learning_rate": 0.000162,
      "loss": 0.7805,
      "step": 81
    },
    {
      "epoch": 0.3371017471736896,
      "grad_norm": 0.1936197429895401,
      "learning_rate": 0.000164,
      "loss": 0.8372,
      "step": 82
    },
    {
      "epoch": 0.34121274409044194,
      "grad_norm": 0.19656707346439362,
      "learning_rate": 0.000166,
      "loss": 0.7773,
      "step": 83
    },
    {
      "epoch": 0.34532374100719426,
      "grad_norm": 0.21295683085918427,
      "learning_rate": 0.000168,
      "loss": 0.7854,
      "step": 84
    },
    {
      "epoch": 0.3494347379239466,
      "grad_norm": 0.19248339533805847,
      "learning_rate": 0.00017,
      "loss": 0.7646,
      "step": 85
    },
    {
      "epoch": 0.35354573484069884,
      "grad_norm": 0.2315826565027237,
      "learning_rate": 0.000172,
      "loss": 0.7105,
      "step": 86
    },
    {
      "epoch": 0.35765673175745116,
      "grad_norm": 0.20218686759471893,
      "learning_rate": 0.000174,
      "loss": 0.7725,
      "step": 87
    },
    {
      "epoch": 0.3617677286742035,
      "grad_norm": 0.20589414238929749,
      "learning_rate": 0.00017600000000000002,
      "loss": 0.7673,
      "step": 88
    },
    {
      "epoch": 0.3658787255909558,
      "grad_norm": 0.19703039526939392,
      "learning_rate": 0.00017800000000000002,
      "loss": 0.8048,
      "step": 89
    },
    {
      "epoch": 0.3699897225077081,
      "grad_norm": 0.19380515813827515,
      "learning_rate": 0.00018,
      "loss": 0.752,
      "step": 90
    },
    {
      "epoch": 0.37410071942446044,
      "grad_norm": 0.184962198138237,
      "learning_rate": 0.000182,
      "loss": 0.7602,
      "step": 91
    },
    {
      "epoch": 0.37821171634121276,
      "grad_norm": 0.22890885174274445,
      "learning_rate": 0.00018400000000000003,
      "loss": 0.7974,
      "step": 92
    },
    {
      "epoch": 0.3823227132579651,
      "grad_norm": 0.18501392006874084,
      "learning_rate": 0.00018600000000000002,
      "loss": 0.6789,
      "step": 93
    },
    {
      "epoch": 0.3864337101747174,
      "grad_norm": 0.18353326618671417,
      "learning_rate": 0.000188,
      "loss": 0.7456,
      "step": 94
    },
    {
      "epoch": 0.39054470709146966,
      "grad_norm": 0.24070213735103607,
      "learning_rate": 0.00019,
      "loss": 0.7647,
      "step": 95
    },
    {
      "epoch": 0.394655704008222,
      "grad_norm": 0.1820479929447174,
      "learning_rate": 0.000192,
      "loss": 0.7448,
      "step": 96
    },
    {
      "epoch": 0.3987667009249743,
      "grad_norm": 0.24182480573654175,
      "learning_rate": 0.000194,
      "loss": 0.7803,
      "step": 97
    },
    {
      "epoch": 0.4028776978417266,
      "grad_norm": 0.24068184196949005,
      "learning_rate": 0.000196,
      "loss": 0.7306,
      "step": 98
    },
    {
      "epoch": 0.40698869475847893,
      "grad_norm": 0.21862854063510895,
      "learning_rate": 0.00019800000000000002,
      "loss": 0.7428,
      "step": 99
    },
    {
      "epoch": 0.41109969167523125,
      "grad_norm": 0.2827813923358917,
      "learning_rate": 0.0002,
      "loss": 0.718,
      "step": 100
    },
    {
      "epoch": 0.41521068859198357,
      "grad_norm": 0.2075868397951126,
      "learning_rate": 0.00019999691576447898,
      "loss": 0.7704,
      "step": 101
    },
    {
      "epoch": 0.4193216855087359,
      "grad_norm": 0.16914305090904236,
      "learning_rate": 0.00019998766324816607,
      "loss": 0.7534,
      "step": 102
    },
    {
      "epoch": 0.4234326824254882,
      "grad_norm": 0.19006559252738953,
      "learning_rate": 0.00019997224302180006,
      "loss": 0.7578,
      "step": 103
    },
    {
      "epoch": 0.42754367934224047,
      "grad_norm": 0.2263256162405014,
      "learning_rate": 0.00019995065603657316,
      "loss": 0.7198,
      "step": 104
    },
    {
      "epoch": 0.4316546762589928,
      "grad_norm": 0.21746039390563965,
      "learning_rate": 0.0001999229036240723,
      "loss": 0.741,
      "step": 105
    },
    {
      "epoch": 0.4357656731757451,
      "grad_norm": 0.19659580290317535,
      "learning_rate": 0.00019988898749619702,
      "loss": 0.7771,
      "step": 106
    },
    {
      "epoch": 0.4398766700924974,
      "grad_norm": 0.24985191226005554,
      "learning_rate": 0.00019984890974505381,
      "loss": 0.7765,
      "step": 107
    },
    {
      "epoch": 0.44398766700924974,
      "grad_norm": 0.2400967925786972,
      "learning_rate": 0.00019980267284282717,
      "loss": 0.8331,
      "step": 108
    },
    {
      "epoch": 0.44809866392600206,
      "grad_norm": 0.19052259624004364,
      "learning_rate": 0.00019975027964162702,
      "loss": 0.7802,
      "step": 109
    },
    {
      "epoch": 0.4522096608427544,
      "grad_norm": 0.207940012216568,
      "learning_rate": 0.0001996917333733128,
      "loss": 0.8083,
      "step": 110
    },
    {
      "epoch": 0.4563206577595067,
      "grad_norm": 0.20245718955993652,
      "learning_rate": 0.00019962703764929413,
      "loss": 0.7424,
      "step": 111
    },
    {
      "epoch": 0.460431654676259,
      "grad_norm": 0.2154696136713028,
      "learning_rate": 0.00019955619646030802,
      "loss": 0.8053,
      "step": 112
    },
    {
      "epoch": 0.4645426515930113,
      "grad_norm": 0.18256965279579163,
      "learning_rate": 0.00019947921417617267,
      "loss": 0.7425,
      "step": 113
    },
    {
      "epoch": 0.4686536485097636,
      "grad_norm": 0.1869020313024521,
      "learning_rate": 0.000199396095545518,
      "loss": 0.7828,
      "step": 114
    },
    {
      "epoch": 0.4727646454265159,
      "grad_norm": 0.2087710052728653,
      "learning_rate": 0.00019930684569549264,
      "loss": 0.7663,
      "step": 115
    },
    {
      "epoch": 0.47687564234326824,
      "grad_norm": 0.20225581526756287,
      "learning_rate": 0.0001992114701314478,
      "loss": 0.7912,
      "step": 116
    },
    {
      "epoch": 0.48098663926002055,
      "grad_norm": 0.2115861177444458,
      "learning_rate": 0.0001991099747365975,
      "loss": 0.8079,
      "step": 117
    },
    {
      "epoch": 0.4850976361767729,
      "grad_norm": 0.2025671750307083,
      "learning_rate": 0.00019900236577165576,
      "loss": 0.7796,
      "step": 118
    },
    {
      "epoch": 0.4892086330935252,
      "grad_norm": 0.21485574543476105,
      "learning_rate": 0.0001988886498744505,
      "loss": 0.7345,
      "step": 119
    },
    {
      "epoch": 0.4933196300102775,
      "grad_norm": 0.20953282713890076,
      "learning_rate": 0.00019876883405951377,
      "loss": 0.7185,
      "step": 120
    },
    {
      "epoch": 0.49743062692702983,
      "grad_norm": 0.18189023435115814,
      "learning_rate": 0.00019864292571764955,
      "loss": 0.7625,
      "step": 121
    },
    {
      "epoch": 0.5015416238437821,
      "grad_norm": 0.19342605769634247,
      "learning_rate": 0.0001985109326154774,
      "loss": 0.7027,
      "step": 122
    },
    {
      "epoch": 0.5056526207605344,
      "grad_norm": 0.18798814713954926,
      "learning_rate": 0.00019837286289495361,
      "loss": 0.7515,
      "step": 123
    },
    {
      "epoch": 0.5097636176772867,
      "grad_norm": 0.19502167403697968,
      "learning_rate": 0.0001982287250728689,
      "loss": 0.7866,
      "step": 124
    },
    {
      "epoch": 0.513874614594039,
      "grad_norm": 0.2178218811750412,
      "learning_rate": 0.00019807852804032305,
      "loss": 0.7886,
      "step": 125
    },
    {
      "epoch": 0.5179856115107914,
      "grad_norm": 0.1815299689769745,
      "learning_rate": 0.00019792228106217658,
      "loss": 0.7238,
      "step": 126
    },
    {
      "epoch": 0.5220966084275437,
      "grad_norm": 0.202816441655159,
      "learning_rate": 0.0001977599937764791,
      "loss": 0.7405,
      "step": 127
    },
    {
      "epoch": 0.526207605344296,
      "grad_norm": 0.17130331695079803,
      "learning_rate": 0.00019759167619387476,
      "loss": 0.6849,
      "step": 128
    },
    {
      "epoch": 0.5303186022610483,
      "grad_norm": 0.18959152698516846,
      "learning_rate": 0.00019741733869698495,
      "loss": 0.8042,
      "step": 129
    },
    {
      "epoch": 0.5344295991778006,
      "grad_norm": 0.17352677881717682,
      "learning_rate": 0.00019723699203976766,
      "loss": 0.734,
      "step": 130
    },
    {
      "epoch": 0.538540596094553,
      "grad_norm": 0.19079051911830902,
      "learning_rate": 0.00019705064734685425,
      "loss": 0.7568,
      "step": 131
    },
    {
      "epoch": 0.5426515930113053,
      "grad_norm": 0.16974347829818726,
      "learning_rate": 0.0001968583161128631,
      "loss": 0.7629,
      "step": 132
    },
    {
      "epoch": 0.5467625899280576,
      "grad_norm": 0.18240852653980255,
      "learning_rate": 0.00019666001020169073,
      "loss": 0.7684,
      "step": 133
    },
    {
      "epoch": 0.5508735868448099,
      "grad_norm": 0.17562364041805267,
      "learning_rate": 0.00019645574184577982,
      "loss": 0.7553,
      "step": 134
    },
    {
      "epoch": 0.5549845837615622,
      "grad_norm": 0.19359007477760315,
