{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 9480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00010548523206751055,
      "grad_norm": 2.011542320251465,
      "learning_rate": 1.5822784810126583e-05,
      "loss": 7.6591,
      "step": 1
    },
    {
      "epoch": 0.0002109704641350211,
      "grad_norm": 1.9964383840560913,
      "learning_rate": 3.1645569620253167e-05,
      "loss": 7.6601,
      "step": 2
    },
    {
      "epoch": 0.00031645569620253165,
      "grad_norm": 1.9953736066818237,
      "learning_rate": 4.746835443037975e-05,
      "loss": 7.6399,
      "step": 3
    },
    {
      "epoch": 0.0004219409282700422,
      "grad_norm": 2.022376537322998,
      "learning_rate": 6.329113924050633e-05,
      "loss": 7.5979,
      "step": 4
    },
    {
      "epoch": 0.0005274261603375527,
      "grad_norm": 2.0246565341949463,
      "learning_rate": 7.911392405063291e-05,
      "loss": 7.532,
      "step": 5
    },
    {
      "epoch": 0.0006329113924050633,
      "grad_norm": 1.8483935594558716,
      "learning_rate": 9.49367088607595e-05,
      "loss": 7.459,
      "step": 6
    },
    {
      "epoch": 0.0007383966244725738,
      "grad_norm": 1.706437110900879,
      "learning_rate": 0.00011075949367088609,
      "loss": 7.3705,
      "step": 7
    },
    {
      "epoch": 0.0008438818565400844,
      "grad_norm": 1.507551908493042,
      "learning_rate": 0.00012658227848101267,
      "loss": 7.283,
      "step": 8
    },
    {
      "epoch": 0.0009493670886075949,
      "grad_norm": 1.3828750848770142,
      "learning_rate": 0.00014240506329113925,
      "loss": 7.191,
      "step": 9
    },
    {
      "epoch": 0.0010548523206751054,
      "grad_norm": 1.251572847366333,
      "learning_rate": 0.00015822784810126583,
      "loss": 7.1113,
      "step": 10
    },
    {
      "epoch": 0.001160337552742616,
      "grad_norm": 1.2011066675186157,
      "learning_rate": 0.0001740506329113924,
      "loss": 7.0247,
      "step": 11
    },
    {
      "epoch": 0.0012658227848101266,
      "grad_norm": 1.1461619138717651,
      "learning_rate": 0.000189873417721519,
      "loss": 6.9572,
      "step": 12
    },
    {
      "epoch": 0.0013713080168776372,
      "grad_norm": 1.1297284364700317,
      "learning_rate": 0.00020569620253164557,
      "loss": 6.8903,
      "step": 13
    },
    {
      "epoch": 0.0014767932489451476,
      "grad_norm": 1.1294082403182983,
      "learning_rate": 0.00022151898734177217,
      "loss": 6.8264,
      "step": 14
    },
    {
      "epoch": 0.0015822784810126582,
      "grad_norm": 1.120693325996399,
      "learning_rate": 0.00023734177215189873,
      "loss": 6.7678,
      "step": 15
    },
    {
      "epoch": 0.0016877637130801688,
      "grad_norm": 1.1134696006774902,
      "learning_rate": 0.00025316455696202533,
      "loss": 6.7163,
      "step": 16
    },
    {
      "epoch": 0.0017932489451476794,
      "grad_norm": 1.1267699003219604,
      "learning_rate": 0.0002689873417721519,
      "loss": 6.6375,
      "step": 17
    },
    {
      "epoch": 0.0018987341772151898,
      "grad_norm": 1.1200734376907349,
      "learning_rate": 0.0002848101265822785,
      "loss": 6.5736,
      "step": 18
    },
    {
      "epoch": 0.0020042194092827004,
      "grad_norm": 1.1191186904907227,
      "learning_rate": 0.00030063291139240507,
      "loss": 6.4996,
      "step": 19
    },
    {
      "epoch": 0.002109704641350211,
      "grad_norm": 1.0853898525238037,
      "learning_rate": 0.00031645569620253165,
      "loss": 6.4426,
      "step": 20
    },
    {
      "epoch": 0.0022151898734177216,
      "grad_norm": 1.0650807619094849,
      "learning_rate": 0.00033227848101265823,
      "loss": 6.3779,
      "step": 21
    },
    {
      "epoch": 0.002320675105485232,
      "grad_norm": 1.0586391687393188,
      "learning_rate": 0.0003481012658227848,
      "loss": 6.311,
      "step": 22
    },
    {
      "epoch": 0.002426160337552743,
      "grad_norm": 1.0491359233856201,
      "learning_rate": 0.00036392405063291145,
      "loss": 6.2385,
      "step": 23
    },
    {
      "epoch": 0.002531645569620253,
      "grad_norm": 1.0341084003448486,
      "learning_rate": 0.000379746835443038,
      "loss": 6.1742,
      "step": 24
    },
    {
      "epoch": 0.0026371308016877636,
      "grad_norm": 0.9996719360351562,
      "learning_rate": 0.00039556962025316455,
      "loss": 6.1213,
      "step": 25
    },
    {
      "epoch": 0.0027426160337552744,
      "grad_norm": 0.9974717497825623,
      "learning_rate": 0.00041139240506329113,
      "loss": 6.048,
      "step": 26
    },
    {
      "epoch": 0.002848101265822785,
      "grad_norm": 0.9489433765411377,
      "learning_rate": 0.00042721518987341776,
      "loss": 6.0148,
      "step": 27
    },
    {
      "epoch": 0.002953586497890295,
      "grad_norm": 0.9366974830627441,
      "learning_rate": 0.00044303797468354434,
      "loss": 5.9313,
      "step": 28
    },
    {
      "epoch": 0.003059071729957806,
      "grad_norm": 0.9141618013381958,
      "learning_rate": 0.0004588607594936709,
      "loss": 5.8789,
      "step": 29
    },
    {
      "epoch": 0.0031645569620253164,
      "grad_norm": 0.8686885237693787,
      "learning_rate": 0.00047468354430379745,
      "loss": 5.8355,
      "step": 30
    },
    {
      "epoch": 0.003270042194092827,
      "grad_norm": 0.8527199029922485,
      "learning_rate": 0.0004905063291139241,
      "loss": 5.7709,
      "step": 31
    },
    {
      "epoch": 0.0033755274261603376,
      "grad_norm": 0.8326303958892822,
      "learning_rate": 0.0005063291139240507,
      "loss": 5.6998,
      "step": 32
    },
    {
      "epoch": 0.003481012658227848,
      "grad_norm": 0.8138381242752075,
      "learning_rate": 0.0005221518987341772,
      "loss": 5.6333,
      "step": 33
    },
    {
      "epoch": 0.003586497890295359,
      "grad_norm": 0.7682183384895325,
      "learning_rate": 0.0005379746835443038,
      "loss": 5.5995,
      "step": 34
    },
    {
      "epoch": 0.003691983122362869,
      "grad_norm": 0.7663596868515015,
      "learning_rate": 0.0005537974683544304,
      "loss": 5.5098,
      "step": 35
    },
    {
      "epoch": 0.0037974683544303796,
      "grad_norm": 0.7409225702285767,
      "learning_rate": 0.000569620253164557,
      "loss": 5.4808,
      "step": 36
    },
    {
      "epoch": 0.0039029535864978904,
      "grad_norm": 0.7263235449790955,
      "learning_rate": 0.0005854430379746836,
      "loss": 5.4119,
      "step": 37
    },
    {
      "epoch": 0.004008438818565401,
      "grad_norm": 0.6572887301445007,
      "learning_rate": 0.0006012658227848101,
      "loss": 5.3797,
      "step": 38
    },
    {
      "epoch": 0.004113924050632912,
      "grad_norm": 0.6338242292404175,
      "learning_rate": 0.0006170886075949367,
      "loss": 5.289,
      "step": 39
    },
    {
      "epoch": 0.004219409282700422,
      "grad_norm": 0.6308150887489319,
      "learning_rate": 0.0006329113924050633,
      "loss": 5.2654,
      "step": 40
    },
    {
      "epoch": 0.004324894514767932,
      "grad_norm": 0.6700575947761536,
      "learning_rate": 0.0006487341772151899,
      "loss": 5.2235,
      "step": 41
    },
    {
      "epoch": 0.004430379746835443,
      "grad_norm": 1.0167239904403687,
      "learning_rate": 0.0006645569620253165,
      "loss": 5.1842,
      "step": 42
    },
    {
      "epoch": 0.004535864978902953,
      "grad_norm": 1.0152013301849365,
      "learning_rate": 0.000680379746835443,
      "loss": 5.1102,
      "step": 43
    },
    {
      "epoch": 0.004641350210970464,
      "grad_norm": 0.5130178332328796,
      "learning_rate": 0.0006962025316455696,
      "loss": 5.0944,
      "step": 44
    },
    {
      "epoch": 0.004746835443037975,
      "grad_norm": 1.6894309520721436,
      "learning_rate": 0.0007120253164556963,
      "loss": 5.0441,
      "step": 45
    },
    {
      "epoch": 0.004852320675105486,
      "grad_norm": 0.4479582905769348,
      "learning_rate": 0.0007278481012658229,
      "loss": 4.9589,
      "step": 46
    },
    {
      "epoch": 0.004957805907172996,
      "grad_norm": 0.7943894267082214,
      "learning_rate": 0.0007436708860759495,
      "loss": 4.9549,
      "step": 47
    },
    {
      "epoch": 0.005063291139240506,
      "grad_norm": 0.48056092858314514,
      "learning_rate": 0.000759493670886076,
      "loss": 4.8983,
      "step": 48
    },
    {
      "epoch": 0.005168776371308017,
      "grad_norm": 0.8553137183189392,
      "learning_rate": 0.0007753164556962025,
      "loss": 4.8396,
      "step": 49
    },
    {
      "epoch": 0.005274261603375527,
      "grad_norm": 0.4161706566810608,
      "learning_rate": 0.0007911392405063291,
      "loss": 4.81,
      "step": 50
    },
    {
      "epoch": 0.005379746835443038,
      "grad_norm": 0.5087617039680481,
      "learning_rate": 0.0008069620253164557,
      "loss": 4.763,
      "step": 51
    },
    {
      "epoch": 0.005485232067510549,
      "grad_norm": 0.4462997019290924,
      "learning_rate": 0.0008227848101265823,
      "loss": 4.7158,
      "step": 52
    },
    {
      "epoch": 0.005590717299578059,
      "grad_norm": 0.4181429445743561,
      "learning_rate": 0.000838607594936709,
      "loss": 4.6819,
      "step": 53
    },
    {
      "epoch": 0.00569620253164557,
      "grad_norm": 0.44886595010757446,
      "learning_rate": 0.0008544303797468355,
      "loss": 4.6129,
      "step": 54
    },
    {
      "epoch": 0.0058016877637130804,
      "grad_norm": 0.3937552571296692,
      "learning_rate": 0.0008702531645569621,
      "loss": 4.599,
      "step": 55
    },
    {
      "epoch": 0.00590717299578059,
      "grad_norm": 0.3460679352283478,
      "learning_rate": 0.0008860759493670887,
      "loss": 4.5437,
      "step": 56
    },
    {
      "epoch": 0.006012658227848101,
      "grad_norm": 0.36171504855155945,
      "learning_rate": 0.0009018987341772153,
      "loss": 4.545,
      "step": 57
    },
    {
      "epoch": 0.006118143459915612,
      "grad_norm": 0.4088135063648224,
      "learning_rate": 0.0009177215189873418,
      "loss": 4.5293,
      "step": 58
    },
    {
      "epoch": 0.006223628691983122,
      "grad_norm": 0.37013325095176697,
      "learning_rate": 0.0009335443037974683,
      "loss": 4.4753,
      "step": 59
    },
    {
      "epoch": 0.006329113924050633,
      "grad_norm": 0.32116636633872986,
      "learning_rate": 0.0009493670886075949,
      "loss": 4.432,
      "step": 60
    },
    {
      "epoch": 0.006434599156118144,
      "grad_norm": 0.30790355801582336,
      "learning_rate": 0.0009651898734177215,
      "loss": 4.4279,
      "step": 61
    },
    {
      "epoch": 0.006540084388185654,
      "grad_norm": 0.33993634581565857,
      "learning_rate": 0.0009810126582278482,
      "loss": 4.3744,
      "step": 62
    },
    {
      "epoch": 0.006645569620253164,
      "grad_norm": 0.29409778118133545,
      "learning_rate": 0.0009968354430379747,
      "loss": 4.3612,
      "step": 63
    },
    {
      "epoch": 0.006751054852320675,
      "grad_norm": 0.35212376713752747,
      "learning_rate": 0.0010126582278481013,
      "loss": 4.3189,
      "step": 64
    },
    {
      "epoch": 0.006856540084388186,
      "grad_norm": 0.4115193486213684,
      "learning_rate": 0.001028481012658228,
      "loss": 4.3196,
      "step": 65
    },
    {
      "epoch": 0.006962025316455696,
      "grad_norm": 0.4627211093902588,
      "learning_rate": 0.0010443037974683545,
      "loss": 4.2708,
      "step": 66
    },
    {
      "epoch": 0.007067510548523207,
      "grad_norm": 0.5270651578903198,
      "learning_rate": 0.001060126582278481,
      "loss": 4.2782,
      "step": 67
    },
    {
      "epoch": 0.007172995780590718,
      "grad_norm": 0.794312596321106,
      "learning_rate": 0.0010759493670886076,
      "loss": 4.2603,
      "step": 68
    },
    {
      "epoch": 0.007278481012658228,
      "grad_norm": 0.6797000169754028,
      "learning_rate": 0.0010917721518987342,
      "loss": 4.205,
      "step": 69
    },
    {
      "epoch": 0.007383966244725738,
      "grad_norm": 0.46117499470710754,
      "learning_rate": 0.0011075949367088608,
      "loss": 4.1972,
      "step": 70
    },
    {
      "epoch": 0.007489451476793249,
      "grad_norm": 0.4507919251918793,
      "learning_rate": 0.0011234177215189874,
      "loss": 4.2155,
      "step": 71
    },
    {
      "epoch": 0.007594936708860759,
      "grad_norm": 0.38559678196907043,
      "learning_rate": 0.001139240506329114,
      "loss": 4.1693,
      "step": 72
    },
    {
      "epoch": 0.00770042194092827,
      "grad_norm": 0.41005730628967285,
      "learning_rate": 0.0011550632911392405,
      "loss": 4.1402,
      "step": 73
    },
    {
      "epoch": 0.007805907172995781,
      "grad_norm": 0.5099040865898132,
      "learning_rate": 0.0011708860759493671,
      "loss": 4.1277,
      "step": 74
    },
    {
      "epoch": 0.007911392405063292,
      "grad_norm": 0.9070386290550232,
      "learning_rate": 0.0011867088607594937,
      "loss": 4.1139,
      "step": 75
    },
    {
      "epoch": 0.008016877637130802,
      "grad_norm": 1.352543592453003,
      "learning_rate": 0.0012025316455696203,
      "loss": 4.13,
      "step": 76
    },
    {
      "epoch": 0.008122362869198312,
      "grad_norm": 0.685218870639801,
      "learning_rate": 0.0012183544303797469,
      "loss": 4.0973,
      "step": 77
    },
    {
      "epoch": 0.008227848101265823,
      "grad_norm": 1.1689188480377197,
      "learning_rate": 0.0012341772151898734,
      "loss": 4.0989,
      "step": 78
    },
    {
      "epoch": 0.008333333333333333,
      "grad_norm": 1.1035869121551514,
      "learning_rate": 0.00125,
      "loss": 4.0717,
      "step": 79
    },
    {
      "epoch": 0.008438818565400843,
      "grad_norm": 0.6442673802375793,
      "learning_rate": 0.0012658227848101266,
      "loss": 4.0578,
      "step": 80
    },
    {
      "epoch": 0.008544303797468355,
      "grad_norm": 0.6536983847618103,
      "learning_rate": 0.0012816455696202532,
      "loss": 4.032,
      "step": 81
    },
    {
      "epoch": 0.008649789029535865,
      "grad_norm": 0.5698292255401611,
      "learning_rate": 0.0012974683544303798,
      "loss": 4.0215,
      "step": 82
    },
    {
      "epoch": 0.008755274261603375,
      "grad_norm": 0.5748029351234436,
      "learning_rate": 0.0013132911392405063,
      "loss": 3.9926,
      "step": 83
    },
    {
      "epoch": 0.008860759493670886,
      "grad_norm": 0.5127552151679993,
      "learning_rate": 0.001329113924050633,
      "loss": 3.9656,
      "step": 84
    },
    {
      "epoch": 0.008966244725738396,
      "grad_norm": 0.5425557494163513,
      "learning_rate": 0.0013449367088607595,
      "loss": 3.9589,
      "step": 85
    },
    {
      "epoch": 0.009071729957805906,
      "grad_norm": 0.38073161244392395,
      "learning_rate": 0.001360759493670886,
      "loss": 3.9674,
      "step": 86
    },
    {
      "epoch": 0.009177215189873418,
      "grad_norm": 0.49425122141838074,
      "learning_rate": 0.0013765822784810127,
      "loss": 3.9609,
      "step": 87
    },
    {
      "epoch": 0.009282700421940928,
      "grad_norm": 0.5223596096038818,
      "learning_rate": 0.0013924050632911392,
      "loss": 3.8881,
      "step": 88
    },
    {
      "epoch": 0.009388185654008438,
      "grad_norm": 0.6041417121887207,
      "learning_rate": 0.001408227848101266,
      "loss": 3.903,
      "step": 89
    },
    {
      "epoch": 0.00949367088607595,
      "grad_norm": 0.6410835385322571,
      "learning_rate": 0.0014240506329113926,
      "loss": 3.9053,
      "step": 90
    },
    {
      "epoch": 0.00959915611814346,
      "grad_norm": 0.6636977791786194,
      "learning_rate": 0.0014398734177215192,
      "loss": 3.8916,
      "step": 91
    },
    {
      "epoch": 0.009704641350210971,
      "grad_norm": 0.6700908541679382,
      "learning_rate": 0.0014556962025316458,
      "loss": 3.874,
      "step": 92
    },
    {
      "epoch": 0.009810126582278481,
      "grad_norm": 0.4951082766056061,
      "learning_rate": 0.0014715189873417724,
      "loss": 3.8548,
      "step": 93
    },
    {
      "epoch": 0.009915611814345991,
      "grad_norm": 0.5571147203445435,
      "learning_rate": 0.001487341772151899,
      "loss": 3.8841,
      "step": 94
    },
    {
      "epoch": 0.010021097046413503,
      "grad_norm": 0.7228207588195801,
      "learning_rate": 0.0015,
      "loss": 3.8444,
      "step": 95
    },
    {
      "epoch": 0.010126582278481013,
      "grad_norm": 0.5196840763092041,
      "learning_rate": 0.0015,
      "loss": 3.8324,
      "step": 96
    },
    {
      "epoch": 0.010232067510548523,
      "grad_norm": 0.6950877904891968,
      "learning_rate": 0.0015,
      "loss": 3.8145,
      "step": 97
    },
    {
      "epoch": 0.010337552742616034,
      "grad_norm": 0.7402157783508301,
      "learning_rate": 0.0015,
      "loss": 3.7988,
      "step": 98
    },
    {
      "epoch": 0.010443037974683544,
      "grad_norm": 0.9650890231132507,
      "learning_rate": 0.0015,
      "loss": 3.8394,
      "step": 99
    },
    {
      "epoch": 0.010548523206751054,
      "grad_norm": 0.6854791641235352,
      "learning_rate": 0.0015,
      "loss": 3.783,
      "step": 100
    },
    {
      "epoch": 0.010654008438818566,
      "grad_norm": 0.5937216877937317,
      "learning_rate": 0.0015,
      "loss": 3.7849,
      "step": 101
    },
    {
      "epoch": 0.010759493670886076,
      "grad_norm": 0.7037617564201355,
      "learning_rate": 0.0015,
      "loss": 3.7504,
      "step": 102
    },
    {
      "epoch": 0.010864978902953586,
      "grad_norm": 0.6718727350234985,
      "learning_rate": 0.0015,
      "loss": 3.7401,
      "step": 103
    },
    {
      "epoch": 0.010970464135021098,
      "grad_norm": 0.6742255091667175,
      "learning_rate": 0.0015,
      "loss": 3.7252,
      "step": 104
    },
    {
      "epoch": 0.011075949367088608,
      "grad_norm": 0.8535324931144714,
      "learning_rate": 0.0015,
      "loss": 3.7297,
      "step": 105
    },
    {
      "epoch": 0.011181434599156118,
      "grad_norm": 0.6025777459144592,
      "learning_rate": 0.0015,
      "loss": 3.7178,
      "step": 106
    },
    {
      "epoch": 0.01128691983122363,
      "grad_norm": 0.5835834741592407,
      "learning_rate": 0.0015,
      "loss": 3.723,
      "step": 107
    },
    {
      "epoch": 0.01139240506329114,
      "grad_norm": 0.6241533756256104,
      "learning_rate": 0.0015,
      "loss": 3.6998,
      "step": 108
    },
    {
      "epoch": 0.01149789029535865,
      "grad_norm": 0.5421590209007263,
      "learning_rate": 0.0015,
      "loss": 3.7155,
      "step": 109
    },
    {
      "epoch": 0.011603375527426161,
      "grad_norm": 0.548008918762207,
      "learning_rate": 0.0015,
      "loss": 3.6711,
      "step": 110
    },
    {
      "epoch": 0.01170886075949367,
      "grad_norm": 0.5887262225151062,
      "learning_rate": 0.0015,
      "loss": 3.6718,
      "step": 111
    },
    {
      "epoch": 0.01181434599156118,
      "grad_norm": 0.5165668725967407,
      "learning_rate": 0.0015,
      "loss": 3.6759,
      "step": 112
    },
    {
      "epoch": 0.011919831223628692,
      "grad_norm": 0.6030138731002808,
      "learning_rate": 0.0015,
      "loss": 3.6532,
      "step": 113
    },
    {
      "epoch": 0.012025316455696202,
      "grad_norm": 0.48852312564849854,
      "learning_rate": 0.0015,
      "loss": 3.6623,
      "step": 114
    },
    {
      "epoch": 0.012130801687763712,
      "grad_norm": 0.49858132004737854,
      "learning_rate": 0.0015,
      "loss": 3.6547,
      "step": 115
    },
    {
      "epoch": 0.012236286919831224,
      "grad_norm": 0.4732319116592407,
      "learning_rate": 0.0015,
      "loss": 3.6331,
      "step": 116
    },
    {
      "epoch": 0.012341772151898734,
      "grad_norm": 0.7504942417144775,
      "learning_rate": 0.0015,
      "loss": 3.6535,
      "step": 117
    },
    {
      "epoch": 0.012447257383966244,
      "grad_norm": 0.8717824816703796,
      "learning_rate": 0.0015,
      "loss": 3.6396,
      "step": 118
    },
    {
      "epoch": 0.012552742616033756,
      "grad_norm": 1.1342066526412964,
      "learning_rate": 0.0015,
      "loss": 3.6024,
      "step": 119
    },
    {
      "epoch": 0.012658227848101266,
      "grad_norm": 0.7692056894302368,
      "learning_rate": 0.0015,
      "loss": 3.6184,
      "step": 120
    },
    {
      "epoch": 0.012763713080168776,
      "grad_norm": 0.9235360026359558,
      "learning_rate": 0.0015,
      "loss": 3.6258,
      "step": 121
    },
    {
      "epoch": 0.012869198312236287,
      "grad_norm": 0.5077849626541138,
      "learning_rate": 0.0015,
      "loss": 3.5949,
      "step": 122
    },
    {
      "epoch": 0.012974683544303797,
      "grad_norm": 0.6044186353683472,
      "learning_rate": 0.0015,
      "loss": 3.5698,
      "step": 123
    },
    {
      "epoch": 0.013080168776371307,
      "grad_norm": 0.47526460886001587,
      "learning_rate": 0.0015,
      "loss": 3.555,
      "step": 124
    },
    {
      "epoch": 0.013185654008438819,
      "grad_norm": 0.46524354815483093,
      "learning_rate": 0.0015,
      "loss": 3.5769,
      "step": 125
    },
    {
      "epoch": 0.013291139240506329,
      "grad_norm": 0.5536606907844543,
      "learning_rate": 0.0015,
      "loss": 3.5524,
      "step": 126
    },
    {
      "epoch": 0.01339662447257384,
      "grad_norm": 0.4977322518825531,
      "learning_rate": 0.0015,
      "loss": 3.5644,
      "step": 127
    },
    {
      "epoch": 0.01350210970464135,
      "grad_norm": 0.481624960899353,
      "learning_rate": 0.0015,
      "loss": 3.5316,
      "step": 128
    },
    {
      "epoch": 0.01360759493670886,
      "grad_norm": 0.4720386564731598,
      "learning_rate": 0.0015,
      "loss": 3.5218,
      "step": 129
    },
    {
      "epoch": 0.013713080168776372,
      "grad_norm": 0.5213106870651245,
      "learning_rate": 0.0015,
      "loss": 3.542,
      "step": 130
    },
    {
      "epoch": 0.013818565400843882,
      "grad_norm": 0.46011459827423096,
      "learning_rate": 0.0015,
      "loss": 3.4923,
      "step": 131
    },
    {
      "epoch": 0.013924050632911392,
      "grad_norm": 0.6758908629417419,
      "learning_rate": 0.0015,
      "loss": 3.5476,
      "step": 132
    },
    {
      "epoch": 0.014029535864978904,
      "grad_norm": 0.977131187915802,
      "learning_rate": 0.0015,
      "loss": 3.5444,
      "step": 133
    },
    {
      "epoch": 0.014135021097046414,
      "grad_norm": 1.2688044309616089,
      "learning_rate": 0.0015,
      "loss": 3.5134,
      "step": 134
    },
    {
      "epoch": 0.014240506329113924,
      "grad_norm": 0.6184185743331909,
      "learning_rate": 0.0015,
      "loss": 3.504,
      "step": 135
    },
    {
      "epoch": 0.014345991561181435,
      "grad_norm": 0.8497546911239624,
      "learning_rate": 0.0015,
      "loss": 3.5002,
      "step": 136
    },
    {
      "epoch": 0.014451476793248945,
      "grad_norm": 0.9705550074577332,
      "learning_rate": 0.0015,
      "loss": 3.5111,
      "step": 137
    },
    {
      "epoch": 0.014556962025316455,
      "grad_norm": 0.7666255831718445,
      "learning_rate": 0.0015,
      "loss": 3.4586,
      "step": 138
    },
    {
      "epoch": 0.014662447257383967,
      "grad_norm": 0.6074670553207397,
      "learning_rate": 0.0015,
      "loss": 3.4511,
      "step": 139
    },
    {
      "epoch": 0.014767932489451477,
      "grad_norm": 0.5568854808807373,
      "learning_rate": 0.0015,
      "loss": 3.4506,
      "step": 140
    },
    {
      "epoch": 0.014873417721518987,
      "grad_norm": 0.5095645785331726,
      "learning_rate": 0.0015,
      "loss": 3.4491,
      "step": 141
    },
    {
      "epoch": 0.014978902953586498,
      "grad_norm": 0.507882297039032,
      "learning_rate": 0.0015,
      "loss": 3.446,
      "step": 142
    },
    {
      "epoch": 0.015084388185654008,
      "grad_norm": 0.5321149230003357,
      "learning_rate": 0.0015,
      "loss": 3.4366,
      "step": 143
    },
    {
      "epoch": 0.015189873417721518,
      "grad_norm": 0.5249614119529724,
      "learning_rate": 0.0015,
      "loss": 3.4379,
      "step": 144
    },
    {
      "epoch": 0.01529535864978903,
      "grad_norm": 0.4674147069454193,
      "learning_rate": 0.0015,
      "loss": 3.4449,
      "step": 145
    },
    {
      "epoch": 0.01540084388185654,
      "grad_norm": 0.47469788789749146,
      "learning_rate": 0.0015,
      "loss": 3.4026,
      "step": 146
    },
    {
      "epoch": 0.01550632911392405,
      "grad_norm": 0.5053249001502991,
      "learning_rate": 0.0015,
      "loss": 3.4355,
      "step": 147
    },
    {
      "epoch": 0.015611814345991562,
      "grad_norm": 0.408283531665802,
      "learning_rate": 0.0015,
      "loss": 3.3723,
      "step": 148
    },
    {
      "epoch": 0.015717299578059073,
      "grad_norm": 0.47718942165374756,
      "learning_rate": 0.0015,
      "loss": 3.3956,
      "step": 149
    },
    {
      "epoch": 0.015822784810126583,
      "grad_norm": 0.5963375568389893,
      "learning_rate": 0.0015,
      "loss": 3.3988,
      "step": 150
    },
    {
      "epoch": 0.015928270042194093,
      "grad_norm": 0.6277444362640381,
      "learning_rate": 0.0015,
      "loss": 3.4276,
      "step": 151
    },
    {
      "epoch": 0.016033755274261603,
      "grad_norm": 0.6306942701339722,
      "learning_rate": 0.0015,
      "loss": 3.3551,
      "step": 152
    },
    {
      "epoch": 0.016139240506329113,
      "grad_norm": 0.5683684945106506,
      "learning_rate": 0.0015,
      "loss": 3.4084,
      "step": 153
    },
    {
      "epoch": 0.016244725738396623,
      "grad_norm": 0.5669676065444946,
      "learning_rate": 0.0015,
      "loss": 3.352,
      "step": 154
    },
    {
      "epoch": 0.016350210970464137,
      "grad_norm": 0.45035621523857117,
      "learning_rate": 0.0015,
      "loss": 3.3635,
      "step": 155
    },
    {
      "epoch": 0.016455696202531647,
      "grad_norm": 0.48036280274391174,
      "learning_rate": 0.0015,
      "loss": 3.3482,
      "step": 156
    },
    {
      "epoch": 0.016561181434599156,
      "grad_norm": 0.5925886631011963,
      "learning_rate": 0.0015,
      "loss": 3.3647,
      "step": 157
    },
    {
      "epoch": 0.016666666666666666,
      "grad_norm": 0.5252505540847778,
      "learning_rate": 0.0015,
      "loss": 3.3313,
      "step": 158
    },
    {
      "epoch": 0.016772151898734176,
      "grad_norm": 0.4907812774181366,
      "learning_rate": 0.0015,
      "loss": 3.3498,
      "step": 159
    },
    {
      "epoch": 0.016877637130801686,
      "grad_norm": 0.5507779121398926,
      "learning_rate": 0.0015,
      "loss": 3.3435,
      "step": 160
    },
    {
      "epoch": 0.0169831223628692,
      "grad_norm": 0.5671168565750122,
      "learning_rate": 0.0015,
      "loss": 3.3011,
      "step": 161
    },
    {
      "epoch": 0.01708860759493671,
      "grad_norm": 0.560148298740387,
      "learning_rate": 0.0015,
      "loss": 3.3534,
      "step": 162
    },
    {
      "epoch": 0.01719409282700422,
      "grad_norm": 0.5426363945007324,
      "learning_rate": 0.0015,
      "loss": 3.3287,
      "step": 163
    },
    {
      "epoch": 0.01729957805907173,
      "grad_norm": 0.48038220405578613,
      "learning_rate": 0.0015,
      "loss": 3.3026,
      "step": 164
    },
    {
      "epoch": 0.01740506329113924,
      "grad_norm": 0.49991828203201294,
      "learning_rate": 0.0015,
      "loss": 3.323,
      "step": 165
    },
    {
      "epoch": 0.01751054852320675,
      "grad_norm": 0.542594313621521,
      "learning_rate": 0.0015,
      "loss": 3.3064,
      "step": 166
    },
    {
      "epoch": 0.017616033755274263,
      "grad_norm": 0.6581192016601562,
      "learning_rate": 0.0015,
      "loss": 3.2768,
      "step": 167
    },
    {
      "epoch": 0.017721518987341773,
      "grad_norm": 1.042526125907898,
      "learning_rate": 0.0015,
      "loss": 3.2859,
      "step": 168
    },
    {
      "epoch": 0.017827004219409283,
      "grad_norm": 1.2364501953125,
      "learning_rate": 0.0015,
      "loss": 3.2897,
      "step": 169
    },
    {
      "epoch": 0.017932489451476793,
      "grad_norm": 0.7119993567466736,
      "learning_rate": 0.0015,
      "loss": 3.2836,
      "step": 170
    },
    {
      "epoch": 0.018037974683544303,
      "grad_norm": 0.9350215792655945,
      "learning_rate": 0.0015,
      "loss": 3.2656,
      "step": 171
    },
    {
      "epoch": 0.018143459915611813,
      "grad_norm": 1.0866774320602417,
      "learning_rate": 0.0015,
      "loss": 3.3286,
      "step": 172
    },
    {
      "epoch": 0.018248945147679326,
      "grad_norm": 0.7057785987854004,
      "learning_rate": 0.0015,
      "loss": 3.2893,
      "step": 173
    },
    {
      "epoch": 0.018354430379746836,
      "grad_norm": 0.8040363192558289,
      "learning_rate": 0.0015,
      "loss": 3.2537,
      "step": 174
    },
    {
      "epoch": 0.018459915611814346,
      "grad_norm": 0.682712733745575,
      "learning_rate": 0.0015,
      "loss": 3.2988,
      "step": 175
    },
    {
      "epoch": 0.018565400843881856,
      "grad_norm": 0.7391072511672974,
      "learning_rate": 0.0015,
      "loss": 3.276,
      "step": 176
    },
    {
      "epoch": 0.018670886075949366,
      "grad_norm": 0.4904000759124756,
      "learning_rate": 0.0015,
      "loss": 3.2685,
      "step": 177
    },
    {
      "epoch": 0.018776371308016876,
      "grad_norm": 0.6686579585075378,
      "learning_rate": 0.0015,
      "loss": 3.2668,
      "step": 178
    },
    {
      "epoch": 0.01888185654008439,
      "grad_norm": 0.7120933532714844,
      "learning_rate": 0.0015,
      "loss": 3.2756,
      "step": 179
    },
    {
      "epoch": 0.0189873417721519,
      "grad_norm": 0.682628870010376,
      "learning_rate": 0.0015,
      "loss": 3.2507,
      "step": 180
    },
    {
      "epoch": 0.01909282700421941,
      "grad_norm": 0.5308851599693298,
      "learning_rate": 0.0015,
      "loss": 3.2523,
      "step": 181
    },
    {
      "epoch": 0.01919831223628692,
      "grad_norm": 0.5541619658470154,
      "learning_rate": 0.0015,
      "loss": 3.2407,
      "step": 182
    },
    {
      "epoch": 0.01930379746835443,
      "grad_norm": 0.5862583518028259,
      "learning_rate": 0.0015,
      "loss": 3.2176,
      "step": 183
    },
    {
      "epoch": 0.019409282700421943,
      "grad_norm": 0.5416473746299744,
      "learning_rate": 0.0015,
      "loss": 3.2443,
      "step": 184
    },
    {
      "epoch": 0.019514767932489453,
      "grad_norm": 0.6392229795455933,
      "learning_rate": 0.0015,
      "loss": 3.1996,
      "step": 185
    },
    {
      "epoch": 0.019620253164556962,
      "grad_norm": 0.6571645736694336,
      "learning_rate": 0.0015,
      "loss": 3.1835,
      "step": 186
    },
    {
      "epoch": 0.019725738396624472,
      "grad_norm": 0.4875413775444031,
      "learning_rate": 0.0015,
      "loss": 3.1681,
      "step": 187
    },
    {
      "epoch": 0.019831223628691982,
      "grad_norm": 0.611209511756897,
      "learning_rate": 0.0015,
      "loss": 3.1712,
      "step": 188
    },
    {
      "epoch": 0.019936708860759492,
      "grad_norm": 0.5870056748390198,
      "learning_rate": 0.0015,
      "loss": 3.1819,
      "step": 189
    },
    {
      "epoch": 0.020042194092827006,
      "grad_norm": 0.5035120248794556,
      "learning_rate": 0.0015,
      "loss": 3.2077,
      "step": 190
    },
    {
      "epoch": 0.020147679324894516,
      "grad_norm": 0.543683648109436,
      "learning_rate": 0.0015,
      "loss": 3.161,
      "step": 191
    },
    {
      "epoch": 0.020253164556962026,
      "grad_norm": 0.5996739864349365,