      "learning_rate": 0.00019624552364536473,
      "loss": 0.758,
      "step": 135
    },
    {
      "epoch": 0.5590955806783144,
      "grad_norm": 0.18970121443271637,
      "learning_rate": 0.0001960293685676943,
      "loss": 0.7949,
      "step": 136
    },
    {
      "epoch": 0.5632065775950668,
      "grad_norm": 0.1877439022064209,
      "learning_rate": 0.00019580728994623195,
      "loss": 0.7295,
      "step": 137
    },
    {
      "epoch": 0.5673175745118191,
      "grad_norm": 0.1851058304309845,
      "learning_rate": 0.00019557930147983302,
      "loss": 0.7264,
      "step": 138
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.1951686590909958,
      "learning_rate": 0.0001953454172319001,
      "loss": 0.7917,
      "step": 139
    },
    {
      "epoch": 0.5755395683453237,
      "grad_norm": 0.20179124176502228,
      "learning_rate": 0.00019510565162951537,
      "loss": 0.7881,
      "step": 140
    },
    {
      "epoch": 0.579650565262076,
      "grad_norm": 0.19038018584251404,
      "learning_rate": 0.00019486001946255046,
      "loss": 0.8396,
      "step": 141
    },
    {
      "epoch": 0.5837615621788284,
      "grad_norm": 0.19785518944263458,
      "learning_rate": 0.00019460853588275454,
      "loss": 0.7312,
      "step": 142
    },
    {
      "epoch": 0.5878725590955807,
      "grad_norm": 0.17820334434509277,
      "learning_rate": 0.00019435121640281938,
      "loss": 0.7572,
      "step": 143
    },
    {
      "epoch": 0.591983556012333,
      "grad_norm": 0.15957815945148468,
      "learning_rate": 0.00019408807689542257,
      "loss": 0.7276,
      "step": 144
    },
    {
      "epoch": 0.5960945529290853,
      "grad_norm": 0.18277224898338318,
      "learning_rate": 0.00019381913359224842,
      "loss": 0.7522,
      "step": 145
    },
    {
      "epoch": 0.6002055498458376,
      "grad_norm": 0.19914941489696503,
      "learning_rate": 0.00019354440308298675,
      "loss": 0.7556,
      "step": 146
    },
    {
      "epoch": 0.60431654676259,
      "grad_norm": 0.19908249378204346,
      "learning_rate": 0.00019326390231430942,
      "loss": 0.9078,
      "step": 147
    },
    {
      "epoch": 0.6084275436793423,
      "grad_norm": 0.19466081261634827,
      "learning_rate": 0.00019297764858882514,
      "loss": 0.8024,
      "step": 148
    },
    {
      "epoch": 0.6125385405960946,
      "grad_norm": 0.17888300120830536,
      "learning_rate": 0.00019268565956401208,
      "loss": 0.6851,
      "step": 149
    },
    {
      "epoch": 0.6166495375128469,
      "grad_norm": 0.18040218949317932,
      "learning_rate": 0.0001923879532511287,
      "loss": 0.7543,
      "step": 150
    },
    {
      "epoch": 0.6207605344295992,
      "grad_norm": 0.1843702495098114,
      "learning_rate": 0.00019208454801410266,
      "loss": 0.7023,
      "step": 151
    },
    {
      "epoch": 0.6248715313463515,
      "grad_norm": 0.19456690549850464,
      "learning_rate": 0.00019177546256839812,
      "loss": 0.7698,
      "step": 152
    },
    {
      "epoch": 0.6289825282631039,
      "grad_norm": 0.19757886230945587,
      "learning_rate": 0.00019146071597986138,
      "loss": 0.7448,
      "step": 153
    },
    {
      "epoch": 0.6330935251798561,
      "grad_norm": 0.17260992527008057,
      "learning_rate": 0.00019114032766354453,
      "loss": 0.7797,
      "step": 154
    },
    {
      "epoch": 0.6372045220966084,
      "grad_norm": 0.17731568217277527,
      "learning_rate": 0.00019081431738250814,
      "loss": 0.8182,
      "step": 155
    },
    {
      "epoch": 0.6413155190133607,
      "grad_norm": 0.21344119310379028,
      "learning_rate": 0.00019048270524660196,
      "loss": 0.7483,
      "step": 156
    },
    {
      "epoch": 0.645426515930113,
      "grad_norm": 0.16134126484394073,
      "learning_rate": 0.00019014551171122457,
      "loss": 0.7009,
      "step": 157
    },
    {
      "epoch": 0.6495375128468653,
      "grad_norm": 0.18650446832180023,
      "learning_rate": 0.00018980275757606157,
      "loss": 0.7873,
      "step": 158
    },
    {
      "epoch": 0.6536485097636177,
      "grad_norm": 0.1889180988073349,
      "learning_rate": 0.0001894544639838025,
      "loss": 0.8256,
      "step": 159
    },
    {
      "epoch": 0.65775950668037,
      "grad_norm": 0.18333201110363007,
      "learning_rate": 0.0001891006524188368,
      "loss": 0.8075,
      "step": 160
    },
    {
      "epoch": 0.6618705035971223,
      "grad_norm": 0.19261512160301208,
      "learning_rate": 0.00018874134470592835,
      "loss": 0.7259,
      "step": 161
    },
    {
      "epoch": 0.6659815005138746,
      "grad_norm": 0.1970406025648117,
      "learning_rate": 0.00018837656300886937,
      "loss": 0.8183,
      "step": 162
    },
    {
      "epoch": 0.6700924974306269,
      "grad_norm": 0.1701861023902893,
      "learning_rate": 0.00018800632982911322,
      "loss": 0.7249,
      "step": 163
    },
    {
      "epoch": 0.6742034943473793,
      "grad_norm": 0.1862119436264038,
      "learning_rate": 0.00018763066800438636,
      "loss": 0.7558,
      "step": 164
    },
    {
      "epoch": 0.6783144912641316,
      "grad_norm": 0.17617414891719818,
      "learning_rate": 0.00018724960070727972,
      "loss": 0.7618,
      "step": 165
    },
    {
      "epoch": 0.6824254881808839,
      "grad_norm": 0.1956210732460022,
      "learning_rate": 0.00018686315144381913,
      "loss": 0.7531,
      "step": 166
    },
    {
      "epoch": 0.6865364850976362,
      "grad_norm": 0.21825283765792847,
      "learning_rate": 0.0001864713440520155,
      "loss": 0.7478,
      "step": 167
    },
    {
      "epoch": 0.6906474820143885,
      "grad_norm": 0.20434311032295227,
      "learning_rate": 0.0001860742027003944,
      "loss": 0.7902,
      "step": 168
    },
    {
      "epoch": 0.6947584789311408,
      "grad_norm": 0.18894919753074646,
      "learning_rate": 0.00018567175188650498,
      "loss": 0.7694,
      "step": 169
    },
    {
      "epoch": 0.6988694758478932,
      "grad_norm": 0.19389449059963226,
      "learning_rate": 0.00018526401643540922,
      "loss": 0.7626,
      "step": 170
    },
    {
      "epoch": 0.7029804727646455,
      "grad_norm": 0.1795973777770996,
      "learning_rate": 0.00018485102149815038,
      "loss": 0.7125,
      "step": 171
    },
    {
      "epoch": 0.7070914696813977,
      "grad_norm": 0.1643115133047104,
      "learning_rate": 0.00018443279255020152,
      "loss": 0.7285,
      "step": 172
    },
    {
      "epoch": 0.71120246659815,
      "grad_norm": 0.17256219685077667,
      "learning_rate": 0.0001840093553898942,
      "loss": 0.7272,
      "step": 173
    },
    {
      "epoch": 0.7153134635149023,
      "grad_norm": 0.16557914018630981,
      "learning_rate": 0.00018358073613682706,
      "loss": 0.7082,
      "step": 174
    },
    {
      "epoch": 0.7194244604316546,
      "grad_norm": 0.19780583679676056,
      "learning_rate": 0.00018314696123025454,
      "loss": 0.7791,
      "step": 175
    },
    {
      "epoch": 0.723535457348407,
      "grad_norm": 0.17353545129299164,
      "learning_rate": 0.00018270805742745617,
      "loss": 0.7728,
      "step": 176
    },
    {
      "epoch": 0.7276464542651593,
      "grad_norm": 0.15986531972885132,
      "learning_rate": 0.000182264051802086,
      "loss": 0.7133,
      "step": 177
    },
    {
      "epoch": 0.7317574511819116,
      "grad_norm": 0.17527668178081512,
      "learning_rate": 0.00018181497174250236,
      "loss": 0.7777,
      "step": 178
    },
    {
      "epoch": 0.7358684480986639,
      "grad_norm": 0.19276300072669983,
      "learning_rate": 0.00018136084495007872,
      "loss": 0.7409,
      "step": 179
    },
    {
      "epoch": 0.7399794450154162,
      "grad_norm": 0.17722974717617035,
      "learning_rate": 0.00018090169943749476,
      "loss": 0.7437,
      "step": 180
    },
    {
      "epoch": 0.7440904419321686,
      "grad_norm": 0.18392924964427948,
      "learning_rate": 0.00018043756352700846,
      "loss": 0.7843,
      "step": 181
    },
    {
      "epoch": 0.7482014388489209,
      "grad_norm": 0.17168018221855164,
      "learning_rate": 0.00017996846584870908,
      "loss": 0.8041,
      "step": 182
    },
    {
      "epoch": 0.7523124357656732,
      "grad_norm": 0.1757081151008606,
      "learning_rate": 0.000179494435338751,
      "loss": 0.755,
      "step": 183
    },
    {
      "epoch": 0.7564234326824255,
      "grad_norm": 0.168577641248703,
      "learning_rate": 0.00017901550123756906,
      "loss": 0.6974,
      "step": 184
    },
    {
      "epoch": 0.7605344295991778,
      "grad_norm": 0.18002143502235413,
      "learning_rate": 0.00017853169308807448,
      "loss": 0.7355,
      "step": 185
    },
    {
      "epoch": 0.7646454265159301,
      "grad_norm": 0.18954350054264069,
      "learning_rate": 0.000178043040733833,
      "loss": 0.7539,
      "step": 186
    },
    {
      "epoch": 0.7687564234326825,
      "grad_norm": 0.1987128108739853,
      "learning_rate": 0.00017754957431722346,
      "loss": 0.747,
      "step": 187
    },
    {
      "epoch": 0.7728674203494348,
      "grad_norm": 0.36224067211151123,
      "learning_rate": 0.00017705132427757895,
      "loss": 0.786,
      "step": 188
    },
    {
      "epoch": 0.7769784172661871,
      "grad_norm": 0.1823827177286148,
      "learning_rate": 0.00017654832134930882,
      "loss": 0.7203,
      "step": 189
    },
    {
      "epoch": 0.7810894141829393,
      "grad_norm": 0.19966542720794678,
      "learning_rate": 0.0001760405965600031,
      "loss": 0.7834,
      "step": 190
    },
    {
      "epoch": 0.7852004110996916,
      "grad_norm": 0.17799125611782074,
      "learning_rate": 0.00017552818122851838,
      "loss": 0.8695,
      "step": 191
    },
    {
      "epoch": 0.789311408016444,
      "grad_norm": 0.1828295886516571,
      "learning_rate": 0.00017501110696304596,
      "loss": 0.8097,
      "step": 192
    },
    {
      "epoch": 0.7934224049331963,
      "grad_norm": 0.17709365487098694,
      "learning_rate": 0.00017448940565916222,
      "loss": 0.7022,
      "step": 193
    },
    {
      "epoch": 0.7975334018499486,
      "grad_norm": 0.1711331456899643,
      "learning_rate": 0.000173963109497861,
      "loss": 0.7711,
      "step": 194
    },
    {
      "epoch": 0.8016443987667009,
      "grad_norm": 0.15606576204299927,
      "learning_rate": 0.00017343225094356855,
      "loss": 0.7364,
      "step": 195
    },
    {
      "epoch": 0.8057553956834532,
      "grad_norm": 0.17102007567882538,
      "learning_rate": 0.00017289686274214118,
      "loss": 0.7671,
      "step": 196
    },
    {
      "epoch": 0.8098663926002055,
      "grad_norm": 0.1857379674911499,
      "learning_rate": 0.00017235697791884494,
      "loss": 0.7663,
      "step": 197
    },
    {
      "epoch": 0.8139773895169579,
      "grad_norm": 0.17865677177906036,
      "learning_rate": 0.00017181262977631888,
      "loss": 0.7189,
      "step": 198
    },
    {
      "epoch": 0.8180883864337102,
      "grad_norm": 0.1891535371541977,
      "learning_rate": 0.00017126385189252053,
      "loss": 0.8047,
      "step": 199
    },
    {
      "epoch": 0.8221993833504625,
      "grad_norm": 0.18282689154148102,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.7902,
      "step": 200
    },
    {
      "epoch": 0.8263103802672148,
      "grad_norm": 0.18731562793254852,
      "learning_rate": 0.0001701531425770856,
      "loss": 0.8048,
      "step": 201
    },
    {
      "epoch": 0.8304213771839671,
      "grad_norm": 0.17205360531806946,
      "learning_rate": 0.00016959127965923142,
      "loss": 0.708,
      "step": 202
    },
    {
      "epoch": 0.8345323741007195,
      "grad_norm": 0.17244985699653625,
      "learning_rate": 0.00016902512402344373,
      "loss": 0.6789,
      "step": 203
    },
    {
      "epoch": 0.8386433710174718,
      "grad_norm": 0.18318970501422882,
      "learning_rate": 0.00016845471059286887,
      "loss": 0.7779,
      "step": 204
    },
    {
      "epoch": 0.8427543679342241,
      "grad_norm": 0.17885752022266388,
      "learning_rate": 0.0001678800745532942,
      "loss": 0.7826,
      "step": 205
    },
    {
      "epoch": 0.8468653648509764,
      "grad_norm": 0.17862874269485474,
      "learning_rate": 0.00016730125135097735,
      "loss": 0.717,
      "step": 206
    },
    {
      "epoch": 0.8509763617677287,
      "grad_norm": 0.1660134345293045,
      "learning_rate": 0.00016671827669045998,
      "loss": 0.7464,
      "step": 207
    },
    {
      "epoch": 0.8550873586844809,
      "grad_norm": 0.1796494722366333,
      "learning_rate": 0.00016613118653236518,
      "loss": 0.7799,
      "step": 208
    },
    {
      "epoch": 0.8591983556012333,
      "grad_norm": 0.15708494186401367,
      "learning_rate": 0.0001655400170911794,
      "loss": 0.6999,
      "step": 209
    },
    {
      "epoch": 0.8633093525179856,
      "grad_norm": 0.17457805573940277,
      "learning_rate": 0.00016494480483301836,
      "loss": 0.8024,
      "step": 210
    },
    {
      "epoch": 0.8674203494347379,
      "grad_norm": 0.1824868619441986,
      "learning_rate": 0.0001643455864733779,
      "loss": 0.7158,
      "step": 211
    },
    {
      "epoch": 0.8715313463514902,
      "grad_norm": 0.17008088529109955,
      "learning_rate": 0.000163742398974869,
      "loss": 0.7086,
      "step": 212
    },
    {
      "epoch": 0.8756423432682425,
      "grad_norm": 0.1706101894378662,
      "learning_rate": 0.00016313527954493778,
      "loss": 0.7605,
      "step": 213
    },
    {
      "epoch": 0.8797533401849948,
      "grad_norm": 0.17592813074588776,
      "learning_rate": 0.00016252426563357055,
      "loss": 0.7234,
      "step": 214
    },
    {
      "epoch": 0.8838643371017472,
      "grad_norm": 0.17459964752197266,
      "learning_rate": 0.00016190939493098344,
      "loss": 0.7865,
      "step": 215
    },
    {
      "epoch": 0.8879753340184995,
      "grad_norm": 0.1685909479856491,
      "learning_rate": 0.00016129070536529766,
      "loss": 0.6992,
      "step": 216
    },
    {
      "epoch": 0.8920863309352518,
      "grad_norm": 0.1832508146762848,
      "learning_rate": 0.00016066823510019998,
      "loss": 0.7943,
      "step": 217
    },
    {
      "epoch": 0.8961973278520041,
      "grad_norm": 0.16831143200397491,
      "learning_rate": 0.00016004202253258842,
      "loss": 0.7289,
      "step": 218
    },
    {
      "epoch": 0.9003083247687564,
      "grad_norm": 0.1783897578716278,
      "learning_rate": 0.00015941210629020388,
      "loss": 0.7563,
      "step": 219
    },
    {
      "epoch": 0.9044193216855088,
      "grad_norm": 0.1713438332080841,
      "learning_rate": 0.00015877852522924732,
      "loss": 0.7541,
      "step": 220
    },
    {
      "epoch": 0.9085303186022611,
      "grad_norm": 0.17131845653057098,
      "learning_rate": 0.00015814131843198308,
      "loss": 0.8142,
      "step": 221
    },
    {
      "epoch": 0.9126413155190134,
      "grad_norm": 0.1550828367471695,
      "learning_rate": 0.00015750052520432787,
      "loss": 0.6872,
      "step": 222
    },
    {
      "epoch": 0.9167523124357657,
      "grad_norm": 0.17278611660003662,
      "learning_rate": 0.0001568561850734264,
      "loss": 0.6932,
      "step": 223
    },
    {
      "epoch": 0.920863309352518,
      "grad_norm": 0.16213761270046234,
      "learning_rate": 0.00015620833778521307,
      "loss": 0.7169,
      "step": 224
    },
    {
      "epoch": 0.9249743062692704,
      "grad_norm": 0.16158106923103333,
      "learning_rate": 0.00015555702330196023,
      "loss": 0.8013,
      "step": 225
    },
    {
      "epoch": 0.9290853031860226,
      "grad_norm": 0.1824110895395279,
      "learning_rate": 0.0001549022817998132,
      "loss": 0.7971,
      "step": 226
    },
    {
      "epoch": 0.9331963001027749,
      "grad_norm": 0.18137367069721222,
      "learning_rate": 0.00015424415366631188,
      "loss": 0.7311,
      "step": 227
    },
    {
      "epoch": 0.9373072970195272,
      "grad_norm": 0.17672066390514374,
      "learning_rate": 0.00015358267949789966,
      "loss": 0.7232,
      "step": 228
    },
    {
      "epoch": 0.9414182939362795,
      "grad_norm": 0.16367360949516296,
      "learning_rate": 0.00015291790009741907,
      "loss": 0.7564,
      "step": 229
    },
    {
      "epoch": 0.9455292908530318,
      "grad_norm": 0.17514045536518097,
      "learning_rate": 0.0001522498564715949,
      "loss": 0.7976,
      "step": 230
    },
    {
      "epoch": 0.9496402877697842,
      "grad_norm": 0.18254303932189941,
      "learning_rate": 0.00015157858982850475,
      "loss": 0.7602,
      "step": 231
    },
    {
      "epoch": 0.9537512846865365,
      "grad_norm": 0.17318600416183472,
      "learning_rate": 0.00015090414157503714,
      "loss": 0.675,
      "step": 232
    },
    {
      "epoch": 0.9578622816032888,
      "grad_norm": 0.1657712310552597,
      "learning_rate": 0.00015022655331433727,
      "loss": 0.7324,
      "step": 233
    },
    {
      "epoch": 0.9619732785200411,
      "grad_norm": 0.17902085185050964,
      "learning_rate": 0.00014954586684324078,
      "loss": 0.808,
      "step": 234
    },
    {
      "epoch": 0.9660842754367934,
      "grad_norm": 0.1783977746963501,
      "learning_rate": 0.00014886212414969553,
      "loss": 0.723,
      "step": 235
    },
    {
      "epoch": 0.9701952723535457,
      "grad_norm": 0.18178479373455048,
      "learning_rate": 0.00014817536741017152,
      "loss": 0.7772,
      "step": 236
    },
    {
      "epoch": 0.9743062692702981,
      "grad_norm": 0.16560284793376923,
      "learning_rate": 0.00014748563898705946,
      "loss": 0.7356,
      "step": 237
    },
    {
      "epoch": 0.9784172661870504,
      "grad_norm": 0.19371147453784943,
      "learning_rate": 0.00014679298142605734,
      "loss": 0.6959,
      "step": 238
    },
    {
      "epoch": 0.9825282631038027,
      "grad_norm": 0.17361164093017578,
      "learning_rate": 0.00014609743745354624,
      "loss": 0.7553,
      "step": 239
    },
    {
      "epoch": 0.986639260020555,
      "grad_norm": 0.17735688388347626,
      "learning_rate": 0.00014539904997395468,