      "learning_rate": 0.0015,
      "loss": 3.203,
      "step": 192
    },
    {
      "epoch": 0.020358649789029536,
      "grad_norm": 0.5495070219039917,
      "learning_rate": 0.0015,
      "loss": 3.135,
      "step": 193
    },
    {
      "epoch": 0.020464135021097046,
      "grad_norm": 0.5591219663619995,
      "learning_rate": 0.0015,
      "loss": 3.1768,
      "step": 194
    },
    {
      "epoch": 0.020569620253164556,
      "grad_norm": 0.5875229239463806,
      "learning_rate": 0.0015,
      "loss": 3.1571,
      "step": 195
    },
    {
      "epoch": 0.02067510548523207,
      "grad_norm": 0.5609016418457031,
      "learning_rate": 0.0015,
      "loss": 3.1541,
      "step": 196
    },
    {
      "epoch": 0.02078059071729958,
      "grad_norm": 0.542759120464325,
      "learning_rate": 0.0015,
      "loss": 3.1458,
      "step": 197
    },
    {
      "epoch": 0.02088607594936709,
      "grad_norm": 0.5104601383209229,
      "learning_rate": 0.0015,
      "loss": 3.1704,
      "step": 198
    },
    {
      "epoch": 0.0209915611814346,
      "grad_norm": 0.738538384437561,
      "learning_rate": 0.0015,
      "loss": 3.1601,
      "step": 199
    },
    {
      "epoch": 0.02109704641350211,
      "grad_norm": 0.9710153341293335,
      "learning_rate": 0.0015,
      "loss": 3.1217,
      "step": 200
    },
    {
      "epoch": 0.02120253164556962,
      "grad_norm": 1.136983871459961,
      "learning_rate": 0.0015,
      "loss": 3.1744,
      "step": 201
    },
    {
      "epoch": 0.021308016877637132,
      "grad_norm": 0.7257745265960693,
      "learning_rate": 0.0015,
      "loss": 3.1412,
      "step": 202
    },
    {
      "epoch": 0.021413502109704642,
      "grad_norm": 0.836239218711853,
      "learning_rate": 0.0015,
      "loss": 3.1502,
      "step": 203
    },
    {
      "epoch": 0.021518987341772152,
      "grad_norm": 0.7749267220497131,
      "learning_rate": 0.0015,
      "loss": 3.1358,
      "step": 204
    },
    {
      "epoch": 0.021624472573839662,
      "grad_norm": 0.6772562861442566,
      "learning_rate": 0.0015,
      "loss": 3.1092,
      "step": 205
    },
    {
      "epoch": 0.021729957805907172,
      "grad_norm": 0.8574336171150208,
      "learning_rate": 0.0015,
      "loss": 3.0991,
      "step": 206
    },
    {
      "epoch": 0.021835443037974682,
      "grad_norm": 0.7635952234268188,
      "learning_rate": 0.0015,
      "loss": 3.1208,
      "step": 207
    },
    {
      "epoch": 0.021940928270042195,
      "grad_norm": 0.8011783361434937,
      "learning_rate": 0.0015,
      "loss": 3.0982,
      "step": 208
    },
    {
      "epoch": 0.022046413502109705,
      "grad_norm": 0.7332499027252197,
      "learning_rate": 0.0015,
      "loss": 3.0964,
      "step": 209
    },
    {
      "epoch": 0.022151898734177215,
      "grad_norm": 0.6593369841575623,
      "learning_rate": 0.0015,
      "loss": 3.1455,
      "step": 210
    },
    {
      "epoch": 0.022257383966244725,
      "grad_norm": 0.5410482883453369,
      "learning_rate": 0.0015,
      "loss": 3.1029,
      "step": 211
    },
    {
      "epoch": 0.022362869198312235,
      "grad_norm": 0.6211299300193787,
      "learning_rate": 0.0015,
      "loss": 3.0993,
      "step": 212
    },
    {
      "epoch": 0.022468354430379745,
      "grad_norm": 0.6471499800682068,
      "learning_rate": 0.0015,
      "loss": 3.1209,
      "step": 213
    },
    {
      "epoch": 0.02257383966244726,
      "grad_norm": 0.7534574270248413,
      "learning_rate": 0.0015,
      "loss": 3.1005,
      "step": 214
    },
    {
      "epoch": 0.02267932489451477,
      "grad_norm": 0.6201465129852295,
      "learning_rate": 0.0015,
      "loss": 3.06,
      "step": 215
    },
    {
      "epoch": 0.02278481012658228,
      "grad_norm": 0.5705814361572266,
      "learning_rate": 0.0015,
      "loss": 3.0798,
      "step": 216
    },
    {
      "epoch": 0.02289029535864979,
      "grad_norm": 0.5413900017738342,
      "learning_rate": 0.0015,
      "loss": 3.0694,
      "step": 217
    },
    {
      "epoch": 0.0229957805907173,
      "grad_norm": 0.6111287474632263,
      "learning_rate": 0.0015,
      "loss": 3.0953,
      "step": 218
    },
    {
      "epoch": 0.023101265822784812,
      "grad_norm": 0.6253988742828369,
      "learning_rate": 0.0015,
      "loss": 3.0854,
      "step": 219
    },
    {
      "epoch": 0.023206751054852322,
      "grad_norm": 0.6183094382286072,
      "learning_rate": 0.0015,
      "loss": 3.0453,
      "step": 220
    },
    {
      "epoch": 0.02331223628691983,
      "grad_norm": 0.5453416109085083,
      "learning_rate": 0.0015,
      "loss": 3.0475,
      "step": 221
    },
    {
      "epoch": 0.02341772151898734,
      "grad_norm": 0.5267513990402222,
      "learning_rate": 0.0015,
      "loss": 3.0213,
      "step": 222
    },
    {
      "epoch": 0.02352320675105485,
      "grad_norm": 0.5483818054199219,
      "learning_rate": 0.0015,
      "loss": 3.0388,
      "step": 223
    },
    {
      "epoch": 0.02362869198312236,
      "grad_norm": 0.5734818577766418,
      "learning_rate": 0.0015,
      "loss": 3.0223,
      "step": 224
    },
    {
      "epoch": 0.023734177215189875,
      "grad_norm": 0.6358630657196045,
      "learning_rate": 0.0015,
      "loss": 3.0151,
      "step": 225
    },
    {
      "epoch": 0.023839662447257385,
      "grad_norm": 0.6588574051856995,
      "learning_rate": 0.0015,
      "loss": 3.059,
      "step": 226
    },
    {
      "epoch": 0.023945147679324895,
      "grad_norm": 0.5852987170219421,
      "learning_rate": 0.0015,
      "loss": 3.0166,
      "step": 227
    },
    {
      "epoch": 0.024050632911392405,
      "grad_norm": 0.5072793364524841,
      "learning_rate": 0.0015,
      "loss": 3.0079,
      "step": 228
    },
    {
      "epoch": 0.024156118143459915,
      "grad_norm": 0.5593225955963135,
      "learning_rate": 0.0015,
      "loss": 2.996,
      "step": 229
    },
    {
      "epoch": 0.024261603375527425,
      "grad_norm": 0.7914431095123291,
      "learning_rate": 0.0015,
      "loss": 3.0295,
      "step": 230
    },
    {
      "epoch": 0.024367088607594938,
      "grad_norm": 0.9069041013717651,
      "learning_rate": 0.0015,
      "loss": 3.0118,
      "step": 231
    },
    {
      "epoch": 0.024472573839662448,
      "grad_norm": 1.1159306764602661,
      "learning_rate": 0.0015,
      "loss": 3.0398,
      "step": 232
    },
    {
      "epoch": 0.024578059071729958,
      "grad_norm": 0.8557447791099548,
      "learning_rate": 0.0015,
      "loss": 2.982,
      "step": 233
    },
    {
      "epoch": 0.024683544303797468,
      "grad_norm": 0.6739777326583862,
      "learning_rate": 0.0015,
      "loss": 3.0187,
      "step": 234
    },
    {
      "epoch": 0.024789029535864978,
      "grad_norm": 0.5493777990341187,
      "learning_rate": 0.0015,
      "loss": 2.9504,
      "step": 235
    },
    {
      "epoch": 0.024894514767932488,
      "grad_norm": 0.6609274744987488,
      "learning_rate": 0.0015,
      "loss": 2.9803,
      "step": 236
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.8530800342559814,
      "learning_rate": 0.0015,
      "loss": 2.9653,
      "step": 237
    },
    {
      "epoch": 0.02510548523206751,
      "grad_norm": 0.9023459553718567,
      "learning_rate": 0.0015,
      "loss": 2.9668,
      "step": 238
    },
    {
      "epoch": 0.02521097046413502,
      "grad_norm": 0.787721574306488,
      "learning_rate": 0.0015,
      "loss": 2.984,
      "step": 239
    },
    {
      "epoch": 0.02531645569620253,
      "grad_norm": 0.7218961119651794,
      "learning_rate": 0.0015,
      "loss": 3.0032,
      "step": 240
    },
    {
      "epoch": 0.02542194092827004,
      "grad_norm": 0.7726406455039978,
      "learning_rate": 0.0015,
      "loss": 2.9452,
      "step": 241
    },
    {
      "epoch": 0.02552742616033755,
      "grad_norm": 0.8458675146102905,
      "learning_rate": 0.0015,
      "loss": 2.9789,
      "step": 242
    },
    {
      "epoch": 0.025632911392405065,
      "grad_norm": 0.6562297344207764,
      "learning_rate": 0.0015,
      "loss": 2.9888,
      "step": 243
    },
    {
      "epoch": 0.025738396624472575,
      "grad_norm": 0.5418792366981506,
      "learning_rate": 0.0015,
      "loss": 2.9712,
      "step": 244
    },
    {
      "epoch": 0.025843881856540084,
      "grad_norm": 0.7549543976783752,
      "learning_rate": 0.0015,
      "loss": 2.9604,
      "step": 245
    },
    {
      "epoch": 0.025949367088607594,
      "grad_norm": 0.6601927280426025,
      "learning_rate": 0.0015,
      "loss": 2.9502,
      "step": 246
    },
    {
      "epoch": 0.026054852320675104,
      "grad_norm": 0.5595108270645142,
      "learning_rate": 0.0015,
      "loss": 2.9067,
      "step": 247
    },
    {
      "epoch": 0.026160337552742614,
      "grad_norm": 0.7016994953155518,
      "learning_rate": 0.0015,
      "loss": 2.9668,
      "step": 248
    },
    {
      "epoch": 0.026265822784810128,
      "grad_norm": 0.7691634893417358,
      "learning_rate": 0.0015,
      "loss": 2.9278,
      "step": 249
    },
    {
      "epoch": 0.026371308016877638,
      "grad_norm": 0.6277539134025574,
      "learning_rate": 0.0015,
      "loss": 2.9002,
      "step": 250
    },
    {
      "epoch": 0.026476793248945148,
      "grad_norm": 0.5682135820388794,
      "learning_rate": 0.0015,
      "loss": 2.946,
      "step": 251
    },
    {
      "epoch": 0.026582278481012658,
      "grad_norm": 0.6085950136184692,
      "learning_rate": 0.0015,
      "loss": 2.9167,
      "step": 252
    },
    {
      "epoch": 0.026687763713080168,
      "grad_norm": 0.7385963797569275,
      "learning_rate": 0.0015,
      "loss": 2.9158,
      "step": 253
    },
    {
      "epoch": 0.02679324894514768,
      "grad_norm": 0.8353692293167114,
      "learning_rate": 0.0015,
      "loss": 2.9078,
      "step": 254
    },
    {
      "epoch": 0.02689873417721519,
      "grad_norm": 0.6944759488105774,
      "learning_rate": 0.0015,
      "loss": 2.8956,
      "step": 255
    },
    {
      "epoch": 0.0270042194092827,
      "grad_norm": 0.6617680191993713,
      "learning_rate": 0.0015,
      "loss": 2.8609,
      "step": 256
    },
    {
      "epoch": 0.02710970464135021,
      "grad_norm": 0.6998510360717773,
      "learning_rate": 0.0015,
      "loss": 2.9136,
      "step": 257
    },
    {
      "epoch": 0.02721518987341772,
      "grad_norm": 0.7599093914031982,
      "learning_rate": 0.0015,
      "loss": 2.9203,
      "step": 258
    },
    {
      "epoch": 0.02732067510548523,
      "grad_norm": 0.594812273979187,
      "learning_rate": 0.0015,
      "loss": 2.9195,
      "step": 259
    },
    {
      "epoch": 0.027426160337552744,
      "grad_norm": 0.6033343076705933,
      "learning_rate": 0.0015,
      "loss": 2.859,
      "step": 260
    },
    {
      "epoch": 0.027531645569620254,
      "grad_norm": 0.7880474328994751,
      "learning_rate": 0.0015,
      "loss": 2.8976,
      "step": 261
    },
    {
      "epoch": 0.027637130801687764,
      "grad_norm": 0.8647599220275879,
      "learning_rate": 0.0015,
      "loss": 2.8831,
      "step": 262
    },
    {
      "epoch": 0.027742616033755274,
      "grad_norm": 0.7458866238594055,
      "learning_rate": 0.0015,
      "loss": 2.881,
      "step": 263
    },
    {
      "epoch": 0.027848101265822784,
      "grad_norm": 0.6117191314697266,
      "learning_rate": 0.0015,
      "loss": 2.8818,
      "step": 264
    },
    {
      "epoch": 0.027953586497890294,
      "grad_norm": 0.5987802147865295,
      "learning_rate": 0.0015,
      "loss": 2.879,
      "step": 265
    },
    {
      "epoch": 0.028059071729957807,
      "grad_norm": 0.7147089838981628,
      "learning_rate": 0.0015,
      "loss": 2.8719,
      "step": 266
    },
    {
      "epoch": 0.028164556962025317,
      "grad_norm": 0.7244475483894348,
      "learning_rate": 0.0015,
      "loss": 2.8613,
      "step": 267
    },
    {
      "epoch": 0.028270042194092827,
      "grad_norm": 0.6152216196060181,
      "learning_rate": 0.0015,
      "loss": 2.8304,
      "step": 268
    },
    {
      "epoch": 0.028375527426160337,
      "grad_norm": 0.9417058229446411,
      "learning_rate": 0.0015,
      "loss": 2.8717,
      "step": 269
    },
    {
      "epoch": 0.028481012658227847,
      "grad_norm": 1.171797513961792,
      "learning_rate": 0.0015,
      "loss": 2.8844,
      "step": 270
    },
    {
      "epoch": 0.028586497890295357,
      "grad_norm": 0.8627143502235413,
      "learning_rate": 0.0015,
      "loss": 2.8732,
      "step": 271
    },
    {
      "epoch": 0.02869198312236287,
      "grad_norm": 0.7298221588134766,
      "learning_rate": 0.0015,
      "loss": 2.8455,
      "step": 272
    },
    {
      "epoch": 0.02879746835443038,
      "grad_norm": 0.8978715538978577,
      "learning_rate": 0.0015,
      "loss": 2.8394,
      "step": 273
    },
    {
      "epoch": 0.02890295358649789,
      "grad_norm": 0.8794608116149902,
      "learning_rate": 0.0015,
      "loss": 2.844,
      "step": 274
    },
    {
      "epoch": 0.0290084388185654,
      "grad_norm": 0.6987318396568298,
      "learning_rate": 0.0015,
      "loss": 2.8633,
      "step": 275
    },
    {
      "epoch": 0.02911392405063291,
      "grad_norm": 0.5574908256530762,
      "learning_rate": 0.0015,
      "loss": 2.8063,
      "step": 276
    },
    {
      "epoch": 0.02921940928270042,
      "grad_norm": 0.6566581130027771,
      "learning_rate": 0.0015,
      "loss": 2.8406,
      "step": 277
    },
    {
      "epoch": 0.029324894514767934,
      "grad_norm": 0.7081720232963562,
      "learning_rate": 0.0015,
      "loss": 2.7729,
      "step": 278
    },
    {
      "epoch": 0.029430379746835444,
      "grad_norm": 0.6616462469100952,
      "learning_rate": 0.0015,
      "loss": 2.8339,
      "step": 279
    },
    {
      "epoch": 0.029535864978902954,
      "grad_norm": 0.7024974822998047,
      "learning_rate": 0.0015,
      "loss": 2.8073,
      "step": 280
    },
    {
      "epoch": 0.029641350210970464,
      "grad_norm": 0.6736257672309875,
      "learning_rate": 0.0015,
      "loss": 2.7869,
      "step": 281
    },
    {
      "epoch": 0.029746835443037974,
      "grad_norm": 0.5996975302696228,
      "learning_rate": 0.0015,
      "loss": 2.854,
      "step": 282
    },
    {
      "epoch": 0.029852320675105484,
      "grad_norm": 0.526852548122406,
      "learning_rate": 0.0015,
      "loss": 2.7721,
      "step": 283
    },
    {
      "epoch": 0.029957805907172997,
      "grad_norm": 0.5094711184501648,
      "learning_rate": 0.0015,
      "loss": 2.7869,
      "step": 284
    },
    {
      "epoch": 0.030063291139240507,
      "grad_norm": 0.5385901927947998,
      "learning_rate": 0.0015,
      "loss": 2.8363,
      "step": 285
    },
    {
      "epoch": 0.030168776371308017,
      "grad_norm": 0.5184576511383057,
      "learning_rate": 0.0015,
      "loss": 2.8153,
      "step": 286
    },
    {
      "epoch": 0.030274261603375527,
      "grad_norm": 0.5651463866233826,
      "learning_rate": 0.0015,
      "loss": 2.812,
      "step": 287
    },
    {
      "epoch": 0.030379746835443037,
      "grad_norm": 0.5776814222335815,
      "learning_rate": 0.0015,
      "loss": 2.7961,
      "step": 288
    },
    {
      "epoch": 0.03048523206751055,
      "grad_norm": 0.6253670454025269,
      "learning_rate": 0.0015,
      "loss": 2.793,
      "step": 289
    },
    {
      "epoch": 0.03059071729957806,
      "grad_norm": 0.5828775763511658,
      "learning_rate": 0.0015,
      "loss": 2.7765,
      "step": 290
    },
    {
      "epoch": 0.03069620253164557,
      "grad_norm": 0.6641796231269836,
      "learning_rate": 0.0015,
      "loss": 2.7911,
      "step": 291
    },
    {
      "epoch": 0.03080168776371308,
      "grad_norm": 0.8532554507255554,
      "learning_rate": 0.0015,
      "loss": 2.8123,
      "step": 292
    },
    {
      "epoch": 0.03090717299578059,
      "grad_norm": 0.7868938446044922,
      "learning_rate": 0.0015,
      "loss": 2.788,
      "step": 293
    },
    {
      "epoch": 0.0310126582278481,
      "grad_norm": 0.7309849858283997,
      "learning_rate": 0.0015,
      "loss": 2.7535,
      "step": 294
    },
    {
      "epoch": 0.031118143459915613,
      "grad_norm": 0.6793031096458435,
      "learning_rate": 0.0015,
      "loss": 2.7548,
      "step": 295
    },
    {
      "epoch": 0.031223628691983123,
      "grad_norm": 0.6210504770278931,
      "learning_rate": 0.0015,
      "loss": 2.7976,
      "step": 296
    },
    {
      "epoch": 0.03132911392405063,
      "grad_norm": 0.5565879940986633,
      "learning_rate": 0.0015,
      "loss": 2.8034,
      "step": 297
    },
    {
      "epoch": 0.03143459915611815,
      "grad_norm": 0.5763741135597229,
      "learning_rate": 0.0015,
      "loss": 2.7692,
      "step": 298
    },
    {
      "epoch": 0.03154008438818565,
      "grad_norm": 0.5793166160583496,
      "learning_rate": 0.0015,
      "loss": 2.7381,
      "step": 299
    },
    {
      "epoch": 0.03164556962025317,
      "grad_norm": 0.5885812640190125,
      "learning_rate": 0.0015,
      "loss": 2.7534,
      "step": 300
    },
    {
      "epoch": 0.03175105485232067,
      "grad_norm": 0.5877847075462341,
      "learning_rate": 0.0015,
      "loss": 2.7264,
      "step": 301
    },
    {
      "epoch": 0.03185654008438819,
      "grad_norm": 0.5303623676300049,
      "learning_rate": 0.0015,
      "loss": 2.7349,
      "step": 302
    },
    {
      "epoch": 0.03196202531645569,
      "grad_norm": 0.5827655792236328,
      "learning_rate": 0.0015,
      "loss": 2.74,
      "step": 303
    },
    {
      "epoch": 0.032067510548523206,
      "grad_norm": 0.6397809982299805,
      "learning_rate": 0.0015,
      "loss": 2.7474,
      "step": 304
    },
    {
      "epoch": 0.03217299578059072,
      "grad_norm": 0.6652061343193054,
      "learning_rate": 0.0015,
      "loss": 2.7568,
      "step": 305
    },
    {
      "epoch": 0.032278481012658226,
      "grad_norm": 0.5894809365272522,
      "learning_rate": 0.0015,
      "loss": 2.7353,
      "step": 306
    },
    {
      "epoch": 0.03238396624472574,
      "grad_norm": 0.7586065530776978,
      "learning_rate": 0.0015,
      "loss": 2.7503,
      "step": 307
    },
    {
      "epoch": 0.032489451476793246,
      "grad_norm": 0.7334324717521667,
      "learning_rate": 0.0015,
      "loss": 2.7292,
      "step": 308
    },
    {
      "epoch": 0.03259493670886076,
      "grad_norm": 0.8651225566864014,
      "learning_rate": 0.0015,
      "loss": 2.7511,
      "step": 309
    },
    {
      "epoch": 0.03270042194092827,
      "grad_norm": 1.0487282276153564,
      "learning_rate": 0.0015,
      "loss": 2.7477,
      "step": 310
    },
    {
      "epoch": 0.03280590717299578,
      "grad_norm": 1.106597900390625,
      "learning_rate": 0.0015,
      "loss": 2.7262,
      "step": 311
    },
    {
      "epoch": 0.03291139240506329,
      "grad_norm": 0.7443594336509705,
      "learning_rate": 0.0015,
      "loss": 2.7328,
      "step": 312
    },
    {
      "epoch": 0.0330168776371308,
      "grad_norm": 0.5400886535644531,
      "learning_rate": 0.0015,
      "loss": 2.7009,
      "step": 313
    },
    {
      "epoch": 0.03312236286919831,
      "grad_norm": 0.5155996084213257,
      "learning_rate": 0.0015,
      "loss": 2.6974,
      "step": 314
    },
    {
      "epoch": 0.03322784810126582,
      "grad_norm": 0.6222251653671265,
      "learning_rate": 0.0015,
      "loss": 2.7196,
      "step": 315
    },
    {
      "epoch": 0.03333333333333333,
      "grad_norm": 0.6961142420768738,
      "learning_rate": 0.0015,
      "loss": 2.6971,
      "step": 316
    },
    {
      "epoch": 0.033438818565400846,
      "grad_norm": 0.6365319490432739,
      "learning_rate": 0.0015,
      "loss": 2.7201,
      "step": 317
    },
    {
      "epoch": 0.03354430379746835,
      "grad_norm": 0.5702477693557739,
      "learning_rate": 0.0015,
      "loss": 2.7052,
      "step": 318
    },
    {
      "epoch": 0.033649789029535866,
      "grad_norm": 0.6683318018913269,
      "learning_rate": 0.0015,
      "loss": 2.7026,
      "step": 319
    },
    {
      "epoch": 0.03375527426160337,
      "grad_norm": 0.7750487923622131,
      "learning_rate": 0.0015,
      "loss": 2.7222,
      "step": 320
    },
    {
      "epoch": 0.033860759493670886,
      "grad_norm": 0.8867079019546509,
      "learning_rate": 0.0015,
      "loss": 2.7187,
      "step": 321
    },
    {
      "epoch": 0.0339662447257384,
      "grad_norm": 0.9916465878486633,
      "learning_rate": 0.0015,
      "loss": 2.7188,
      "step": 322
    },
    {
      "epoch": 0.034071729957805906,
      "grad_norm": 0.9273934960365295,
      "learning_rate": 0.0015,
      "loss": 2.7302,
      "step": 323
    },
    {
      "epoch": 0.03417721518987342,
      "grad_norm": 0.6221110224723816,
      "learning_rate": 0.0015,
      "loss": 2.7487,
      "step": 324
    },
    {
      "epoch": 0.034282700421940926,
      "grad_norm": 0.6257839798927307,
      "learning_rate": 0.0015,
      "loss": 2.6852,
      "step": 325
    },
    {
      "epoch": 0.03438818565400844,
      "grad_norm": 0.9795200824737549,
      "learning_rate": 0.0015,
      "loss": 2.7023,
      "step": 326
    },
    {
      "epoch": 0.03449367088607595,
      "grad_norm": 0.9704818725585938,
      "learning_rate": 0.0015,
      "loss": 2.7275,
      "step": 327
    },
    {
      "epoch": 0.03459915611814346,
      "grad_norm": 0.6575543284416199,
      "learning_rate": 0.0015,
      "loss": 2.6751,
      "step": 328
    },
    {
      "epoch": 0.03470464135021097,
      "grad_norm": 0.5703964233398438,
      "learning_rate": 0.0015,
      "loss": 2.6609,
      "step": 329
    },
    {
      "epoch": 0.03481012658227848,
      "grad_norm": 0.7702048420906067,
      "learning_rate": 0.0015,
      "loss": 2.6545,
      "step": 330
    },
    {
      "epoch": 0.03491561181434599,
      "grad_norm": 0.6877147555351257,
      "learning_rate": 0.0015,
      "loss": 2.6597,
      "step": 331
    },
    {
      "epoch": 0.0350210970464135,
      "grad_norm": 0.666469156742096,
      "learning_rate": 0.0015,
      "loss": 2.6652,
      "step": 332
    },
    {
      "epoch": 0.03512658227848101,
      "grad_norm": 1.0524582862854004,
      "learning_rate": 0.0015,
      "loss": 2.7067,
      "step": 333
    },
    {
      "epoch": 0.035232067510548526,
      "grad_norm": 1.1469674110412598,
      "learning_rate": 0.0015,
      "loss": 2.6914,
      "step": 334
    },
    {
      "epoch": 0.03533755274261603,
      "grad_norm": 0.8497218489646912,
      "learning_rate": 0.0015,
      "loss": 2.6765,
      "step": 335
    },
    {
      "epoch": 0.035443037974683546,
      "grad_norm": 0.8292915225028992,
      "learning_rate": 0.0015,
      "loss": 2.6794,
      "step": 336
    },
    {
      "epoch": 0.03554852320675105,
      "grad_norm": 0.713080644607544,
      "learning_rate": 0.0015,
      "loss": 2.6623,
      "step": 337
    },
    {
      "epoch": 0.035654008438818566,
      "grad_norm": 0.6908189058303833,
      "learning_rate": 0.0015,
      "loss": 2.6872,
      "step": 338
    },
    {
      "epoch": 0.03575949367088608,
      "grad_norm": 0.7819079756736755,
      "learning_rate": 0.0015,
      "loss": 2.7063,
      "step": 339
    },
    {
      "epoch": 0.035864978902953586,
      "grad_norm": 0.6286224126815796,
      "learning_rate": 0.0015,
      "loss": 2.6636,
      "step": 340
    },
    {
      "epoch": 0.0359704641350211,
      "grad_norm": 0.6791321039199829,
      "learning_rate": 0.0015,
      "loss": 2.6431,
      "step": 341
    },
    {
      "epoch": 0.036075949367088606,
      "grad_norm": 0.6508675813674927,
      "learning_rate": 0.0015,
      "loss": 2.6151,
      "step": 342
    },
    {
      "epoch": 0.03618143459915612,
      "grad_norm": 0.6431472897529602,
      "learning_rate": 0.0015,
      "loss": 2.656,
      "step": 343
    },
    {
      "epoch": 0.036286919831223625,
      "grad_norm": 0.5608028769493103,
      "learning_rate": 0.0015,
      "loss": 2.6329,
      "step": 344
    },
    {
      "epoch": 0.03639240506329114,
      "grad_norm": 0.5455867052078247,
      "learning_rate": 0.0015,
      "loss": 2.6485,
      "step": 345
    },
    {
      "epoch": 0.03649789029535865,
      "grad_norm": 0.5543195009231567,
      "learning_rate": 0.0015,
      "loss": 2.6162,
      "step": 346
    },
    {
      "epoch": 0.03660337552742616,
      "grad_norm": 0.5591506958007812,
      "learning_rate": 0.0015,
      "loss": 2.6481,
      "step": 347
    },
    {
      "epoch": 0.03670886075949367,
      "grad_norm": 0.6832466125488281,
      "learning_rate": 0.0015,
      "loss": 2.64,
      "step": 348
    },
    {
      "epoch": 0.03681434599156118,
      "grad_norm": 0.7821243405342102,
      "learning_rate": 0.0015,
      "loss": 2.5994,
      "step": 349
    },
    {
      "epoch": 0.03691983122362869,
      "grad_norm": 0.9078012108802795,
      "learning_rate": 0.0015,
      "loss": 2.6168,
      "step": 350
    },
    {
      "epoch": 0.037025316455696206,
      "grad_norm": 1.0436315536499023,
      "learning_rate": 0.0015,
      "loss": 2.6131,
      "step": 351
    },
    {
      "epoch": 0.03713080168776371,
      "grad_norm": 0.7729336619377136,
      "learning_rate": 0.0015,
      "loss": 2.6366,
      "step": 352
    },
    {
      "epoch": 0.037236286919831225,
      "grad_norm": 0.6074826717376709,
      "learning_rate": 0.0015,
      "loss": 2.6429,
      "step": 353
    },
    {
      "epoch": 0.03734177215189873,
      "grad_norm": 0.5426937341690063,
      "learning_rate": 0.0015,
      "loss": 2.5742,
      "step": 354
    },
    {
      "epoch": 0.037447257383966245,
      "grad_norm": 0.6311236023902893,
      "learning_rate": 0.0015,
      "loss": 2.6135,
      "step": 355
    },
    {
      "epoch": 0.03755274261603375,
      "grad_norm": 0.6573986411094666,
      "learning_rate": 0.0015,
      "loss": 2.6319,
      "step": 356
    },
    {
      "epoch": 0.037658227848101265,
      "grad_norm": 0.5465301871299744,
      "learning_rate": 0.0015,
      "loss": 2.5794,
      "step": 357
    },
    {
      "epoch": 0.03776371308016878,
      "grad_norm": 0.5390920639038086,
      "learning_rate": 0.0015,
      "loss": 2.6012,
      "step": 358
    },
    {
      "epoch": 0.037869198312236285,
      "grad_norm": 0.5990561842918396,
      "learning_rate": 0.0015,
      "loss": 2.5803,
      "step": 359
    },
    {
      "epoch": 0.0379746835443038,
      "grad_norm": 0.6378905177116394,
      "learning_rate": 0.0015,
      "loss": 2.628,
      "step": 360
    },
    {
      "epoch": 0.038080168776371305,
      "grad_norm": 0.5791345834732056,
      "learning_rate": 0.0015,
      "loss": 2.5944,
      "step": 361
    },
    {
      "epoch": 0.03818565400843882,
      "grad_norm": 0.5968369841575623,
      "learning_rate": 0.0015,
      "loss": 2.6213,
      "step": 362
    },
    {
      "epoch": 0.03829113924050633,
      "grad_norm": 0.7104408144950867,
      "learning_rate": 0.0015,
      "loss": 2.5727,
      "step": 363
    },
    {
      "epoch": 0.03839662447257384,
      "grad_norm": 0.6978356242179871,
      "learning_rate": 0.0015,
      "loss": 2.582,
      "step": 364
    },
    {
      "epoch": 0.03850210970464135,
      "grad_norm": 0.67247074842453,
      "learning_rate": 0.0015,
      "loss": 2.6225,
      "step": 365
    },
    {
      "epoch": 0.03860759493670886,
      "grad_norm": 0.6189120411872864,
      "learning_rate": 0.0015,
      "loss": 2.5861,
      "step": 366
    },
    {
      "epoch": 0.03871308016877637,
      "grad_norm": 0.5502013564109802,
      "learning_rate": 0.0015,
      "loss": 2.5867,
      "step": 367
    },
    {
      "epoch": 0.038818565400843885,
      "grad_norm": 0.6945213675498962,
      "learning_rate": 0.0015,
      "loss": 2.6102,
      "step": 368
    },
    {
      "epoch": 0.03892405063291139,
      "grad_norm": 0.8436279296875,
      "learning_rate": 0.0015,
      "loss": 2.5969,
      "step": 369
    },
    {
      "epoch": 0.039029535864978905,
      "grad_norm": 0.8353848457336426,
      "learning_rate": 0.0015,
      "loss": 2.5668,
      "step": 370
    },
    {
      "epoch": 0.03913502109704641,
      "grad_norm": 0.7062360048294067,
      "learning_rate": 0.0015,
      "loss": 2.5801,
      "step": 371
    },
    {
      "epoch": 0.039240506329113925,
      "grad_norm": 0.5779558420181274,
      "learning_rate": 0.0015,
      "loss": 2.5597,
      "step": 372
    },
    {
      "epoch": 0.03934599156118143,
      "grad_norm": 0.7748365998268127,
      "learning_rate": 0.0015,
      "loss": 2.6049,
      "step": 373
    },
    {
      "epoch": 0.039451476793248945,
      "grad_norm": 1.1137992143630981,
      "learning_rate": 0.0015,
      "loss": 2.625,
      "step": 374
    },
    {
      "epoch": 0.03955696202531646,
      "grad_norm": 0.7565868496894836,
      "learning_rate": 0.0015,
      "loss": 2.5573,
      "step": 375
    },
    {
      "epoch": 0.039662447257383965,
      "grad_norm": 0.690518856048584,
      "learning_rate": 0.0015,
      "loss": 2.5875,
      "step": 376
    },
    {
      "epoch": 0.03976793248945148,
      "grad_norm": 1.1048948764801025,
      "learning_rate": 0.0015,
      "loss": 2.5974,
      "step": 377
    },
    {
      "epoch": 0.039873417721518985,
      "grad_norm": 0.913773775100708,
      "learning_rate": 0.0015,
      "loss": 2.5692,
      "step": 378
    },
    {
      "epoch": 0.0399789029535865,
      "grad_norm": 0.599328339099884,
      "learning_rate": 0.0015,
      "loss": 2.586,
      "step": 379
    },
    {
      "epoch": 0.04008438818565401,
      "grad_norm": 1.0448973178863525,
      "learning_rate": 0.0015,
      "loss": 2.5786,
      "step": 380
    },
    {
      "epoch": 0.04018987341772152,
      "grad_norm": 0.9382826685905457,
      "learning_rate": 0.0015,
      "loss": 2.5854,
      "step": 381
    },
    {
      "epoch": 0.04029535864978903,
      "grad_norm": 0.6715505123138428,
      "learning_rate": 0.0015,
      "loss": 2.5878,
      "step": 382
    },
    {
      "epoch": 0.04040084388185654,
      "grad_norm": 0.6978819966316223,
      "learning_rate": 0.0015,
      "loss": 2.5783,
      "step": 383
    },
    {
      "epoch": 0.04050632911392405,
      "grad_norm": 0.7260177135467529,
      "learning_rate": 0.0015,
      "loss": 2.559,
      "step": 384
    },
    {
      "epoch": 0.04061181434599156,
      "grad_norm": 0.630085825920105,
      "learning_rate": 0.0015,
      "loss": 2.5605,
      "step": 385
    },
    {
      "epoch": 0.04071729957805907,
      "grad_norm": 0.6087722778320312,
      "learning_rate": 0.0015,
      "loss": 2.5364,
      "step": 386
    },
    {
      "epoch": 0.040822784810126585,
      "grad_norm": 0.599683403968811,
      "learning_rate": 0.0015,
      "loss": 2.5776,
      "step": 387
    },
    {
      "epoch": 0.04092827004219409,
      "grad_norm": 0.4998455345630646,
      "learning_rate": 0.0015,
      "loss": 2.5425,
      "step": 388
    },
    {
      "epoch": 0.041033755274261605,
      "grad_norm": 0.6637672185897827,
      "learning_rate": 0.0015,
      "loss": 2.5436,
      "step": 389
    },
    {
      "epoch": 0.04113924050632911,
      "grad_norm": 0.8620326519012451,
      "learning_rate": 0.0015,
      "loss": 2.5473,
      "step": 390
    },
    {
      "epoch": 0.041244725738396625,
      "grad_norm": 0.8207728266716003,
      "learning_rate": 0.0015,
      "loss": 2.5478,
      "step": 391
    },
    {
      "epoch": 0.04135021097046414,
      "grad_norm": 0.6227527260780334,
      "learning_rate": 0.0015,
      "loss": 2.5378,
      "step": 392
    },
    {
      "epoch": 0.041455696202531644,
      "grad_norm": 0.8634827136993408,
      "learning_rate": 0.0015,
      "loss": 2.536,
      "step": 393
    },
    {
      "epoch": 0.04156118143459916,
      "grad_norm": 0.934406578540802,
      "learning_rate": 0.0015,
      "loss": 2.5233,
      "step": 394
    },
    {
      "epoch": 0.041666666666666664,
      "grad_norm": 1.0203802585601807,
      "learning_rate": 0.0015,
      "loss": 2.5296,
      "step": 395
    },
    {
      "epoch": 0.04177215189873418,
      "grad_norm": 1.086199402809143,
      "learning_rate": 0.0015,
      "loss": 2.55,
      "step": 396
    },
    {
      "epoch": 0.04187763713080169,
      "grad_norm": 0.6120628714561462,
      "learning_rate": 0.0015,
      "loss": 2.5153,
      "step": 397
    },
    {
      "epoch": 0.0419831223628692,
      "grad_norm": 1.0643750429153442,
      "learning_rate": 0.0015,
      "loss": 2.5341,
      "step": 398
    },
    {
      "epoch": 0.04208860759493671,
      "grad_norm": 1.1236447095870972,
      "learning_rate": 0.0015,
      "loss": 2.5576,
      "step": 399
    },
    {
      "epoch": 0.04219409282700422,
      "grad_norm": 0.6694768071174622,
      "learning_rate": 0.0015,
      "loss": 2.548,
      "step": 400
    },
    {
      "epoch": 0.04229957805907173,
      "grad_norm": 0.772110641002655,
      "learning_rate": 0.0015,
      "loss": 2.5275,
      "step": 401
    },
    {
      "epoch": 0.04240506329113924,
      "grad_norm": 0.7637317776679993,
      "learning_rate": 0.0015,
      "loss": 2.5056,
      "step": 402
    },
    {
      "epoch": 0.04251054852320675,
      "grad_norm": 0.6758496761322021,
      "learning_rate": 0.0015,
      "loss": 2.5498,
      "step": 403
    },
    {
      "epoch": 0.042616033755274264,
      "grad_norm": 0.7265697717666626,
      "learning_rate": 0.0015,
      "loss": 2.5457,
      "step": 404
    },
    {
      "epoch": 0.04272151898734177,
      "grad_norm": 0.5827400088310242,
      "learning_rate": 0.0015,
      "loss": 2.4919,
      "step": 405
    },
    {
      "epoch": 0.042827004219409284,
      "grad_norm": 0.6745532155036926,
      "learning_rate": 0.0015,
      "loss": 2.5369,
      "step": 406
    },
    {
      "epoch": 0.04293248945147679,
      "grad_norm": 0.5512591004371643,
      "learning_rate": 0.0015,
      "loss": 2.5534,
      "step": 407
    },
    {
      "epoch": 0.043037974683544304,
      "grad_norm": 0.6537434458732605,
      "learning_rate": 0.0015,
      "loss": 2.52,
      "step": 408
    },
    {
      "epoch": 0.04314345991561182,
      "grad_norm": 0.8097763657569885,
      "learning_rate": 0.0015,
      "loss": 2.5116,
      "step": 409
    },
    {
      "epoch": 0.043248945147679324,
      "grad_norm": 0.8432343006134033,
      "learning_rate": 0.0015,
      "loss": 2.5115,
      "step": 410
    },
    {
      "epoch": 0.04335443037974684,
      "grad_norm": 0.6861345767974854,
      "learning_rate": 0.0015,
      "loss": 2.4932,
      "step": 411
    },
    {
      "epoch": 0.043459915611814344,
      "grad_norm": 0.6173989176750183,
      "learning_rate": 0.0015,
      "loss": 2.5086,
      "step": 412
    },
    {
      "epoch": 0.04356540084388186,
      "grad_norm": 0.5517957806587219,
      "learning_rate": 0.0015,
      "loss": 2.5073,
      "step": 413
    },
    {
      "epoch": 0.043670886075949364,
      "grad_norm": 0.5692659020423889,
      "learning_rate": 0.0015,
      "loss": 2.4853,
      "step": 414
    },
    {
      "epoch": 0.04377637130801688,
      "grad_norm": 0.5736432075500488,
      "learning_rate": 0.0015,
      "loss": 2.5176,
      "step": 415
    },
    {
      "epoch": 0.04388185654008439,
      "grad_norm": 0.5319284796714783,
      "learning_rate": 0.0015,
      "loss": 2.4839,
      "step": 416
    },
    {
      "epoch": 0.0439873417721519,
      "grad_norm": 0.5271692276000977,
      "learning_rate": 0.0015,
      "loss": 2.5005,
      "step": 417
    },
    {
      "epoch": 0.04409282700421941,
      "grad_norm": 0.5658661723136902,
      "learning_rate": 0.0015,
      "loss": 2.5209,
      "step": 418
    },
    {
      "epoch": 0.04419831223628692,
      "grad_norm": 0.5643138289451599,
      "learning_rate": 0.0015,
      "loss": 2.5047,
      "step": 419
    },
    {
      "epoch": 0.04430379746835443,
      "grad_norm": 0.6114088892936707,
      "learning_rate": 0.0015,
      "loss": 2.4671,
      "step": 420
    },
    {
      "epoch": 0.044409282700421944,
      "grad_norm": 0.7342972755432129,
      "learning_rate": 0.0015,
      "loss": 2.4929,
      "step": 421
    },
    {
      "epoch": 0.04451476793248945,
      "grad_norm": 0.6202294230461121,
      "learning_rate": 0.0015,
      "loss": 2.4719,
      "step": 422
    },
    {
      "epoch": 0.044620253164556964,
      "grad_norm": 0.5686100125312805,
      "learning_rate": 0.0015,
      "loss": 2.482,
      "step": 423
    },
    {
      "epoch": 0.04472573839662447,
      "grad_norm": 0.7075520753860474,
      "learning_rate": 0.0015,
      "loss": 2.4976,
      "step": 424
    },
    {
      "epoch": 0.044831223628691984,
      "grad_norm": 0.8364173173904419,
      "learning_rate": 0.0015,
      "loss": 2.4605,
      "step": 425
    },
    {
      "epoch": 0.04493670886075949,
      "grad_norm": 0.6680495142936707,
      "learning_rate": 0.0015,
      "loss": 2.4886,
      "step": 426
    },
    {
      "epoch": 0.045042194092827004,
      "grad_norm": 0.5401420593261719,
      "learning_rate": 0.0015,
      "loss": 2.4795,
      "step": 427
    },
    {
      "epoch": 0.04514767932489452,
      "grad_norm": 0.5875886678695679,
      "learning_rate": 0.0015,
      "loss": 2.4747,
      "step": 428
    },
    {
      "epoch": 0.045253164556962024,
      "grad_norm": 0.5946866869926453,
      "learning_rate": 0.0015,
      "loss": 2.4539,
      "step": 429
    },
    {
      "epoch": 0.04535864978902954,
      "grad_norm": 0.5908014178276062,
      "learning_rate": 0.0015,
      "loss": 2.4399,
      "step": 430
    },
    {
      "epoch": 0.045464135021097044,
      "grad_norm": 0.7210457921028137,
      "learning_rate": 0.0015,
      "loss": 2.4281,
      "step": 431
    },
    {
      "epoch": 0.04556962025316456,
      "grad_norm": 0.6678378582000732,
      "learning_rate": 0.0015,
      "loss": 2.4607,
      "step": 432
    },
    {
      "epoch": 0.04567510548523207,
      "grad_norm": 0.6293447613716125,
      "learning_rate": 0.0015,
      "loss": 2.4535,
      "step": 433
    },
    {
      "epoch": 0.04578059071729958,
      "grad_norm": 0.5875449180603027,
      "learning_rate": 0.0015,
      "loss": 2.4784,
      "step": 434
    },
    {
      "epoch": 0.04588607594936709,
      "grad_norm": 0.5475998520851135,
      "learning_rate": 0.0015,
      "loss": 2.4629,
      "step": 435
    },
    {
      "epoch": 0.0459915611814346,
      "grad_norm": 0.5864022970199585,
      "learning_rate": 0.0015,
      "loss": 2.466,
      "step": 436
    },
    {
      "epoch": 0.04609704641350211,
      "grad_norm": 0.6136614084243774,
      "learning_rate": 0.0015,
      "loss": 2.4491,
      "step": 437
    },
    {
      "epoch": 0.046202531645569624,
      "grad_norm": 0.5596399307250977,
      "learning_rate": 0.0015,
      "loss": 2.4851,
      "step": 438
    },
    {
      "epoch": 0.04630801687763713,
      "grad_norm": 0.617128849029541,
      "learning_rate": 0.0015,
      "loss": 2.4432,
      "step": 439
    },
    {
      "epoch": 0.046413502109704644,
      "grad_norm": 0.6627237200737,
      "learning_rate": 0.0015,
      "loss": 2.4641,
      "step": 440
    },
    {
      "epoch": 0.04651898734177215,
      "grad_norm": 0.6823180317878723,
      "learning_rate": 0.0015,
      "loss": 2.457,
      "step": 441
    },
    {
      "epoch": 0.04662447257383966,
      "grad_norm": 0.6112765669822693,
      "learning_rate": 0.0015,
      "loss": 2.4667,
      "step": 442
    },
    {
      "epoch": 0.04672995780590717,
      "grad_norm": 0.7016992568969727,
      "learning_rate": 0.0015,
      "loss": 2.4383,
      "step": 443
    },
    {
      "epoch": 0.04683544303797468,
      "grad_norm": 0.7539768218994141,
      "learning_rate": 0.0015,
      "loss": 2.4564,
      "step": 444
    },
    {
      "epoch": 0.0469409282700422,
      "grad_norm": 0.7324343323707581,
      "learning_rate": 0.0015,
      "loss": 2.4494,
      "step": 445
    },
    {
      "epoch": 0.0470464135021097,
      "grad_norm": 0.5209283232688904,
      "learning_rate": 0.0015,
      "loss": 2.4662,
      "step": 446
    },
    {
      "epoch": 0.04715189873417722,
      "grad_norm": 0.5297408699989319,
      "learning_rate": 0.0015,
      "loss": 2.4531,
      "step": 447
    },
    {
      "epoch": 0.04725738396624472,
      "grad_norm": 0.5961151719093323,
      "learning_rate": 0.0015,
      "loss": 2.4268,
      "step": 448
    },
    {
      "epoch": 0.04736286919831224,
      "grad_norm": 0.6886147260665894,
      "learning_rate": 0.0015,
      "loss": 2.4761,
      "step": 449
    },
    {
      "epoch": 0.04746835443037975,
      "grad_norm": 0.5722856521606445,
      "learning_rate": 0.0015,
      "loss": 2.4366,
      "step": 450
    },
    {
      "epoch": 0.047573839662447256,
      "grad_norm": 0.5168962478637695,
      "learning_rate": 0.0015,
      "loss": 2.4501,
      "step": 451
    },
    {
      "epoch": 0.04767932489451477,
      "grad_norm": 0.646081268787384,
      "learning_rate": 0.0015,
      "loss": 2.4133,
      "step": 452
    },
    {
      "epoch": 0.047784810126582276,
      "grad_norm": 0.7620195746421814,
      "learning_rate": 0.0015,
      "loss": 2.4844,
      "step": 453
    },
    {
      "epoch": 0.04789029535864979,
      "grad_norm": 0.701268196105957,
      "learning_rate": 0.0015,
      "loss": 2.4486,
      "step": 454
    },
    {
      "epoch": 0.047995780590717296,
      "grad_norm": 0.6020218729972839,
      "learning_rate": 0.0015,
      "loss": 2.4346,
      "step": 455
    },
    {
      "epoch": 0.04810126582278481,
      "grad_norm": 0.5474327802658081,
      "learning_rate": 0.0015,
      "loss": 2.3995,
      "step": 456
    },
    {
      "epoch": 0.04820675105485232,
      "grad_norm": 0.5468835830688477,
      "learning_rate": 0.0015,
      "loss": 2.4123,
      "step": 457
    },
    {
      "epoch": 0.04831223628691983,
      "grad_norm": 0.5273303985595703,
      "learning_rate": 0.0015,
      "loss": 2.4013,
      "step": 458
    },
    {
      "epoch": 0.04841772151898734,
      "grad_norm": 0.5838407874107361,
      "learning_rate": 0.0015,
      "loss": 2.4056,
      "step": 459
    },
    {
      "epoch": 0.04852320675105485,
      "grad_norm": 0.5841299295425415,
      "learning_rate": 0.0015,
      "loss": 2.4535,
      "step": 460
    },
    {
      "epoch": 0.04862869198312236,
      "grad_norm": 0.5731238722801208,
      "learning_rate": 0.0015,
      "loss": 2.4097,
      "step": 461
    },
    {
      "epoch": 0.048734177215189876,
      "grad_norm": 0.614305317401886,
      "learning_rate": 0.0015,
      "loss": 2.4327,
      "step": 462
    },
    {
      "epoch": 0.04883966244725738,
      "grad_norm": 0.666114330291748,
      "learning_rate": 0.0015,
      "loss": 2.4071,
      "step": 463
    },
    {
      "epoch": 0.048945147679324896,
      "grad_norm": 0.7254586219787598,
      "learning_rate": 0.0015,
      "loss": 2.4103,
      "step": 464
    },
    {
      "epoch": 0.0490506329113924,
      "grad_norm": 0.6337289214134216,
      "learning_rate": 0.0015,
      "loss": 2.4079,
      "step": 465
    },
    {
      "epoch": 0.049156118143459916,
      "grad_norm": 0.5618485808372498,
      "learning_rate": 0.0015,
      "loss": 2.3842,
      "step": 466
    },
    {
      "epoch": 0.04926160337552743,
      "grad_norm": 0.5461037158966064,
      "learning_rate": 0.0015,
      "loss": 2.4154,
      "step": 467
    },
    {
      "epoch": 0.049367088607594936,
      "grad_norm": 0.5691366195678711,
      "learning_rate": 0.0015,
      "loss": 2.4129,
      "step": 468
    },
    {
      "epoch": 0.04947257383966245,
      "grad_norm": 0.5997835993766785,
      "learning_rate": 0.0015,
      "loss": 2.3984,
      "step": 469
    },
    {
      "epoch": 0.049578059071729956,
      "grad_norm": 0.6511602997779846,
      "learning_rate": 0.0015,
      "loss": 2.3911,
      "step": 470
    },
    {
      "epoch": 0.04968354430379747,
      "grad_norm": 0.6190035343170166,
      "learning_rate": 0.0015,
      "loss": 2.4017,
      "step": 471
    },
    {
      "epoch": 0.049789029535864976,
      "grad_norm": 0.5229429602622986,
      "learning_rate": 0.0015,
      "loss": 2.4059,
      "step": 472
    },
    {
      "epoch": 0.04989451476793249,
      "grad_norm": 0.6145848631858826,
      "learning_rate": 0.0015,
      "loss": 2.4286,
      "step": 473
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7568457126617432,
      "learning_rate": 0.0015,
      "loss": 2.4268,
      "step": 474
    },
    {
      "epoch": 0.05010548523206751,
      "grad_norm": 0.698989987373352,
      "learning_rate": 0.0015,
      "loss": 2.4168,
      "step": 475
    },
    {
      "epoch": 0.05021097046413502,
      "grad_norm": 0.6458556652069092,
      "learning_rate": 0.0015,
      "loss": 2.4086,
      "step": 476
    },
    {
      "epoch": 0.05031645569620253,
      "grad_norm": 0.6524352431297302,
      "learning_rate": 0.0015,
      "loss": 2.3677,
      "step": 477
    },
    {
      "epoch": 0.05042194092827004,
      "grad_norm": 0.6493321061134338,
      "learning_rate": 0.0015,
      "loss": 2.4448,
      "step": 478
    },
    {
      "epoch": 0.050527426160337556,
      "grad_norm": 0.6462939977645874,
      "learning_rate": 0.0015,
      "loss": 2.3956,
      "step": 479
    },
    {
      "epoch": 0.05063291139240506,
      "grad_norm": 0.6222774386405945,
      "learning_rate": 0.0015,
      "loss": 2.4013,
      "step": 480
    },
    {
      "epoch": 0.050738396624472576,
      "grad_norm": 0.658645510673523,
      "learning_rate": 0.0015,
      "loss": 2.3639,
      "step": 481
    },
    {
      "epoch": 0.05084388185654008,
      "grad_norm": 0.6878094673156738,
      "learning_rate": 0.0015,
      "loss": 2.4076,
      "step": 482
    },
    {
      "epoch": 0.050949367088607596,
      "grad_norm": 0.8742771148681641,
      "learning_rate": 0.0015,
      "loss": 2.4382,
      "step": 483
    },
    {
      "epoch": 0.0510548523206751,
      "grad_norm": 1.0188567638397217,
      "learning_rate": 0.0015,
      "loss": 2.402,
      "step": 484
    },
    {
      "epoch": 0.051160337552742616,
      "grad_norm": 0.9459471702575684,
      "learning_rate": 0.0015,
      "loss": 2.3598,
      "step": 485
    },
    {
      "epoch": 0.05126582278481013,
      "grad_norm": 0.5860833525657654,
      "learning_rate": 0.0015,
      "loss": 2.3536,
      "step": 486
    },
    {
      "epoch": 0.051371308016877636,
      "grad_norm": 0.8695175647735596,
      "learning_rate": 0.0015,
      "loss": 2.4133,
      "step": 487
    },
    {
      "epoch": 0.05147679324894515,
      "grad_norm": 1.435496211051941,
      "learning_rate": 0.0015,
      "loss": 2.3908,
      "step": 488
    },
    {
      "epoch": 0.051582278481012656,
      "grad_norm": 0.6248399615287781,
      "learning_rate": 0.0015,
      "loss": 2.3795,
      "step": 489
    },
    {
      "epoch": 0.05168776371308017,
      "grad_norm": 1.3923648595809937,
      "learning_rate": 0.0015,
      "loss": 2.4195,
      "step": 490
    },
    {
      "epoch": 0.05179324894514768,
      "grad_norm": 0.8495714068412781,
      "learning_rate": 0.0015,
      "loss": 2.4117,
      "step": 491
    },
    {
      "epoch": 0.05189873417721519,
      "grad_norm": 0.7323932647705078,
      "learning_rate": 0.0015,
      "loss": 2.3952,
      "step": 492
    },
    {
      "epoch": 0.0520042194092827,
      "grad_norm": 1.0794618129730225,
      "learning_rate": 0.0015,
      "loss": 2.391,
      "step": 493
    },
    {
      "epoch": 0.05210970464135021,
      "grad_norm": 0.7117097973823547,
      "learning_rate": 0.0015,
      "loss": 2.3928,
      "step": 494
    },
    {
      "epoch": 0.05221518987341772,
      "grad_norm": 0.5951404571533203,
      "learning_rate": 0.0015,
      "loss": 2.3742,
      "step": 495
    },
    {
      "epoch": 0.05232067510548523,
      "grad_norm": 0.7290152907371521,
      "learning_rate": 0.0015,
      "loss": 2.3557,
      "step": 496
    },
    {
      "epoch": 0.05242616033755274,
      "grad_norm": 0.7287200689315796,
      "learning_rate": 0.0015,
      "loss": 2.3915,
      "step": 497
    },
    {
      "epoch": 0.052531645569620256,
      "grad_norm": 0.5445609092712402,
      "learning_rate": 0.0015,
      "loss": 2.3464,
      "step": 498
    },
    {
      "epoch": 0.05263713080168776,
      "grad_norm": 0.6681470274925232,
      "learning_rate": 0.0015,
      "loss": 2.3818,
      "step": 499
    },
    {
      "epoch": 0.052742616033755275,
      "grad_norm": 0.8860105276107788,
      "learning_rate": 0.0015,
      "loss": 2.3743,
      "step": 500
    },
    {
      "epoch": 0.05284810126582278,
      "grad_norm": 0.7466394305229187,
      "learning_rate": 0.0015,
      "loss": 2.3796,
      "step": 501
    },
    {
      "epoch": 0.052953586497890295,
      "grad_norm": 0.5135951638221741,
      "learning_rate": 0.0015,
      "loss": 2.4019,
      "step": 502
    },
    {
      "epoch": 0.05305907172995781,
      "grad_norm": 0.7751052975654602,
      "learning_rate": 0.0015,
      "loss": 2.3433,
      "step": 503
    },
    {
      "epoch": 0.053164556962025315,
      "grad_norm": 0.8892273902893066,
      "learning_rate": 0.0015,
      "loss": 2.3632,
      "step": 504
    },
    {
      "epoch": 0.05327004219409283,
      "grad_norm": 0.6392226815223694,
      "learning_rate": 0.0015,
      "loss": 2.3638,
      "step": 505
    },
    {
      "epoch": 0.053375527426160335,
      "grad_norm": 0.7267099618911743,
      "learning_rate": 0.0015,
      "loss": 2.4042,
      "step": 506
    },
    {
      "epoch": 0.05348101265822785,
      "grad_norm": 0.7698792815208435,
      "learning_rate": 0.0015,
      "loss": 2.3684,
      "step": 507
    },
    {
      "epoch": 0.05358649789029536,
      "grad_norm": 0.5992477536201477,
      "learning_rate": 0.0015,
      "loss": 2.3566,
      "step": 508
    },
    {
      "epoch": 0.05369198312236287,
      "grad_norm": 0.67427659034729,
      "learning_rate": 0.0015,
      "loss": 2.3174,
      "step": 509
    },
    {
      "epoch": 0.05379746835443038,
      "grad_norm": 0.6546165943145752,
      "learning_rate": 0.0015,
      "loss": 2.3764,
      "step": 510
    },
    {
      "epoch": 0.05390295358649789,
      "grad_norm": 0.5334204435348511,
      "learning_rate": 0.0015,
      "loss": 2.3481,
      "step": 511
    },
    {
      "epoch": 0.0540084388185654,
      "grad_norm": 0.807060718536377,
      "learning_rate": 0.0015,
      "loss": 2.3659,
      "step": 512
    },
    {
      "epoch": 0.05411392405063291,
      "grad_norm": 0.6117734313011169,
      "learning_rate": 0.0015,
      "loss": 2.3503,
      "step": 513
    },
    {
      "epoch": 0.05421940928270042,
      "grad_norm": 0.6080499887466431,
      "learning_rate": 0.0015,
      "loss": 2.3636,
      "step": 514
    },
    {
      "epoch": 0.054324894514767935,
      "grad_norm": 0.7046027779579163,
      "learning_rate": 0.0015,
      "loss": 2.328,
      "step": 515
    },
    {
      "epoch": 0.05443037974683544,
      "grad_norm": 0.6732563376426697,
      "learning_rate": 0.0015,
      "loss": 2.3559,
      "step": 516
    },
    {
      "epoch": 0.054535864978902955,
      "grad_norm": 0.6331255435943604,
      "learning_rate": 0.0015,
      "loss": 2.3518,
      "step": 517
    },
    {
      "epoch": 0.05464135021097046,
      "grad_norm": 0.560720682144165,
      "learning_rate": 0.0015,
      "loss": 2.3017,
      "step": 518
    },
    {
      "epoch": 0.054746835443037975,
      "grad_norm": 0.6451405882835388,
      "learning_rate": 0.0015,
      "loss": 2.3548,
      "step": 519
    },
    {
      "epoch": 0.05485232067510549,
      "grad_norm": 0.7499489784240723,
      "learning_rate": 0.0015,
      "loss": 2.3601,
      "step": 520
    },
    {
      "epoch": 0.054957805907172995,
      "grad_norm": 0.7935493588447571,
      "learning_rate": 0.0015,
      "loss": 2.3588,
      "step": 521
    },
    {
      "epoch": 0.05506329113924051,
      "grad_norm": 0.6315867900848389,
      "learning_rate": 0.0015,
      "loss": 2.3242,
      "step": 522
    },
    {
      "epoch": 0.055168776371308015,
      "grad_norm": 0.5828781127929688,
      "learning_rate": 0.0015,
      "loss": 2.3563,
      "step": 523
    },
    {
      "epoch": 0.05527426160337553,
      "grad_norm": 0.8364554047584534,
      "learning_rate": 0.0015,
      "loss": 2.3566,
      "step": 524
    },
    {
      "epoch": 0.055379746835443035,
      "grad_norm": 0.6991103291511536,
      "learning_rate": 0.0015,
      "loss": 2.335,
      "step": 525
    },
    {
      "epoch": 0.05548523206751055,
      "grad_norm": 0.5167779922485352,
      "learning_rate": 0.0015,
      "loss": 2.3269,
      "step": 526
    },
    {
      "epoch": 0.05559071729957806,
      "grad_norm": 0.7246729135513306,
      "learning_rate": 0.0015,
      "loss": 2.3199,
      "step": 527
    },
    {
      "epoch": 0.05569620253164557,
      "grad_norm": 0.6558444499969482,
      "learning_rate": 0.0015,
      "loss": 2.3397,
      "step": 528
    },
    {
      "epoch": 0.05580168776371308,
      "grad_norm": 0.5521800518035889,
      "learning_rate": 0.0015,
      "loss": 2.3423,
      "step": 529
    },
    {
      "epoch": 0.05590717299578059,
      "grad_norm": 0.5958268046379089,
      "learning_rate": 0.0015,
      "loss": 2.3282,
      "step": 530
    },
    {
      "epoch": 0.0560126582278481,
      "grad_norm": 0.7040470838546753,
      "learning_rate": 0.0015,
      "loss": 2.3378,
      "step": 531
    },
    {
      "epoch": 0.056118143459915615,
      "grad_norm": 0.665379524230957,
      "learning_rate": 0.0015,
      "loss": 2.2979,
      "step": 532
    },
    {
      "epoch": 0.05622362869198312,
      "grad_norm": 0.5010139346122742,
      "learning_rate": 0.0015,
      "loss": 2.3223,
      "step": 533
    },
    {
      "epoch": 0.056329113924050635,
      "grad_norm": 0.5269133448600769,
      "learning_rate": 0.0015,
      "loss": 2.317,
      "step": 534
    },
    {
      "epoch": 0.05643459915611814,
      "grad_norm": 0.5116428732872009,
      "learning_rate": 0.0015,
      "loss": 2.3151,
      "step": 535
    },
    {
      "epoch": 0.056540084388185655,
      "grad_norm": 0.5242851376533508,
      "learning_rate": 0.0015,
      "loss": 2.3138,
      "step": 536
    },
    {
      "epoch": 0.05664556962025316,
      "grad_norm": 0.5800849795341492,
      "learning_rate": 0.0015,
      "loss": 2.2844,
      "step": 537
    },
    {
      "epoch": 0.056751054852320675,
      "grad_norm": 0.5497094392776489,
      "learning_rate": 0.0015,
      "loss": 2.3159,
      "step": 538
    },
    {
      "epoch": 0.05685654008438819,
      "grad_norm": 0.4886211156845093,
      "learning_rate": 0.0015,
      "loss": 2.3201,
      "step": 539
    },
    {
      "epoch": 0.056962025316455694,
      "grad_norm": 0.636679470539093,
      "learning_rate": 0.0015,
      "loss": 2.3284,
      "step": 540
    },
    {
      "epoch": 0.05706751054852321,
      "grad_norm": 0.525220513343811,
      "learning_rate": 0.0015,
      "loss": 2.3409,
      "step": 541
    },
    {
      "epoch": 0.057172995780590714,
      "grad_norm": 0.6222096085548401,
      "learning_rate": 0.0015,
      "loss": 2.3222,
      "step": 542
    },
    {
      "epoch": 0.05727848101265823,
      "grad_norm": 0.592156708240509,
      "learning_rate": 0.0015,
      "loss": 2.2913,
      "step": 543
    },
    {
      "epoch": 0.05738396624472574,
      "grad_norm": 0.6042721271514893,
      "learning_rate": 0.0015,
      "loss": 2.3167,
      "step": 544
    },
    {
      "epoch": 0.05748945147679325,
      "grad_norm": 0.5723241567611694,
      "learning_rate": 0.0015,
      "loss": 2.3211,
      "step": 545
    },
    {
      "epoch": 0.05759493670886076,
      "grad_norm": 0.5514154434204102,
      "learning_rate": 0.0015,
      "loss": 2.314,
      "step": 546
    },
    {
      "epoch": 0.05770042194092827,
      "grad_norm": 0.5217574238777161,
      "learning_rate": 0.0015,
      "loss": 2.3222,
      "step": 547
    },
    {
      "epoch": 0.05780590717299578,
      "grad_norm": 0.5978240966796875,
      "learning_rate": 0.0015,
      "loss": 2.3298,
      "step": 548
    },
    {
      "epoch": 0.057911392405063294,
      "grad_norm": 0.5819392800331116,
      "learning_rate": 0.0015,
      "loss": 2.3219,
      "step": 549
    },
    {
      "epoch": 0.0580168776371308,
      "grad_norm": 0.5390170812606812,
      "learning_rate": 0.0015,
      "loss": 2.2765,
      "step": 550
    },
    {
      "epoch": 0.058122362869198314,
      "grad_norm": 0.6666958332061768,
      "learning_rate": 0.0015,
      "loss": 2.2777,
      "step": 551
    },
    {
      "epoch": 0.05822784810126582,
      "grad_norm": 0.8040550947189331,
      "learning_rate": 0.0015,
      "loss": 2.3051,
      "step": 552
    },
    {
      "epoch": 0.058333333333333334,
      "grad_norm": 0.8092217445373535,
      "learning_rate": 0.0015,
      "loss": 2.2997,
      "step": 553
    },
    {
      "epoch": 0.05843881856540084,
      "grad_norm": 0.7098361253738403,
      "learning_rate": 0.0015,
      "loss": 2.247,
      "step": 554
    },
    {
      "epoch": 0.058544303797468354,
      "grad_norm": 0.5931481719017029,
      "learning_rate": 0.0015,
      "loss": 2.2511,
      "step": 555
    },
    {
      "epoch": 0.05864978902953587,
      "grad_norm": 0.6600017547607422,
      "learning_rate": 0.0015,
      "loss": 2.306,
      "step": 556
    },
    {
      "epoch": 0.058755274261603374,
      "grad_norm": 0.7226420044898987,
      "learning_rate": 0.0015,
      "loss": 2.3279,
      "step": 557
    },
    {
      "epoch": 0.05886075949367089,
      "grad_norm": 0.854891300201416,
      "learning_rate": 0.0015,
      "loss": 2.2823,
      "step": 558
    },
    {
      "epoch": 0.058966244725738394,
      "grad_norm": 0.7509938478469849,
      "learning_rate": 0.0015,
      "loss": 2.2597,
      "step": 559
    },
    {
      "epoch": 0.05907172995780591,
      "grad_norm": 0.5163098573684692,
      "learning_rate": 0.0015,
      "loss": 2.2749,
      "step": 560
    },
    {
      "epoch": 0.05917721518987342,
      "grad_norm": 0.752998948097229,
      "learning_rate": 0.0015,
      "loss": 2.2796,
      "step": 561
    },
    {
      "epoch": 0.05928270042194093,
      "grad_norm": 0.7569834589958191,
      "learning_rate": 0.0015,
      "loss": 2.286,
      "step": 562
    },
    {
      "epoch": 0.05938818565400844,
      "grad_norm": 0.5669600367546082,
      "learning_rate": 0.0015,
      "loss": 2.3074,
      "step": 563
    },
    {
      "epoch": 0.05949367088607595,
      "grad_norm": 0.6263278126716614,
      "learning_rate": 0.0015,
      "loss": 2.3007,
      "step": 564
    },
    {
      "epoch": 0.05959915611814346,
      "grad_norm": 0.7167249321937561,
      "learning_rate": 0.0015,
      "loss": 2.295,
      "step": 565
    },
    {
      "epoch": 0.05970464135021097,
      "grad_norm": 0.657244086265564,
      "learning_rate": 0.0015,
      "loss": 2.3088,
      "step": 566
    },
    {
      "epoch": 0.05981012658227848,
      "grad_norm": 0.5568730235099792,
      "learning_rate": 0.0015,
      "loss": 2.3145,
      "step": 567
    },
    {
      "epoch": 0.059915611814345994,
      "grad_norm": 0.6420565843582153,
      "learning_rate": 0.0015,
      "loss": 2.2538,
      "step": 568
    },
    {
      "epoch": 0.0600210970464135,
      "grad_norm": 0.6171932816505432,
      "learning_rate": 0.0015,
      "loss": 2.2294,
      "step": 569
    },
    {
      "epoch": 0.060126582278481014,
      "grad_norm": 0.630375862121582,
      "learning_rate": 0.0015,
      "loss": 2.3236,
      "step": 570
    },
    {
      "epoch": 0.06023206751054852,
      "grad_norm": 0.5438060760498047,
      "learning_rate": 0.0015,
      "loss": 2.2984,
      "step": 571
    },
    {
      "epoch": 0.060337552742616034,
      "grad_norm": 0.5490220785140991,
      "learning_rate": 0.0015,
      "loss": 2.2881,
      "step": 572
    },
    {
      "epoch": 0.06044303797468355,
      "grad_norm": 0.6096124053001404,
      "learning_rate": 0.0015,
      "loss": 2.3037,
      "step": 573
    },
    {
      "epoch": 0.060548523206751054,
      "grad_norm": 0.5740655660629272,
      "learning_rate": 0.0015,
      "loss": 2.2615,
      "step": 574
    },
    {
      "epoch": 0.06065400843881857,
      "grad_norm": 0.5798143744468689,
      "learning_rate": 0.0015,
      "loss": 2.3062,
      "step": 575
    },
    {
      "epoch": 0.060759493670886074,
      "grad_norm": 0.6170733571052551,
      "learning_rate": 0.0015,
      "loss": 2.2874,
      "step": 576
    },
    {
      "epoch": 0.06086497890295359,
      "grad_norm": 0.5171892046928406,
      "learning_rate": 0.0015,
      "loss": 2.2718,
      "step": 577
    },
    {
      "epoch": 0.0609704641350211,
      "grad_norm": 0.6215243935585022,
      "learning_rate": 0.0015,
      "loss": 2.2765,
      "step": 578
    },
    {
      "epoch": 0.06107594936708861,
      "grad_norm": 0.7542302012443542,
      "learning_rate": 0.0015,
      "loss": 2.2812,
      "step": 579
    },
    {
      "epoch": 0.06118143459915612,
      "grad_norm": 0.7645319104194641,
      "learning_rate": 0.0015,
      "loss": 2.289,
      "step": 580
    },
    {
      "epoch": 0.06128691983122363,
      "grad_norm": 0.7993782758712769,
      "learning_rate": 0.0015,
      "loss": 2.284,
      "step": 581
    },
    {
      "epoch": 0.06139240506329114,
      "grad_norm": 0.7777169346809387,
      "learning_rate": 0.0015,
      "loss": 2.2568,
      "step": 582
    },
    {
      "epoch": 0.06149789029535865,
      "grad_norm": 0.5615309476852417,
      "learning_rate": 0.0015,
      "loss": 2.2898,
      "step": 583
    },
    {
      "epoch": 0.06160337552742616,
      "grad_norm": 0.6185851693153381,
      "learning_rate": 0.0015,
      "loss": 2.2565,
      "step": 584
    },
    {
      "epoch": 0.061708860759493674,
      "grad_norm": 0.7018452882766724,
      "learning_rate": 0.0015,
      "loss": 2.2963,
      "step": 585
    },
    {
      "epoch": 0.06181434599156118,
      "grad_norm": 0.6029337048530579,
      "learning_rate": 0.0015,
      "loss": 2.2682,
      "step": 586
    },
    {
      "epoch": 0.061919831223628694,
      "grad_norm": 0.6678157448768616,
      "learning_rate": 0.0015,
      "loss": 2.2443,
      "step": 587
    },
    {
      "epoch": 0.0620253164556962,
      "grad_norm": 0.7533937096595764,
      "learning_rate": 0.0015,
      "loss": 2.2334,
      "step": 588
    },
    {
      "epoch": 0.06213080168776371,
      "grad_norm": 0.7184306979179382,
      "learning_rate": 0.0015,
      "loss": 2.2771,
      "step": 589
    },
    {
      "epoch": 0.06223628691983123,
      "grad_norm": 0.7327088117599487,
      "learning_rate": 0.0015,
      "loss": 2.2224,
      "step": 590
    },
    {
      "epoch": 0.06234177215189873,
      "grad_norm": 0.6438800692558289,
      "learning_rate": 0.0015,
      "loss": 2.2418,
      "step": 591
    },
    {
      "epoch": 0.06244725738396625,
      "grad_norm": 0.5882759094238281,
      "learning_rate": 0.0015,
      "loss": 2.2468,
      "step": 592
    },
    {
      "epoch": 0.06255274261603376,
      "grad_norm": 0.5381282567977905,
      "learning_rate": 0.0015,
      "loss": 2.2718,
      "step": 593
    },
    {
      "epoch": 0.06265822784810127,
      "grad_norm": 0.7376332879066467,
      "learning_rate": 0.0015,
      "loss": 2.23,
      "step": 594
    },
    {
      "epoch": 0.06276371308016877,