      "loss": 0.7532,
      "step": 240
    },
    {
      "epoch": 0.9907502569373073,
      "grad_norm": 0.17273449897766113,
      "learning_rate": 0.00014469786206711214,
      "loss": 0.6887,
      "step": 241
    },
    {
      "epoch": 0.9948612538540597,
      "grad_norm": 0.16542835533618927,
      "learning_rate": 0.00014399391698559152,
      "loss": 0.7181,
      "step": 242
    },
    {
      "epoch": 0.998972250770812,
      "grad_norm": 0.16543813049793243,
      "learning_rate": 0.00014328725815204144,
      "loss": 0.6641,
      "step": 243
    },
    {
      "epoch": 0.998972250770812,
      "eval_loss": 0.7379785180091858,
      "eval_runtime": 22.0253,
      "eval_samples_per_second": 55.8,
      "eval_steps_per_second": 2.361,
      "step": 243
    },
    {
      "epoch": 1.0030832476875642,
      "grad_norm": 0.39137569069862366,
      "learning_rate": 0.00014257792915650728,
      "loss": 1.2335,
      "step": 244
    },
    {
      "epoch": 1.0071942446043165,
      "grad_norm": 0.1728934645652771,
      "learning_rate": 0.0001418659737537428,
      "loss": 0.729,
      "step": 245
    },
    {
      "epoch": 1.0113052415210688,
      "grad_norm": 0.1766151785850525,
      "learning_rate": 0.00014115143586051088,
      "loss": 0.6647,
      "step": 246
    },
    {
      "epoch": 1.0154162384378211,
      "grad_norm": 0.18938228487968445,
      "learning_rate": 0.00014043435955287452,
      "loss": 0.7435,
      "step": 247
    },
    {
      "epoch": 1.0195272353545735,
      "grad_norm": 0.1688462197780609,
      "learning_rate": 0.00013971478906347806,
      "loss": 0.7373,
      "step": 248
    },
    {
      "epoch": 1.0236382322713258,
      "grad_norm": 0.16754169762134552,
      "learning_rate": 0.00013899276877881884,
      "loss": 0.6979,
      "step": 249
    },
    {
      "epoch": 1.027749229188078,
      "grad_norm": 0.18979588150978088,
      "learning_rate": 0.000138268343236509,
      "loss": 0.7347,
      "step": 250
    },
    {
      "epoch": 1.0318602261048304,
      "grad_norm": 0.19292105734348297,
      "learning_rate": 0.00013754155712252832,
      "loss": 0.6692,
      "step": 251
    },
    {
      "epoch": 1.0359712230215827,
      "grad_norm": 0.18789665400981903,
      "learning_rate": 0.00013681245526846783,
      "loss": 0.7093,
      "step": 252
    },
    {
      "epoch": 1.040082219938335,
      "grad_norm": 0.18097302317619324,
      "learning_rate": 0.0001360810826487642,
      "loss": 0.6592,
      "step": 253
    },
    {
      "epoch": 1.0441932168550874,
      "grad_norm": 0.19411665201187134,
      "learning_rate": 0.00013534748437792573,
      "loss": 0.7602,
      "step": 254
    },
    {
      "epoch": 1.0483042137718397,
      "grad_norm": 0.21550589799880981,
      "learning_rate": 0.0001346117057077493,
      "loss": 0.8553,
      "step": 255
    },
    {
      "epoch": 1.052415210688592,
      "grad_norm": 0.17013515532016754,
      "learning_rate": 0.00013387379202452917,
      "loss": 0.6454,
      "step": 256
    },
    {
      "epoch": 1.0565262076053443,
      "grad_norm": 0.18332229554653168,
      "learning_rate": 0.0001331337888462571,
      "loss": 0.7672,
      "step": 257
    },
    {
      "epoch": 1.0606372045220966,
      "grad_norm": 0.20885473489761353,
      "learning_rate": 0.00013239174181981495,
      "loss": 0.7312,
      "step": 258
    },
    {
      "epoch": 1.064748201438849,
      "grad_norm": 0.2096792310476303,
      "learning_rate": 0.00013164769671815862,
      "loss": 0.653,
      "step": 259
    },
    {
      "epoch": 1.0688591983556013,
      "grad_norm": 0.21655458211898804,
      "learning_rate": 0.00013090169943749476,
      "loss": 0.7269,
      "step": 260
    },
    {
      "epoch": 1.0729701952723536,
      "grad_norm": 0.20201562345027924,
      "learning_rate": 0.00013015379599444957,
      "loss": 0.7138,
      "step": 261
    },
    {
      "epoch": 1.077081192189106,
      "grad_norm": 0.24191242456436157,
      "learning_rate": 0.0001294040325232304,
      "loss": 0.7303,
      "step": 262
    },
    {
      "epoch": 1.0811921891058582,
      "grad_norm": 0.2298012375831604,
      "learning_rate": 0.00012865245527277986,
      "loss": 0.6369,
      "step": 263
    },
    {
      "epoch": 1.0853031860226106,
      "grad_norm": 0.19918785989284515,
      "learning_rate": 0.00012789911060392294,
      "loss": 0.6771,
      "step": 264
    },
    {
      "epoch": 1.0894141829393629,
      "grad_norm": 0.21279063820838928,
      "learning_rate": 0.00012714404498650743,
      "loss": 0.7147,
      "step": 265
    },
    {
      "epoch": 1.0935251798561152,
      "grad_norm": 0.19962945580482483,
      "learning_rate": 0.0001263873049965373,
      "loss": 0.6821,
      "step": 266
    },
    {
      "epoch": 1.0976361767728675,
      "grad_norm": 0.22876287996768951,
      "learning_rate": 0.00012562893731329967,
      "loss": 0.6001,
      "step": 267
    },
    {
      "epoch": 1.1017471736896198,
      "grad_norm": 0.22757205367088318,
      "learning_rate": 0.0001248689887164855,
      "loss": 0.6901,
      "step": 268
    },
    {
      "epoch": 1.1058581706063721,
      "grad_norm": 0.19157056510448456,
      "learning_rate": 0.00012410750608330388,
      "loss": 0.6423,
      "step": 269
    },
    {
      "epoch": 1.1099691675231242,
      "grad_norm": 0.18848693370819092,
      "learning_rate": 0.00012334453638559057,
      "loss": 0.7074,
      "step": 270
    },
    {
      "epoch": 1.1140801644398768,
      "grad_norm": 0.1879797875881195,
      "learning_rate": 0.0001225801266869104,
      "loss": 0.6496,
      "step": 271
    },
    {
      "epoch": 1.1181911613566289,
      "grad_norm": 0.19743949174880981,
      "learning_rate": 0.00012181432413965428,
      "loss": 0.5992,
      "step": 272
    },
    {
      "epoch": 1.1223021582733812,
      "grad_norm": 0.21017858386039734,
      "learning_rate": 0.00012104717598213056,
      "loss": 0.6979,
      "step": 273
    },
    {
      "epoch": 1.1264131551901335,
      "grad_norm": 0.18878193199634552,
      "learning_rate": 0.00012027872953565125,
      "loss": 0.802,
      "step": 274
    },
    {
      "epoch": 1.1305241521068858,
      "grad_norm": 0.2085169404745102,
      "learning_rate": 0.00011950903220161285,
      "loss": 0.7286,
      "step": 275
    },
    {
      "epoch": 1.1346351490236382,
      "grad_norm": 0.21597957611083984,
      "learning_rate": 0.00011873813145857249,
      "loss": 0.657,
      "step": 276
    },
    {
      "epoch": 1.1387461459403905,
      "grad_norm": 0.21635057032108307,
      "learning_rate": 0.00011796607485931928,
      "loss": 0.7045,
      "step": 277
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.19274653494358063,
      "learning_rate": 0.00011719291002794096,
      "loss": 0.6372,
      "step": 278
    },
    {
      "epoch": 1.1469681397738951,
      "grad_norm": 0.2068728506565094,
      "learning_rate": 0.0001164186846568863,
      "loss": 0.7435,
      "step": 279
    },
    {
      "epoch": 1.1510791366906474,
      "grad_norm": 0.2124863862991333,
      "learning_rate": 0.0001156434465040231,
      "loss": 0.7567,
      "step": 280
    },
    {
      "epoch": 1.1551901336073997,
      "grad_norm": 0.18096277117729187,
      "learning_rate": 0.00011486724338969232,
      "loss": 0.6533,
      "step": 281
    },
    {
      "epoch": 1.159301130524152,
      "grad_norm": 0.21124780178070068,
      "learning_rate": 0.00011409012319375827,
      "loss": 0.7253,
      "step": 282
    },
    {
      "epoch": 1.1634121274409044,
      "grad_norm": 0.20828507840633392,
      "learning_rate": 0.00011331213385265524,
      "loss": 0.6197,
      "step": 283
    },
    {
      "epoch": 1.1675231243576567,
      "grad_norm": 0.20803777873516083,
      "learning_rate": 0.00011253332335643043,
      "loss": 0.655,
      "step": 284
    },
    {
      "epoch": 1.171634121274409,
      "grad_norm": 0.1842370480298996,
      "learning_rate": 0.00011175373974578378,
      "loss": 0.6897,
      "step": 285
    },
    {
      "epoch": 1.1757451181911613,
      "grad_norm": 0.20155522227287292,
      "learning_rate": 0.00011097343110910452,
      "loss": 0.6041,
      "step": 286
    },
    {
      "epoch": 1.1798561151079137,
      "grad_norm": 0.21121251583099365,
      "learning_rate": 0.000110192445579505,
      "loss": 0.7351,
      "step": 287
    },
    {
      "epoch": 1.183967112024666,
      "grad_norm": 0.20202356576919556,
      "learning_rate": 0.00010941083133185146,
      "loss": 0.6428,
      "step": 288
    },
    {
      "epoch": 1.1880781089414183,
      "grad_norm": 0.20538434386253357,
      "learning_rate": 0.00010862863657979237,
      "loss": 0.6914,
      "step": 289
    },
    {
      "epoch": 1.1921891058581706,
      "grad_norm": 0.20957379043102264,
      "learning_rate": 0.0001078459095727845,
      "loss": 0.7057,
      "step": 290
    },
    {
      "epoch": 1.196300102774923,
      "grad_norm": 0.21016211807727814,
      "learning_rate": 0.00010706269859311669,
      "loss": 0.6091,
      "step": 291
    },
    {
      "epoch": 1.2004110996916753,