      "grad_norm": 0.8881618976593018,
      "learning_rate": 0.0015,
      "loss": 2.2605,
      "step": 595
    },
    {
      "epoch": 0.0628691983122363,
      "grad_norm": 0.576248049736023,
      "learning_rate": 0.0015,
      "loss": 2.2125,
      "step": 596
    },
    {
      "epoch": 0.0629746835443038,
      "grad_norm": 0.5991083979606628,
      "learning_rate": 0.0015,
      "loss": 2.2441,
      "step": 597
    },
    {
      "epoch": 0.0630801687763713,
      "grad_norm": 0.6734524369239807,
      "learning_rate": 0.0015,
      "loss": 2.2927,
      "step": 598
    },
    {
      "epoch": 0.06318565400843881,
      "grad_norm": 0.5658050179481506,
      "learning_rate": 0.0015,
      "loss": 2.2358,
      "step": 599
    },
    {
      "epoch": 0.06329113924050633,
      "grad_norm": 0.5681853294372559,
      "learning_rate": 0.0015,
      "loss": 2.2486,
      "step": 600
    },
    {
      "epoch": 0.06339662447257384,
      "grad_norm": 0.7272197008132935,
      "learning_rate": 0.0015,
      "loss": 2.2683,
      "step": 601
    },
    {
      "epoch": 0.06350210970464135,
      "grad_norm": 0.5892140865325928,
      "learning_rate": 0.0015,
      "loss": 2.2398,
      "step": 602
    },
    {
      "epoch": 0.06360759493670887,
      "grad_norm": 0.545028567314148,
      "learning_rate": 0.0015,
      "loss": 2.263,
      "step": 603
    },
    {
      "epoch": 0.06371308016877637,
      "grad_norm": 0.5640975832939148,
      "learning_rate": 0.0015,
      "loss": 2.2047,
      "step": 604
    },
    {
      "epoch": 0.06381856540084388,
      "grad_norm": 0.6265382766723633,
      "learning_rate": 0.0015,
      "loss": 2.2459,
      "step": 605
    },
    {
      "epoch": 0.06392405063291139,
      "grad_norm": 0.6744657754898071,
      "learning_rate": 0.0015,
      "loss": 2.2274,
      "step": 606
    },
    {
      "epoch": 0.0640295358649789,
      "grad_norm": 0.46932274103164673,
      "learning_rate": 0.0015,
      "loss": 2.2616,
      "step": 607
    },
    {
      "epoch": 0.06413502109704641,
      "grad_norm": 0.6074360013008118,
      "learning_rate": 0.0015,
      "loss": 2.2918,
      "step": 608
    },
    {
      "epoch": 0.06424050632911392,
      "grad_norm": 0.5361587405204773,
      "learning_rate": 0.0015,
      "loss": 2.2424,
      "step": 609
    },
    {
      "epoch": 0.06434599156118144,
      "grad_norm": 0.5556535124778748,
      "learning_rate": 0.0015,
      "loss": 2.2512,
      "step": 610
    },
    {
      "epoch": 0.06445147679324895,
      "grad_norm": 0.5393496751785278,
      "learning_rate": 0.0015,
      "loss": 2.2317,
      "step": 611
    },
    {
      "epoch": 0.06455696202531645,
      "grad_norm": 0.5839556455612183,
      "learning_rate": 0.0015,
      "loss": 2.239,
      "step": 612
    },
    {
      "epoch": 0.06466244725738397,
      "grad_norm": 0.5499026775360107,
      "learning_rate": 0.0015,
      "loss": 2.2222,
      "step": 613
    },
    {
      "epoch": 0.06476793248945148,
      "grad_norm": 0.569526731967926,
      "learning_rate": 0.0015,
      "loss": 2.1879,
      "step": 614
    },
    {
      "epoch": 0.06487341772151899,
      "grad_norm": 0.49739882349967957,
      "learning_rate": 0.0015,
      "loss": 2.24,
      "step": 615
    },
    {
      "epoch": 0.06497890295358649,
      "grad_norm": 0.5533305406570435,
      "learning_rate": 0.0015,
      "loss": 2.2358,
      "step": 616
    },
    {
      "epoch": 0.06508438818565401,
      "grad_norm": 0.739514172077179,
      "learning_rate": 0.0015,
      "loss": 2.2399,
      "step": 617
    },
    {
      "epoch": 0.06518987341772152,
      "grad_norm": 0.7596186399459839,
      "learning_rate": 0.0015,
      "loss": 2.2509,
      "step": 618
    },
    {
      "epoch": 0.06529535864978903,
      "grad_norm": 0.6795740127563477,
      "learning_rate": 0.0015,
      "loss": 2.2705,
      "step": 619
    },
    {
      "epoch": 0.06540084388185655,
      "grad_norm": 0.5516452789306641,
      "learning_rate": 0.0015,
      "loss": 2.2454,
      "step": 620
    },
    {
      "epoch": 0.06550632911392405,
      "grad_norm": 0.5583570003509521,
      "learning_rate": 0.0015,
      "loss": 2.2198,
      "step": 621
    },
    {
      "epoch": 0.06561181434599156,
      "grad_norm": 0.6461924910545349,
      "learning_rate": 0.0015,
      "loss": 2.2505,
      "step": 622
    },
    {
      "epoch": 0.06571729957805907,
      "grad_norm": 0.8149744868278503,
      "learning_rate": 0.0015,
      "loss": 2.2306,
      "step": 623
    },
    {
      "epoch": 0.06582278481012659,
      "grad_norm": 0.7241102457046509,
      "learning_rate": 0.0015,
      "loss": 2.1932,
      "step": 624
    },
    {
      "epoch": 0.06592827004219409,
      "grad_norm": 0.7473236918449402,
      "learning_rate": 0.0015,
      "loss": 2.223,
      "step": 625
    },
    {
      "epoch": 0.0660337552742616,
      "grad_norm": 0.9002881646156311,
      "learning_rate": 0.0015,
      "loss": 2.2247,
      "step": 626
    },
    {
      "epoch": 0.06613924050632912,
      "grad_norm": 0.806682288646698,
      "learning_rate": 0.0015,
      "loss": 2.2434,
      "step": 627
    },
    {
      "epoch": 0.06624472573839663,
      "grad_norm": 0.5749506950378418,
      "learning_rate": 0.0015,
      "loss": 2.2224,
      "step": 628
    },
    {
      "epoch": 0.06635021097046413,
      "grad_norm": 0.6432097554206848,
      "learning_rate": 0.0015,
      "loss": 2.2443,
      "step": 629
    },
    {
      "epoch": 0.06645569620253164,
      "grad_norm": 0.7201072573661804,
      "learning_rate": 0.0015,
      "loss": 2.1948,
      "step": 630
    },
    {
      "epoch": 0.06656118143459916,
      "grad_norm": 0.6522536873817444,
      "learning_rate": 0.0015,
      "loss": 2.1953,
      "step": 631
    },
    {
      "epoch": 0.06666666666666667,
      "grad_norm": 0.5203250646591187,
      "learning_rate": 0.0015,
      "loss": 2.2295,
      "step": 632
    },
    {
      "epoch": 0.06677215189873417,
      "grad_norm": 0.8893463015556335,
      "learning_rate": 0.0015,
      "loss": 2.2317,
      "step": 633
    },
    {
      "epoch": 0.06687763713080169,
      "grad_norm": 1.216801643371582,
      "learning_rate": 0.0015,
      "loss": 2.2222,
      "step": 634
    },
    {
      "epoch": 0.0669831223628692,
      "grad_norm": 0.5740790963172913,
      "learning_rate": 0.0015,
      "loss": 2.2802,
      "step": 635
    },
    {
      "epoch": 0.0670886075949367,
      "grad_norm": 0.9687669277191162,
      "learning_rate": 0.0015,
      "loss": 2.2258,
      "step": 636
    },
    {
      "epoch": 0.06719409282700423,
      "grad_norm": 1.0738919973373413,
      "learning_rate": 0.0015,
      "loss": 2.2869,
      "step": 637
    },
    {
      "epoch": 0.06729957805907173,
      "grad_norm": 0.5578457117080688,
      "learning_rate": 0.0015,
      "loss": 2.2387,
      "step": 638
    },
    {
      "epoch": 0.06740506329113924,
      "grad_norm": 1.198937177658081,
      "learning_rate": 0.0015,
      "loss": 2.2381,
      "step": 639
    },
    {
      "epoch": 0.06751054852320675,
      "grad_norm": 0.7291653752326965,
      "learning_rate": 0.0015,
      "loss": 2.2371,
      "step": 640
    },
    {
      "epoch": 0.06761603375527427,
      "grad_norm": 0.6903422474861145,
      "learning_rate": 0.0015,
      "loss": 2.219,
      "step": 641
    },
    {
      "epoch": 0.06772151898734177,
      "grad_norm": 0.8727262020111084,
      "learning_rate": 0.0015,
      "loss": 2.2328,
      "step": 642
    },
    {
      "epoch": 0.06782700421940928,
      "grad_norm": 0.6038119196891785,
      "learning_rate": 0.0015,
      "loss": 2.2282,
      "step": 643
    },
    {
      "epoch": 0.0679324894514768,
      "grad_norm": 0.6179218888282776,
      "learning_rate": 0.0015,
      "loss": 2.1871,
      "step": 644
    },
    {
      "epoch": 0.0680379746835443,
      "grad_norm": 0.8188096880912781,
      "learning_rate": 0.0015,
      "loss": 2.1958,
      "step": 645
    },
    {
      "epoch": 0.06814345991561181,
      "grad_norm": 0.5943591594696045,
      "learning_rate": 0.0015,
      "loss": 2.1807,
      "step": 646
    },
    {
      "epoch": 0.06824894514767932,
      "grad_norm": 0.603207528591156,
      "learning_rate": 0.0015,
      "loss": 2.2571,
      "step": 647
    },
    {
      "epoch": 0.06835443037974684,
      "grad_norm": 0.8463202714920044,
      "learning_rate": 0.0015,
      "loss": 2.2727,
      "step": 648
    },
    {
      "epoch": 0.06845991561181435,
      "grad_norm": 0.732502818107605,
      "learning_rate": 0.0015,
      "loss": 2.2261,
      "step": 649
    },
    {
      "epoch": 0.06856540084388185,
      "grad_norm": 0.5361315608024597,
      "learning_rate": 0.0015,
      "loss": 2.2414,
      "step": 650
    },
    {
      "epoch": 0.06867088607594937,
      "grad_norm": 0.7492843866348267,
      "learning_rate": 0.0015,
      "loss": 2.2033,
      "step": 651
    },
    {
      "epoch": 0.06877637130801688,
      "grad_norm": 0.8218656182289124,
      "learning_rate": 0.0015,
      "loss": 2.2163,
      "step": 652
    },
    {
      "epoch": 0.06888185654008439,
      "grad_norm": 0.5739061236381531,
      "learning_rate": 0.0015,
      "loss": 2.2198,
      "step": 653
    },
    {
      "epoch": 0.0689873417721519,
      "grad_norm": 0.7323062419891357,
      "learning_rate": 0.0015,
      "loss": 2.2298,
      "step": 654
    },
    {
      "epoch": 0.06909282700421941,
      "grad_norm": 0.9246823787689209,
      "learning_rate": 0.0015,
      "loss": 2.2114,
      "step": 655
    },
    {
      "epoch": 0.06919831223628692,
      "grad_norm": 0.6526539921760559,
      "learning_rate": 0.0015,
      "loss": 2.2249,
      "step": 656
    },
    {
      "epoch": 0.06930379746835443,
      "grad_norm": 0.506503164768219,
      "learning_rate": 0.0015,
      "loss": 2.1921,
      "step": 657
    },
    {
      "epoch": 0.06940928270042195,
      "grad_norm": 0.7326666116714478,
      "learning_rate": 0.0015,
      "loss": 2.1825,
      "step": 658
    },
    {
      "epoch": 0.06951476793248945,
      "grad_norm": 0.6644753217697144,
      "learning_rate": 0.0015,
      "loss": 2.2153,
      "step": 659
    },
    {
      "epoch": 0.06962025316455696,
      "grad_norm": 0.5262641310691833,
      "learning_rate": 0.0015,
      "loss": 2.1952,
      "step": 660
    },
    {
      "epoch": 0.06972573839662448,
      "grad_norm": 0.4978020489215851,
      "learning_rate": 0.0015,
      "loss": 2.1827,
      "step": 661
    },
    {
      "epoch": 0.06983122362869199,
      "grad_norm": 0.6205664873123169,
      "learning_rate": 0.0015,
      "loss": 2.1931,
      "step": 662
    },
    {
      "epoch": 0.06993670886075949,
      "grad_norm": 0.6488565802574158,
      "learning_rate": 0.0015,
      "loss": 2.1365,
      "step": 663
    },
    {
      "epoch": 0.070042194092827,
      "grad_norm": 0.5634413957595825,
      "learning_rate": 0.0015,
      "loss": 2.2136,
      "step": 664
    },
    {
      "epoch": 0.07014767932489452,
      "grad_norm": 0.5196778178215027,
      "learning_rate": 0.0015,
      "loss": 2.1981,
      "step": 665
    },
    {
      "epoch": 0.07025316455696203,
      "grad_norm": 0.6007410287857056,
      "learning_rate": 0.0015,
      "loss": 2.1712,
      "step": 666
    },
    {
      "epoch": 0.07035864978902953,
      "grad_norm": 0.5979984402656555,
      "learning_rate": 0.0015,
      "loss": 2.2109,
      "step": 667
    },
    {
      "epoch": 0.07046413502109705,
      "grad_norm": 0.5532450079917908,
      "learning_rate": 0.0015,
      "loss": 2.1711,
      "step": 668
    },
    {
      "epoch": 0.07056962025316456,
      "grad_norm": 0.5314489603042603,
      "learning_rate": 0.0015,
      "loss": 2.2164,
      "step": 669
    },
    {
      "epoch": 0.07067510548523206,
      "grad_norm": 0.541140615940094,
      "learning_rate": 0.0015,
      "loss": 2.2041,
      "step": 670
    },
    {
      "epoch": 0.07078059071729957,
      "grad_norm": 0.6494712233543396,
      "learning_rate": 0.0015,
      "loss": 2.2075,
      "step": 671
    },
    {
      "epoch": 0.07088607594936709,
      "grad_norm": 0.7607830762863159,
      "learning_rate": 0.0015,
      "loss": 2.1846,
      "step": 672
    },
    {
      "epoch": 0.0709915611814346,
      "grad_norm": 0.5456535220146179,
      "learning_rate": 0.0015,
      "loss": 2.208,
      "step": 673
    },
    {
      "epoch": 0.0710970464135021,
      "grad_norm": 0.5576298832893372,
      "learning_rate": 0.0015,
      "loss": 2.1365,
      "step": 674
    },
    {
      "epoch": 0.07120253164556962,
      "grad_norm": 0.5766642689704895,
      "learning_rate": 0.0015,
      "loss": 2.1986,
      "step": 675
    },
    {
      "epoch": 0.07130801687763713,
      "grad_norm": 0.51692134141922,
      "learning_rate": 0.0015,
      "loss": 2.1853,
      "step": 676
    },
    {
      "epoch": 0.07141350210970464,
      "grad_norm": 0.517126202583313,
      "learning_rate": 0.0015,
      "loss": 2.1648,
      "step": 677
    },
    {
      "epoch": 0.07151898734177216,
      "grad_norm": 0.582494854927063,
      "learning_rate": 0.0015,
      "loss": 2.1938,
      "step": 678
    },
    {
      "epoch": 0.07162447257383966,
      "grad_norm": 0.5192087292671204,
      "learning_rate": 0.0015,
      "loss": 2.1605,
      "step": 679
    },
    {
      "epoch": 0.07172995780590717,
      "grad_norm": 0.532579779624939,
      "learning_rate": 0.0015,
      "loss": 2.195,
      "step": 680
    },
    {
      "epoch": 0.07183544303797468,
      "grad_norm": 0.5732529759407043,
      "learning_rate": 0.0015,
      "loss": 2.1955,
      "step": 681
    },
    {
      "epoch": 0.0719409282700422,
      "grad_norm": 0.6465937495231628,
      "learning_rate": 0.0015,
      "loss": 2.214,
      "step": 682
    },
    {
      "epoch": 0.0720464135021097,
      "grad_norm": 0.911175012588501,
      "learning_rate": 0.0015,
      "loss": 2.2002,
      "step": 683
    },
    {
      "epoch": 0.07215189873417721,
      "grad_norm": 0.9529163837432861,
      "learning_rate": 0.0015,
      "loss": 2.2194,
      "step": 684
    },
    {
      "epoch": 0.07225738396624473,
      "grad_norm": 0.6736530065536499,
      "learning_rate": 0.0015,
      "loss": 2.1713,
      "step": 685
    },
    {
      "epoch": 0.07236286919831224,
      "grad_norm": 0.5988576412200928,
      "learning_rate": 0.0015,
      "loss": 2.2258,
      "step": 686
    },
    {
      "epoch": 0.07246835443037974,
      "grad_norm": 0.7737632393836975,
      "learning_rate": 0.0015,
      "loss": 2.1885,
      "step": 687
    },
    {
      "epoch": 0.07257383966244725,
      "grad_norm": 0.8874591588973999,
      "learning_rate": 0.0015,
      "loss": 2.2235,
      "step": 688
    },
    {
      "epoch": 0.07267932489451477,
      "grad_norm": 0.7639130353927612,
      "learning_rate": 0.0015,
      "loss": 2.145,
      "step": 689
    },
    {
      "epoch": 0.07278481012658228,
      "grad_norm": 0.5340981483459473,
      "learning_rate": 0.0015,
      "loss": 2.218,
      "step": 690
    },
    {
      "epoch": 0.07289029535864978,
      "grad_norm": 0.6690497994422913,
      "learning_rate": 0.0015,
      "loss": 2.199,
      "step": 691
    },
    {
      "epoch": 0.0729957805907173,
      "grad_norm": 0.7696680426597595,
      "learning_rate": 0.0015,
      "loss": 2.1728,
      "step": 692
    },
    {
      "epoch": 0.07310126582278481,
      "grad_norm": 0.6508092880249023,
      "learning_rate": 0.0015,
      "loss": 2.2021,
      "step": 693
    },
    {
      "epoch": 0.07320675105485232,
      "grad_norm": 0.5294222235679626,
      "learning_rate": 0.0015,
      "loss": 2.2278,
      "step": 694
    },
    {
      "epoch": 0.07331223628691984,
      "grad_norm": 0.5852862000465393,
      "learning_rate": 0.0015,
      "loss": 2.1901,
      "step": 695
    },
    {
      "epoch": 0.07341772151898734,
      "grad_norm": 0.7184546589851379,
      "learning_rate": 0.0015,
      "loss": 2.1666,
      "step": 696
    },
    {
      "epoch": 0.07352320675105485,
      "grad_norm": 0.6798696517944336,
      "learning_rate": 0.0015,
      "loss": 2.1779,
      "step": 697
    },
    {
      "epoch": 0.07362869198312236,
      "grad_norm": 0.5243903398513794,
      "learning_rate": 0.0015,
      "loss": 2.1461,
      "step": 698
    },
    {
      "epoch": 0.07373417721518988,
      "grad_norm": 0.5244808197021484,
      "learning_rate": 0.0015,
      "loss": 2.1401,
      "step": 699
    },
    {
      "epoch": 0.07383966244725738,
      "grad_norm": 0.5956200361251831,
      "learning_rate": 0.0015,
      "loss": 2.1455,
      "step": 700
    },
    {
      "epoch": 0.07394514767932489,
      "grad_norm": 0.6188979744911194,
      "learning_rate": 0.0015,
      "loss": 2.1398,
      "step": 701
    },
    {
      "epoch": 0.07405063291139241,
      "grad_norm": 0.5388018488883972,
      "learning_rate": 0.0015,
      "loss": 2.154,
      "step": 702
    },
    {
      "epoch": 0.07415611814345992,
      "grad_norm": 0.5440879464149475,
      "learning_rate": 0.0015,
      "loss": 2.1976,
      "step": 703
    },
    {
      "epoch": 0.07426160337552742,
      "grad_norm": 0.7074193954467773,
      "learning_rate": 0.0015,
      "loss": 2.1673,
      "step": 704
    },
    {
      "epoch": 0.07436708860759493,
      "grad_norm": 0.675395131111145,
      "learning_rate": 0.0015,
      "loss": 2.1536,
      "step": 705
    },
    {
      "epoch": 0.07447257383966245,
      "grad_norm": 0.5538883805274963,
      "learning_rate": 0.0015,
      "loss": 2.1816,
      "step": 706
    },
    {
      "epoch": 0.07457805907172996,
      "grad_norm": 0.5245557427406311,
      "learning_rate": 0.0015,
      "loss": 2.1515,
      "step": 707
    },
    {
      "epoch": 0.07468354430379746,
      "grad_norm": 0.5336805582046509,
      "learning_rate": 0.0015,
      "loss": 2.181,
      "step": 708
    },
    {
      "epoch": 0.07478902953586498,
      "grad_norm": 0.5505400896072388,
      "learning_rate": 0.0015,
      "loss": 2.1744,
      "step": 709
    },
    {
      "epoch": 0.07489451476793249,
      "grad_norm": 0.5089576840400696,
      "learning_rate": 0.0015,
      "loss": 2.1609,
      "step": 710
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.5851828455924988,
      "learning_rate": 0.0015,
      "loss": 2.1624,
      "step": 711
    },
    {
      "epoch": 0.0751054852320675,
      "grad_norm": 0.7139943838119507,
      "learning_rate": 0.0015,
      "loss": 2.1741,
      "step": 712
    },
    {
      "epoch": 0.07521097046413502,
      "grad_norm": 0.6880309581756592,
      "learning_rate": 0.0015,
      "loss": 2.1633,
      "step": 713
    },
    {
      "epoch": 0.07531645569620253,
      "grad_norm": 0.5787697434425354,
      "learning_rate": 0.0015,
      "loss": 2.1401,
      "step": 714
    },
    {
      "epoch": 0.07542194092827004,
      "grad_norm": 0.6487759351730347,
      "learning_rate": 0.0015,
      "loss": 2.1647,
      "step": 715
    },
    {
      "epoch": 0.07552742616033756,
      "grad_norm": 0.8057007789611816,
      "learning_rate": 0.0015,
      "loss": 2.1567,
      "step": 716
    },
    {
      "epoch": 0.07563291139240506,
      "grad_norm": 0.566325843334198,
      "learning_rate": 0.0015,
      "loss": 2.1354,
      "step": 717
    },
    {
      "epoch": 0.07573839662447257,
      "grad_norm": 0.6920539140701294,
      "learning_rate": 0.0015,
      "loss": 2.1069,
      "step": 718
    },
    {
      "epoch": 0.07584388185654009,
      "grad_norm": 0.7430959939956665,
      "learning_rate": 0.0015,
      "loss": 2.1486,
      "step": 719
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 0.7641369700431824,
      "learning_rate": 0.0015,
      "loss": 2.1426,
      "step": 720
    },
    {
      "epoch": 0.0760548523206751,
      "grad_norm": 0.7014267444610596,
      "learning_rate": 0.0015,
      "loss": 2.1756,
      "step": 721
    },
    {
      "epoch": 0.07616033755274261,
      "grad_norm": 0.5831820368766785,
      "learning_rate": 0.0015,
      "loss": 2.1519,
      "step": 722
    },
    {
      "epoch": 0.07626582278481013,
      "grad_norm": 0.5453060269355774,
      "learning_rate": 0.0015,
      "loss": 2.1494,
      "step": 723
    },
    {
      "epoch": 0.07637130801687764,
      "grad_norm": 0.5728681683540344,
      "learning_rate": 0.0015,
      "loss": 2.1202,
      "step": 724
    },
    {
      "epoch": 0.07647679324894514,
      "grad_norm": 0.5387490391731262,
      "learning_rate": 0.0015,
      "loss": 2.1438,
      "step": 725
    },
    {
      "epoch": 0.07658227848101266,
      "grad_norm": 0.6138736009597778,
      "learning_rate": 0.0015,
      "loss": 2.1363,
      "step": 726
    },
    {
      "epoch": 0.07668776371308017,
      "grad_norm": 0.7266629934310913,
      "learning_rate": 0.0015,
      "loss": 2.1411,
      "step": 727
    },
    {
      "epoch": 0.07679324894514768,
      "grad_norm": 0.6275491118431091,
      "learning_rate": 0.0015,
      "loss": 2.1555,
      "step": 728
    },
    {
      "epoch": 0.07689873417721518,
      "grad_norm": 0.6663520932197571,
      "learning_rate": 0.0015,
      "loss": 2.16,
      "step": 729
    },
    {
      "epoch": 0.0770042194092827,
      "grad_norm": 0.5328286290168762,
      "learning_rate": 0.0015,
      "loss": 2.126,
      "step": 730
    },
    {
      "epoch": 0.07710970464135021,
      "grad_norm": 0.5819591283798218,
      "learning_rate": 0.0015,
      "loss": 2.1471,
      "step": 731
    },
    {
      "epoch": 0.07721518987341772,
      "grad_norm": 0.5698686242103577,
      "learning_rate": 0.0015,
      "loss": 2.1406,
      "step": 732
    },
    {
      "epoch": 0.07732067510548524,
      "grad_norm": 0.542801022529602,
      "learning_rate": 0.0015,
      "loss": 2.1671,
      "step": 733
    },
    {
      "epoch": 0.07742616033755274,
      "grad_norm": 0.5827435255050659,
      "learning_rate": 0.0015,
      "loss": 2.1873,
      "step": 734
    },
    {
      "epoch": 0.07753164556962025,
      "grad_norm": 0.5806185603141785,
      "learning_rate": 0.0015,
      "loss": 2.1574,
      "step": 735
    },
    {
      "epoch": 0.07763713080168777,
      "grad_norm": 0.6766160130500793,
      "learning_rate": 0.0015,
      "loss": 2.1232,
      "step": 736
    },
    {
      "epoch": 0.07774261603375528,
      "grad_norm": 0.6849648356437683,
      "learning_rate": 0.0015,
      "loss": 2.1708,
      "step": 737
    },
    {
      "epoch": 0.07784810126582278,
      "grad_norm": 0.5955662727355957,
      "learning_rate": 0.0015,
      "loss": 2.1286,
      "step": 738
    },
    {
      "epoch": 0.07795358649789029,
      "grad_norm": 0.5330286622047424,
      "learning_rate": 0.0015,
      "loss": 2.1891,
      "step": 739
    },
    {
      "epoch": 0.07805907172995781,
      "grad_norm": 0.6239925622940063,
      "learning_rate": 0.0015,
      "loss": 2.1241,
      "step": 740
    },
    {
      "epoch": 0.07816455696202532,
      "grad_norm": 0.5498470067977905,
      "learning_rate": 0.0015,
      "loss": 2.1374,
      "step": 741
    },
    {
      "epoch": 0.07827004219409282,
      "grad_norm": 0.5445939302444458,
      "learning_rate": 0.0015,
      "loss": 2.1607,
      "step": 742
    },
    {
      "epoch": 0.07837552742616034,
      "grad_norm": 0.558857262134552,
      "learning_rate": 0.0015,
      "loss": 2.1321,
      "step": 743
    },
    {
      "epoch": 0.07848101265822785,
      "grad_norm": 0.5508167147636414,
      "learning_rate": 0.0015,
      "loss": 2.1469,
      "step": 744
    },
    {
      "epoch": 0.07858649789029536,
      "grad_norm": 0.48142459988594055,
      "learning_rate": 0.0015,
      "loss": 2.1532,
      "step": 745
    },
    {
      "epoch": 0.07869198312236286,
      "grad_norm": 0.5125391483306885,
      "learning_rate": 0.0015,
      "loss": 2.1399,
      "step": 746
    },
    {
      "epoch": 0.07879746835443038,
      "grad_norm": 0.5317972302436829,
      "learning_rate": 0.0015,
      "loss": 2.1238,
      "step": 747
    },
    {
      "epoch": 0.07890295358649789,
      "grad_norm": 0.49319085478782654,
      "learning_rate": 0.0015,
      "loss": 2.0949,
      "step": 748
    },
    {
      "epoch": 0.0790084388185654,
      "grad_norm": 0.6260822415351868,
      "learning_rate": 0.0015,
      "loss": 2.1365,
      "step": 749
    },
    {
      "epoch": 0.07911392405063292,
      "grad_norm": 0.8042246699333191,
      "learning_rate": 0.0015,
      "loss": 2.1318,
      "step": 750
    },
    {
      "epoch": 0.07921940928270042,
      "grad_norm": 0.748389482498169,
      "learning_rate": 0.0015,
      "loss": 2.1359,
      "step": 751
    },
    {
      "epoch": 0.07932489451476793,
      "grad_norm": 0.5691909790039062,
      "learning_rate": 0.0015,
      "loss": 2.161,
      "step": 752
    },
    {
      "epoch": 0.07943037974683544,
      "grad_norm": 0.5758563280105591,
      "learning_rate": 0.0015,
      "loss": 2.1311,
      "step": 753
    },
    {
      "epoch": 0.07953586497890296,
      "grad_norm": 0.8811555504798889,
      "learning_rate": 0.0015,
      "loss": 2.1346,
      "step": 754
    },
    {
      "epoch": 0.07964135021097046,
      "grad_norm": 0.9433655738830566,
      "learning_rate": 0.0015,
      "loss": 2.15,
      "step": 755
    },
    {
      "epoch": 0.07974683544303797,
      "grad_norm": 0.528934895992279,
      "learning_rate": 0.0015,
      "loss": 2.1164,
      "step": 756
    },
    {
      "epoch": 0.07985232067510549,
      "grad_norm": 0.8696373701095581,
      "learning_rate": 0.0015,
      "loss": 2.1111,
      "step": 757
    },
    {
      "epoch": 0.079957805907173,
      "grad_norm": 1.1396772861480713,
      "learning_rate": 0.0015,
      "loss": 2.1701,
      "step": 758
    },
    {
      "epoch": 0.0800632911392405,
      "grad_norm": 0.49097946286201477,
      "learning_rate": 0.0015,
      "loss": 2.1009,
      "step": 759
    },
    {
      "epoch": 0.08016877637130802,
      "grad_norm": 1.039196252822876,
      "learning_rate": 0.0015,
      "loss": 2.1188,
      "step": 760
    },
    {
      "epoch": 0.08027426160337553,
      "grad_norm": 0.916569173336029,
      "learning_rate": 0.0015,
      "loss": 2.1709,
      "step": 761
    },
    {
      "epoch": 0.08037974683544304,
      "grad_norm": 0.5536020398139954,
      "learning_rate": 0.0015,
      "loss": 2.1158,
      "step": 762
    },
    {
      "epoch": 0.08048523206751054,
      "grad_norm": 1.2422629594802856,
      "learning_rate": 0.0015,
      "loss": 2.1076,
      "step": 763
    },
    {
      "epoch": 0.08059071729957806,
      "grad_norm": 0.6607741117477417,
      "learning_rate": 0.0015,
      "loss": 2.1299,
      "step": 764
    },
    {
      "epoch": 0.08069620253164557,
      "grad_norm": 0.6373748183250427,
      "learning_rate": 0.0015,
      "loss": 2.1238,
      "step": 765
    },
    {
      "epoch": 0.08080168776371308,
      "grad_norm": 0.7773301005363464,
      "learning_rate": 0.0015,
      "loss": 2.1345,
      "step": 766
    },
    {
      "epoch": 0.0809071729957806,
      "grad_norm": 0.535194993019104,
      "learning_rate": 0.0015,
      "loss": 2.1051,
      "step": 767
    },
    {
      "epoch": 0.0810126582278481,
      "grad_norm": 0.6248049139976501,
      "learning_rate": 0.0015,
      "loss": 2.1182,
      "step": 768
    },
    {
      "epoch": 0.08111814345991561,
      "grad_norm": 0.7658226490020752,
      "learning_rate": 0.0015,
      "loss": 2.1528,
      "step": 769
    },
    {
      "epoch": 0.08122362869198312,
      "grad_norm": 0.5887392163276672,
      "learning_rate": 0.0015,
      "loss": 2.1008,
      "step": 770
    },
    {
      "epoch": 0.08132911392405064,
      "grad_norm": 0.5774050354957581,
      "learning_rate": 0.0015,
      "loss": 2.0934,
      "step": 771
    },
    {
      "epoch": 0.08143459915611814,
      "grad_norm": 0.595152735710144,
      "learning_rate": 0.0015,
      "loss": 2.1748,
      "step": 772
    },
    {
      "epoch": 0.08154008438818565,
      "grad_norm": 0.5118763446807861,
      "learning_rate": 0.0015,
      "loss": 2.1056,
      "step": 773
    },
    {
      "epoch": 0.08164556962025317,
      "grad_norm": 0.5855534076690674,
      "learning_rate": 0.0015,
      "loss": 2.1198,
      "step": 774
    },
    {
      "epoch": 0.08175105485232068,
      "grad_norm": 0.5367034673690796,
      "learning_rate": 0.0015,
      "loss": 2.0888,
      "step": 775
    },
    {
      "epoch": 0.08185654008438818,
      "grad_norm": 0.5104321837425232,
      "learning_rate": 0.0015,
      "loss": 2.0662,
      "step": 776
    },
    {
      "epoch": 0.0819620253164557,
      "grad_norm": 0.4698849320411682,
      "learning_rate": 0.0015,
      "loss": 2.1104,
      "step": 777
    },
    {
      "epoch": 0.08206751054852321,
      "grad_norm": 0.6529117226600647,
      "learning_rate": 0.0015,
      "loss": 2.1377,
      "step": 778
    },
    {
      "epoch": 0.08217299578059072,
      "grad_norm": 0.8502539396286011,
      "learning_rate": 0.0015,
      "loss": 2.1527,
      "step": 779
    },
    {
      "epoch": 0.08227848101265822,
      "grad_norm": 0.9378929138183594,
      "learning_rate": 0.0015,
      "loss": 2.1388,
      "step": 780
    },
    {
      "epoch": 0.08238396624472574,
      "grad_norm": 0.567205011844635,
      "learning_rate": 0.0015,
      "loss": 2.0874,
      "step": 781
    },
    {
      "epoch": 0.08248945147679325,
      "grad_norm": 0.8217090368270874,
      "learning_rate": 0.0015,
      "loss": 2.1141,
      "step": 782
    },
    {
      "epoch": 0.08259493670886076,
      "grad_norm": 1.048482894897461,
      "learning_rate": 0.0015,
      "loss": 2.1185,
      "step": 783
    },
    {
      "epoch": 0.08270042194092828,
      "grad_norm": 0.5079817771911621,
      "learning_rate": 0.0015,
      "loss": 2.1207,
      "step": 784
    },
    {
      "epoch": 0.08280590717299578,
      "grad_norm": 0.9008298516273499,
      "learning_rate": 0.0015,
      "loss": 2.1355,
      "step": 785
    },
    {
      "epoch": 0.08291139240506329,
      "grad_norm": 0.8511000275611877,
      "learning_rate": 0.0015,
      "loss": 2.1183,
      "step": 786
    },
    {
      "epoch": 0.0830168776371308,
      "grad_norm": 0.5419773459434509,
      "learning_rate": 0.0015,
      "loss": 2.1527,
      "step": 787
    },
    {
      "epoch": 0.08312236286919832,
      "grad_norm": 0.9313815236091614,
      "learning_rate": 0.0015,
      "loss": 2.1413,
      "step": 788
    },
    {
      "epoch": 0.08322784810126582,
      "grad_norm": 0.7607855200767517,
      "learning_rate": 0.0015,
      "loss": 2.1122,
      "step": 789
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.6888298392295837,
      "learning_rate": 0.0015,
      "loss": 2.1706,
      "step": 790
    },
    {
      "epoch": 0.08343881856540085,
      "grad_norm": 0.6333563923835754,
      "learning_rate": 0.0015,
      "loss": 2.1394,
      "step": 791
    },
    {
      "epoch": 0.08354430379746836,
      "grad_norm": 0.6713033318519592,
      "learning_rate": 0.0015,
      "loss": 2.1206,
      "step": 792
    },
    {
      "epoch": 0.08364978902953586,
      "grad_norm": 0.7254213690757751,
      "learning_rate": 0.0015,
      "loss": 2.0938,
      "step": 793
    },
    {
      "epoch": 0.08375527426160338,
      "grad_norm": 0.562282145023346,
      "learning_rate": 0.0015,
      "loss": 2.1005,
      "step": 794
    },
    {
      "epoch": 0.08386075949367089,
      "grad_norm": 0.6040173768997192,
      "learning_rate": 0.0015,
      "loss": 2.0771,
      "step": 795
    },
    {
      "epoch": 0.0839662447257384,
      "grad_norm": 0.5453705191612244,
      "learning_rate": 0.0015,
      "loss": 2.1052,
      "step": 796
    },
    {
      "epoch": 0.0840717299578059,
      "grad_norm": 0.7957865595817566,
      "learning_rate": 0.0015,
      "loss": 2.1356,
      "step": 797
    },
    {
      "epoch": 0.08417721518987342,
      "grad_norm": 0.8070213794708252,
      "learning_rate": 0.0015,
      "loss": 2.0855,
      "step": 798
    },
    {
      "epoch": 0.08428270042194093,
      "grad_norm": 0.5396027565002441,
      "learning_rate": 0.0015,
      "loss": 2.0852,
      "step": 799
    },
    {
      "epoch": 0.08438818565400844,
      "grad_norm": 0.6908420324325562,
      "learning_rate": 0.0015,
      "loss": 2.067,
      "step": 800
    },
    {
      "epoch": 0.08449367088607596,
      "grad_norm": 0.787321150302887,
      "learning_rate": 0.0015,
      "loss": 2.0946,
      "step": 801
    },
    {
      "epoch": 0.08459915611814346,
      "grad_norm": 0.5643385648727417,
      "learning_rate": 0.0015,
      "loss": 2.0993,
      "step": 802
    },
    {
      "epoch": 0.08470464135021097,
      "grad_norm": 0.7755754590034485,
      "learning_rate": 0.0015,
      "loss": 2.0839,
      "step": 803
    },
    {
      "epoch": 0.08481012658227848,
      "grad_norm": 0.7757706046104431,
      "learning_rate": 0.0015,
      "loss": 2.0479,
      "step": 804
    },
    {
      "epoch": 0.084915611814346,
      "grad_norm": 0.520271360874176,
      "learning_rate": 0.0015,
      "loss": 2.0966,
      "step": 805
    },
    {
      "epoch": 0.0850210970464135,
      "grad_norm": 0.8074872493743896,
      "learning_rate": 0.0015,
      "loss": 2.1286,
      "step": 806
    },
    {
      "epoch": 0.08512658227848101,
      "grad_norm": 0.9122884273529053,
      "learning_rate": 0.0015,
      "loss": 2.0998,
      "step": 807
    },
    {
      "epoch": 0.08523206751054853,
      "grad_norm": 0.5138601660728455,
      "learning_rate": 0.0015,
      "loss": 2.1151,
      "step": 808
    },
    {
      "epoch": 0.08533755274261604,
      "grad_norm": 0.8120249509811401,
      "learning_rate": 0.0015,
      "loss": 2.0593,
      "step": 809
    },
    {
      "epoch": 0.08544303797468354,
      "grad_norm": 0.8636590242385864,
      "learning_rate": 0.0015,
      "loss": 2.0836,
      "step": 810
    },
    {
      "epoch": 0.08554852320675105,
      "grad_norm": 0.4965047538280487,
      "learning_rate": 0.0015,
      "loss": 2.1041,
      "step": 811
    },
    {
      "epoch": 0.08565400843881857,
      "grad_norm": 0.778988778591156,
      "learning_rate": 0.0015,
      "loss": 2.0745,
      "step": 812
    },
    {
      "epoch": 0.08575949367088608,
      "grad_norm": 0.7225834727287292,
      "learning_rate": 0.0015,
      "loss": 2.1396,
      "step": 813
    },
    {
      "epoch": 0.08586497890295358,
      "grad_norm": 0.5558802485466003,
      "learning_rate": 0.0015,
      "loss": 2.1128,
      "step": 814
    },
    {
      "epoch": 0.0859704641350211,
      "grad_norm": 0.8078777194023132,
      "learning_rate": 0.0015,
      "loss": 2.0878,
      "step": 815
    },
    {
      "epoch": 0.08607594936708861,
      "grad_norm": 0.5530787706375122,
      "learning_rate": 0.0015,
      "loss": 2.0787,
      "step": 816
    },
    {
      "epoch": 0.08618143459915611,
      "grad_norm": 0.633363664150238,
      "learning_rate": 0.0015,
      "loss": 2.106,
      "step": 817
    },
    {
      "epoch": 0.08628691983122364,
      "grad_norm": 0.7179210782051086,
      "learning_rate": 0.0015,
      "loss": 2.1051,
      "step": 818
    },
    {
      "epoch": 0.08639240506329114,
      "grad_norm": 0.5488991737365723,
      "learning_rate": 0.0015,
      "loss": 2.0767,
      "step": 819
    },
    {
      "epoch": 0.08649789029535865,
      "grad_norm": 0.7378754615783691,
      "learning_rate": 0.0015,
      "loss": 2.074,
      "step": 820
    },
    {
      "epoch": 0.08660337552742615,
      "grad_norm": 0.7365741729736328,
      "learning_rate": 0.0015,
      "loss": 2.1063,
      "step": 821
    },
    {
      "epoch": 0.08670886075949367,
      "grad_norm": 0.5890570282936096,
      "learning_rate": 0.0015,
      "loss": 2.0978,
      "step": 822
    },
    {
      "epoch": 0.08681434599156118,
      "grad_norm": 0.5736057758331299,
      "learning_rate": 0.0015,
      "loss": 2.0928,
      "step": 823
    },
    {
      "epoch": 0.08691983122362869,
      "grad_norm": 0.6182436347007751,
      "learning_rate": 0.0015,
      "loss": 2.1404,
      "step": 824
    },
    {
      "epoch": 0.08702531645569621,
      "grad_norm": 0.564915657043457,
      "learning_rate": 0.0015,
      "loss": 2.0751,
      "step": 825
    },
    {
      "epoch": 0.08713080168776371,
      "grad_norm": 0.4753126800060272,
      "learning_rate": 0.0015,
      "loss": 2.1099,
      "step": 826
    },
    {
      "epoch": 0.08723628691983122,
      "grad_norm": 0.5777750015258789,
      "learning_rate": 0.0015,
      "loss": 2.1002,
      "step": 827
    },
    {
      "epoch": 0.08734177215189873,
      "grad_norm": 0.5815672874450684,
      "learning_rate": 0.0015,
      "loss": 2.0849,
      "step": 828
    },
    {
      "epoch": 0.08744725738396625,
      "grad_norm": 0.5881022214889526,
      "learning_rate": 0.0015,
      "loss": 2.0816,
      "step": 829
    },
    {
      "epoch": 0.08755274261603375,
      "grad_norm": 0.5091956257820129,
      "learning_rate": 0.0015,
      "loss": 2.0973,
      "step": 830
    },
    {
      "epoch": 0.08765822784810126,
      "grad_norm": 0.5754399299621582,
      "learning_rate": 0.0015,
      "loss": 2.052,
      "step": 831
    },
    {
      "epoch": 0.08776371308016878,
      "grad_norm": 0.4812692105770111,
      "learning_rate": 0.0015,
      "loss": 2.0468,
      "step": 832
    },
    {
      "epoch": 0.08786919831223629,
      "grad_norm": 0.5575996041297913,
      "learning_rate": 0.0015,
      "loss": 2.0753,
      "step": 833
    },
    {
      "epoch": 0.0879746835443038,
      "grad_norm": 0.797473132610321,
      "learning_rate": 0.0015,
      "loss": 2.0403,
      "step": 834
    },
    {
      "epoch": 0.08808016877637131,
      "grad_norm": 0.6777927875518799,
      "learning_rate": 0.0015,
      "loss": 2.0985,
      "step": 835
    },
    {
      "epoch": 0.08818565400843882,
      "grad_norm": 0.48958081007003784,
      "learning_rate": 0.0015,
      "loss": 2.0704,
      "step": 836
    },
    {
      "epoch": 0.08829113924050633,
      "grad_norm": 0.6332954168319702,
      "learning_rate": 0.0015,
      "loss": 2.0903,
      "step": 837
    },
    {
      "epoch": 0.08839662447257383,
      "grad_norm": 0.6067173480987549,
      "learning_rate": 0.0015,
      "loss": 2.1112,
      "step": 838
    },
    {
      "epoch": 0.08850210970464135,
      "grad_norm": 0.4982474744319916,
      "learning_rate": 0.0015,
      "loss": 2.0722,
      "step": 839
    },
    {
      "epoch": 0.08860759493670886,
      "grad_norm": 0.6219644546508789,
      "learning_rate": 0.0015,
      "loss": 2.0967,
      "step": 840
    },
    {
      "epoch": 0.08871308016877637,
      "grad_norm": 0.7281220555305481,
      "learning_rate": 0.0015,
      "loss": 2.0884,
      "step": 841
    },
    {
      "epoch": 0.08881856540084389,
      "grad_norm": 0.5716444849967957,
      "learning_rate": 0.0015,
      "loss": 2.0399,
      "step": 842
    },
    {
      "epoch": 0.0889240506329114,
      "grad_norm": 0.5442397594451904,
      "learning_rate": 0.0015,
      "loss": 2.0879,
      "step": 843
    },
    {
      "epoch": 0.0890295358649789,
      "grad_norm": 0.6892992854118347,
      "learning_rate": 0.0015,
      "loss": 2.0685,
      "step": 844
    },
    {
      "epoch": 0.08913502109704641,
      "grad_norm": 0.578584611415863,
      "learning_rate": 0.0015,
      "loss": 2.0866,
      "step": 845
    },
    {
      "epoch": 0.08924050632911393,
      "grad_norm": 0.6582306623458862,
      "learning_rate": 0.0015,
      "loss": 2.1037,
      "step": 846
    },
    {
      "epoch": 0.08934599156118143,
      "grad_norm": 0.6320885419845581,
      "learning_rate": 0.0015,
      "loss": 2.0964,
      "step": 847
    },
    {
      "epoch": 0.08945147679324894,
      "grad_norm": 0.6004634499549866,
      "learning_rate": 0.0015,
      "loss": 2.0891,
      "step": 848
    },
    {
      "epoch": 0.08955696202531646,
      "grad_norm": 0.5187250971794128,
      "learning_rate": 0.0015,
      "loss": 2.054,
      "step": 849
    },
    {
      "epoch": 0.08966244725738397,
      "grad_norm": 0.5002400875091553,
      "learning_rate": 0.0015,
      "loss": 2.0343,
      "step": 850
    },
    {
      "epoch": 0.08976793248945147,
      "grad_norm": 0.5926933884620667,
      "learning_rate": 0.0015,
      "loss": 2.0767,
      "step": 851
    },
    {
      "epoch": 0.08987341772151898,
      "grad_norm": 0.5073577165603638,
      "learning_rate": 0.0015,
      "loss": 2.091,
      "step": 852
    },
    {
      "epoch": 0.0899789029535865,
      "grad_norm": 0.5054778456687927,
      "learning_rate": 0.0015,
      "loss": 2.0622,
      "step": 853
    },
    {
      "epoch": 0.09008438818565401,
      "grad_norm": 0.5435723066329956,
      "learning_rate": 0.0015,
      "loss": 2.1132,
      "step": 854
    },
    {
      "epoch": 0.09018987341772151,
      "grad_norm": 0.4906049370765686,
      "learning_rate": 0.0015,
      "loss": 2.0811,
      "step": 855
    },
    {
      "epoch": 0.09029535864978903,
      "grad_norm": 0.5326725840568542,
      "learning_rate": 0.0015,
      "loss": 2.0879,
      "step": 856
    },
    {
      "epoch": 0.09040084388185654,
      "grad_norm": 0.5024213790893555,
      "learning_rate": 0.0015,
      "loss": 2.0634,
      "step": 857
    },
    {
      "epoch": 0.09050632911392405,
      "grad_norm": 0.6202135682106018,
      "learning_rate": 0.0015,
      "loss": 2.0841,
      "step": 858
    },
    {
      "epoch": 0.09061181434599157,
      "grad_norm": 0.6742455959320068,
      "learning_rate": 0.0015,
      "loss": 2.0819,
      "step": 859
    },
    {
      "epoch": 0.09071729957805907,
      "grad_norm": 0.6315163373947144,
      "learning_rate": 0.0015,
      "loss": 2.0284,
      "step": 860
    },
    {
      "epoch": 0.09082278481012658,
      "grad_norm": 0.6156589388847351,
      "learning_rate": 0.0015,
      "loss": 2.0886,
      "step": 861
    },
    {
      "epoch": 0.09092827004219409,
      "grad_norm": 0.4940195381641388,
      "learning_rate": 0.0015,
      "loss": 2.0542,
      "step": 862
    },
    {
      "epoch": 0.09103375527426161,
      "grad_norm": 0.5877742767333984,
      "learning_rate": 0.0015,
      "loss": 2.0654,
      "step": 863
    },
    {
      "epoch": 0.09113924050632911,
      "grad_norm": 0.5197194814682007,
      "learning_rate": 0.0015,
      "loss": 2.0727,
      "step": 864
    },
    {
      "epoch": 0.09124472573839662,
      "grad_norm": 0.5381242632865906,
      "learning_rate": 0.0015,
      "loss": 2.0749,
      "step": 865
    },
    {
      "epoch": 0.09135021097046414,
      "grad_norm": 0.5649956464767456,
      "learning_rate": 0.0015,
      "loss": 2.0322,
      "step": 866
    },
    {
      "epoch": 0.09145569620253165,
      "grad_norm": 0.5806474089622498,
      "learning_rate": 0.0015,
      "loss": 2.0715,
      "step": 867
    },
    {
      "epoch": 0.09156118143459915,
      "grad_norm": 0.5830804109573364,
      "learning_rate": 0.0015,
      "loss": 2.1055,
      "step": 868
    },
    {
      "epoch": 0.09166666666666666,
      "grad_norm": 0.47413384914398193,
      "learning_rate": 0.0015,
      "loss": 2.0386,
      "step": 869
    },
    {
      "epoch": 0.09177215189873418,
      "grad_norm": 0.5385271906852722,
      "learning_rate": 0.0015,
      "loss": 2.0903,
      "step": 870
    },
    {
      "epoch": 0.09187763713080169,
      "grad_norm": 0.5487086772918701,
      "learning_rate": 0.0015,
      "loss": 2.073,
      "step": 871
    },
    {
      "epoch": 0.0919831223628692,
      "grad_norm": 0.5364266633987427,
      "learning_rate": 0.0015,
      "loss": 2.0621,
      "step": 872
    },
    {
      "epoch": 0.09208860759493671,
      "grad_norm": 0.5214003324508667,
      "learning_rate": 0.0015,
      "loss": 2.0578,
      "step": 873
    },
    {
      "epoch": 0.09219409282700422,
      "grad_norm": 0.5408740639686584,
      "learning_rate": 0.0015,
      "loss": 2.0689,
      "step": 874
    },
    {
      "epoch": 0.09229957805907173,
      "grad_norm": 0.5028771758079529,
      "learning_rate": 0.0015,
      "loss": 2.0566,
      "step": 875
    },
    {
      "epoch": 0.09240506329113925,
      "grad_norm": 0.46585795283317566,
      "learning_rate": 0.0015,
      "loss": 2.0709,
      "step": 876
    },
    {
      "epoch": 0.09251054852320675,
      "grad_norm": 0.5307130217552185,
      "learning_rate": 0.0015,
      "loss": 2.0632,
      "step": 877
    },
    {
      "epoch": 0.09261603375527426,
      "grad_norm": 0.47115400433540344,
      "learning_rate": 0.0015,
      "loss": 2.0358,
      "step": 878
    },
    {
      "epoch": 0.09272151898734177,
      "grad_norm": 0.5778770446777344,
      "learning_rate": 0.0015,
      "loss": 2.0956,
      "step": 879
    },
    {
      "epoch": 0.09282700421940929,
      "grad_norm": 0.6261560320854187,
      "learning_rate": 0.0015,
      "loss": 2.0703,
      "step": 880
    },
    {
      "epoch": 0.0929324894514768,
      "grad_norm": 0.5335504412651062,
      "learning_rate": 0.0015,
      "loss": 2.0924,
      "step": 881
    },
    {
      "epoch": 0.0930379746835443,
      "grad_norm": 0.5628005266189575,
      "learning_rate": 0.0015,
      "loss": 2.0395,
      "step": 882
    },
    {
      "epoch": 0.09314345991561182,
      "grad_norm": 0.7454283833503723,
      "learning_rate": 0.0015,
      "loss": 2.0085,
      "step": 883
    },
    {
      "epoch": 0.09324894514767933,
      "grad_norm": 0.6037567257881165,
      "learning_rate": 0.0015,
      "loss": 2.0398,
      "step": 884
    },
    {
      "epoch": 0.09335443037974683,
      "grad_norm": 0.5915200710296631,
      "learning_rate": 0.0015,
      "loss": 2.103,
      "step": 885
    },
    {
      "epoch": 0.09345991561181434,
      "grad_norm": 0.8386354446411133,
      "learning_rate": 0.0015,
      "loss": 2.0275,
      "step": 886
    },
    {
      "epoch": 0.09356540084388186,
      "grad_norm": 0.9508600831031799,
      "learning_rate": 0.0015,
      "loss": 2.0471,
      "step": 887
    },
    {
      "epoch": 0.09367088607594937,
      "grad_norm": 0.7215097546577454,
      "learning_rate": 0.0015,
      "loss": 2.0822,
      "step": 888
    },
    {
      "epoch": 0.09377637130801687,
      "grad_norm": 0.7270636558532715,
      "learning_rate": 0.0015,
      "loss": 2.0547,
      "step": 889
    },
    {
      "epoch": 0.0938818565400844,
      "grad_norm": 0.5731157660484314,
      "learning_rate": 0.0015,
      "loss": 2.0668,
      "step": 890
    },
    {
      "epoch": 0.0939873417721519,
      "grad_norm": 0.6870090365409851,
      "learning_rate": 0.0015,
      "loss": 2.1074,
      "step": 891
    },
    {
      "epoch": 0.0940928270042194,
      "grad_norm": 0.6307251453399658,
      "learning_rate": 0.0015,
      "loss": 2.0586,
      "step": 892
    },
    {
      "epoch": 0.09419831223628691,
      "grad_norm": 0.5434120893478394,
      "learning_rate": 0.0015,
      "loss": 2.0349,
      "step": 893
    },
    {
      "epoch": 0.09430379746835443,
      "grad_norm": 0.5502089858055115,
      "learning_rate": 0.0015,
      "loss": 2.0808,
      "step": 894
    },
    {
      "epoch": 0.09440928270042194,
      "grad_norm": 0.5304689407348633,
      "learning_rate": 0.0015,
      "loss": 2.035,
      "step": 895
    },
    {
      "epoch": 0.09451476793248945,
      "grad_norm": 0.5041806101799011,
      "learning_rate": 0.0015,
      "loss": 2.0407,
      "step": 896
    },
    {
      "epoch": 0.09462025316455697,
      "grad_norm": 0.5043999552726746,
      "learning_rate": 0.0015,
      "loss": 2.0427,
      "step": 897
    },
    {
      "epoch": 0.09472573839662447,
      "grad_norm": 0.6511165499687195,
      "learning_rate": 0.0015,
      "loss": 2.0476,
      "step": 898
    },
    {
      "epoch": 0.09483122362869198,
      "grad_norm": 0.6559220552444458,
      "learning_rate": 0.0015,
      "loss": 2.0277,
      "step": 899
    },
    {
      "epoch": 0.0949367088607595,
      "grad_norm": 0.6053939461708069,
      "learning_rate": 0.0015,
      "loss": 2.05,
      "step": 900
    },
    {
      "epoch": 0.095042194092827,
      "grad_norm": 0.5160669684410095,
      "learning_rate": 0.0015,
      "loss": 2.0707,
      "step": 901
    },
    {
      "epoch": 0.09514767932489451,
      "grad_norm": 0.5985272526741028,
      "learning_rate": 0.0015,
      "loss": 2.0565,
      "step": 902
    },
    {
      "epoch": 0.09525316455696202,
      "grad_norm": 0.6856181621551514,
      "learning_rate": 0.0015,
      "loss": 2.0514,
      "step": 903
    },
    {
      "epoch": 0.09535864978902954,
      "grad_norm": 0.6011475920677185,
      "learning_rate": 0.0015,
      "loss": 2.0498,
      "step": 904
    },
    {
      "epoch": 0.09546413502109705,
      "grad_norm": 0.6291939616203308,
      "learning_rate": 0.0015,
      "loss": 2.0307,
      "step": 905
    },
    {
      "epoch": 0.09556962025316455,
      "grad_norm": 0.7939316630363464,
      "learning_rate": 0.0015,
      "loss": 2.0714,
      "step": 906
    },
    {
      "epoch": 0.09567510548523207,
      "grad_norm": 0.8090095520019531,
      "learning_rate": 0.0015,
      "loss": 2.0856,
      "step": 907
    },
    {
      "epoch": 0.09578059071729958,
      "grad_norm": 0.7017444968223572,
      "learning_rate": 0.0015,
      "loss": 2.0333,
      "step": 908
    },
    {
      "epoch": 0.09588607594936709,
      "grad_norm": 0.47305548191070557,
      "learning_rate": 0.0015,
      "loss": 2.0538,
      "step": 909
    },
    {
      "epoch": 0.09599156118143459,
      "grad_norm": 0.7019219398498535,
      "learning_rate": 0.0015,
      "loss": 2.1068,
      "step": 910
    },
    {
      "epoch": 0.09609704641350211,
      "grad_norm": 0.6788797974586487,
      "learning_rate": 0.0015,
      "loss": 2.0596,
      "step": 911
    },
    {
      "epoch": 0.09620253164556962,
      "grad_norm": 0.5691693425178528,
      "learning_rate": 0.0015,
      "loss": 2.0453,
      "step": 912
    },
    {
      "epoch": 0.09630801687763713,
      "grad_norm": 0.6917991638183594,
      "learning_rate": 0.0015,
      "loss": 2.039,
      "step": 913
    },
    {
      "epoch": 0.09641350210970465,
      "grad_norm": 0.5481536388397217,
      "learning_rate": 0.0015,
      "loss": 2.0631,
      "step": 914
    },
    {
      "epoch": 0.09651898734177215,
      "grad_norm": 0.5559115409851074,
      "learning_rate": 0.0015,
      "loss": 2.0885,
      "step": 915
    },
    {
      "epoch": 0.09662447257383966,
      "grad_norm": 0.6355902552604675,
      "learning_rate": 0.0015,
      "loss": 2.0648,
      "step": 916
    },
    {
      "epoch": 0.09672995780590718,
      "grad_norm": 0.6880175471305847,
      "learning_rate": 0.0015,
      "loss": 2.0277,
      "step": 917
    },
    {
      "epoch": 0.09683544303797469,
      "grad_norm": 0.650114119052887,
      "learning_rate": 0.0015,
      "loss": 2.0384,
      "step": 918
    },
    {
      "epoch": 0.09694092827004219,
      "grad_norm": 0.5070211291313171,
      "learning_rate": 0.0015,
      "loss": 2.0686,
      "step": 919
    },
    {
      "epoch": 0.0970464135021097,
      "grad_norm": 0.559771716594696,
      "learning_rate": 0.0015,
      "loss": 2.0018,
      "step": 920
    },
    {
      "epoch": 0.09715189873417722,
      "grad_norm": 0.6372707486152649,
      "learning_rate": 0.0015,
      "loss": 2.0719,
      "step": 921
    },
    {
      "epoch": 0.09725738396624473,
      "grad_norm": 0.44606539607048035,
      "learning_rate": 0.0015,
      "loss": 2.0349,
      "step": 922
    },
    {
      "epoch": 0.09736286919831223,
      "grad_norm": 0.583096444606781,
      "learning_rate": 0.0015,
      "loss": 2.0134,
      "step": 923
    },
    {
      "epoch": 0.09746835443037975,
      "grad_norm": 0.5403363108634949,
      "learning_rate": 0.0015,
      "loss": 2.0489,
      "step": 924
    },
    {
      "epoch": 0.09757383966244726,
      "grad_norm": 0.47966498136520386,
      "learning_rate": 0.0015,
      "loss": 2.0077,
      "step": 925
    },
    {
      "epoch": 0.09767932489451477,
      "grad_norm": 0.6291763782501221,
      "learning_rate": 0.0015,
      "loss": 2.0395,
      "step": 926
    },
    {
      "epoch": 0.09778481012658227,
      "grad_norm": 0.575806736946106,
      "learning_rate": 0.0015,
      "loss": 2.0404,
      "step": 927
    },
    {
      "epoch": 0.09789029535864979,
      "grad_norm": 0.4905599355697632,
      "learning_rate": 0.0015,
      "loss": 1.9916,
      "step": 928
    },
    {
      "epoch": 0.0979957805907173,
      "grad_norm": 0.461168110370636,
      "learning_rate": 0.0015,
      "loss": 2.045,
      "step": 929
    },
    {
      "epoch": 0.0981012658227848,
      "grad_norm": 0.4989011883735657,
      "learning_rate": 0.0015,
      "loss": 2.042,
      "step": 930
    },
    {
      "epoch": 0.09820675105485233,
      "grad_norm": 0.6231623888015747,
      "learning_rate": 0.0015,
      "loss": 2.0148,
      "step": 931
    },
    {
      "epoch": 0.09831223628691983,
      "grad_norm": 0.7172541618347168,
      "learning_rate": 0.0015,
      "loss": 2.0736,
      "step": 932
    },
    {
      "epoch": 0.09841772151898734,
      "grad_norm": 0.7043025493621826,
      "learning_rate": 0.0015,
      "loss": 2.0488,
      "step": 933
    },
    {
      "epoch": 0.09852320675105486,
      "grad_norm": 0.5088157057762146,
      "learning_rate": 0.0015,
      "loss": 2.0493,
      "step": 934
    },
    {
      "epoch": 0.09862869198312237,
      "grad_norm": 0.5786383748054504,
      "learning_rate": 0.0015,
      "loss": 2.0489,
      "step": 935
    },
    {
      "epoch": 0.09873417721518987,
      "grad_norm": 0.7820794582366943,
      "learning_rate": 0.0015,
      "loss": 2.0251,
      "step": 936
    },
    {
      "epoch": 0.09883966244725738,
      "grad_norm": 0.7000012993812561,
      "learning_rate": 0.0015,
      "loss": 2.0558,
      "step": 937
    },
    {
      "epoch": 0.0989451476793249,
      "grad_norm": 0.5158394575119019,
      "learning_rate": 0.0015,
      "loss": 2.0317,
      "step": 938
    },
    {
      "epoch": 0.0990506329113924,
      "grad_norm": 0.5724575519561768,
      "learning_rate": 0.0015,
      "loss": 2.0387,
      "step": 939
    },
    {
      "epoch": 0.09915611814345991,
      "grad_norm": 0.6447023749351501,
      "learning_rate": 0.0015,
      "loss": 2.0268,
      "step": 940
    },
    {
      "epoch": 0.09926160337552743,
      "grad_norm": 0.603764533996582,
      "learning_rate": 0.0015,
      "loss": 2.0097,
      "step": 941
    },
    {
      "epoch": 0.09936708860759494,
      "grad_norm": 0.49410590529441833,
      "learning_rate": 0.0015,
      "loss": 2.0437,
      "step": 942
    },
    {
      "epoch": 0.09947257383966245,
      "grad_norm": 0.5381091833114624,
      "learning_rate": 0.0015,
      "loss": 2.0101,
      "step": 943
    },
    {
      "epoch": 0.09957805907172995,
      "grad_norm": 0.6123061180114746,
      "learning_rate": 0.0015,
      "loss": 1.9906,
      "step": 944
    },
    {
      "epoch": 0.09968354430379747,
      "grad_norm": 0.5386263132095337,
      "learning_rate": 0.0015,
      "loss": 2.0218,
      "step": 945
    },
    {
      "epoch": 0.09978902953586498,
      "grad_norm": 0.4787702262401581,
      "learning_rate": 0.0015,
      "loss": 2.019,
      "step": 946
    },
    {
      "epoch": 0.09989451476793249,
      "grad_norm": 0.5042153000831604,
      "learning_rate": 0.0015,
      "loss": 1.9806,
      "step": 947
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4878217875957489,
      "learning_rate": 0.0015,
      "loss": 2.0517,
      "step": 948
    },
    {
      "epoch": 0.10010548523206751,
      "grad_norm": 0.47430136799812317,
      "learning_rate": 0.0015,
      "loss": 2.0142,
      "step": 949
    },
    {
      "epoch": 0.10021097046413502,
      "grad_norm": 0.5746744871139526,
      "learning_rate": 0.0015,
      "loss": 2.0445,
      "step": 950
    },
    {
      "epoch": 0.10031645569620253,
      "grad_norm": 0.5339949727058411,
      "learning_rate": 0.0015,
      "loss": 2.0465,
      "step": 951
    },
    {
      "epoch": 0.10042194092827005,
      "grad_norm": 0.5987122654914856,
      "learning_rate": 0.0015,
      "loss": 2.0319,
      "step": 952
    },
    {
      "epoch": 0.10052742616033755,
      "grad_norm": 0.6438586711883545,
      "learning_rate": 0.0015,
      "loss": 2.055,
      "step": 953
    },
    {
      "epoch": 0.10063291139240506,
      "grad_norm": 0.6770201921463013,
      "learning_rate": 0.0015,
      "loss": 2.0432,
      "step": 954
    },
    {
      "epoch": 0.10073839662447258,
      "grad_norm": 0.5997467637062073,
      "learning_rate": 0.0015,
      "loss": 2.0345,
      "step": 955
    },
    {
      "epoch": 0.10084388185654009,
      "grad_norm": 0.5038588047027588,
      "learning_rate": 0.0015,
      "loss": 2.0118,
      "step": 956
    },
    {
      "epoch": 0.10094936708860759,
      "grad_norm": 0.6786378622055054,
      "learning_rate": 0.0015,
      "loss": 2.0024,
      "step": 957
    },
    {
      "epoch": 0.10105485232067511,
      "grad_norm": 0.9915666580200195,
      "learning_rate": 0.0015,
      "loss": 2.0532,
      "step": 958
    },
    {
      "epoch": 0.10116033755274262,
      "grad_norm": 0.8501437306404114,
      "learning_rate": 0.0015,
      "loss": 1.9929,
      "step": 959
    },
    {
      "epoch": 0.10126582278481013,
      "grad_norm": 0.5623571872711182,
      "learning_rate": 0.0015,
      "loss": 2.012,
      "step": 960
    },
    {
      "epoch": 0.10137130801687763,
      "grad_norm": 0.5599110126495361,
      "learning_rate": 0.0015,
      "loss": 2.0275,
      "step": 961
    },
    {
      "epoch": 0.10147679324894515,
      "grad_norm": 0.72342449426651,
      "learning_rate": 0.0015,
      "loss": 2.0046,
      "step": 962
    },
    {
      "epoch": 0.10158227848101266,
      "grad_norm": 0.6908513903617859,
      "learning_rate": 0.0015,
      "loss": 2.0556,
      "step": 963
    },
    {
      "epoch": 0.10168776371308016,
      "grad_norm": 0.531358540058136,
      "learning_rate": 0.0015,
      "loss": 2.0225,
      "step": 964
    },
    {
      "epoch": 0.10179324894514769,
      "grad_norm": 0.5810999274253845,
      "learning_rate": 0.0015,
      "loss": 2.0637,
      "step": 965
    },
    {
      "epoch": 0.10189873417721519,
      "grad_norm": 0.7688680291175842,
      "learning_rate": 0.0015,
      "loss": 2.0004,
      "step": 966
    },
    {
      "epoch": 0.1020042194092827,
      "grad_norm": 0.7815285921096802,
      "learning_rate": 0.0015,
      "loss": 2.0317,
      "step": 967
    },
    {
      "epoch": 0.1021097046413502,
      "grad_norm": 0.5666143894195557,
      "learning_rate": 0.0015,
      "loss": 2.0622,
      "step": 968
    },
    {
      "epoch": 0.10221518987341772,
      "grad_norm": 0.821260929107666,
      "learning_rate": 0.0015,
      "loss": 2.0122,
      "step": 969
    },
    {
      "epoch": 0.10232067510548523,
      "grad_norm": 1.2794028520584106,
      "learning_rate": 0.0015,
      "loss": 2.0467,
      "step": 970
    },
    {
      "epoch": 0.10242616033755274,
      "grad_norm": 0.5275937914848328,
      "learning_rate": 0.0015,
      "loss": 2.002,
      "step": 971
    },
    {
      "epoch": 0.10253164556962026,
      "grad_norm": 0.9917479753494263,
      "learning_rate": 0.0015,
      "loss": 2.0123,
      "step": 972
    },
    {
      "epoch": 0.10263713080168776,
      "grad_norm": 1.0854936838150024,
      "learning_rate": 0.0015,
      "loss": 2.0247,
      "step": 973
    },
    {
      "epoch": 0.10274261603375527,
      "grad_norm": 0.5119172930717468,
      "learning_rate": 0.0015,
      "loss": 2.0184,
      "step": 974
    },
    {
      "epoch": 0.10284810126582279,
      "grad_norm": 0.7493650317192078,
      "learning_rate": 0.0015,
      "loss": 2.0069,
      "step": 975
    },
    {
      "epoch": 0.1029535864978903,
      "grad_norm": 0.7433720827102661,
      "learning_rate": 0.0015,
      "loss": 2.0068,
      "step": 976
    },
    {
      "epoch": 0.1030590717299578,
      "grad_norm": 0.5080604553222656,
      "learning_rate": 0.0015,
      "loss": 2.0213,
      "step": 977
    },
    {
      "epoch": 0.10316455696202531,
      "grad_norm": 0.6605928540229797,
      "learning_rate": 0.0015,
      "loss": 2.0108,
      "step": 978
    },
    {
      "epoch": 0.10327004219409283,
      "grad_norm": 0.7366412878036499,
      "learning_rate": 0.0015,
      "loss": 1.9982,
      "step": 979
    },
    {
      "epoch": 0.10337552742616034,
      "grad_norm": 0.5461682677268982,
      "learning_rate": 0.0015,
      "loss": 2.0353,
      "step": 980
    },
    {
      "epoch": 0.10348101265822784,
      "grad_norm": 0.5941941142082214,
      "learning_rate": 0.0015,
      "loss": 2.0194,
      "step": 981
    },
    {
      "epoch": 0.10358649789029536,
      "grad_norm": 0.6931915283203125,
      "learning_rate": 0.0015,
      "loss": 2.0428,
      "step": 982
    },
    {
      "epoch": 0.10369198312236287,
      "grad_norm": 0.4936782419681549,
      "learning_rate": 0.0015,
      "loss": 1.9833,
      "step": 983
    },
    {
      "epoch": 0.10379746835443038,
      "grad_norm": 0.641244113445282,
      "learning_rate": 0.0015,
      "loss": 1.9947,
      "step": 984
    },
    {
      "epoch": 0.10390295358649788,
      "grad_norm": 0.8523558378219604,
      "learning_rate": 0.0015,
      "loss": 2.0132,
      "step": 985
    },
    {
      "epoch": 0.1040084388185654,
      "grad_norm": 0.5719506740570068,
      "learning_rate": 0.0015,
      "loss": 2.0082,
      "step": 986
    },
    {
      "epoch": 0.10411392405063291,
      "grad_norm": 0.6469573974609375,
      "learning_rate": 0.0015,
      "loss": 2.0403,
      "step": 987
    },
    {
      "epoch": 0.10421940928270042,
      "grad_norm": 0.8099203109741211,
      "learning_rate": 0.0015,
      "loss": 1.9987,
      "step": 988
    },
    {
      "epoch": 0.10432489451476794,
      "grad_norm": 0.5929587483406067,
      "learning_rate": 0.0015,
      "loss": 1.9888,
      "step": 989
    },
    {
      "epoch": 0.10443037974683544,
      "grad_norm": 0.5847246050834656,
      "learning_rate": 0.0015,
      "loss": 2.0016,
      "step": 990
    },
    {
      "epoch": 0.10453586497890295,
      "grad_norm": 0.7067970037460327,
      "learning_rate": 0.0015,
      "loss": 2.0808,
      "step": 991
    },
    {
      "epoch": 0.10464135021097046,
      "grad_norm": 0.5860495567321777,
      "learning_rate": 0.0015,
      "loss": 2.012,
      "step": 992
    },
    {
      "epoch": 0.10474683544303798,
      "grad_norm": 0.5571528077125549,
      "learning_rate": 0.0015,
      "loss": 1.9845,
      "step": 993
    },
    {
      "epoch": 0.10485232067510548,
      "grad_norm": 0.6108555197715759,
      "learning_rate": 0.0015,
      "loss": 2.035,
      "step": 994
    },
    {
      "epoch": 0.10495780590717299,
      "grad_norm": 0.5549131035804749,
      "learning_rate": 0.0015,
      "loss": 1.9854,
      "step": 995
    },
    {
      "epoch": 0.10506329113924051,
      "grad_norm": 0.48402878642082214,
      "learning_rate": 0.0015,
      "loss": 2.0152,
      "step": 996
    },
    {
      "epoch": 0.10516877637130802,
      "grad_norm": 0.5344575643539429,
      "learning_rate": 0.0015,
      "loss": 1.9893,
      "step": 997
    },
    {
      "epoch": 0.10527426160337552,
      "grad_norm": 0.5143426656723022,
      "learning_rate": 0.0015,
      "loss": 2.0294,
      "step": 998