      "grad_norm": 0.20619118213653564,
      "learning_rate": 0.00010627905195293135,
      "loss": 0.6076,
      "step": 292
    },
    {
      "epoch": 1.2045220966084276,
      "grad_norm": 0.22167161107063293,
      "learning_rate": 0.0001054950179912446,
      "loss": 0.8816,
      "step": 293
    },
    {
      "epoch": 1.20863309352518,
      "grad_norm": 0.2072238177061081,
      "learning_rate": 0.00010471064507096426,
      "loss": 0.7214,
      "step": 294
    },
    {
      "epoch": 1.2127440904419322,
      "grad_norm": 0.2088334858417511,
      "learning_rate": 0.00010392598157590688,
      "loss": 0.77,
      "step": 295
    },
    {
      "epoch": 1.2168550873586845,
      "grad_norm": 0.20157490670681,
      "learning_rate": 0.00010314107590781284,
      "loss": 0.566,
      "step": 296
    },
    {
      "epoch": 1.2209660842754368,
      "grad_norm": 0.21017199754714966,
      "learning_rate": 0.00010235597648336104,
      "loss": 0.7441,
      "step": 297
    },
    {
      "epoch": 1.2250770811921892,
      "grad_norm": 0.19077669084072113,
      "learning_rate": 0.00010157073173118208,
      "loss": 0.6781,
      "step": 298
    },
    {
      "epoch": 1.2291880781089415,
      "grad_norm": 0.2080603539943695,
      "learning_rate": 0.00010078539008887114,
      "loss": 0.7883,
      "step": 299
    },
    {
      "epoch": 1.2332990750256938,
      "grad_norm": 0.20922887325286865,
      "learning_rate": 0.0001,
      "loss": 0.6855,
      "step": 300
    },
    {
      "epoch": 1.2374100719424461,
      "grad_norm": 0.20584046840667725,
      "learning_rate": 9.921460991112891e-05,
      "loss": 0.7107,
      "step": 301
    },
    {
      "epoch": 1.2415210688591984,
      "grad_norm": 0.19233423471450806,
      "learning_rate": 9.842926826881796e-05,
      "loss": 0.6749,
      "step": 302
    },
    {
      "epoch": 1.2456320657759508,
      "grad_norm": 0.21580633521080017,
      "learning_rate": 9.764402351663901e-05,
      "loss": 0.6595,
      "step": 303
    },
    {
      "epoch": 1.249743062692703,
      "grad_norm": 0.21304890513420105,
      "learning_rate": 9.685892409218717e-05,
      "loss": 0.6387,
      "step": 304
    },
    {
      "epoch": 1.2538540596094552,
      "grad_norm": 0.21136440336704254,
      "learning_rate": 9.607401842409317e-05,
      "loss": 0.6474,
      "step": 305
    },
    {
      "epoch": 1.2579650565262077,
      "grad_norm": 0.199236661195755,
      "learning_rate": 9.528935492903575e-05,
      "loss": 0.6313,
      "step": 306
    },
    {
      "epoch": 1.2620760534429598,
      "grad_norm": 0.22413790225982666,
      "learning_rate": 9.450498200875546e-05,
      "loss": 0.7085,
      "step": 307
    },
    {
      "epoch": 1.2661870503597124,
      "grad_norm": 0.2168358415365219,
      "learning_rate": 9.372094804706867e-05,
      "loss": 0.6339,
      "step": 308
    },
    {
      "epoch": 1.2702980472764644,
      "grad_norm": 0.21489964425563812,
      "learning_rate": 9.293730140688336e-05,
      "loss": 0.7211,
      "step": 309
    },
    {
      "epoch": 1.274409044193217,
      "grad_norm": 0.19973471760749817,
      "learning_rate": 9.215409042721552e-05,
      "loss": 0.7066,
      "step": 310
    },
    {
      "epoch": 1.278520041109969,
      "grad_norm": 0.2147323042154312,
      "learning_rate": 9.137136342020768e-05,
      "loss": 0.6607,
      "step": 311
    },
    {
      "epoch": 1.2826310380267214,
      "grad_norm": 0.2042505294084549,
      "learning_rate": 9.058916866814858e-05,
      "loss": 0.6156,
      "step": 312
    },
    {
      "epoch": 1.2867420349434737,
      "grad_norm": 0.22227294743061066,
      "learning_rate": 8.980755442049502e-05,
      "loss": 0.6916,
      "step": 313
    },
    {
      "epoch": 1.290853031860226,
      "grad_norm": 0.21164259314537048,
      "learning_rate": 8.902656889089548e-05,
      "loss": 0.6994,
      "step": 314
    },
    {
      "epoch": 1.2949640287769784,
      "grad_norm": 0.20284223556518555,
      "learning_rate": 8.824626025421626e-05,
      "loss": 0.6307,
      "step": 315
    },
    {
      "epoch": 1.2990750256937307,
      "grad_norm": 0.2196999192237854,
      "learning_rate": 8.746667664356956e-05,
      "loss": 0.7666,
      "step": 316
    },
    {
      "epoch": 1.303186022610483,
      "grad_norm": 0.21564753353595734,
      "learning_rate": 8.668786614734478e-05,
      "loss": 0.6618,
      "step": 317
    },
    {
      "epoch": 1.3072970195272353,
      "grad_norm": 0.2207728624343872,
      "learning_rate": 8.590987680624174e-05,
      "loss": 0.6531,
      "step": 318
    },
    {
      "epoch": 1.3114080164439876,
      "grad_norm": 0.20650914311408997,
      "learning_rate": 8.51327566103077e-05,
      "loss": 0.6698,
      "step": 319
    },
    {
      "epoch": 1.31551901336074,
      "grad_norm": 0.20808032155036926,
      "learning_rate": 8.435655349597689e-05,
      "loss": 0.6661,
      "step": 320
    },
    {
      "epoch": 1.3196300102774923,
      "grad_norm": 0.19636279344558716,
      "learning_rate": 8.358131534311372e-05,
      "loss": 0.6238,
      "step": 321
    },
    {
      "epoch": 1.3237410071942446,
      "grad_norm": 0.21458536386489868,
      "learning_rate": 8.280708997205904e-05,
      "loss": 0.7495,
      "step": 322
    },
    {
      "epoch": 1.327852004110997,
      "grad_norm": 0.21887531876564026,
      "learning_rate": 8.203392514068074e-05,
      "loss": 0.6865,
      "step": 323
    },
    {
      "epoch": 1.3319630010277492,
      "grad_norm": 0.20179466903209686,
      "learning_rate": 8.126186854142752e-05,
      "loss": 0.6137,
      "step": 324
    },
    {
      "epoch": 1.3360739979445015,
      "grad_norm": 0.2168751358985901,
      "learning_rate": 8.049096779838719e-05,
      "loss": 0.755,
      "step": 325
    },
    {
      "epoch": 1.3401849948612539,
      "grad_norm": 0.21560409665107727,
      "learning_rate": 7.972127046434878e-05,
      "loss": 0.716,
      "step": 326
    },
    {
      "epoch": 1.3442959917780062,
      "grad_norm": 0.19912707805633545,
      "learning_rate": 7.895282401786945e-05,
      "loss": 0.6221,
      "step": 327
    },
    {
      "epoch": 1.3484069886947585,
      "grad_norm": 0.2023993283510208,
      "learning_rate": 7.818567586034577e-05,
      "loss": 0.6791,
      "step": 328
    },
    {
      "epoch": 1.3525179856115108,
      "grad_norm": 0.19967655837535858,
      "learning_rate": 7.741987331308964e-05,
      "loss": 0.6379,
      "step": 329
    },
    {
      "epoch": 1.3566289825282631,
      "grad_norm": 0.2044181376695633,
      "learning_rate": 7.66554636144095e-05,
      "loss": 0.7104,
      "step": 330
    },
    {
      "epoch": 1.3607399794450155,
      "grad_norm": 0.2115282118320465,
      "learning_rate": 7.589249391669616e-05,
      "loss": 0.6931,
      "step": 331
    },
    {
      "epoch": 1.3648509763617678,
      "grad_norm": 0.21434563398361206,
      "learning_rate": 7.513101128351454e-05,
      "loss": 0.6416,
      "step": 332
    },
    {
      "epoch": 1.36896197327852,
      "grad_norm": 0.20948341488838196,
      "learning_rate": 7.437106268670034e-05,
      "loss": 0.617,
      "step": 333
    },
    {
      "epoch": 1.3730729701952724,
      "grad_norm": 0.22513225674629211,
      "learning_rate": 7.361269500346274e-05,
      "loss": 0.6452,
      "step": 334
    },
    {
      "epoch": 1.3771839671120247,
      "grad_norm": 0.21809957921504974,
      "learning_rate": 7.285595501349258e-05,
      "loss": 0.669,
      "step": 335
    },
    {
      "epoch": 1.381294964028777,
      "grad_norm": 0.2174353450536728,
      "learning_rate": 7.210088939607708e-05,
      "loss": 0.6669,
      "step": 336
    },
    {
      "epoch": 1.3854059609455294,
      "grad_norm": 0.2154754251241684,
      "learning_rate": 7.134754472722017e-05,
      "loss": 0.6595,
      "step": 337
    },
    {
      "epoch": 1.3895169578622815,
      "grad_norm": 0.21969543397426605,
      "learning_rate": 7.059596747676962e-05,
      "loss": 0.6375,
      "step": 338
    },
    {
      "epoch": 1.393627954779034,
      "grad_norm": 0.2284114807844162,
      "learning_rate": 6.984620400555044e-05,
      "loss": 0.7243,
      "step": 339
    },
    {
      "epoch": 1.397738951695786,
      "grad_norm": 0.1979370266199112,
      "learning_rate": 6.909830056250527e-05,
      "loss": 0.6366,
      "step": 340
    },
    {
      "epoch": 1.4018499486125386,
      "grad_norm": 0.2233443707227707,
      "learning_rate": 6.835230328184138e-05,
      "loss": 0.7023,
      "step": 341
    },
    {
      "epoch": 1.4059609455292907,
      "grad_norm": 0.20319107174873352,
      "learning_rate": 6.760825818018508e-05,
      "loss": 0.7099,
      "step": 342
    },
    {
      "epoch": 1.4100719424460433,
      "grad_norm": 0.20188498497009277,
      "learning_rate": 6.68662111537429e-05,
      "loss": 0.6882,
      "step": 343
    },
    {
      "epoch": 1.4141829393627954,
      "grad_norm": 0.22043703496456146,
      "learning_rate": 6.612620797547087e-05,
      "loss": 0.7133,
      "step": 344
    },
    {
      "epoch": 1.418293936279548,