    },
    {
      "epoch": 0.10537974683544304,
      "grad_norm": 0.5781170129776001,
      "learning_rate": 0.0015,
      "loss": 2.0395,
      "step": 999
    },
    {
      "epoch": 0.10548523206751055,
      "grad_norm": 0.4923545718193054,
      "learning_rate": 0.0015,
      "loss": 2.0296,
      "step": 1000
    },
    {
      "epoch": 0.10559071729957806,
      "grad_norm": 0.6493373513221741,
      "learning_rate": 0.0015,
      "loss": 1.9926,
      "step": 1001
    },
    {
      "epoch": 0.10569620253164556,
      "grad_norm": 0.7961377501487732,
      "learning_rate": 0.0015,
      "loss": 2.0014,
      "step": 1002
    },
    {
      "epoch": 0.10580168776371308,
      "grad_norm": 0.5933529734611511,
      "learning_rate": 0.0015,
      "loss": 2.0503,
      "step": 1003
    },
    {
      "epoch": 0.10590717299578059,
      "grad_norm": 0.5460384488105774,
      "learning_rate": 0.0015,
      "loss": 1.9732,
      "step": 1004
    },
    {
      "epoch": 0.1060126582278481,
      "grad_norm": 0.8536837697029114,
      "learning_rate": 0.0015,
      "loss": 1.9999,
      "step": 1005
    },
    {
      "epoch": 0.10611814345991562,
      "grad_norm": 0.7791167497634888,
      "learning_rate": 0.0015,
      "loss": 2.014,
      "step": 1006
    },
    {
      "epoch": 0.10622362869198312,
      "grad_norm": 0.49463993310928345,
      "learning_rate": 0.0015,
      "loss": 2.0405,
      "step": 1007
    },
    {
      "epoch": 0.10632911392405063,
      "grad_norm": 0.6463145613670349,
      "learning_rate": 0.0015,
      "loss": 2.0008,
      "step": 1008
    },
    {
      "epoch": 0.10643459915611814,
      "grad_norm": 0.7075051665306091,
      "learning_rate": 0.0015,
      "loss": 2.0169,
      "step": 1009
    },
    {
      "epoch": 0.10654008438818566,
      "grad_norm": 0.5285488963127136,
      "learning_rate": 0.0015,
      "loss": 2.0235,
      "step": 1010
    },
    {
      "epoch": 0.10664556962025316,
      "grad_norm": 0.6681978702545166,
      "learning_rate": 0.0015,
      "loss": 1.9893,
      "step": 1011
    },
    {
      "epoch": 0.10675105485232067,
      "grad_norm": 0.7121663689613342,
      "learning_rate": 0.0015,
      "loss": 1.9869,
      "step": 1012
    },
    {
      "epoch": 0.10685654008438819,
      "grad_norm": 0.5381683111190796,
      "learning_rate": 0.0015,
      "loss": 1.9709,
      "step": 1013
    },
    {
      "epoch": 0.1069620253164557,
      "grad_norm": 0.5448917150497437,
      "learning_rate": 0.0015,
      "loss": 2.0114,
      "step": 1014
    },
    {
      "epoch": 0.1070675105485232,
      "grad_norm": 0.5569539666175842,
      "learning_rate": 0.0015,
      "loss": 1.993,
      "step": 1015
    },
    {
      "epoch": 0.10717299578059072,
      "grad_norm": 0.48897144198417664,
      "learning_rate": 0.0015,
      "loss": 1.9559,
      "step": 1016
    },
    {
      "epoch": 0.10727848101265823,
      "grad_norm": 0.5529647469520569,
      "learning_rate": 0.0015,
      "loss": 2.0235,
      "step": 1017
    },
    {
      "epoch": 0.10738396624472574,
      "grad_norm": 0.5180500149726868,
      "learning_rate": 0.0015,
      "loss": 2.0314,
      "step": 1018
    },
    {
      "epoch": 0.10748945147679324,
      "grad_norm": 0.7292158603668213,
      "learning_rate": 0.0015,
      "loss": 2.0739,
      "step": 1019
    },
    {
      "epoch": 0.10759493670886076,
      "grad_norm": 0.9114202857017517,
      "learning_rate": 0.0015,
      "loss": 1.9931,
      "step": 1020
    },
    {
      "epoch": 0.10770042194092827,
      "grad_norm": 0.6082236766815186,
      "learning_rate": 0.0015,
      "loss": 1.9947,
      "step": 1021
    },
    {
      "epoch": 0.10780590717299578,
      "grad_norm": 0.5877995491027832,
      "learning_rate": 0.0015,
      "loss": 2.0045,
      "step": 1022
    },
    {
      "epoch": 0.1079113924050633,
      "grad_norm": 0.7676417231559753,
      "learning_rate": 0.0015,
      "loss": 1.9798,
      "step": 1023
    },
    {
      "epoch": 0.1080168776371308,
      "grad_norm": 0.6970999240875244,
      "learning_rate": 0.0015,
      "loss": 2.012,
      "step": 1024
    },
    {
      "epoch": 0.10812236286919831,
      "grad_norm": 0.5083110928535461,
      "learning_rate": 0.0015,
      "loss": 2.0343,
      "step": 1025
    },
    {
      "epoch": 0.10822784810126582,
      "grad_norm": 0.7133620381355286,
      "learning_rate": 0.0015,
      "loss": 1.9744,
      "step": 1026
    },
    {
      "epoch": 0.10833333333333334,
      "grad_norm": 0.6370148062705994,
      "learning_rate": 0.0015,
      "loss": 2.0279,
      "step": 1027
    },
    {
      "epoch": 0.10843881856540084,
      "grad_norm": 0.4984666109085083,
      "learning_rate": 0.0015,
      "loss": 1.9719,
      "step": 1028
    },
    {
      "epoch": 0.10854430379746835,
      "grad_norm": 0.7172163724899292,
      "learning_rate": 0.0015,
      "loss": 2.0103,
      "step": 1029
    },
    {
      "epoch": 0.10864978902953587,
      "grad_norm": 0.5677787065505981,
      "learning_rate": 0.0015,
      "loss": 2.0073,
      "step": 1030
    },
    {
      "epoch": 0.10875527426160338,
      "grad_norm": 0.5745548009872437,
      "learning_rate": 0.0015,
      "loss": 1.9782,
      "step": 1031
    },
    {
      "epoch": 0.10886075949367088,
      "grad_norm": 0.7175948619842529,
      "learning_rate": 0.0015,
      "loss": 1.966,
      "step": 1032
    },
    {
      "epoch": 0.10896624472573839,
      "grad_norm": 0.5983565449714661,
      "learning_rate": 0.0015,
      "loss": 1.9896,
      "step": 1033
    },
    {
      "epoch": 0.10907172995780591,
      "grad_norm": 0.5486365556716919,
      "learning_rate": 0.0015,
      "loss": 2.0331,
      "step": 1034
    },
    {
      "epoch": 0.10917721518987342,
      "grad_norm": 0.5033898949623108,
      "learning_rate": 0.0015,
      "loss": 2.0143,
      "step": 1035
    },
    {
      "epoch": 0.10928270042194092,
      "grad_norm": 0.5592193603515625,
      "learning_rate": 0.0015,
      "loss": 1.9721,
      "step": 1036
    },
    {
      "epoch": 0.10938818565400844,
      "grad_norm": 0.4769430458545685,
      "learning_rate": 0.0015,
      "loss": 1.9864,
      "step": 1037
    },
    {
      "epoch": 0.10949367088607595,
      "grad_norm": 0.5129791498184204,
      "learning_rate": 0.0015,
      "loss": 2.0264,
      "step": 1038
    },
    {
      "epoch": 0.10959915611814346,
      "grad_norm": 0.5057224035263062,
      "learning_rate": 0.0015,
      "loss": 1.9787,
      "step": 1039
    },
    {
      "epoch": 0.10970464135021098,
      "grad_norm": 0.623877763748169,
      "learning_rate": 0.0015,
      "loss": 1.9988,
      "step": 1040
    },
    {
      "epoch": 0.10981012658227848,
      "grad_norm": 0.5065798759460449,
      "learning_rate": 0.0015,
      "loss": 2.0087,
      "step": 1041
    },
    {
      "epoch": 0.10991561181434599,
      "grad_norm": 0.5388501286506653,
      "learning_rate": 0.0015,
      "loss": 1.9967,
      "step": 1042
    },
    {
      "epoch": 0.1100210970464135,
      "grad_norm": 0.589823305606842,
      "learning_rate": 0.0015,
      "loss": 1.9957,
      "step": 1043
    },
    {
      "epoch": 0.11012658227848102,
      "grad_norm": 0.6062374114990234,
      "learning_rate": 0.0015,
      "loss": 1.9748,
      "step": 1044
    },
    {
      "epoch": 0.11023206751054852,
      "grad_norm": 0.5757289528846741,
      "learning_rate": 0.0015,
      "loss": 2.0066,
      "step": 1045
    },
    {
      "epoch": 0.11033755274261603,
      "grad_norm": 0.5318777561187744,
      "learning_rate": 0.0015,
      "loss": 2.0156,
      "step": 1046
    },
    {
      "epoch": 0.11044303797468355,
      "grad_norm": 0.4598942697048187,
      "learning_rate": 0.0015,
      "loss": 1.9953,
      "step": 1047
    },
    {
      "epoch": 0.11054852320675106,
      "grad_norm": 0.5211374759674072,
      "learning_rate": 0.0015,
      "loss": 1.9958,
      "step": 1048
    },
    {
      "epoch": 0.11065400843881856,
      "grad_norm": 0.6289006471633911,
      "learning_rate": 0.0015,
      "loss": 1.9837,
      "step": 1049
    },
    {
      "epoch": 0.11075949367088607,
      "grad_norm": 0.7800363302230835,
      "learning_rate": 0.0015,
      "loss": 2.0197,
      "step": 1050
    },
    {
      "epoch": 0.11086497890295359,
      "grad_norm": 0.5729717016220093,
      "learning_rate": 0.0015,
      "loss": 1.981,
      "step": 1051
    },
    {
      "epoch": 0.1109704641350211,
      "grad_norm": 0.5333393216133118,
      "learning_rate": 0.0015,
      "loss": 2.0084,
      "step": 1052
    },
    {
      "epoch": 0.1110759493670886,
      "grad_norm": 0.8517758250236511,
      "learning_rate": 0.0015,
      "loss": 2.01,
      "step": 1053
    },
    {
      "epoch": 0.11118143459915612,
      "grad_norm": 0.8174675703048706,
      "learning_rate": 0.0015,
      "loss": 2.005,
      "step": 1054
    },
    {
      "epoch": 0.11128691983122363,
      "grad_norm": 0.5155147910118103,
      "learning_rate": 0.0015,
      "loss": 1.979,
      "step": 1055
    },
    {
      "epoch": 0.11139240506329114,
      "grad_norm": 0.6293675303459167,
      "learning_rate": 0.0015,
      "loss": 1.9687,
      "step": 1056
    },
    {
      "epoch": 0.11149789029535866,
      "grad_norm": 0.7848137617111206,
      "learning_rate": 0.0015,
      "loss": 2.0003,
      "step": 1057
    },
    {
      "epoch": 0.11160337552742616,
      "grad_norm": 0.6546721458435059,
      "learning_rate": 0.0015,
      "loss": 1.9847,
      "step": 1058
    },
    {
      "epoch": 0.11170886075949367,
      "grad_norm": 0.5005208849906921,
      "learning_rate": 0.0015,
      "loss": 1.9546,
      "step": 1059
    },
    {
      "epoch": 0.11181434599156118,
      "grad_norm": 0.660610020160675,
      "learning_rate": 0.0015,
      "loss": 1.9856,
      "step": 1060
    },
    {
      "epoch": 0.1119198312236287,
      "grad_norm": 0.7893019914627075,
      "learning_rate": 0.0015,
      "loss": 2.0161,
      "step": 1061
    },
    {
      "epoch": 0.1120253164556962,
      "grad_norm": 0.5391327142715454,
      "learning_rate": 0.0015,
      "loss": 1.9456,
      "step": 1062
    },
    {
      "epoch": 0.11213080168776371,
      "grad_norm": 0.611042857170105,
      "learning_rate": 0.0015,
      "loss": 1.9991,
      "step": 1063
    },
    {
      "epoch": 0.11223628691983123,
      "grad_norm": 0.8975500464439392,
      "learning_rate": 0.0015,
      "loss": 1.9622,
      "step": 1064
    },
    {
      "epoch": 0.11234177215189874,
      "grad_norm": 0.8975341320037842,
      "learning_rate": 0.0015,
      "loss": 1.9706,
      "step": 1065
    },
    {
      "epoch": 0.11244725738396624,
      "grad_norm": 0.5800900459289551,
      "learning_rate": 0.0015,
      "loss": 1.9549,
      "step": 1066
    },
    {
      "epoch": 0.11255274261603375,
      "grad_norm": 0.5761541724205017,
      "learning_rate": 0.0015,
      "loss": 1.9832,
      "step": 1067
    },
    {
      "epoch": 0.11265822784810127,
      "grad_norm": 0.777255654335022,
      "learning_rate": 0.0015,
      "loss": 1.9604,
      "step": 1068
    },
    {
      "epoch": 0.11276371308016878,
      "grad_norm": 0.5889245271682739,
      "learning_rate": 0.0015,
      "loss": 1.9752,
      "step": 1069
    },
    {
      "epoch": 0.11286919831223628,
      "grad_norm": 0.5593006610870361,
      "learning_rate": 0.0015,
      "loss": 1.992,
      "step": 1070
    },
    {
      "epoch": 0.1129746835443038,
      "grad_norm": 0.7435798048973083,
      "learning_rate": 0.0015,
      "loss": 1.9515,
      "step": 1071
    },
    {
      "epoch": 0.11308016877637131,
      "grad_norm": 0.6535780429840088,
      "learning_rate": 0.0015,
      "loss": 2.0214,
      "step": 1072
    },
    {
      "epoch": 0.11318565400843882,
      "grad_norm": 0.5546253323554993,
      "learning_rate": 0.0015,
      "loss": 1.959,
      "step": 1073
    },
    {
      "epoch": 0.11329113924050632,
      "grad_norm": 0.695321261882782,
      "learning_rate": 0.0015,
      "loss": 1.9537,
      "step": 1074
    },
    {
      "epoch": 0.11339662447257384,
      "grad_norm": 0.5591767430305481,
      "learning_rate": 0.0015,
      "loss": 1.9822,
      "step": 1075
    },
    {
      "epoch": 0.11350210970464135,
      "grad_norm": 0.6114763021469116,
      "learning_rate": 0.0015,
      "loss": 2.0125,
      "step": 1076
    },
    {
      "epoch": 0.11360759493670886,
      "grad_norm": 0.7554244995117188,
      "learning_rate": 0.0015,
      "loss": 2.0057,
      "step": 1077
    },
    {
      "epoch": 0.11371308016877638,
      "grad_norm": 0.7061241269111633,
      "learning_rate": 0.0015,
      "loss": 1.9577,
      "step": 1078
    },
    {
      "epoch": 0.11381856540084388,
      "grad_norm": 0.551810085773468,
      "learning_rate": 0.0015,
      "loss": 1.9885,
      "step": 1079
    },
    {
      "epoch": 0.11392405063291139,
      "grad_norm": 0.6357426643371582,
      "learning_rate": 0.0015,
      "loss": 1.9633,
      "step": 1080
    },
    {
      "epoch": 0.11402953586497891,
      "grad_norm": 0.736291229724884,
      "learning_rate": 0.0015,
      "loss": 1.9794,
      "step": 1081
    },
    {
      "epoch": 0.11413502109704642,
      "grad_norm": 0.6027610301971436,
      "learning_rate": 0.0015,
      "loss": 1.9985,
      "step": 1082
    },
    {
      "epoch": 0.11424050632911392,
      "grad_norm": 0.5908238887786865,
      "learning_rate": 0.0015,
      "loss": 1.9655,
      "step": 1083
    },
    {
      "epoch": 0.11434599156118143,
      "grad_norm": 0.609118640422821,
      "learning_rate": 0.0015,
      "loss": 1.9944,
      "step": 1084
    },
    {
      "epoch": 0.11445147679324895,
      "grad_norm": 0.5698200464248657,
      "learning_rate": 0.0015,
      "loss": 1.9589,
      "step": 1085
    },
    {
      "epoch": 0.11455696202531646,
      "grad_norm": 0.5287510752677917,
      "learning_rate": 0.0015,
      "loss": 1.9422,
      "step": 1086
    },
    {
      "epoch": 0.11466244725738396,
      "grad_norm": 0.545129120349884,
      "learning_rate": 0.0015,
      "loss": 2.0234,
      "step": 1087
    },
    {
      "epoch": 0.11476793248945148,
      "grad_norm": 0.5150210857391357,
      "learning_rate": 0.0015,
      "loss": 2.0034,
      "step": 1088
    },
    {
      "epoch": 0.11487341772151899,
      "grad_norm": 0.5504254698753357,
      "learning_rate": 0.0015,
      "loss": 2.014,
      "step": 1089
    },
    {
      "epoch": 0.1149789029535865,
      "grad_norm": 0.5121006369590759,
      "learning_rate": 0.0015,
      "loss": 1.9867,
      "step": 1090
    },
    {
      "epoch": 0.115084388185654,
      "grad_norm": 0.5371829271316528,
      "learning_rate": 0.0015,
      "loss": 1.9441,
      "step": 1091
    },
    {
      "epoch": 0.11518987341772152,
      "grad_norm": 0.5726335048675537,
      "learning_rate": 0.0015,
      "loss": 1.9548,
      "step": 1092
    },
    {
      "epoch": 0.11529535864978903,
      "grad_norm": 0.615773618221283,
      "learning_rate": 0.0015,
      "loss": 1.9593,
      "step": 1093
    },
    {
      "epoch": 0.11540084388185654,
      "grad_norm": 0.5490939617156982,
      "learning_rate": 0.0015,
      "loss": 1.9929,
      "step": 1094
    },
    {
      "epoch": 0.11550632911392406,
      "grad_norm": 0.5754554271697998,
      "learning_rate": 0.0015,
      "loss": 1.9863,
      "step": 1095
    },
    {
      "epoch": 0.11561181434599156,
      "grad_norm": 0.5992894172668457,
      "learning_rate": 0.0015,
      "loss": 1.9975,
      "step": 1096
    },
    {
      "epoch": 0.11571729957805907,
      "grad_norm": 0.5731102228164673,
      "learning_rate": 0.0015,
      "loss": 1.9731,
      "step": 1097
    },
    {
      "epoch": 0.11582278481012659,
      "grad_norm": 0.4969916343688965,
      "learning_rate": 0.0015,
      "loss": 1.9942,
      "step": 1098
    },
    {
      "epoch": 0.1159282700421941,
      "grad_norm": 0.5169114470481873,
      "learning_rate": 0.0015,
      "loss": 1.9716,
      "step": 1099
    },
    {
      "epoch": 0.1160337552742616,
      "grad_norm": 0.5371574759483337,
      "learning_rate": 0.0015,
      "loss": 1.9765,
      "step": 1100
    },
    {
      "epoch": 0.11613924050632911,
      "grad_norm": 0.4847295880317688,
      "learning_rate": 0.0015,
      "loss": 1.9667,
      "step": 1101
    },
    {
      "epoch": 0.11624472573839663,
      "grad_norm": 0.4888870120048523,
      "learning_rate": 0.0015,
      "loss": 1.9542,
      "step": 1102
    },
    {
      "epoch": 0.11635021097046414,
      "grad_norm": 0.47460025548934937,
      "learning_rate": 0.0015,
      "loss": 1.9795,
      "step": 1103
    },
    {
      "epoch": 0.11645569620253164,
      "grad_norm": 0.48211368918418884,
      "learning_rate": 0.0015,
      "loss": 1.9859,
      "step": 1104
    },
    {
      "epoch": 0.11656118143459916,
      "grad_norm": 0.4726405441761017,
      "learning_rate": 0.0015,
      "loss": 2.001,
      "step": 1105
    },
    {
      "epoch": 0.11666666666666667,
      "grad_norm": 0.48614928126335144,
      "learning_rate": 0.0015,
      "loss": 1.9683,
      "step": 1106
    },
    {
      "epoch": 0.11677215189873418,
      "grad_norm": 0.47507819533348083,
      "learning_rate": 0.0015,
      "loss": 1.9645,
      "step": 1107
    },
    {
      "epoch": 0.11687763713080168,
      "grad_norm": 0.47301697731018066,
      "learning_rate": 0.0015,
      "loss": 1.9654,
      "step": 1108
    },
    {
      "epoch": 0.1169831223628692,
      "grad_norm": 0.5072380900382996,
      "learning_rate": 0.0015,
      "loss": 1.9898,
      "step": 1109
    },
    {
      "epoch": 0.11708860759493671,
      "grad_norm": 0.47868582606315613,
      "learning_rate": 0.0015,
      "loss": 1.952,
      "step": 1110
    },
    {
      "epoch": 0.11719409282700421,
      "grad_norm": 0.4702244699001312,
      "learning_rate": 0.0015,
      "loss": 1.9978,
      "step": 1111
    },
    {
      "epoch": 0.11729957805907174,
      "grad_norm": 0.5345383286476135,
      "learning_rate": 0.0015,
      "loss": 1.9474,
      "step": 1112
    },
    {
      "epoch": 0.11740506329113924,
      "grad_norm": 0.5054839849472046,
      "learning_rate": 0.0015,
      "loss": 1.9918,
      "step": 1113
    },
    {
      "epoch": 0.11751054852320675,
      "grad_norm": 0.5359642505645752,
      "learning_rate": 0.0015,
      "loss": 1.9513,
      "step": 1114
    },
    {
      "epoch": 0.11761603375527427,
      "grad_norm": 0.6218224167823792,
      "learning_rate": 0.0015,
      "loss": 2.0022,
      "step": 1115
    },
    {
      "epoch": 0.11772151898734177,
      "grad_norm": 0.585112452507019,
      "learning_rate": 0.0015,
      "loss": 1.9663,
      "step": 1116
    },
    {
      "epoch": 0.11782700421940928,
      "grad_norm": 0.5225456953048706,
      "learning_rate": 0.0015,
      "loss": 1.9649,
      "step": 1117
    },
    {
      "epoch": 0.11793248945147679,
      "grad_norm": 0.45402202010154724,
      "learning_rate": 0.0015,
      "loss": 1.9892,
      "step": 1118
    },
    {
      "epoch": 0.11803797468354431,
      "grad_norm": 0.5021755695343018,
      "learning_rate": 0.0015,
      "loss": 1.9542,
      "step": 1119
    },
    {
      "epoch": 0.11814345991561181,
      "grad_norm": 0.4669731557369232,
      "learning_rate": 0.0015,
      "loss": 1.9293,
      "step": 1120
    },
    {
      "epoch": 0.11824894514767932,
      "grad_norm": 0.4613353908061981,
      "learning_rate": 0.0015,
      "loss": 1.9253,
      "step": 1121
    },
    {
      "epoch": 0.11835443037974684,
      "grad_norm": 0.48189011216163635,
      "learning_rate": 0.0015,
      "loss": 1.972,
      "step": 1122
    },
    {
      "epoch": 0.11845991561181435,
      "grad_norm": 0.5315321087837219,
      "learning_rate": 0.0015,
      "loss": 1.9697,
      "step": 1123
    },
    {
      "epoch": 0.11856540084388185,
      "grad_norm": 0.5763983130455017,
      "learning_rate": 0.0015,
      "loss": 1.942,
      "step": 1124
    },
    {
      "epoch": 0.11867088607594936,
      "grad_norm": 0.5278778672218323,
      "learning_rate": 0.0015,
      "loss": 1.9655,
      "step": 1125
    },
    {
      "epoch": 0.11877637130801688,
      "grad_norm": 0.5579981803894043,
      "learning_rate": 0.0015,
      "loss": 1.9401,
      "step": 1126
    },
    {
      "epoch": 0.11888185654008439,
      "grad_norm": 0.5369895100593567,
      "learning_rate": 0.0015,
      "loss": 1.9643,
      "step": 1127
    },
    {
      "epoch": 0.1189873417721519,
      "grad_norm": 0.5380120277404785,
      "learning_rate": 0.0015,
      "loss": 1.9634,
      "step": 1128
    },
    {
      "epoch": 0.11909282700421941,
      "grad_norm": 0.6157765984535217,
      "learning_rate": 0.0015,
      "loss": 1.9678,
      "step": 1129
    },
    {
      "epoch": 0.11919831223628692,
      "grad_norm": 0.734170138835907,
      "learning_rate": 0.0015,
      "loss": 1.9468,
      "step": 1130
    },
    {
      "epoch": 0.11930379746835443,
      "grad_norm": 0.7338579297065735,
      "learning_rate": 0.0015,
      "loss": 2.0127,
      "step": 1131
    },
    {
      "epoch": 0.11940928270042193,
      "grad_norm": 0.6663212180137634,
      "learning_rate": 0.0015,
      "loss": 1.9487,
      "step": 1132
    },
    {
      "epoch": 0.11951476793248945,
      "grad_norm": 0.5774522423744202,
      "learning_rate": 0.0015,
      "loss": 1.9725,
      "step": 1133
    },
    {
      "epoch": 0.11962025316455696,
      "grad_norm": 0.5378499031066895,
      "learning_rate": 0.0015,
      "loss": 1.9819,
      "step": 1134
    },
    {
      "epoch": 0.11972573839662447,
      "grad_norm": 0.7140102982521057,
      "learning_rate": 0.0015,
      "loss": 1.9674,
      "step": 1135
    },
    {
      "epoch": 0.11983122362869199,
      "grad_norm": 0.7592074871063232,
      "learning_rate": 0.0015,
      "loss": 1.9326,
      "step": 1136
    },
    {
      "epoch": 0.1199367088607595,
      "grad_norm": 0.5613864660263062,
      "learning_rate": 0.0015,
      "loss": 1.9879,
      "step": 1137
    },
    {
      "epoch": 0.120042194092827,
      "grad_norm": 0.6440332531929016,
      "learning_rate": 0.0015,
      "loss": 1.9775,
      "step": 1138
    },
    {
      "epoch": 0.12014767932489452,
      "grad_norm": 0.7823119759559631,
      "learning_rate": 0.0015,
      "loss": 1.9609,
      "step": 1139
    },
    {
      "epoch": 0.12025316455696203,
      "grad_norm": 0.5680260062217712,
      "learning_rate": 0.0015,
      "loss": 1.9837,
      "step": 1140
    },
    {
      "epoch": 0.12035864978902953,
      "grad_norm": 0.5426461696624756,
      "learning_rate": 0.0015,
      "loss": 1.9091,
      "step": 1141
    },
    {
      "epoch": 0.12046413502109704,
      "grad_norm": 0.5475901961326599,
      "learning_rate": 0.0015,
      "loss": 1.9542,
      "step": 1142
    },
    {
      "epoch": 0.12056962025316456,
      "grad_norm": 0.6051155924797058,
      "learning_rate": 0.0015,
      "loss": 1.9994,
      "step": 1143
    },
    {
      "epoch": 0.12067510548523207,
      "grad_norm": 0.6208446621894836,
      "learning_rate": 0.0015,
      "loss": 1.9243,
      "step": 1144
    },
    {
      "epoch": 0.12078059071729957,
      "grad_norm": 0.5487984418869019,
      "learning_rate": 0.0015,
      "loss": 1.9616,
      "step": 1145
    },
    {
      "epoch": 0.1208860759493671,
      "grad_norm": 0.5089645981788635,
      "learning_rate": 0.0015,
      "loss": 1.9577,
      "step": 1146
    },
    {
      "epoch": 0.1209915611814346,
      "grad_norm": 0.547914981842041,
      "learning_rate": 0.0015,
      "loss": 1.9444,
      "step": 1147
    },
    {
      "epoch": 0.12109704641350211,
      "grad_norm": 0.569980800151825,
      "learning_rate": 0.0015,
      "loss": 1.9948,
      "step": 1148
    },
    {
      "epoch": 0.12120253164556961,
      "grad_norm": 0.541676938533783,
      "learning_rate": 0.0015,
      "loss": 1.9639,
      "step": 1149
    },
    {
      "epoch": 0.12130801687763713,
      "grad_norm": 0.6242750883102417,
      "learning_rate": 0.0015,
      "loss": 1.9558,
      "step": 1150
    },
    {
      "epoch": 0.12141350210970464,
      "grad_norm": 0.6878270506858826,
      "learning_rate": 0.0015,
      "loss": 1.9578,
      "step": 1151
    },
    {
      "epoch": 0.12151898734177215,
      "grad_norm": 0.5642980337142944,
      "learning_rate": 0.0015,
      "loss": 2.0148,
      "step": 1152
    },
    {
      "epoch": 0.12162447257383967,
      "grad_norm": 0.5831053853034973,
      "learning_rate": 0.0015,
      "loss": 1.9544,
      "step": 1153
    },
    {
      "epoch": 0.12172995780590717,
      "grad_norm": 0.478061705827713,
      "learning_rate": 0.0015,
      "loss": 1.9535,
      "step": 1154
    },
    {
      "epoch": 0.12183544303797468,
      "grad_norm": 0.5408726334571838,
      "learning_rate": 0.0015,
      "loss": 1.9616,
      "step": 1155
    },
    {
      "epoch": 0.1219409282700422,
      "grad_norm": 0.5360070466995239,
      "learning_rate": 0.0015,
      "loss": 1.9697,
      "step": 1156
    },
    {
      "epoch": 0.12204641350210971,
      "grad_norm": 0.6306201219558716,
      "learning_rate": 0.0015,
      "loss": 1.9458,
      "step": 1157
    },
    {
      "epoch": 0.12215189873417721,
      "grad_norm": 0.7493512034416199,
      "learning_rate": 0.0015,
      "loss": 1.9565,
      "step": 1158
    },
    {
      "epoch": 0.12225738396624472,
      "grad_norm": 0.6081839203834534,
      "learning_rate": 0.0015,
      "loss": 1.9812,
      "step": 1159
    },
    {
      "epoch": 0.12236286919831224,
      "grad_norm": 0.4983415901660919,
      "learning_rate": 0.0015,
      "loss": 1.9505,
      "step": 1160
    },
    {
      "epoch": 0.12246835443037975,
      "grad_norm": 0.7041601538658142,
      "learning_rate": 0.0015,
      "loss": 1.9813,
      "step": 1161
    },
    {
      "epoch": 0.12257383966244725,
      "grad_norm": 0.9435353875160217,
      "learning_rate": 0.0015,
      "loss": 1.9892,
      "step": 1162
    },
    {
      "epoch": 0.12267932489451477,
      "grad_norm": 0.8395876884460449,
      "learning_rate": 0.0015,
      "loss": 1.9444,
      "step": 1163
    },
    {
      "epoch": 0.12278481012658228,
      "grad_norm": 0.5393134951591492,
      "learning_rate": 0.0015,
      "loss": 1.9603,
      "step": 1164
    },
    {
      "epoch": 0.12289029535864979,
      "grad_norm": 0.7198797464370728,
      "learning_rate": 0.0015,
      "loss": 1.9438,
      "step": 1165
    },
    {
      "epoch": 0.1229957805907173,
      "grad_norm": 0.6611842513084412,
      "learning_rate": 0.0015,
      "loss": 1.9457,
      "step": 1166
    },
    {
      "epoch": 0.12310126582278481,
      "grad_norm": 0.5443592667579651,
      "learning_rate": 0.0015,
      "loss": 1.9891,
      "step": 1167
    },
    {
      "epoch": 0.12320675105485232,
      "grad_norm": 0.5471925735473633,
      "learning_rate": 0.0015,
      "loss": 1.9243,
      "step": 1168
    },
    {
      "epoch": 0.12331223628691983,
      "grad_norm": 0.5817985534667969,
      "learning_rate": 0.0015,
      "loss": 1.9644,
      "step": 1169
    },
    {
      "epoch": 0.12341772151898735,
      "grad_norm": 0.6553073525428772,
      "learning_rate": 0.0015,
      "loss": 1.9297,
      "step": 1170
    },
    {
      "epoch": 0.12352320675105485,
      "grad_norm": 0.5561738014221191,
      "learning_rate": 0.0015,
      "loss": 1.972,
      "step": 1171
    },
    {
      "epoch": 0.12362869198312236,
      "grad_norm": 0.5268250107765198,
      "learning_rate": 0.0015,
      "loss": 1.9412,
      "step": 1172
    },
    {
      "epoch": 0.12373417721518987,
      "grad_norm": 0.6380401849746704,
      "learning_rate": 0.0015,
      "loss": 1.9007,
      "step": 1173
    },
    {
      "epoch": 0.12383966244725739,
      "grad_norm": 0.5170300006866455,
      "learning_rate": 0.0015,
      "loss": 1.9554,
      "step": 1174
    },
    {
      "epoch": 0.1239451476793249,
      "grad_norm": 0.6789668798446655,
      "learning_rate": 0.0015,
      "loss": 1.9243,
      "step": 1175
    },
    {
      "epoch": 0.1240506329113924,
      "grad_norm": 0.6632237434387207,
      "learning_rate": 0.0015,
      "loss": 1.8938,
      "step": 1176
    },
    {
      "epoch": 0.12415611814345992,
      "grad_norm": 0.5542116761207581,
      "learning_rate": 0.0015,
      "loss": 1.9172,
      "step": 1177
    },
    {
      "epoch": 0.12426160337552743,
      "grad_norm": 0.5060286521911621,
      "learning_rate": 0.0015,
      "loss": 1.9651,
      "step": 1178
    },
    {
      "epoch": 0.12436708860759493,
      "grad_norm": 0.6038116216659546,
      "learning_rate": 0.0015,
      "loss": 1.9736,
      "step": 1179
    },
    {
      "epoch": 0.12447257383966245,
      "grad_norm": 0.7100387811660767,
      "learning_rate": 0.0015,
      "loss": 1.9515,
      "step": 1180
    },
    {
      "epoch": 0.12457805907172996,
      "grad_norm": 0.5594522953033447,
      "learning_rate": 0.0015,
      "loss": 1.924,
      "step": 1181
    },
    {
      "epoch": 0.12468354430379747,
      "grad_norm": 0.5433719754219055,
      "learning_rate": 0.0015,
      "loss": 1.9346,
      "step": 1182
    },
    {
      "epoch": 0.12478902953586497,
      "grad_norm": 0.7844252586364746,
      "learning_rate": 0.0015,
      "loss": 1.9514,
      "step": 1183
    },
    {
      "epoch": 0.1248945147679325,
      "grad_norm": 0.7507432103157043,
      "learning_rate": 0.0015,
      "loss": 1.9621,
      "step": 1184
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.4914301037788391,
      "learning_rate": 0.0015,
      "loss": 1.953,
      "step": 1185
    },
    {
      "epoch": 0.12510548523206752,
      "grad_norm": 0.5079389214515686,
      "learning_rate": 0.0015,
      "loss": 1.9482,
      "step": 1186
    },
    {
      "epoch": 0.125210970464135,
      "grad_norm": 0.5197246670722961,
      "learning_rate": 0.0015,
      "loss": 1.9596,
      "step": 1187
    },
    {
      "epoch": 0.12531645569620253,
      "grad_norm": 0.48691096901893616,
      "learning_rate": 0.0015,
      "loss": 1.9611,
      "step": 1188
    },
    {
      "epoch": 0.12542194092827005,
      "grad_norm": 0.5134087800979614,
      "learning_rate": 0.0015,
      "loss": 1.9179,
      "step": 1189
    },
    {
      "epoch": 0.12552742616033755,
      "grad_norm": 0.5778923630714417,
      "learning_rate": 0.0015,
      "loss": 1.9153,
      "step": 1190
    },
    {
      "epoch": 0.12563291139240507,
      "grad_norm": 0.6021434664726257,
      "learning_rate": 0.0015,
      "loss": 1.9434,
      "step": 1191
    },
    {
      "epoch": 0.1257383966244726,
      "grad_norm": 0.567496120929718,
      "learning_rate": 0.0015,
      "loss": 1.9278,
      "step": 1192
    },
    {
      "epoch": 0.12584388185654008,
      "grad_norm": 0.7252230644226074,
      "learning_rate": 0.0015,
      "loss": 1.9859,
      "step": 1193
    },
    {
      "epoch": 0.1259493670886076,
      "grad_norm": 0.6520013809204102,
      "learning_rate": 0.0015,
      "loss": 2.0026,
      "step": 1194
    },
    {
      "epoch": 0.1260548523206751,
      "grad_norm": 0.593677818775177,
      "learning_rate": 0.0015,
      "loss": 1.9466,
      "step": 1195
    },
    {
      "epoch": 0.1261603375527426,
      "grad_norm": 1.0265676975250244,
      "learning_rate": 0.0015,
      "loss": 1.9382,
      "step": 1196
    },
    {
      "epoch": 0.12626582278481013,
      "grad_norm": 1.0938862562179565,
      "learning_rate": 0.0015,
      "loss": 1.9382,
      "step": 1197
    },
    {
      "epoch": 0.12637130801687763,
      "grad_norm": 0.5909872055053711,
      "learning_rate": 0.0015,
      "loss": 1.9739,
      "step": 1198
    },
    {
      "epoch": 0.12647679324894515,
      "grad_norm": 0.8484125137329102,