      "grad_norm": 0.20449554920196533,
      "learning_rate": 6.538829429225069e-05,
      "loss": 0.6838,
      "step": 345
    },
    {
      "epoch": 1.4224049331963,
      "grad_norm": 0.2155667394399643,
      "learning_rate": 6.465251562207431e-05,
      "loss": 0.712,
      "step": 346
    },
    {
      "epoch": 1.4265159301130523,
      "grad_norm": 0.21181143820285797,
      "learning_rate": 6.391891735123582e-05,
      "loss": 0.6925,
      "step": 347
    },
    {
      "epoch": 1.4306269270298047,
      "grad_norm": 0.2073758840560913,
      "learning_rate": 6.318754473153221e-05,
      "loss": 0.6784,
      "step": 348
    },
    {
      "epoch": 1.434737923946557,
      "grad_norm": 0.2193324863910675,
      "learning_rate": 6.245844287747168e-05,
      "loss": 0.7427,
      "step": 349
    },
    {
      "epoch": 1.4388489208633093,
      "grad_norm": 0.20216886699199677,
      "learning_rate": 6.173165676349103e-05,
      "loss": 0.6601,
      "step": 350
    },
    {
      "epoch": 1.4429599177800616,
      "grad_norm": 0.20811676979064941,
      "learning_rate": 6.1007231221181206e-05,
      "loss": 0.6826,
      "step": 351
    },
    {
      "epoch": 1.447070914696814,
      "grad_norm": 0.2313561737537384,
      "learning_rate": 6.0285210936521955e-05,
      "loss": 0.7001,
      "step": 352
    },
    {
      "epoch": 1.4511819116135662,
      "grad_norm": 0.21192209422588348,
      "learning_rate": 5.956564044712551e-05,
      "loss": 0.7647,
      "step": 353
    },
    {
      "epoch": 1.4552929085303186,
      "grad_norm": 0.21655315160751343,
      "learning_rate": 5.884856413948913e-05,
      "loss": 0.7176,
      "step": 354
    },
    {
      "epoch": 1.4594039054470709,
      "grad_norm": 0.22142696380615234,
      "learning_rate": 5.8134026246257225e-05,
      "loss": 0.7129,
      "step": 355
    },
    {
      "epoch": 1.4635149023638232,
      "grad_norm": 0.22208990156650543,
      "learning_rate": 5.7422070843492734e-05,
      "loss": 0.6819,
      "step": 356
    },
    {
      "epoch": 1.4676258992805755,
      "grad_norm": 0.19597405195236206,
      "learning_rate": 5.671274184795865e-05,
      "loss": 0.6963,
      "step": 357
    },
    {
      "epoch": 1.4717368961973278,
      "grad_norm": 0.18282102048397064,
      "learning_rate": 5.6006083014408484e-05,
      "loss": 0.6516,
      "step": 358
    },
    {
      "epoch": 1.4758478931140802,
      "grad_norm": 0.21373850107192993,
      "learning_rate": 5.53021379328879e-05,
      "loss": 0.676,
      "step": 359
    },
    {
      "epoch": 1.4799588900308325,
      "grad_norm": 0.2060038447380066,
      "learning_rate": 5.4600950026045326e-05,
      "loss": 0.7064,
      "step": 360
    },
    {
      "epoch": 1.4840698869475848,
      "grad_norm": 0.19545258581638336,
      "learning_rate": 5.390256254645378e-05,
      "loss": 0.6314,
      "step": 361
    },
    {
      "epoch": 1.4881808838643371,
      "grad_norm": 0.21232843399047852,
      "learning_rate": 5.320701857394268e-05,
      "loss": 0.6883,
      "step": 362
    },
    {
      "epoch": 1.4922918807810894,
      "grad_norm": 0.2111021876335144,
      "learning_rate": 5.251436101294056e-05,
      "loss": 0.6969,
      "step": 363
    },
    {
      "epoch": 1.4964028776978417,
      "grad_norm": 0.20870216190814972,
      "learning_rate": 5.182463258982846e-05,
      "loss": 0.6739,
      "step": 364
    },
    {
      "epoch": 1.500513874614594,
      "grad_norm": 0.22273269295692444,
      "learning_rate": 5.113787585030454e-05,
      "loss": 0.7781,
      "step": 365
    },
    {
      "epoch": 1.5046248715313464,
      "grad_norm": 0.22421270608901978,
      "learning_rate": 5.045413315675924e-05,
      "loss": 0.6341,
      "step": 366
    },
    {
      "epoch": 1.5087358684480987,
      "grad_norm": 0.2206403613090515,
      "learning_rate": 4.977344668566275e-05,
      "loss": 0.712,
      "step": 367
    },
    {
      "epoch": 1.512846865364851,
      "grad_norm": 0.210668683052063,
      "learning_rate": 4.909585842496287e-05,
      "loss": 0.6691,
      "step": 368
    },
    {
      "epoch": 1.5169578622816033,
      "grad_norm": 0.19762299954891205,
      "learning_rate": 4.842141017149526e-05,
      "loss": 0.6698,
      "step": 369
    },
    {
      "epoch": 1.5210688591983557,
      "grad_norm": 0.2029179334640503,
      "learning_rate": 4.7750143528405126e-05,
      "loss": 0.7191,
      "step": 370
    },
    {
      "epoch": 1.5251798561151078,
      "grad_norm": 0.21504949033260345,
      "learning_rate": 4.708209990258095e-05,
      "loss": 0.6326,
      "step": 371
    },
    {
      "epoch": 1.5292908530318603,
      "grad_norm": 0.22788691520690918,
      "learning_rate": 4.6417320502100316e-05,
      "loss": 0.6279,
      "step": 372
    },
    {
      "epoch": 1.5334018499486124,
      "grad_norm": 0.2158738374710083,
      "learning_rate": 4.575584633368815e-05,
      "loss": 0.6246,
      "step": 373
    },
    {
      "epoch": 1.537512846865365,
      "grad_norm": 0.21760185062885284,
      "learning_rate": 4.5097718200186814e-05,
      "loss": 0.6362,
      "step": 374
    },
    {
      "epoch": 1.541623843782117,
      "grad_norm": 0.22105571627616882,
      "learning_rate": 4.444297669803981e-05,
      "loss": 0.6717,
      "step": 375
    },
    {
      "epoch": 1.5457348406988696,
      "grad_norm": 0.20178484916687012,
      "learning_rate": 4.379166221478697e-05,
      "loss": 0.6622,
      "step": 376
    },
    {
      "epoch": 1.5498458376156217,
      "grad_norm": 0.2052689641714096,
      "learning_rate": 4.31438149265736e-05,
      "loss": 0.715,
      "step": 377
    },
    {
      "epoch": 1.5539568345323742,
      "grad_norm": 0.19674015045166016,
      "learning_rate": 4.249947479567218e-05,
      "loss": 0.6022,
      "step": 378
    },
    {
      "epoch": 1.5580678314491263,
      "grad_norm": 0.21909905970096588,
      "learning_rate": 4.185868156801694e-05,
      "loss": 0.6999,
      "step": 379
    },
    {
      "epoch": 1.5621788283658788,
      "grad_norm": 0.2254362851381302,
      "learning_rate": 4.12214747707527e-05,
      "loss": 0.6939,
      "step": 380
    },
    {
      "epoch": 1.566289825282631,
      "grad_norm": 0.2040894329547882,
      "learning_rate": 4.058789370979615e-05,
      "loss": 0.7336,
      "step": 381
    },
    {
      "epoch": 1.5704008221993835,
      "grad_norm": 0.22476311028003693,
      "learning_rate": 3.9957977467411615e-05,
      "loss": 0.7356,
      "step": 382
    },
    {
      "epoch": 1.5745118191161356,
      "grad_norm": 0.21693895757198334,
      "learning_rate": 3.933176489980005e-05,
      "loss": 0.6224,
      "step": 383
    },
    {
      "epoch": 1.5786228160328881,
      "grad_norm": 0.220389723777771,
      "learning_rate": 3.8709294634702376e-05,
      "loss": 0.7293,
      "step": 384
    },
    {
      "epoch": 1.5827338129496402,
      "grad_norm": 0.21954071521759033,
      "learning_rate": 3.8090605069016595e-05,
      "loss": 0.7157,
      "step": 385
    },
    {
      "epoch": 1.5868448098663928,
      "grad_norm": 0.21520665287971497,
      "learning_rate": 3.747573436642951e-05,
      "loss": 0.671,
      "step": 386
    },
    {
      "epoch": 1.5909558067831449,
      "grad_norm": 0.20397382974624634,
      "learning_rate": 3.686472045506223e-05,
      "loss": 0.6693,
      "step": 387
    },
    {
      "epoch": 1.5950668036998972,
      "grad_norm": 0.22096459567546844,
      "learning_rate": 3.6257601025131026e-05,
      "loss": 0.7277,
      "step": 388
    },
    {
      "epoch": 1.5991778006166495,
      "grad_norm": 0.2036741077899933,
      "learning_rate": 3.565441352662211e-05,
      "loss": 0.5967,
      "step": 389
    },
    {
      "epoch": 1.6032887975334018,
      "grad_norm": 0.21363277733325958,
      "learning_rate": 3.5055195166981645e-05,
      "loss": 0.6765,
      "step": 390
    },
    {
      "epoch": 1.6073997944501541,
      "grad_norm": 0.21069061756134033,
      "learning_rate": 3.445998290882062e-05,
      "loss": 0.7003,
      "step": 391
    },
    {
      "epoch": 1.6115107913669064,
      "grad_norm": 0.19123807549476624,
      "learning_rate": 3.386881346763483e-05,
      "loss": 0.6354,
      "step": 392
    },
    {
      "epoch": 1.6156217882836588,
      "grad_norm": 0.2067817896604538,
      "learning_rate": 3.328172330954001e-05,
      "loss": 0.7407,
      "step": 393
    },
    {
      "epoch": 1.619732785200411,
      "grad_norm": 0.2092856615781784,
      "learning_rate": 3.269874864902269e-05,
      "loss": 0.6656,
      "step": 394
    },
    {
      "epoch": 1.6238437821171634,
      "grad_norm": 0.21701788902282715,
      "learning_rate": 3.211992544670582e-05,
      "loss": 0.6847,
      "step": 395
    },
    {
      "epoch": 1.6279547790339157,
      "grad_norm": 0.20100919902324677,
      "learning_rate": 3.154528940713113e-05,
      "loss": 0.6239,
      "step": 396
    },
    {
      "epoch": 1.632065775950668,
      "grad_norm": 0.2200203686952591,
      "learning_rate": 3.0974875976556284e-05,
      "loss": 0.7482,
      "step": 397
    },
    {