      "learning_rate": 0.0015,
      "loss": 1.9813,
      "step": 1199
    },
    {
      "epoch": 0.12658227848101267,
      "grad_norm": 0.8648718595504761,
      "learning_rate": 0.0015,
      "loss": 1.952,
      "step": 1200
    },
    {
      "epoch": 0.12668776371308016,
      "grad_norm": 0.6355069279670715,
      "learning_rate": 0.0015,
      "loss": 1.9441,
      "step": 1201
    },
    {
      "epoch": 0.12679324894514768,
      "grad_norm": 0.6418319344520569,
      "learning_rate": 0.0015,
      "loss": 1.9418,
      "step": 1202
    },
    {
      "epoch": 0.1268987341772152,
      "grad_norm": 0.9209024906158447,
      "learning_rate": 0.0015,
      "loss": 1.9329,
      "step": 1203
    },
    {
      "epoch": 0.1270042194092827,
      "grad_norm": 0.8455290198326111,
      "learning_rate": 0.0015,
      "loss": 1.9469,
      "step": 1204
    },
    {
      "epoch": 0.1271097046413502,
      "grad_norm": 0.5296677350997925,
      "learning_rate": 0.0015,
      "loss": 1.9771,
      "step": 1205
    },
    {
      "epoch": 0.12721518987341773,
      "grad_norm": 0.5680837631225586,
      "learning_rate": 0.0015,
      "loss": 1.9523,
      "step": 1206
    },
    {
      "epoch": 0.12732067510548523,
      "grad_norm": 0.6870173215866089,
      "learning_rate": 0.0015,
      "loss": 1.939,
      "step": 1207
    },
    {
      "epoch": 0.12742616033755275,
      "grad_norm": 0.584763765335083,
      "learning_rate": 0.0015,
      "loss": 1.9413,
      "step": 1208
    },
    {
      "epoch": 0.12753164556962027,
      "grad_norm": 0.5821747183799744,
      "learning_rate": 0.0015,
      "loss": 1.9455,
      "step": 1209
    },
    {
      "epoch": 0.12763713080168776,
      "grad_norm": 0.6364511847496033,
      "learning_rate": 0.0015,
      "loss": 1.9723,
      "step": 1210
    },
    {
      "epoch": 0.12774261603375528,
      "grad_norm": 0.5813069939613342,
      "learning_rate": 0.0015,
      "loss": 1.9321,
      "step": 1211
    },
    {
      "epoch": 0.12784810126582277,
      "grad_norm": 0.4982289671897888,
      "learning_rate": 0.0015,
      "loss": 1.9126,
      "step": 1212
    },
    {
      "epoch": 0.1279535864978903,
      "grad_norm": 0.4817267954349518,
      "learning_rate": 0.0015,
      "loss": 1.927,
      "step": 1213
    },
    {
      "epoch": 0.1280590717299578,
      "grad_norm": 0.5128777623176575,
      "learning_rate": 0.0015,
      "loss": 1.9373,
      "step": 1214
    },
    {
      "epoch": 0.1281645569620253,
      "grad_norm": 0.5634309649467468,
      "learning_rate": 0.0015,
      "loss": 1.9241,
      "step": 1215
    },
    {
      "epoch": 0.12827004219409283,
      "grad_norm": 0.6341714859008789,
      "learning_rate": 0.0015,
      "loss": 1.9532,
      "step": 1216
    },
    {
      "epoch": 0.12837552742616035,
      "grad_norm": 0.5437911152839661,
      "learning_rate": 0.0015,
      "loss": 1.939,
      "step": 1217
    },
    {
      "epoch": 0.12848101265822784,
      "grad_norm": 0.5569978952407837,
      "learning_rate": 0.0015,
      "loss": 1.9899,
      "step": 1218
    },
    {
      "epoch": 0.12858649789029536,
      "grad_norm": 0.6110464334487915,
      "learning_rate": 0.0015,
      "loss": 1.9051,
      "step": 1219
    },
    {
      "epoch": 0.12869198312236288,
      "grad_norm": 0.7041509747505188,
      "learning_rate": 0.0015,
      "loss": 1.9077,
      "step": 1220
    },
    {
      "epoch": 0.12879746835443037,
      "grad_norm": 0.6820054054260254,
      "learning_rate": 0.0015,
      "loss": 1.9507,
      "step": 1221
    },
    {
      "epoch": 0.1289029535864979,
      "grad_norm": 0.5716399550437927,
      "learning_rate": 0.0015,
      "loss": 1.9591,
      "step": 1222
    },
    {
      "epoch": 0.1290084388185654,
      "grad_norm": 0.5545234680175781,
      "learning_rate": 0.0015,
      "loss": 1.9115,
      "step": 1223
    },
    {
      "epoch": 0.1291139240506329,
      "grad_norm": 0.6347628831863403,
      "learning_rate": 0.0015,
      "loss": 1.9709,
      "step": 1224
    },
    {
      "epoch": 0.12921940928270043,
      "grad_norm": 0.5690528154373169,
      "learning_rate": 0.0015,
      "loss": 1.8906,
      "step": 1225
    },
    {
      "epoch": 0.12932489451476795,
      "grad_norm": 0.5223212838172913,
      "learning_rate": 0.0015,
      "loss": 1.9083,
      "step": 1226
    },
    {
      "epoch": 0.12943037974683544,
      "grad_norm": 0.5073110461235046,
      "learning_rate": 0.0015,
      "loss": 1.9062,
      "step": 1227
    },
    {
      "epoch": 0.12953586497890296,
      "grad_norm": 0.5078768134117126,
      "learning_rate": 0.0015,
      "loss": 1.9045,
      "step": 1228
    },
    {
      "epoch": 0.12964135021097045,
      "grad_norm": 0.5362216830253601,
      "learning_rate": 0.0015,
      "loss": 1.9488,
      "step": 1229
    },
    {
      "epoch": 0.12974683544303797,
      "grad_norm": 0.5915535092353821,
      "learning_rate": 0.0015,
      "loss": 1.8828,
      "step": 1230
    },
    {
      "epoch": 0.1298523206751055,
      "grad_norm": 0.5478518605232239,
      "learning_rate": 0.0015,
      "loss": 1.9493,
      "step": 1231
    },
    {
      "epoch": 0.12995780590717299,
      "grad_norm": 0.48701831698417664,
      "learning_rate": 0.0015,
      "loss": 1.9755,
      "step": 1232
    },
    {
      "epoch": 0.1300632911392405,
      "grad_norm": 0.6849209070205688,
      "learning_rate": 0.0015,
      "loss": 1.9118,
      "step": 1233
    },
    {
      "epoch": 0.13016877637130803,
      "grad_norm": 0.7918779850006104,
      "learning_rate": 0.0015,
      "loss": 1.9584,
      "step": 1234
    },
    {
      "epoch": 0.13027426160337552,
      "grad_norm": 0.6584113240242004,
      "learning_rate": 0.0015,
      "loss": 1.9594,
      "step": 1235
    },
    {
      "epoch": 0.13037974683544304,
      "grad_norm": 0.5166054964065552,
      "learning_rate": 0.0015,
      "loss": 1.9275,
      "step": 1236
    },
    {
      "epoch": 0.13048523206751056,
      "grad_norm": 0.5061794519424438,
      "learning_rate": 0.0015,
      "loss": 1.9268,
      "step": 1237
    },
    {
      "epoch": 0.13059071729957805,
      "grad_norm": 0.6233305931091309,
      "learning_rate": 0.0015,
      "loss": 1.9293,
      "step": 1238
    },
    {
      "epoch": 0.13069620253164557,
      "grad_norm": 0.6041268706321716,
      "learning_rate": 0.0015,
      "loss": 1.9525,
      "step": 1239
    },
    {
      "epoch": 0.1308016877637131,
      "grad_norm": 0.49778875708580017,
      "learning_rate": 0.0015,
      "loss": 1.9124,
      "step": 1240
    },
    {
      "epoch": 0.13090717299578059,
      "grad_norm": 0.606499195098877,
      "learning_rate": 0.0015,
      "loss": 1.8972,
      "step": 1241
    },
    {
      "epoch": 0.1310126582278481,
      "grad_norm": 0.6507505774497986,
      "learning_rate": 0.0015,
      "loss": 1.947,
      "step": 1242
    },
    {
      "epoch": 0.1311181434599156,
      "grad_norm": 0.6108731031417847,
      "learning_rate": 0.0015,
      "loss": 1.9446,
      "step": 1243
    },
    {
      "epoch": 0.13122362869198312,
      "grad_norm": 0.47272953391075134,
      "learning_rate": 0.0015,
      "loss": 1.9318,
      "step": 1244
    },
    {
      "epoch": 0.13132911392405064,
      "grad_norm": 0.647142767906189,
      "learning_rate": 0.0015,
      "loss": 1.9676,
      "step": 1245
    },
    {
      "epoch": 0.13143459915611813,
      "grad_norm": 0.529202938079834,
      "learning_rate": 0.0015,
      "loss": 1.9716,
      "step": 1246
    },
    {
      "epoch": 0.13154008438818565,
      "grad_norm": 0.6392340064048767,
      "learning_rate": 0.0015,
      "loss": 1.9349,
      "step": 1247
    },
    {
      "epoch": 0.13164556962025317,
      "grad_norm": 0.8022511601448059,
      "learning_rate": 0.0015,
      "loss": 1.9426,
      "step": 1248
    },
    {
      "epoch": 0.13175105485232066,
      "grad_norm": 0.6682711243629456,
      "learning_rate": 0.0015,
      "loss": 1.9289,
      "step": 1249
    },
    {
      "epoch": 0.13185654008438819,
      "grad_norm": 0.572135329246521,
      "learning_rate": 0.0015,
      "loss": 1.9282,
      "step": 1250
    },
    {
      "epoch": 0.1319620253164557,
      "grad_norm": 0.6414625644683838,
      "learning_rate": 0.0015,
      "loss": 1.9652,
      "step": 1251
    },
    {
      "epoch": 0.1320675105485232,
      "grad_norm": 0.47704997658729553,
      "learning_rate": 0.0015,
      "loss": 1.9204,
      "step": 1252
    },
    {
      "epoch": 0.13217299578059072,
      "grad_norm": 0.6268481016159058,
      "learning_rate": 0.0015,
      "loss": 1.9174,
      "step": 1253
    },
    {
      "epoch": 0.13227848101265824,
      "grad_norm": 0.5927824378013611,
      "learning_rate": 0.0015,
      "loss": 1.9014,
      "step": 1254
    },
    {
      "epoch": 0.13238396624472573,
      "grad_norm": 0.5019801259040833,
      "learning_rate": 0.0015,
      "loss": 1.9107,
      "step": 1255
    },
    {
      "epoch": 0.13248945147679325,
      "grad_norm": 0.7533797025680542,
      "learning_rate": 0.0015,
      "loss": 1.9139,
      "step": 1256
    },
    {
      "epoch": 0.13259493670886077,
      "grad_norm": 0.7740539908409119,
      "learning_rate": 0.0015,
      "loss": 1.9471,
      "step": 1257
    },
    {
      "epoch": 0.13270042194092826,
      "grad_norm": 0.5386258363723755,
      "learning_rate": 0.0015,
      "loss": 1.9229,
      "step": 1258
    },
    {
      "epoch": 0.13280590717299579,
      "grad_norm": 0.7877216339111328,
      "learning_rate": 0.0015,
      "loss": 1.9412,
      "step": 1259
    },
    {
      "epoch": 0.13291139240506328,
      "grad_norm": 0.9018682241439819,
      "learning_rate": 0.0015,
      "loss": 1.9276,
      "step": 1260
    },
    {
      "epoch": 0.1330168776371308,
      "grad_norm": 0.6250298619270325,
      "learning_rate": 0.0015,
      "loss": 1.9353,
      "step": 1261
    },
    {
      "epoch": 0.13312236286919832,
      "grad_norm": 0.7231791615486145,
      "learning_rate": 0.0015,
      "loss": 1.9128,
      "step": 1262
    },
    {
      "epoch": 0.1332278481012658,
      "grad_norm": 0.7344803810119629,
      "learning_rate": 0.0015,
      "loss": 1.9061,
      "step": 1263
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 0.7405490875244141,
      "learning_rate": 0.0015,
      "loss": 1.9064,
      "step": 1264
    },
    {
      "epoch": 0.13343881856540085,
      "grad_norm": 0.6008923649787903,
      "learning_rate": 0.0015,
      "loss": 1.9235,
      "step": 1265
    },
    {
      "epoch": 0.13354430379746834,
      "grad_norm": 0.5715137124061584,
      "learning_rate": 0.0015,
      "loss": 1.925,
      "step": 1266
    },
    {
      "epoch": 0.13364978902953586,
      "grad_norm": 0.6669307947158813,
      "learning_rate": 0.0015,
      "loss": 1.9494,
      "step": 1267
    },
    {
      "epoch": 0.13375527426160339,
      "grad_norm": 0.6033222675323486,
      "learning_rate": 0.0015,
      "loss": 1.9016,
      "step": 1268
    },
    {
      "epoch": 0.13386075949367088,
      "grad_norm": 0.6311067938804626,
      "learning_rate": 0.0015,
      "loss": 1.953,
      "step": 1269
    },
    {
      "epoch": 0.1339662447257384,
      "grad_norm": 0.5897722244262695,
      "learning_rate": 0.0015,
      "loss": 1.9264,
      "step": 1270
    },
    {
      "epoch": 0.13407172995780592,
      "grad_norm": 0.5945970416069031,
      "learning_rate": 0.0015,
      "loss": 1.9584,
      "step": 1271
    },
    {
      "epoch": 0.1341772151898734,
      "grad_norm": 0.703152060508728,
      "learning_rate": 0.0015,
      "loss": 1.9897,
      "step": 1272
    },
    {
      "epoch": 0.13428270042194093,
      "grad_norm": 0.5912957787513733,
      "learning_rate": 0.0015,
      "loss": 1.9146,
      "step": 1273
    },
    {
      "epoch": 0.13438818565400845,
      "grad_norm": 0.5079948902130127,
      "learning_rate": 0.0015,
      "loss": 1.9265,
      "step": 1274
    },
    {
      "epoch": 0.13449367088607594,
      "grad_norm": 0.5293281078338623,
      "learning_rate": 0.0015,
      "loss": 1.9319,
      "step": 1275
    },
    {
      "epoch": 0.13459915611814346,
      "grad_norm": 0.5533709526062012,
      "learning_rate": 0.0015,
      "loss": 1.9128,
      "step": 1276
    },
    {
      "epoch": 0.13470464135021096,
      "grad_norm": 0.5535594820976257,
      "learning_rate": 0.0015,
      "loss": 1.948,
      "step": 1277
    },
    {
      "epoch": 0.13481012658227848,
      "grad_norm": 0.6553687453269958,
      "learning_rate": 0.0015,
      "loss": 1.9219,
      "step": 1278
    },
    {
      "epoch": 0.134915611814346,
      "grad_norm": 0.5242647528648376,
      "learning_rate": 0.0015,
      "loss": 1.9505,
      "step": 1279
    },
    {
      "epoch": 0.1350210970464135,
      "grad_norm": 0.4557071924209595,
      "learning_rate": 0.0015,
      "loss": 1.8867,
      "step": 1280
    },
    {
      "epoch": 0.135126582278481,
      "grad_norm": 0.49745699763298035,
      "learning_rate": 0.0015,
      "loss": 1.9194,
      "step": 1281
    },
    {
      "epoch": 0.13523206751054853,
      "grad_norm": 0.49396854639053345,
      "learning_rate": 0.0015,
      "loss": 1.874,
      "step": 1282
    },
    {
      "epoch": 0.13533755274261602,
      "grad_norm": 0.5221226215362549,
      "learning_rate": 0.0015,
      "loss": 1.9331,
      "step": 1283
    },
    {
      "epoch": 0.13544303797468354,
      "grad_norm": 0.5184764266014099,
      "learning_rate": 0.0015,
      "loss": 1.9056,
      "step": 1284
    },
    {
      "epoch": 0.13554852320675106,
      "grad_norm": 0.5110886096954346,
      "learning_rate": 0.0015,
      "loss": 1.9058,
      "step": 1285
    },
    {
      "epoch": 0.13565400843881856,
      "grad_norm": 0.5074360966682434,
      "learning_rate": 0.0015,
      "loss": 1.9169,
      "step": 1286
    },
    {
      "epoch": 0.13575949367088608,
      "grad_norm": 0.5905272364616394,
      "learning_rate": 0.0015,
      "loss": 1.9116,
      "step": 1287
    },
    {
      "epoch": 0.1358649789029536,
      "grad_norm": 1.001373291015625,
      "learning_rate": 0.0015,
      "loss": 1.9335,
      "step": 1288
    },
    {
      "epoch": 0.1359704641350211,
      "grad_norm": 0.9810734391212463,
      "learning_rate": 0.0015,
      "loss": 1.977,
      "step": 1289
    },
    {
      "epoch": 0.1360759493670886,
      "grad_norm": 0.7405872344970703,
      "learning_rate": 0.0015,
      "loss": 1.9381,
      "step": 1290
    },
    {
      "epoch": 0.13618143459915613,
      "grad_norm": 0.5617262721061707,
      "learning_rate": 0.0015,
      "loss": 1.9498,
      "step": 1291
    },
    {
      "epoch": 0.13628691983122362,
      "grad_norm": 0.6589044332504272,
      "learning_rate": 0.0015,
      "loss": 1.9188,
      "step": 1292
    },
    {
      "epoch": 0.13639240506329114,
      "grad_norm": 0.62932288646698,
      "learning_rate": 0.0015,
      "loss": 1.9082,
      "step": 1293
    },
    {
      "epoch": 0.13649789029535864,
      "grad_norm": 0.6271987557411194,
      "learning_rate": 0.0015,
      "loss": 1.9473,
      "step": 1294
    },
    {
      "epoch": 0.13660337552742616,
      "grad_norm": 0.567391574382782,
      "learning_rate": 0.0015,
      "loss": 1.8498,
      "step": 1295
    },
    {
      "epoch": 0.13670886075949368,
      "grad_norm": 0.5319356918334961,
      "learning_rate": 0.0015,
      "loss": 1.9092,
      "step": 1296
    },
    {
      "epoch": 0.13681434599156117,
      "grad_norm": 0.6010887622833252,
      "learning_rate": 0.0015,
      "loss": 1.9257,
      "step": 1297
    },
    {
      "epoch": 0.1369198312236287,
      "grad_norm": 0.4996672570705414,
      "learning_rate": 0.0015,
      "loss": 1.865,
      "step": 1298
    },
    {
      "epoch": 0.1370253164556962,
      "grad_norm": 0.6371192932128906,
      "learning_rate": 0.0015,
      "loss": 1.9058,
      "step": 1299
    },
    {
      "epoch": 0.1371308016877637,
      "grad_norm": 0.6034473776817322,
      "learning_rate": 0.0015,
      "loss": 1.9081,
      "step": 1300
    },
    {
      "epoch": 0.13723628691983122,
      "grad_norm": 0.5497844815254211,
      "learning_rate": 0.0015,
      "loss": 1.9084,
      "step": 1301
    },
    {
      "epoch": 0.13734177215189874,
      "grad_norm": 0.4697115123271942,
      "learning_rate": 0.0015,
      "loss": 1.9346,
      "step": 1302
    },
    {
      "epoch": 0.13744725738396624,
      "grad_norm": 0.5742257237434387,
      "learning_rate": 0.0015,
      "loss": 1.91,
      "step": 1303
    },
    {
      "epoch": 0.13755274261603376,
      "grad_norm": 0.5243496298789978,
      "learning_rate": 0.0015,
      "loss": 1.9071,
      "step": 1304
    },
    {
      "epoch": 0.13765822784810128,
      "grad_norm": 0.5664542317390442,
      "learning_rate": 0.0015,
      "loss": 1.9405,
      "step": 1305
    },
    {
      "epoch": 0.13776371308016877,
      "grad_norm": 0.5212581753730774,
      "learning_rate": 0.0015,
      "loss": 1.9191,
      "step": 1306
    },
    {
      "epoch": 0.1378691983122363,
      "grad_norm": 0.540576696395874,
      "learning_rate": 0.0015,
      "loss": 1.8893,
      "step": 1307
    },
    {
      "epoch": 0.1379746835443038,
      "grad_norm": 0.511799156665802,
      "learning_rate": 0.0015,
      "loss": 1.9379,
      "step": 1308
    },
    {
      "epoch": 0.1380801687763713,
      "grad_norm": 0.5222600698471069,
      "learning_rate": 0.0015,
      "loss": 1.8976,
      "step": 1309
    },
    {
      "epoch": 0.13818565400843882,
      "grad_norm": 0.5310981273651123,
      "learning_rate": 0.0015,
      "loss": 1.8895,
      "step": 1310
    },
    {
      "epoch": 0.13829113924050632,
      "grad_norm": 0.5159386396408081,
      "learning_rate": 0.0015,
      "loss": 1.9094,
      "step": 1311
    },
    {
      "epoch": 0.13839662447257384,
      "grad_norm": 0.5052205324172974,
      "learning_rate": 0.0015,
      "loss": 1.9066,
      "step": 1312
    },
    {
      "epoch": 0.13850210970464136,
      "grad_norm": 0.598240852355957,
      "learning_rate": 0.0015,
      "loss": 1.9448,
      "step": 1313
    },
    {
      "epoch": 0.13860759493670885,
      "grad_norm": 0.5641024112701416,
      "learning_rate": 0.0015,
      "loss": 1.9506,
      "step": 1314
    },
    {
      "epoch": 0.13871308016877637,
      "grad_norm": 0.5119905471801758,
      "learning_rate": 0.0015,
      "loss": 1.9432,
      "step": 1315
    },
    {
      "epoch": 0.1388185654008439,
      "grad_norm": 0.5300618410110474,
      "learning_rate": 0.0015,
      "loss": 1.8719,
      "step": 1316
    },
    {
      "epoch": 0.13892405063291138,
      "grad_norm": 0.5458492636680603,
      "learning_rate": 0.0015,
      "loss": 1.9017,
      "step": 1317
    },
    {
      "epoch": 0.1390295358649789,
      "grad_norm": 0.696159839630127,
      "learning_rate": 0.0015,
      "loss": 1.9102,
      "step": 1318
    },
    {
      "epoch": 0.13913502109704642,
      "grad_norm": 0.6056944131851196,
      "learning_rate": 0.0015,
      "loss": 1.9299,
      "step": 1319
    },
    {
      "epoch": 0.13924050632911392,
      "grad_norm": 0.5543187260627747,
      "learning_rate": 0.0015,
      "loss": 1.9144,
      "step": 1320
    },
    {
      "epoch": 0.13934599156118144,
      "grad_norm": 0.6323696374893188,
      "learning_rate": 0.0015,
      "loss": 1.8901,
      "step": 1321
    },
    {
      "epoch": 0.13945147679324896,
      "grad_norm": 0.7577414512634277,
      "learning_rate": 0.0015,
      "loss": 1.9112,
      "step": 1322
    },
    {
      "epoch": 0.13955696202531645,
      "grad_norm": 0.6322816014289856,
      "learning_rate": 0.0015,
      "loss": 1.9211,
      "step": 1323
    },
    {
      "epoch": 0.13966244725738397,
      "grad_norm": 0.558798611164093,
      "learning_rate": 0.0015,
      "loss": 1.9221,
      "step": 1324
    },
    {
      "epoch": 0.13976793248945146,
      "grad_norm": 0.561659038066864,
      "learning_rate": 0.0015,
      "loss": 1.8909,
      "step": 1325
    },
    {
      "epoch": 0.13987341772151898,
      "grad_norm": 0.637718141078949,
      "learning_rate": 0.0015,
      "loss": 1.9256,
      "step": 1326
    },
    {
      "epoch": 0.1399789029535865,
      "grad_norm": 0.6083945631980896,
      "learning_rate": 0.0015,
      "loss": 1.9133,
      "step": 1327
    },
    {
      "epoch": 0.140084388185654,
      "grad_norm": 0.5388321280479431,
      "learning_rate": 0.0015,
      "loss": 1.8933,
      "step": 1328
    },
    {
      "epoch": 0.14018987341772152,
      "grad_norm": 0.8044694662094116,
      "learning_rate": 0.0015,
      "loss": 1.9152,
      "step": 1329
    },
    {
      "epoch": 0.14029535864978904,
      "grad_norm": 1.0448036193847656,
      "learning_rate": 0.0015,
      "loss": 1.9463,
      "step": 1330
    },
    {
      "epoch": 0.14040084388185653,
      "grad_norm": 0.723845362663269,
      "learning_rate": 0.0015,
      "loss": 1.9355,
      "step": 1331
    },
    {
      "epoch": 0.14050632911392405,
      "grad_norm": 0.4954807162284851,
      "learning_rate": 0.0015,
      "loss": 1.9053,
      "step": 1332
    },
    {
      "epoch": 0.14061181434599157,
      "grad_norm": 0.7370380163192749,
      "learning_rate": 0.0015,
      "loss": 1.8835,
      "step": 1333
    },
    {
      "epoch": 0.14071729957805906,
      "grad_norm": 0.5626806616783142,
      "learning_rate": 0.0015,
      "loss": 1.8999,
      "step": 1334
    },
    {
      "epoch": 0.14082278481012658,
      "grad_norm": 0.642727792263031,
      "learning_rate": 0.0015,
      "loss": 1.8812,
      "step": 1335
    },
    {
      "epoch": 0.1409282700421941,
      "grad_norm": 0.7532495856285095,
      "learning_rate": 0.0015,
      "loss": 1.9145,
      "step": 1336
    },
    {
      "epoch": 0.1410337552742616,
      "grad_norm": 0.520085871219635,
      "learning_rate": 0.0015,
      "loss": 1.9333,
      "step": 1337
    },
    {
      "epoch": 0.14113924050632912,
      "grad_norm": 0.6775652766227722,
      "learning_rate": 0.0015,
      "loss": 1.9143,
      "step": 1338
    },
    {
      "epoch": 0.14124472573839664,
      "grad_norm": 0.5690031051635742,
      "learning_rate": 0.0015,
      "loss": 1.95,
      "step": 1339
    },
    {
      "epoch": 0.14135021097046413,
      "grad_norm": 0.5454837679862976,
      "learning_rate": 0.0015,
      "loss": 1.8997,
      "step": 1340
    },
    {
      "epoch": 0.14145569620253165,
      "grad_norm": 0.49931493401527405,
      "learning_rate": 0.0015,
      "loss": 1.9232,
      "step": 1341
    },
    {
      "epoch": 0.14156118143459914,
      "grad_norm": 0.49317747354507446,
      "learning_rate": 0.0015,
      "loss": 1.9455,
      "step": 1342
    },
    {
      "epoch": 0.14166666666666666,
      "grad_norm": 0.5960569381713867,
      "learning_rate": 0.0015,
      "loss": 1.8954,
      "step": 1343
    },
    {
      "epoch": 0.14177215189873418,
      "grad_norm": 0.46698129177093506,
      "learning_rate": 0.0015,
      "loss": 1.9203,
      "step": 1344
    },
    {
      "epoch": 0.14187763713080168,
      "grad_norm": 0.5159361362457275,
      "learning_rate": 0.0015,
      "loss": 1.9051,
      "step": 1345
    },
    {
      "epoch": 0.1419831223628692,
      "grad_norm": 0.5780317783355713,
      "learning_rate": 0.0015,
      "loss": 1.8936,
      "step": 1346
    },
    {
      "epoch": 0.14208860759493672,
      "grad_norm": 0.5957590341567993,
      "learning_rate": 0.0015,
      "loss": 1.9178,
      "step": 1347
    },
    {
      "epoch": 0.1421940928270042,
      "grad_norm": 0.5916475653648376,
      "learning_rate": 0.0015,
      "loss": 1.9047,
      "step": 1348
    },
    {
      "epoch": 0.14229957805907173,
      "grad_norm": 0.5349442958831787,
      "learning_rate": 0.0015,
      "loss": 1.9249,
      "step": 1349
    },
    {
      "epoch": 0.14240506329113925,
      "grad_norm": 0.5401943325996399,
      "learning_rate": 0.0015,
      "loss": 1.9202,
      "step": 1350
    },
    {
      "epoch": 0.14251054852320674,
      "grad_norm": 0.6015076041221619,
      "learning_rate": 0.0015,
      "loss": 1.9001,
      "step": 1351
    },
    {
      "epoch": 0.14261603375527426,
      "grad_norm": 0.5438253879547119,
      "learning_rate": 0.0015,
      "loss": 1.8802,
      "step": 1352
    },
    {
      "epoch": 0.14272151898734178,
      "grad_norm": 0.5660523176193237,
      "learning_rate": 0.0015,
      "loss": 1.9255,
      "step": 1353
    },
    {
      "epoch": 0.14282700421940928,
      "grad_norm": 0.530584454536438,
      "learning_rate": 0.0015,
      "loss": 1.925,
      "step": 1354
    },
    {
      "epoch": 0.1429324894514768,
      "grad_norm": 0.5304757952690125,
      "learning_rate": 0.0015,
      "loss": 1.8891,
      "step": 1355
    },
    {
      "epoch": 0.14303797468354432,
      "grad_norm": 0.5094669461250305,
      "learning_rate": 0.0015,
      "loss": 1.886,
      "step": 1356
    },
    {
      "epoch": 0.1431434599156118,
      "grad_norm": 0.5024034976959229,
      "learning_rate": 0.0015,
      "loss": 1.9161,
      "step": 1357
    },
    {
      "epoch": 0.14324894514767933,
      "grad_norm": 0.4967750608921051,
      "learning_rate": 0.0015,
      "loss": 1.9029,
      "step": 1358
    },
    {
      "epoch": 0.14335443037974682,
      "grad_norm": 0.5492123365402222,
      "learning_rate": 0.0015,
      "loss": 1.8813,
      "step": 1359
    },
    {
      "epoch": 0.14345991561181434,
      "grad_norm": 0.554287314414978,
      "learning_rate": 0.0015,
      "loss": 1.9306,
      "step": 1360
    },
    {
      "epoch": 0.14356540084388186,
      "grad_norm": 0.5291145443916321,
      "learning_rate": 0.0015,
      "loss": 1.8936,
      "step": 1361
    },
    {
      "epoch": 0.14367088607594936,
      "grad_norm": 0.49103015661239624,
      "learning_rate": 0.0015,
      "loss": 1.9026,
      "step": 1362
    },
    {
      "epoch": 0.14377637130801688,
      "grad_norm": 0.44740259647369385,
      "learning_rate": 0.0015,
      "loss": 1.9061,
      "step": 1363
    },
    {
      "epoch": 0.1438818565400844,
      "grad_norm": 0.536920428276062,
      "learning_rate": 0.0015,
      "loss": 1.9036,
      "step": 1364
    },
    {
      "epoch": 0.1439873417721519,
      "grad_norm": 0.5714421272277832,
      "learning_rate": 0.0015,
      "loss": 1.8859,
      "step": 1365
    },
    {
      "epoch": 0.1440928270042194,
      "grad_norm": 0.6274572014808655,
      "learning_rate": 0.0015,
      "loss": 1.8955,
      "step": 1366
    },
    {
      "epoch": 0.14419831223628693,
      "grad_norm": 0.6626944541931152,
      "learning_rate": 0.0015,
      "loss": 1.9207,
      "step": 1367
    },
    {
      "epoch": 0.14430379746835442,
      "grad_norm": 0.6287270188331604,
      "learning_rate": 0.0015,
      "loss": 1.8994,
      "step": 1368
    },
    {
      "epoch": 0.14440928270042194,
      "grad_norm": 0.5640893578529358,
      "learning_rate": 0.0015,
      "loss": 1.9263,
      "step": 1369
    },
    {
      "epoch": 0.14451476793248946,
      "grad_norm": 0.5738046169281006,
      "learning_rate": 0.0015,
      "loss": 1.8986,
      "step": 1370
    },
    {
      "epoch": 0.14462025316455696,
      "grad_norm": 0.5989475846290588,
      "learning_rate": 0.0015,
      "loss": 1.8974,
      "step": 1371
    },
    {
      "epoch": 0.14472573839662448,
      "grad_norm": 0.700721263885498,
      "learning_rate": 0.0015,
      "loss": 1.8893,
      "step": 1372
    },
    {
      "epoch": 0.144831223628692,
      "grad_norm": 0.6120259761810303,
      "learning_rate": 0.0015,
      "loss": 1.9043,
      "step": 1373
    },
    {
      "epoch": 0.1449367088607595,
      "grad_norm": 0.496288925409317,
      "learning_rate": 0.0015,
      "loss": 1.886,
      "step": 1374
    },
    {
      "epoch": 0.145042194092827,
      "grad_norm": 0.5448711514472961,
      "learning_rate": 0.0015,
      "loss": 1.9054,
      "step": 1375
    },
    {
      "epoch": 0.1451476793248945,
      "grad_norm": 0.6740733981132507,
      "learning_rate": 0.0015,
      "loss": 1.8896,
      "step": 1376
    },
    {
      "epoch": 0.14525316455696202,
      "grad_norm": 0.506820797920227,
      "learning_rate": 0.0015,
      "loss": 1.8615,
      "step": 1377
    },
    {
      "epoch": 0.14535864978902954,
      "grad_norm": 0.6523921489715576,
      "learning_rate": 0.0015,
      "loss": 1.8576,
      "step": 1378
    },
    {
      "epoch": 0.14546413502109704,
      "grad_norm": 0.8767607808113098,
      "learning_rate": 0.0015,
      "loss": 1.9352,
      "step": 1379
    },
    {
      "epoch": 0.14556962025316456,
      "grad_norm": 0.7612723112106323,
      "learning_rate": 0.0015,
      "loss": 1.8807,
      "step": 1380
    },
    {
      "epoch": 0.14567510548523208,
      "grad_norm": 0.5680716037750244,
      "learning_rate": 0.0015,
      "loss": 1.9319,
      "step": 1381
    },
    {
      "epoch": 0.14578059071729957,
      "grad_norm": 0.6475235223770142,
      "learning_rate": 0.0015,
      "loss": 1.9018,
      "step": 1382
    },
    {
      "epoch": 0.1458860759493671,
      "grad_norm": 0.9306323528289795,
      "learning_rate": 0.0015,
      "loss": 1.8967,
      "step": 1383
    },
    {
      "epoch": 0.1459915611814346,
      "grad_norm": 0.5690900683403015,
      "learning_rate": 0.0015,
      "loss": 1.88,
      "step": 1384
    },
    {
      "epoch": 0.1460970464135021,
      "grad_norm": 0.5900740027427673,
      "learning_rate": 0.0015,
      "loss": 1.9248,
      "step": 1385
    },
    {
      "epoch": 0.14620253164556962,
      "grad_norm": 0.6930634379386902,
      "learning_rate": 0.0015,
      "loss": 1.9081,
      "step": 1386
    },
    {
      "epoch": 0.14630801687763714,
      "grad_norm": 0.6711006760597229,
      "learning_rate": 0.0015,
      "loss": 1.9043,
      "step": 1387
    },
    {
      "epoch": 0.14641350210970464,
      "grad_norm": 0.5288648009300232,
      "learning_rate": 0.0015,
      "loss": 1.894,
      "step": 1388
    },
    {
      "epoch": 0.14651898734177216,
      "grad_norm": 0.5598739981651306,
      "learning_rate": 0.0015,
      "loss": 1.8999,
      "step": 1389
    },
    {
      "epoch": 0.14662447257383968,
      "grad_norm": 0.7916274070739746,
      "learning_rate": 0.0015,
      "loss": 1.9217,
      "step": 1390
    },
    {
      "epoch": 0.14672995780590717,
      "grad_norm": 0.775980532169342,
      "learning_rate": 0.0015,
      "loss": 1.9335,
      "step": 1391
    },
    {
      "epoch": 0.1468354430379747,
      "grad_norm": 0.5397216081619263,
      "learning_rate": 0.0015,
      "loss": 1.9054,
      "step": 1392
    },
    {
      "epoch": 0.14694092827004218,
      "grad_norm": 0.544588029384613,
      "learning_rate": 0.0015,
      "loss": 1.8826,
      "step": 1393
    },
    {
      "epoch": 0.1470464135021097,
      "grad_norm": 0.7192983031272888,
      "learning_rate": 0.0015,
      "loss": 1.8856,
      "step": 1394
    },
    {
      "epoch": 0.14715189873417722,
      "grad_norm": 0.7209638953208923,
      "learning_rate": 0.0015,
      "loss": 1.9181,
      "step": 1395
    },
    {
      "epoch": 0.14725738396624471,
      "grad_norm": 0.4494272470474243,
      "learning_rate": 0.0015,
      "loss": 1.8951,
      "step": 1396
    },
    {
      "epoch": 0.14736286919831224,
      "grad_norm": 0.5227426290512085,
      "learning_rate": 0.0015,
      "loss": 1.8495,
      "step": 1397
    },
    {
      "epoch": 0.14746835443037976,
      "grad_norm": 0.4833694398403168,
      "learning_rate": 0.0015,
      "loss": 1.8527,
      "step": 1398
    },
    {
      "epoch": 0.14757383966244725,
      "grad_norm": 0.5250594615936279,
      "learning_rate": 0.0015,
      "loss": 1.8901,
      "step": 1399
    },
    {
      "epoch": 0.14767932489451477,