      "epoch": 1.6361767728674204,
      "grad_norm": 0.2217399626970291,
      "learning_rate": 3.0408720340768572e-05,
      "loss": 0.6808,
      "step": 398
    },
    {
      "epoch": 1.6402877697841727,
      "grad_norm": 0.22481383383274078,
      "learning_rate": 2.9846857422914433e-05,
      "loss": 0.7116,
      "step": 399
    },
    {
      "epoch": 1.644398766700925,
      "grad_norm": 0.2123376727104187,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.6897,
      "step": 400
    },
    {
      "epoch": 1.6485097636176773,
      "grad_norm": 0.2224406749010086,
      "learning_rate": 2.8736148107479467e-05,
      "loss": 0.705,
      "step": 401
    },
    {
      "epoch": 1.6526207605344296,
      "grad_norm": 0.1946137398481369,
      "learning_rate": 2.8187370223681132e-05,
      "loss": 0.5942,
      "step": 402
    },
    {
      "epoch": 1.656731757451182,
      "grad_norm": 0.22985008358955383,
      "learning_rate": 2.764302208115508e-05,
      "loss": 0.7314,
      "step": 403
    },
    {
      "epoch": 1.6608427543679343,
      "grad_norm": 0.223613902926445,
      "learning_rate": 2.7103137257858868e-05,
      "loss": 0.7334,
      "step": 404
    },
    {
      "epoch": 1.6649537512846866,
      "grad_norm": 0.20055550336837769,
      "learning_rate": 2.6567749056431467e-05,
      "loss": 0.6186,
      "step": 405
    },
    {
      "epoch": 1.6690647482014387,
      "grad_norm": 0.21312382817268372,
      "learning_rate": 2.603689050213902e-05,
      "loss": 0.6441,
      "step": 406
    },
    {
      "epoch": 1.6731757451181912,
      "grad_norm": 0.2099653035402298,
      "learning_rate": 2.5510594340837824e-05,
      "loss": 0.6522,
      "step": 407
    },
    {
      "epoch": 1.6772867420349433,
      "grad_norm": 0.21277178823947906,
      "learning_rate": 2.4988893036954043e-05,
      "loss": 0.7224,
      "step": 408
    },
    {
      "epoch": 1.6813977389516959,
      "grad_norm": 0.19120965898036957,
      "learning_rate": 2.4471818771481648e-05,
      "loss": 0.6124,
      "step": 409
    },
    {
      "epoch": 1.685508735868448,
      "grad_norm": 0.19973115622997284,
      "learning_rate": 2.3959403439996907e-05,
      "loss": 0.6481,
      "step": 410
    },
    {
      "epoch": 1.6896197327852005,
      "grad_norm": 0.20857864618301392,
      "learning_rate": 2.34516786506912e-05,
      "loss": 0.6761,
      "step": 411
    },
    {
      "epoch": 1.6937307297019526,
      "grad_norm": 0.20877736806869507,
      "learning_rate": 2.2948675722421086e-05,
      "loss": 0.661,
      "step": 412
    },
    {
      "epoch": 1.6978417266187051,
      "grad_norm": 0.22759252786636353,
      "learning_rate": 2.2450425682776565e-05,
      "loss": 0.7186,
      "step": 413
    },
    {
      "epoch": 1.7019527235354572,
      "grad_norm": 0.2148085981607437,
      "learning_rate": 2.195695926616702e-05,
      "loss": 0.6914,
      "step": 414
    },
    {
      "epoch": 1.7060637204522098,
      "grad_norm": 0.2166018933057785,
      "learning_rate": 2.146830691192553e-05,
      "loss": 0.6557,
      "step": 415
    },
    {
      "epoch": 1.7101747173689619,
      "grad_norm": 0.21270379424095154,
      "learning_rate": 2.098449876243096e-05,
      "loss": 0.6932,
      "step": 416
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.20171096920967102,
      "learning_rate": 2.0505564661249e-05,
      "loss": 0.6497,
      "step": 417
    },
    {
      "epoch": 1.7183967112024665,
      "grad_norm": 0.2250605672597885,
      "learning_rate": 2.0031534151290943e-05,
      "loss": 0.6639,
      "step": 418
    },
    {
      "epoch": 1.722507708119219,
      "grad_norm": 0.21208657324314117,
      "learning_rate": 1.9562436472991552e-05,
      "loss": 0.7111,
      "step": 419
    },
    {
      "epoch": 1.7266187050359711,
      "grad_norm": 0.21967846155166626,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 0.6989,
      "step": 420
    },
    {
      "epoch": 1.7307297019527237,
      "grad_norm": 0.21354608237743378,
      "learning_rate": 1.863915504992131e-05,
      "loss": 0.6644,
      "step": 421
    },
    {
      "epoch": 1.7348406988694758,
      "grad_norm": 0.22123783826828003,
      "learning_rate": 1.818502825749764e-05,
      "loss": 0.7171,
      "step": 422
    },
    {
      "epoch": 1.7389516957862283,
      "grad_norm": 0.22253422439098358,
      "learning_rate": 1.773594819791403e-05,
      "loss": 0.6772,
      "step": 423
    },
    {
      "epoch": 1.7430626927029804,
      "grad_norm": 0.20984973013401031,
      "learning_rate": 1.7291942572543807e-05,
      "loss": 0.6228,
      "step": 424
    },
    {
      "epoch": 1.7471736896197327,
      "grad_norm": 0.22086894512176514,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 0.6943,
      "step": 425
    },
    {
      "epoch": 1.751284686536485,
      "grad_norm": 0.23192021250724792,
      "learning_rate": 1.6419263863172997e-05,
      "loss": 0.6851,
      "step": 426
    },
    {
      "epoch": 1.7553956834532374,
      "grad_norm": 0.22143663465976715,
      "learning_rate": 1.599064461010581e-05,
      "loss": 0.714,
      "step": 427
    },
    {
      "epoch": 1.7595066803699897,
      "grad_norm": 0.21095366775989532,
      "learning_rate": 1.5567207449798515e-05,
      "loss": 0.6482,
      "step": 428
    },
    {
      "epoch": 1.763617677286742,
      "grad_norm": 0.22433477640151978,
      "learning_rate": 1.5148978501849642e-05,
      "loss": 0.665,
      "step": 429
    },
    {
      "epoch": 1.7677286742034943,
      "grad_norm": 0.20697914063930511,
      "learning_rate": 1.4735983564590783e-05,
      "loss": 0.6062,
      "step": 430
    },
    {
      "epoch": 1.7718396711202467,
      "grad_norm": 0.2220154106616974,
      "learning_rate": 1.4328248113495047e-05,
      "loss": 0.6616,
      "step": 431
    },
    {
      "epoch": 1.775950668036999,
      "grad_norm": 0.23051774501800537,
      "learning_rate": 1.3925797299605647e-05,
      "loss": 0.7011,
      "step": 432
    },
    {
      "epoch": 1.7800616649537513,
      "grad_norm": 0.23257797956466675,
      "learning_rate": 1.3528655947984504e-05,
      "loss": 0.7214,
      "step": 433
    },
    {
      "epoch": 1.7841726618705036,
      "grad_norm": 0.21937896311283112,
      "learning_rate": 1.3136848556180892e-05,
      "loss": 0.6699,
      "step": 434
    },
    {
      "epoch": 1.788283658787256,
      "grad_norm": 0.22290131449699402,
      "learning_rate": 1.2750399292720283e-05,
      "loss": 0.6753,
      "step": 435
    },
    {
      "epoch": 1.7923946557040082,
      "grad_norm": 0.23099978268146515,
      "learning_rate": 1.2369331995613665e-05,
      "loss": 0.7554,
      "step": 436
    },
    {
      "epoch": 1.7965056526207606,
      "grad_norm": 0.21958616375923157,
      "learning_rate": 1.1993670170886806e-05,
      "loss": 0.7929,
      "step": 437
    },
    {
      "epoch": 1.8006166495375129,
      "grad_norm": 0.2076365202665329,
      "learning_rate": 1.1623436991130654e-05,
      "loss": 0.6315,
      "step": 438
    },
    {
      "epoch": 1.8047276464542652,
      "grad_norm": 0.20589125156402588,
      "learning_rate": 1.1258655294071685e-05,
      "loss": 0.5885,
      "step": 439
    },
    {
      "epoch": 1.8088386433710175,
      "grad_norm": 0.22324712574481964,
      "learning_rate": 1.0899347581163221e-05,
      "loss": 0.6675,
      "step": 440
    },
    {
      "epoch": 1.8129496402877698,
      "grad_norm": 0.21829169988632202,
      "learning_rate": 1.0545536016197499e-05,
      "loss": 0.6778,
      "step": 441
    },
    {
      "epoch": 1.8170606372045222,
      "grad_norm": 0.20735836029052734,
      "learning_rate": 1.0197242423938446e-05,
      "loss": 0.6285,
      "step": 442
    },
    {
      "epoch": 1.8211716341212743,
      "grad_norm": 0.2346031367778778,
      "learning_rate": 9.854488288775422e-06,
      "loss": 0.6879,
      "step": 443
    },
    {
      "epoch": 1.8252826310380268,
      "grad_norm": 0.20201553404331207,
      "learning_rate": 9.517294753398064e-06,
      "loss": 0.6394,
      "step": 444
    },
    {
      "epoch": 1.829393627954779,
      "grad_norm": 0.20408907532691956,
      "learning_rate": 9.185682617491863e-06,
      "loss": 0.5979,
      "step": 445
    },
    {
      "epoch": 1.8335046248715314,
      "grad_norm": 0.2089148461818695,
      "learning_rate": 8.85967233645547e-06,
      "loss": 0.6033,
      "step": 446
    },
    {
      "epoch": 1.8376156217882835,
      "grad_norm": 0.24372489750385284,
      "learning_rate": 8.539284020138638e-06,
      "loss": 0.7283,
      "step": 447
    },
    {
      "epoch": 1.841726618705036,
      "grad_norm": 0.23384876549243927,
      "learning_rate": 8.224537431601886e-06,
      "loss": 0.6958,
      "step": 448
    },
    {
      "epoch": 1.8458376156217882,
      "grad_norm": 0.22088314592838287,
      "learning_rate": 7.915451985897382e-06,
      "loss": 0.6806,
      "step": 449
    },
    {
      "epoch": 1.8499486125385407,
      "grad_norm": 0.2152312994003296,
      "learning_rate": 7.612046748871327e-06,
      "loss": 0.6843,
      "step": 450
    },
    {
      "epoch": 1.8540596094552928,
      "grad_norm": 0.2192438691854477,
      "learning_rate": 7.314340435987921e-06,
      "loss": 0.6708,
      "step": 451
    },
    {
      "epoch": 1.8581706063720453,
      "grad_norm": 0.22158104181289673,
      "learning_rate": 7.022351411174866e-06,
      "loss": 0.6602,
      "step": 452
    },
    {
      "epoch": 1.8622816032887974,
      "grad_norm": 0.20825935900211334,
      "learning_rate": 6.736097685690601e-06,
      "loss": 0.6707,
      "step": 453
    },
    {
      "epoch": 1.86639260020555,
      "grad_norm": 0.20734339952468872,
      "learning_rate": 6.455596917013273e-06,
      "loss": 0.7042,
      "step": 454
    },
    {
      "epoch": 1.870503597122302,
      "grad_norm": 0.18766121566295624,
      "learning_rate": 6.180866407751595e-06,
      "loss": 0.6123,
      "step": 455
    },
    {
      "epoch": 1.8746145940390546,
      "grad_norm": 0.19911910593509674,
      "learning_rate": 5.911923104577455e-06,
      "loss": 0.6345,
      "step": 456
    },
    {
      "epoch": 1.8787255909558067,
      "grad_norm": 0.20555348694324493,
      "learning_rate": 5.6487835971806515e-06,
      "loss": 0.6796,
      "step": 457
    },
    {
      "epoch": 1.8828365878725593,
      "grad_norm": 0.22171315550804138,
      "learning_rate": 5.39146411724547e-06,
      "loss": 0.6782,
      "step": 458
    },
    {
      "epoch": 1.8869475847893113,
      "grad_norm": 0.20477399230003357,
      "learning_rate": 5.13998053744954e-06,
      "loss": 0.6542,
      "step": 459
    },
    {
      "epoch": 1.8910585817060637,
      "grad_norm": 0.211378812789917,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 0.6869,
      "step": 460
    },
    {
      "epoch": 1.895169578622816,
      "grad_norm": 0.19824330508708954,
      "learning_rate": 4.65458276809988e-06,
      "loss": 0.6009,
      "step": 461
    },
    {
      "epoch": 1.8992805755395683,
      "grad_norm": 0.20775100588798523,
      "learning_rate": 4.420698520166988e-06,
      "loss": 0.6299,
      "step": 462
    },
    {
      "epoch": 1.9033915724563206,
      "grad_norm": 0.23011107742786407,
      "learning_rate": 4.19271005376809e-06,
      "loss": 0.6573,
      "step": 463
    },
    {
      "epoch": 1.907502569373073,
      "grad_norm": 0.2232479453086853,
      "learning_rate": 3.970631432305694e-06,
      "loss": 0.675,
      "step": 464
    },
    {
      "epoch": 1.9116135662898253,
      "grad_norm": 0.2247730940580368,
      "learning_rate": 3.7544763546352834e-06,
      "loss": 0.7775,
      "step": 465
    },
    {
      "epoch": 1.9157245632065776,
      "grad_norm": 0.21717530488967896,
      "learning_rate": 3.5442581542201923e-06,
      "loss": 0.6305,
      "step": 466
    },
    {
      "epoch": 1.91983556012333,
      "grad_norm": 0.23125427961349487,
      "learning_rate": 3.339989798309273e-06,
      "loss": 0.7336,
      "step": 467
    },
    {
      "epoch": 1.9239465570400822,
      "grad_norm": 0.2131197601556778,
      "learning_rate": 3.1416838871368924e-06,
      "loss": 0.6839,
      "step": 468
    },
    {
      "epoch": 1.9280575539568345,
      "grad_norm": 0.2215229570865631,
      "learning_rate": 2.949352653145754e-06,
      "loss": 0.6604,
      "step": 469
    },
    {
      "epoch": 1.9321685508735869,
      "grad_norm": 0.21065653860569,
      "learning_rate": 2.7630079602323442e-06,
      "loss": 0.6712,
      "step": 470
    },
    {
      "epoch": 1.9362795477903392,
      "grad_norm": 0.2188219279050827,
      "learning_rate": 2.582661303015066e-06,
      "loss": 0.6734,
      "step": 471
    },
    {
      "epoch": 1.9403905447070915,
      "grad_norm": 0.21708612143993378,
      "learning_rate": 2.4083238061252567e-06,
      "loss": 0.7141,
      "step": 472
    },
    {
      "epoch": 1.9445015416238438,
      "grad_norm": 0.21707835793495178,
      "learning_rate": 2.240006223520941e-06,
      "loss": 0.6693,
      "step": 473
    },
    {
      "epoch": 1.9486125385405961,
      "grad_norm": 0.21637198328971863,
      "learning_rate": 2.0777189378234143e-06,
      "loss": 0.6397,
      "step": 474
    },
    {
      "epoch": 1.9527235354573484,
      "grad_norm": 0.2045229971408844,
      "learning_rate": 1.921471959676957e-06,
      "loss": 0.6079,
      "step": 475
    },
    {
      "epoch": 1.9568345323741008,
      "grad_norm": 0.22080466151237488,
      "learning_rate": 1.771274927131139e-06,
      "loss": 0.7531,
      "step": 476
    },
    {
      "epoch": 1.960945529290853,
      "grad_norm": 0.21688653528690338,
      "learning_rate": 1.6271371050464056e-06,
      "loss": 0.6342,
      "step": 477
    },
    {
      "epoch": 1.9650565262076052,
      "grad_norm": 0.21895740926265717,
      "learning_rate": 1.4890673845226133e-06,
      "loss": 0.6558,
      "step": 478
    },
    {
      "epoch": 1.9691675231243577,
      "grad_norm": 0.21992015838623047,
      "learning_rate": 1.3570742823504567e-06,
      "loss": 0.704,
      "step": 479
    },
    {
      "epoch": 1.9732785200411098,
      "grad_norm": 0.21318915486335754,
      "learning_rate": 1.231165940486234e-06,
      "loss": 0.6551,
      "step": 480
    },
    {
      "epoch": 1.9773895169578624,
      "grad_norm": 0.23126220703125,
      "learning_rate": 1.1113501255495485e-06,
      "loss": 0.6678,
      "step": 481
    },
    {
      "epoch": 1.9815005138746145,
      "grad_norm": 0.23048514127731323,
      "learning_rate": 9.976342283442463e-07,
      "loss": 0.7386,
      "step": 482
    },
    {
      "epoch": 1.985611510791367,
      "grad_norm": 0.20240825414657593,
      "learning_rate": 8.900252634025274e-07,
      "loss": 0.6322,
      "step": 483
    },
    {
      "epoch": 1.989722507708119,
      "grad_norm": 0.22034034132957458,
      "learning_rate": 7.885298685522235e-07,
      "loss": 0.7067,
      "step": 484
    },
    {
      "epoch": 1.9938335046248716,
      "grad_norm": 0.2250751405954361,
      "learning_rate": 6.931543045073708e-07,
      "loss": 0.7403,
      "step": 485
    },
    {
      "epoch": 1.9979445015416237,
      "grad_norm": 0.2361016869544983,
      "learning_rate": 6.039044544820404e-07,
      "loss": 0.7316,
      "step": 486
    },
    {
      "epoch": 1.9979445015416237,
      "eval_loss": 0.734196126461029,
      "eval_runtime": 18.5187,
      "eval_samples_per_second": 66.365,
      "eval_steps_per_second": 2.808,
      "step": 486
    },
    {
      "epoch": 2.0020554984583763,
      "grad_norm": 0.43072688579559326,
      "learning_rate": 5.207858238273522e-07,
      "loss": 1.0921,
      "step": 487
    },
    {
      "epoch": 2.0061664953751284,
      "grad_norm": 0.23005196452140808,
      "learning_rate": 4.438035396920004e-07,
      "loss": 0.7754,
      "step": 488
    },
    {
      "epoch": 2.010277492291881,
      "grad_norm": 0.19805458188056946,
      "learning_rate": 3.7296235070587435e-07,
      "loss": 0.5467,
      "step": 489
    },
    {
      "epoch": 2.014388489208633,
      "grad_norm": 0.21197497844696045,
      "learning_rate": 3.0826662668720364e-07,
      "loss": 0.665,
      "step": 490
    },
    {
      "epoch": 2.0184994861253855,
      "grad_norm": 0.19797663390636444,
      "learning_rate": 2.497203583729957e-07,
      "loss": 0.6127,
      "step": 491
    },
    {
      "epoch": 2.0226104830421376,
      "grad_norm": 0.2046641856431961,
      "learning_rate": 1.973271571728441e-07,
      "loss": 0.607,
      "step": 492
    },
    {
      "epoch": 2.02672147995889,
      "grad_norm": 0.20540988445281982,
      "learning_rate": 1.5109025494620677e-07,
      "loss": 0.6685,
      "step": 493
    },
    {
      "epoch": 2.0308324768756423,
      "grad_norm": 0.1942788064479828,
      "learning_rate": 1.1101250380300965e-07,
      "loss": 0.5893,
      "step": 494
    },
    {
      "epoch": 2.034943473792395,
      "grad_norm": 0.2019982933998108,
      "learning_rate": 7.709637592770991e-08,
      "loss": 0.6838,
      "step": 495
    },
    {
      "epoch": 2.039054470709147,
      "grad_norm": 0.19681663811206818,
      "learning_rate": 4.934396342684e-08,
      "loss": 0.5974,
      "step": 496
    },
    {
      "epoch": 2.0431654676258995,
      "grad_norm": 0.20149853825569153,
      "learning_rate": 2.7756978199944273e-08,
      "loss": 0.6318,
      "step": 497
    },
    {
      "epoch": 2.0472764645426516,
      "grad_norm": 0.18899139761924744,
      "learning_rate": 1.2336751833941229e-08,
      "loss": 0.5549,
      "step": 498
    },
    {
      "epoch": 2.051387461459404,
      "grad_norm": 0.21279366314411163,
      "learning_rate": 3.0842355210336515e-09,
      "loss": 0.6963,
      "step": 499
    },
    {
      "epoch": 2.055498458376156,
      "grad_norm": 0.18985603749752045,
      "learning_rate": 0.0,
      "loss": 0.4875,
      "step": 500
    },
    {
      "epoch": 2.055498458376156,
      "eval_loss": 0.7341954112052917,
      "eval_runtime": 18.4158,
      "eval_samples_per_second": 66.736,
      "eval_steps_per_second": 2.824,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.283442719022121e+17,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}