      "grad_norm": 0.5358574986457825,
      "learning_rate": 0.0015,
      "loss": 1.855,
      "step": 1400
    },
    {
      "epoch": 0.1477848101265823,
      "grad_norm": 0.6389911770820618,
      "learning_rate": 0.0015,
      "loss": 1.9013,
      "step": 1401
    },
    {
      "epoch": 0.14789029535864978,
      "grad_norm": 0.6424770355224609,
      "learning_rate": 0.0015,
      "loss": 1.8813,
      "step": 1402
    },
    {
      "epoch": 0.1479957805907173,
      "grad_norm": 0.5069495439529419,
      "learning_rate": 0.0015,
      "loss": 1.9264,
      "step": 1403
    },
    {
      "epoch": 0.14810126582278482,
      "grad_norm": 0.5170621871948242,
      "learning_rate": 0.0015,
      "loss": 1.8949,
      "step": 1404
    },
    {
      "epoch": 0.14820675105485231,
      "grad_norm": 0.5447293519973755,
      "learning_rate": 0.0015,
      "loss": 1.8721,
      "step": 1405
    },
    {
      "epoch": 0.14831223628691984,
      "grad_norm": 0.5404070019721985,
      "learning_rate": 0.0015,
      "loss": 1.8762,
      "step": 1406
    },
    {
      "epoch": 0.14841772151898736,
      "grad_norm": 0.5344438552856445,
      "learning_rate": 0.0015,
      "loss": 1.8184,
      "step": 1407
    },
    {
      "epoch": 0.14852320675105485,
      "grad_norm": 0.542682409286499,
      "learning_rate": 0.0015,
      "loss": 1.9025,
      "step": 1408
    },
    {
      "epoch": 0.14862869198312237,
      "grad_norm": 0.5511244535446167,
      "learning_rate": 0.0015,
      "loss": 1.9206,
      "step": 1409
    },
    {
      "epoch": 0.14873417721518986,
      "grad_norm": 0.5384794473648071,
      "learning_rate": 0.0015,
      "loss": 1.9052,
      "step": 1410
    },
    {
      "epoch": 0.14883966244725738,
      "grad_norm": 0.5632681250572205,
      "learning_rate": 0.0015,
      "loss": 1.9285,
      "step": 1411
    },
    {
      "epoch": 0.1489451476793249,
      "grad_norm": 0.6840286254882812,
      "learning_rate": 0.0015,
      "loss": 1.8769,
      "step": 1412
    },
    {
      "epoch": 0.1490506329113924,
      "grad_norm": 0.5219169855117798,
      "learning_rate": 0.0015,
      "loss": 1.8857,
      "step": 1413
    },
    {
      "epoch": 0.14915611814345991,
      "grad_norm": 0.6446149945259094,
      "learning_rate": 0.0015,
      "loss": 1.9076,
      "step": 1414
    },
    {
      "epoch": 0.14926160337552744,
      "grad_norm": 0.8110989928245544,
      "learning_rate": 0.0015,
      "loss": 1.8654,
      "step": 1415
    },
    {
      "epoch": 0.14936708860759493,
      "grad_norm": 0.7284207344055176,
      "learning_rate": 0.0015,
      "loss": 1.9381,
      "step": 1416
    },
    {
      "epoch": 0.14947257383966245,
      "grad_norm": 0.49806055426597595,
      "learning_rate": 0.0015,
      "loss": 1.8676,
      "step": 1417
    },
    {
      "epoch": 0.14957805907172997,
      "grad_norm": 0.5483485460281372,
      "learning_rate": 0.0015,
      "loss": 1.9014,
      "step": 1418
    },
    {
      "epoch": 0.14968354430379746,
      "grad_norm": 0.6617444753646851,
      "learning_rate": 0.0015,
      "loss": 1.8951,
      "step": 1419
    },
    {
      "epoch": 0.14978902953586498,
      "grad_norm": 0.5816461443901062,
      "learning_rate": 0.0015,
      "loss": 1.9154,
      "step": 1420
    },
    {
      "epoch": 0.1498945147679325,
      "grad_norm": 0.6530291438102722,
      "learning_rate": 0.0015,
      "loss": 1.8919,
      "step": 1421
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7094903588294983,
      "learning_rate": 0.0015,
      "loss": 1.8866,
      "step": 1422
    },
    {
      "epoch": 0.15010548523206751,
      "grad_norm": 0.5501838326454163,
      "learning_rate": 0.0015,
      "loss": 1.8919,
      "step": 1423
    },
    {
      "epoch": 0.150210970464135,
      "grad_norm": 0.47818896174430847,
      "learning_rate": 0.0015,
      "loss": 1.8814,
      "step": 1424
    },
    {
      "epoch": 0.15031645569620253,
      "grad_norm": 0.47611212730407715,
      "learning_rate": 0.0015,
      "loss": 1.8664,
      "step": 1425
    },
    {
      "epoch": 0.15042194092827005,
      "grad_norm": 0.4526898264884949,
      "learning_rate": 0.0015,
      "loss": 1.9122,
      "step": 1426
    },
    {
      "epoch": 0.15052742616033754,
      "grad_norm": 0.5256379246711731,
      "learning_rate": 0.0015,
      "loss": 1.9339,
      "step": 1427
    },
    {
      "epoch": 0.15063291139240506,
      "grad_norm": 0.5785458087921143,
      "learning_rate": 0.0015,
      "loss": 1.8744,
      "step": 1428
    },
    {
      "epoch": 0.15073839662447258,
      "grad_norm": 0.6067950129508972,
      "learning_rate": 0.0015,
      "loss": 1.8688,
      "step": 1429
    },
    {
      "epoch": 0.15084388185654007,
      "grad_norm": 0.6320534348487854,
      "learning_rate": 0.0015,
      "loss": 1.8747,
      "step": 1430
    },
    {
      "epoch": 0.1509493670886076,
      "grad_norm": 0.49489179253578186,
      "learning_rate": 0.0015,
      "loss": 1.8906,
      "step": 1431
    },
    {
      "epoch": 0.15105485232067511,
      "grad_norm": 0.6265627145767212,
      "learning_rate": 0.0015,
      "loss": 1.8953,
      "step": 1432
    },
    {
      "epoch": 0.1511603375527426,
      "grad_norm": 0.6951988339424133,
      "learning_rate": 0.0015,
      "loss": 1.8505,
      "step": 1433
    },
    {
      "epoch": 0.15126582278481013,
      "grad_norm": 0.5689625144004822,
      "learning_rate": 0.0015,
      "loss": 1.8699,
      "step": 1434
    },
    {
      "epoch": 0.15137130801687765,
      "grad_norm": 0.5780248641967773,
      "learning_rate": 0.0015,
      "loss": 1.9143,
      "step": 1435
    },
    {
      "epoch": 0.15147679324894514,
      "grad_norm": 0.7600285410881042,
      "learning_rate": 0.0015,
      "loss": 1.8772,
      "step": 1436
    },
    {
      "epoch": 0.15158227848101266,
      "grad_norm": 0.6233757138252258,
      "learning_rate": 0.0015,
      "loss": 1.9013,
      "step": 1437
    },
    {
      "epoch": 0.15168776371308018,
      "grad_norm": 0.5610736012458801,
      "learning_rate": 0.0015,
      "loss": 1.8579,
      "step": 1438
    },
    {
      "epoch": 0.15179324894514767,
      "grad_norm": 0.6786007285118103,
      "learning_rate": 0.0015,
      "loss": 1.9008,
      "step": 1439
    },
    {
      "epoch": 0.1518987341772152,
      "grad_norm": 0.7043277025222778,
      "learning_rate": 0.0015,
      "loss": 1.8911,
      "step": 1440
    },
    {
      "epoch": 0.1520042194092827,
      "grad_norm": 0.5467262864112854,
      "learning_rate": 0.0015,
      "loss": 1.8553,
      "step": 1441
    },
    {
      "epoch": 0.1521097046413502,
      "grad_norm": 0.5686880946159363,
      "learning_rate": 0.0015,
      "loss": 1.8951,
      "step": 1442
    },
    {
      "epoch": 0.15221518987341773,
      "grad_norm": 0.6156380772590637,
      "learning_rate": 0.0015,
      "loss": 1.8887,
      "step": 1443
    },
    {
      "epoch": 0.15232067510548522,
      "grad_norm": 0.5663610696792603,
      "learning_rate": 0.0015,
      "loss": 1.896,
      "step": 1444
    },
    {
      "epoch": 0.15242616033755274,
      "grad_norm": 0.4784590005874634,
      "learning_rate": 0.0015,
      "loss": 1.8684,
      "step": 1445
    },
    {
      "epoch": 0.15253164556962026,
      "grad_norm": 0.5938230156898499,
      "learning_rate": 0.0015,
      "loss": 1.8793,
      "step": 1446
    },
    {
      "epoch": 0.15263713080168775,
      "grad_norm": 0.49647626280784607,
      "learning_rate": 0.0015,
      "loss": 1.8941,
      "step": 1447
    },
    {
      "epoch": 0.15274261603375527,
      "grad_norm": 0.4847690463066101,
      "learning_rate": 0.0015,
      "loss": 1.8821,
      "step": 1448
    },
    {
      "epoch": 0.1528481012658228,
      "grad_norm": 0.4634542465209961,
      "learning_rate": 0.0015,
      "loss": 1.9033,
      "step": 1449
    },
    {
      "epoch": 0.1529535864978903,
      "grad_norm": 0.4846337139606476,
      "learning_rate": 0.0015,
      "loss": 1.9016,
      "step": 1450
    },
    {
      "epoch": 0.1530590717299578,
      "grad_norm": 0.46131351590156555,
      "learning_rate": 0.0015,
      "loss": 1.8913,
      "step": 1451
    },
    {
      "epoch": 0.15316455696202533,
      "grad_norm": 0.4795179069042206,
      "learning_rate": 0.0015,
      "loss": 1.8861,
      "step": 1452
    },
    {
      "epoch": 0.15327004219409282,
      "grad_norm": 0.5590512156486511,
      "learning_rate": 0.0015,
      "loss": 1.861,
      "step": 1453
    },
    {
      "epoch": 0.15337552742616034,
      "grad_norm": 0.6019673943519592,
      "learning_rate": 0.0015,
      "loss": 1.8494,
      "step": 1454
    },
    {
      "epoch": 0.15348101265822786,
      "grad_norm": 0.5595062375068665,
      "learning_rate": 0.0015,
      "loss": 1.8629,
      "step": 1455
    },
    {
      "epoch": 0.15358649789029535,
      "grad_norm": 0.5171561241149902,
      "learning_rate": 0.0015,
      "loss": 1.9071,
      "step": 1456
    },
    {
      "epoch": 0.15369198312236287,
      "grad_norm": 0.6185160279273987,
      "learning_rate": 0.0015,
      "loss": 1.8534,
      "step": 1457
    },
    {
      "epoch": 0.15379746835443037,
      "grad_norm": 0.6364607214927673,
      "learning_rate": 0.0015,
      "loss": 1.8662,
      "step": 1458
    },
    {
      "epoch": 0.1539029535864979,
      "grad_norm": 0.4733855426311493,
      "learning_rate": 0.0015,
      "loss": 1.8811,
      "step": 1459
    },
    {
      "epoch": 0.1540084388185654,
      "grad_norm": 0.5263296365737915,
      "learning_rate": 0.0015,
      "loss": 1.9313,
      "step": 1460
    },
    {
      "epoch": 0.1541139240506329,
      "grad_norm": 0.5156674981117249,
      "learning_rate": 0.0015,
      "loss": 1.8937,
      "step": 1461
    },
    {
      "epoch": 0.15421940928270042,
      "grad_norm": 0.4853241741657257,
      "learning_rate": 0.0015,
      "loss": 1.8315,
      "step": 1462
    },
    {
      "epoch": 0.15432489451476794,
      "grad_norm": 0.49924007058143616,
      "learning_rate": 0.0015,
      "loss": 1.8542,
      "step": 1463
    },
    {
      "epoch": 0.15443037974683543,
      "grad_norm": 0.5335673093795776,
      "learning_rate": 0.0015,
      "loss": 1.8749,
      "step": 1464
    },
    {
      "epoch": 0.15453586497890295,
      "grad_norm": 0.5205351114273071,
      "learning_rate": 0.0015,
      "loss": 1.8589,
      "step": 1465
    },
    {
      "epoch": 0.15464135021097047,
      "grad_norm": 0.5043001770973206,
      "learning_rate": 0.0015,
      "loss": 1.8689,
      "step": 1466
    },
    {
      "epoch": 0.15474683544303797,
      "grad_norm": 0.5932759642601013,
      "learning_rate": 0.0015,
      "loss": 1.9155,
      "step": 1467
    },
    {
      "epoch": 0.1548523206751055,
      "grad_norm": 0.6116816401481628,
      "learning_rate": 0.0015,
      "loss": 1.8978,
      "step": 1468
    },
    {
      "epoch": 0.154957805907173,
      "grad_norm": 0.43347838521003723,
      "learning_rate": 0.0015,
      "loss": 1.9024,
      "step": 1469
    },
    {
      "epoch": 0.1550632911392405,
      "grad_norm": 0.6800468564033508,
      "learning_rate": 0.0015,
      "loss": 1.8802,
      "step": 1470
    },
    {
      "epoch": 0.15516877637130802,
      "grad_norm": 0.7267336845397949,
      "learning_rate": 0.0015,
      "loss": 1.9061,
      "step": 1471
    },
    {
      "epoch": 0.15527426160337554,
      "grad_norm": 0.5035741925239563,
      "learning_rate": 0.0015,
      "loss": 1.9023,
      "step": 1472
    },
    {
      "epoch": 0.15537974683544303,
      "grad_norm": 0.55813068151474,
      "learning_rate": 0.0015,
      "loss": 1.8653,
      "step": 1473
    },
    {
      "epoch": 0.15548523206751055,
      "grad_norm": 0.6493954658508301,
      "learning_rate": 0.0015,
      "loss": 1.8819,
      "step": 1474
    },
    {
      "epoch": 0.15559071729957805,
      "grad_norm": 0.5621296763420105,
      "learning_rate": 0.0015,
      "loss": 1.8897,
      "step": 1475
    },
    {
      "epoch": 0.15569620253164557,
      "grad_norm": 0.5570229887962341,
      "learning_rate": 0.0015,
      "loss": 1.88,
      "step": 1476
    },
    {
      "epoch": 0.1558016877637131,
      "grad_norm": 0.6718025207519531,
      "learning_rate": 0.0015,
      "loss": 1.8785,
      "step": 1477
    },
    {
      "epoch": 0.15590717299578058,
      "grad_norm": 0.56767737865448,
      "learning_rate": 0.0015,
      "loss": 1.8757,
      "step": 1478
    },
    {
      "epoch": 0.1560126582278481,
      "grad_norm": 0.6203089952468872,
      "learning_rate": 0.0015,
      "loss": 1.9055,
      "step": 1479
    },
    {
      "epoch": 0.15611814345991562,
      "grad_norm": 0.6166823506355286,
      "learning_rate": 0.0015,
      "loss": 1.8435,
      "step": 1480
    },
    {
      "epoch": 0.1562236286919831,
      "grad_norm": 0.5030540823936462,
      "learning_rate": 0.0015,
      "loss": 1.895,
      "step": 1481
    },
    {
      "epoch": 0.15632911392405063,
      "grad_norm": 0.6684638261795044,
      "learning_rate": 0.0015,
      "loss": 1.8784,
      "step": 1482
    },
    {
      "epoch": 0.15643459915611815,
      "grad_norm": 0.6513040661811829,
      "learning_rate": 0.0015,
      "loss": 1.8804,
      "step": 1483
    },
    {
      "epoch": 0.15654008438818565,
      "grad_norm": 0.5312937498092651,
      "learning_rate": 0.0015,
      "loss": 1.9073,
      "step": 1484
    },
    {
      "epoch": 0.15664556962025317,
      "grad_norm": 0.5202449560165405,
      "learning_rate": 0.0015,
      "loss": 1.8449,
      "step": 1485
    },
    {
      "epoch": 0.1567510548523207,
      "grad_norm": 0.47056058049201965,
      "learning_rate": 0.0015,
      "loss": 1.9037,
      "step": 1486
    },
    {
      "epoch": 0.15685654008438818,
      "grad_norm": 0.49510762095451355,
      "learning_rate": 0.0015,
      "loss": 1.8363,
      "step": 1487
    },
    {
      "epoch": 0.1569620253164557,
      "grad_norm": 0.4986557960510254,
      "learning_rate": 0.0015,
      "loss": 1.8992,
      "step": 1488
    },
    {
      "epoch": 0.15706751054852322,
      "grad_norm": 0.4750652015209198,
      "learning_rate": 0.0015,
      "loss": 1.8632,
      "step": 1489
    },
    {
      "epoch": 0.1571729957805907,
      "grad_norm": 0.5059555172920227,
      "learning_rate": 0.0015,
      "loss": 1.8714,
      "step": 1490
    },
    {
      "epoch": 0.15727848101265823,
      "grad_norm": 0.4990314245223999,
      "learning_rate": 0.0015,
      "loss": 1.8459,
      "step": 1491
    },
    {
      "epoch": 0.15738396624472573,
      "grad_norm": 0.49329227209091187,
      "learning_rate": 0.0015,
      "loss": 1.8777,
      "step": 1492
    },
    {
      "epoch": 0.15748945147679325,
      "grad_norm": 0.5516387224197388,
      "learning_rate": 0.0015,
      "loss": 1.8695,
      "step": 1493
    },
    {
      "epoch": 0.15759493670886077,
      "grad_norm": 0.6168501973152161,
      "learning_rate": 0.0015,
      "loss": 1.8702,
      "step": 1494
    },
    {
      "epoch": 0.15770042194092826,
      "grad_norm": 0.5879580974578857,
      "learning_rate": 0.0015,
      "loss": 1.8553,
      "step": 1495
    },
    {
      "epoch": 0.15780590717299578,
      "grad_norm": 0.5255633592605591,
      "learning_rate": 0.0015,
      "loss": 1.9239,
      "step": 1496
    },
    {
      "epoch": 0.1579113924050633,
      "grad_norm": 0.49261853098869324,
      "learning_rate": 0.0015,
      "loss": 1.8797,
      "step": 1497
    },
    {
      "epoch": 0.1580168776371308,
      "grad_norm": 0.552263617515564,
      "learning_rate": 0.0015,
      "loss": 1.8586,
      "step": 1498
    },
    {
      "epoch": 0.1581223628691983,
      "grad_norm": 0.574817955493927,
      "learning_rate": 0.0015,
      "loss": 1.9088,
      "step": 1499
    },
    {
      "epoch": 0.15822784810126583,
      "grad_norm": 0.49150896072387695,
      "learning_rate": 0.0015,
      "loss": 1.8756,
      "step": 1500
    },
    {
      "epoch": 0.15833333333333333,
      "grad_norm": 0.5402496457099915,
      "learning_rate": 0.0015,
      "loss": 1.8705,
      "step": 1501
    },
    {
      "epoch": 0.15843881856540085,
      "grad_norm": 0.6506123542785645,
      "learning_rate": 0.0015,
      "loss": 1.866,
      "step": 1502
    },
    {
      "epoch": 0.15854430379746837,
      "grad_norm": 0.7406247854232788,
      "learning_rate": 0.0015,
      "loss": 1.8821,
      "step": 1503
    },
    {
      "epoch": 0.15864978902953586,
      "grad_norm": 0.6048151850700378,
      "learning_rate": 0.0015,
      "loss": 1.9199,
      "step": 1504
    },
    {
      "epoch": 0.15875527426160338,
      "grad_norm": 0.5365409851074219,
      "learning_rate": 0.0015,
      "loss": 1.8691,
      "step": 1505
    },
    {
      "epoch": 0.15886075949367087,
      "grad_norm": 0.8235931396484375,
      "learning_rate": 0.0015,
      "loss": 1.859,
      "step": 1506
    },
    {
      "epoch": 0.1589662447257384,
      "grad_norm": 1.0617879629135132,
      "learning_rate": 0.0015,
      "loss": 1.9236,
      "step": 1507
    },
    {
      "epoch": 0.1590717299578059,
      "grad_norm": 0.7661322355270386,
      "learning_rate": 0.0015,
      "loss": 1.8518,
      "step": 1508
    },
    {
      "epoch": 0.1591772151898734,
      "grad_norm": 0.5323702096939087,
      "learning_rate": 0.0015,
      "loss": 1.8727,
      "step": 1509
    },
    {
      "epoch": 0.15928270042194093,
      "grad_norm": 0.9609919786453247,
      "learning_rate": 0.0015,
      "loss": 1.8842,
      "step": 1510
    },
    {
      "epoch": 0.15938818565400845,
      "grad_norm": 0.8012824654579163,
      "learning_rate": 0.0015,
      "loss": 1.8607,
      "step": 1511
    },
    {
      "epoch": 0.15949367088607594,
      "grad_norm": 0.5442456603050232,
      "learning_rate": 0.0015,
      "loss": 1.8757,
      "step": 1512
    },
    {
      "epoch": 0.15959915611814346,
      "grad_norm": 0.8813095092773438,
      "learning_rate": 0.0015,
      "loss": 1.9034,
      "step": 1513
    },
    {
      "epoch": 0.15970464135021098,
      "grad_norm": 0.704605758190155,
      "learning_rate": 0.0015,
      "loss": 1.8746,
      "step": 1514
    },
    {
      "epoch": 0.15981012658227847,
      "grad_norm": 0.6096099615097046,
      "learning_rate": 0.0015,
      "loss": 1.9057,
      "step": 1515
    },
    {
      "epoch": 0.159915611814346,
      "grad_norm": 0.6169876456260681,
      "learning_rate": 0.0015,
      "loss": 1.8855,
      "step": 1516
    },
    {
      "epoch": 0.1600210970464135,
      "grad_norm": 0.7339420914649963,
      "learning_rate": 0.0015,
      "loss": 1.8827,
      "step": 1517
    },
    {
      "epoch": 0.160126582278481,
      "grad_norm": 0.718132495880127,
      "learning_rate": 0.0015,
      "loss": 1.8824,
      "step": 1518
    },
    {
      "epoch": 0.16023206751054853,
      "grad_norm": 0.5238819122314453,
      "learning_rate": 0.0015,
      "loss": 1.8892,
      "step": 1519
    },
    {
      "epoch": 0.16033755274261605,
      "grad_norm": 0.5656552314758301,
      "learning_rate": 0.0015,
      "loss": 1.8712,
      "step": 1520
    },
    {
      "epoch": 0.16044303797468354,
      "grad_norm": 0.6397966146469116,
      "learning_rate": 0.0015,
      "loss": 1.8874,
      "step": 1521
    },
    {
      "epoch": 0.16054852320675106,
      "grad_norm": 0.5664899349212646,
      "learning_rate": 0.0015,
      "loss": 1.8662,
      "step": 1522
    },
    {
      "epoch": 0.16065400843881855,
      "grad_norm": 0.5070544481277466,
      "learning_rate": 0.0015,
      "loss": 1.8484,
      "step": 1523
    },
    {
      "epoch": 0.16075949367088607,
      "grad_norm": 0.5437641143798828,
      "learning_rate": 0.0015,
      "loss": 1.8599,
      "step": 1524
    },
    {
      "epoch": 0.1608649789029536,
      "grad_norm": 0.5785396695137024,
      "learning_rate": 0.0015,
      "loss": 1.8813,
      "step": 1525
    },
    {
      "epoch": 0.16097046413502109,
      "grad_norm": 0.5251486301422119,
      "learning_rate": 0.0015,
      "loss": 1.8095,
      "step": 1526
    },
    {
      "epoch": 0.1610759493670886,
      "grad_norm": 0.4845963716506958,
      "learning_rate": 0.0015,
      "loss": 1.9025,
      "step": 1527
    },
    {
      "epoch": 0.16118143459915613,
      "grad_norm": 0.5619175434112549,
      "learning_rate": 0.0015,
      "loss": 1.8383,
      "step": 1528
    },
    {
      "epoch": 0.16128691983122362,
      "grad_norm": 0.5256401896476746,
      "learning_rate": 0.0015,
      "loss": 1.9047,
      "step": 1529
    },
    {
      "epoch": 0.16139240506329114,
      "grad_norm": 0.4932977259159088,
      "learning_rate": 0.0015,
      "loss": 1.841,
      "step": 1530
    },
    {
      "epoch": 0.16149789029535866,
      "grad_norm": 0.6641926169395447,
      "learning_rate": 0.0015,
      "loss": 1.8212,
      "step": 1531
    },
    {
      "epoch": 0.16160337552742615,
      "grad_norm": 0.561372697353363,
      "learning_rate": 0.0015,
      "loss": 1.8225,
      "step": 1532
    },
    {
      "epoch": 0.16170886075949367,
      "grad_norm": 0.4468382000923157,
      "learning_rate": 0.0015,
      "loss": 1.8606,
      "step": 1533
    },
    {
      "epoch": 0.1618143459915612,
      "grad_norm": 0.5442577600479126,
      "learning_rate": 0.0015,
      "loss": 1.8562,
      "step": 1534
    },
    {
      "epoch": 0.16191983122362869,
      "grad_norm": 0.5323686003684998,
      "learning_rate": 0.0015,
      "loss": 1.8833,
      "step": 1535
    },
    {
      "epoch": 0.1620253164556962,
      "grad_norm": 0.5348008871078491,
      "learning_rate": 0.0015,
      "loss": 1.8188,
      "step": 1536
    },
    {
      "epoch": 0.16213080168776373,
      "grad_norm": 0.841894268989563,
      "learning_rate": 0.0015,
      "loss": 1.8405,
      "step": 1537
    },
    {
      "epoch": 0.16223628691983122,
      "grad_norm": 0.7206017374992371,
      "learning_rate": 0.0015,
      "loss": 1.8518,
      "step": 1538
    },
    {
      "epoch": 0.16234177215189874,
      "grad_norm": 0.5118170380592346,
      "learning_rate": 0.0015,
      "loss": 1.8518,
      "step": 1539
    },
    {
      "epoch": 0.16244725738396623,
      "grad_norm": 0.6494106650352478,
      "learning_rate": 0.0015,
      "loss": 1.8818,
      "step": 1540
    },
    {
      "epoch": 0.16255274261603375,
      "grad_norm": 0.7285601496696472,
      "learning_rate": 0.0015,
      "loss": 1.8676,
      "step": 1541
    },
    {
      "epoch": 0.16265822784810127,
      "grad_norm": 0.45697057247161865,
      "learning_rate": 0.0015,
      "loss": 1.8189,
      "step": 1542
    },
    {
      "epoch": 0.16276371308016876,
      "grad_norm": 0.6642175316810608,
      "learning_rate": 0.0015,
      "loss": 1.8592,
      "step": 1543
    },
    {
      "epoch": 0.16286919831223629,
      "grad_norm": 0.7059324979782104,
      "learning_rate": 0.0015,
      "loss": 1.8697,
      "step": 1544
    },
    {
      "epoch": 0.1629746835443038,
      "grad_norm": 0.5232691168785095,
      "learning_rate": 0.0015,
      "loss": 1.8293,
      "step": 1545
    },
    {
      "epoch": 0.1630801687763713,
      "grad_norm": 0.5357704758644104,
      "learning_rate": 0.0015,
      "loss": 1.8315,
      "step": 1546
    },
    {
      "epoch": 0.16318565400843882,
      "grad_norm": 0.5508078932762146,
      "learning_rate": 0.0015,
      "loss": 1.855,
      "step": 1547
    },
    {
      "epoch": 0.16329113924050634,
      "grad_norm": 0.582895815372467,
      "learning_rate": 0.0015,
      "loss": 1.8947,
      "step": 1548
    },
    {
      "epoch": 0.16339662447257383,
      "grad_norm": 0.5417135953903198,
      "learning_rate": 0.0015,
      "loss": 1.8855,
      "step": 1549
    },
    {
      "epoch": 0.16350210970464135,
      "grad_norm": 0.5199078321456909,
      "learning_rate": 0.0015,
      "loss": 1.8524,
      "step": 1550
    },
    {
      "epoch": 0.16360759493670887,
      "grad_norm": 0.5891591310501099,
      "learning_rate": 0.0015,
      "loss": 1.8633,
      "step": 1551
    },
    {
      "epoch": 0.16371308016877636,
      "grad_norm": 0.5260766744613647,
      "learning_rate": 0.0015,
      "loss": 1.8756,
      "step": 1552
    },
    {
      "epoch": 0.16381856540084389,
      "grad_norm": 0.532564640045166,
      "learning_rate": 0.0015,
      "loss": 1.9029,
      "step": 1553
    },
    {
      "epoch": 0.1639240506329114,
      "grad_norm": 0.5645464062690735,
      "learning_rate": 0.0015,
      "loss": 1.8243,
      "step": 1554
    },
    {
      "epoch": 0.1640295358649789,
      "grad_norm": 0.6165211796760559,
      "learning_rate": 0.0015,
      "loss": 1.8533,
      "step": 1555
    },
    {
      "epoch": 0.16413502109704642,
      "grad_norm": 0.592799186706543,
      "learning_rate": 0.0015,
      "loss": 1.8479,
      "step": 1556
    },
    {
      "epoch": 0.1642405063291139,
      "grad_norm": 0.7791257500648499,
      "learning_rate": 0.0015,
      "loss": 1.8865,
      "step": 1557
    },
    {
      "epoch": 0.16434599156118143,
      "grad_norm": 0.9933264851570129,
      "learning_rate": 0.0015,
      "loss": 1.8491,
      "step": 1558
    },
    {
      "epoch": 0.16445147679324895,
      "grad_norm": 0.8477607369422913,
      "learning_rate": 0.0015,
      "loss": 1.828,
      "step": 1559
    },
    {
      "epoch": 0.16455696202531644,
      "grad_norm": 0.6159766912460327,
      "learning_rate": 0.0015,
      "loss": 1.8583,
      "step": 1560
    },
    {
      "epoch": 0.16466244725738396,
      "grad_norm": 0.8783729076385498,
      "learning_rate": 0.0015,
      "loss": 1.8076,
      "step": 1561
    },
    {
      "epoch": 0.16476793248945149,
      "grad_norm": 0.8979581594467163,
      "learning_rate": 0.0015,
      "loss": 1.8854,
      "step": 1562
    },
    {
      "epoch": 0.16487341772151898,
      "grad_norm": 0.5556858777999878,
      "learning_rate": 0.0015,
      "loss": 1.8422,
      "step": 1563
    },
    {
      "epoch": 0.1649789029535865,
      "grad_norm": 0.9216030240058899,
      "learning_rate": 0.0015,
      "loss": 1.8519,
      "step": 1564
    },
    {
      "epoch": 0.16508438818565402,
      "grad_norm": 1.0365732908248901,
      "learning_rate": 0.0015,
      "loss": 1.8103,
      "step": 1565
    },
    {
      "epoch": 0.1651898734177215,
      "grad_norm": 0.4660235047340393,
      "learning_rate": 0.0015,
      "loss": 1.8632,
      "step": 1566
    },
    {
      "epoch": 0.16529535864978903,
      "grad_norm": 0.9182472825050354,
      "learning_rate": 0.0015,
      "loss": 1.8678,
      "step": 1567
    },
    {
      "epoch": 0.16540084388185655,
      "grad_norm": 0.7967995405197144,
      "learning_rate": 0.0015,
      "loss": 1.8962,
      "step": 1568
    },
    {
      "epoch": 0.16550632911392404,
      "grad_norm": 0.5126940608024597,
      "learning_rate": 0.0015,
      "loss": 1.8474,
      "step": 1569
    },
    {
      "epoch": 0.16561181434599156,
      "grad_norm": 0.7454519867897034,
      "learning_rate": 0.0015,
      "loss": 1.8575,
      "step": 1570
    },
    {
      "epoch": 0.16571729957805909,
      "grad_norm": 0.5309233069419861,
      "learning_rate": 0.0015,
      "loss": 1.829,
      "step": 1571
    },
    {
      "epoch": 0.16582278481012658,
      "grad_norm": 0.596048891544342,
      "learning_rate": 0.0015,
      "loss": 1.877,
      "step": 1572
    },
    {
      "epoch": 0.1659282700421941,
      "grad_norm": 0.5971057415008545,
      "learning_rate": 0.0015,
      "loss": 1.8776,
      "step": 1573
    },
    {
      "epoch": 0.1660337552742616,
      "grad_norm": 0.5350774526596069,
      "learning_rate": 0.0015,
      "loss": 1.8381,
      "step": 1574
    },
    {
      "epoch": 0.1661392405063291,
      "grad_norm": 0.6208733320236206,
      "learning_rate": 0.0015,
      "loss": 1.8453,
      "step": 1575
    },
    {
      "epoch": 0.16624472573839663,
      "grad_norm": 0.5253459811210632,
      "learning_rate": 0.0015,
      "loss": 1.8794,
      "step": 1576
    },
    {
      "epoch": 0.16635021097046412,
      "grad_norm": 0.566020131111145,
      "learning_rate": 0.0015,
      "loss": 1.8807,
      "step": 1577
    },
    {
      "epoch": 0.16645569620253164,
      "grad_norm": 0.6361575126647949,
      "learning_rate": 0.0015,
      "loss": 1.8412,
      "step": 1578
    },
    {
      "epoch": 0.16656118143459916,
      "grad_norm": 0.6150435209274292,
      "learning_rate": 0.0015,
      "loss": 1.9096,
      "step": 1579
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.6432693004608154,
      "learning_rate": 0.0015,
      "loss": 1.8587,
      "step": 1580
    },
    {
      "epoch": 0.16677215189873418,
      "grad_norm": 0.5596107840538025,
      "learning_rate": 0.0015,
      "loss": 1.8737,
      "step": 1581
    },
    {
      "epoch": 0.1668776371308017,
      "grad_norm": 0.6534042358398438,
      "learning_rate": 0.0015,
      "loss": 1.867,
      "step": 1582
    },
    {
      "epoch": 0.1669831223628692,
      "grad_norm": 0.6573490500450134,
      "learning_rate": 0.0015,
      "loss": 1.8356,
      "step": 1583
    },
    {
      "epoch": 0.1670886075949367,
      "grad_norm": 0.5070683360099792,
      "learning_rate": 0.0015,
      "loss": 1.8489,
      "step": 1584
    },
    {
      "epoch": 0.16719409282700423,
      "grad_norm": 0.62180095911026,
      "learning_rate": 0.0015,
      "loss": 1.897,
      "step": 1585
    },
    {
      "epoch": 0.16729957805907172,
      "grad_norm": 0.5332339406013489,
      "learning_rate": 0.0015,
      "loss": 1.8668,
      "step": 1586
    },
    {
      "epoch": 0.16740506329113924,
      "grad_norm": 0.6476864814758301,
      "learning_rate": 0.0015,
      "loss": 1.8272,
      "step": 1587
    },
    {
      "epoch": 0.16751054852320676,
      "grad_norm": 0.5089560151100159,
      "learning_rate": 0.0015,
      "loss": 1.8459,
      "step": 1588
    },
    {
      "epoch": 0.16761603375527426,
      "grad_norm": 0.4792965054512024,
      "learning_rate": 0.0015,
      "loss": 1.8695,
      "step": 1589
    },
    {
      "epoch": 0.16772151898734178,
      "grad_norm": 0.5402447581291199,
      "learning_rate": 0.0015,
      "loss": 1.8333,
      "step": 1590
    },
    {
      "epoch": 0.16782700421940927,
      "grad_norm": 0.508988082408905,
      "learning_rate": 0.0015,
      "loss": 1.8442,
      "step": 1591
    },
    {
      "epoch": 0.1679324894514768,
      "grad_norm": 0.5005342960357666,
      "learning_rate": 0.0015,
      "loss": 1.8281,
      "step": 1592
    },
    {
      "epoch": 0.1680379746835443,
      "grad_norm": 0.4904800057411194,
      "learning_rate": 0.0015,
      "loss": 1.8654,
      "step": 1593
    },
    {
      "epoch": 0.1681434599156118,
      "grad_norm": 0.5084935426712036,
      "learning_rate": 0.0015,
      "loss": 1.8517,
      "step": 1594
    },
    {
      "epoch": 0.16824894514767932,
      "grad_norm": 0.5018736720085144,
      "learning_rate": 0.0015,
      "loss": 1.8989,
      "step": 1595
    },
    {
      "epoch": 0.16835443037974684,
      "grad_norm": 0.4767528772354126,
      "learning_rate": 0.0015,
      "loss": 1.8713,
      "step": 1596
    },
    {
      "epoch": 0.16845991561181434,
      "grad_norm": 0.5333482623100281,
      "learning_rate": 0.0015,
      "loss": 1.8476,
      "step": 1597
    },
    {
      "epoch": 0.16856540084388186,
      "grad_norm": 0.4845268726348877,
      "learning_rate": 0.0015,
      "loss": 1.8608,
      "step": 1598
    },
    {
      "epoch": 0.16867088607594938,
      "grad_norm": 0.6233534812927246,
      "learning_rate": 0.0015,
      "loss": 1.8541,
      "step": 1599
    },
    {
      "epoch": 0.16877637130801687,
      "grad_norm": 0.5676378607749939,
      "learning_rate": 0.0015,
      "loss": 1.8727,
      "step": 1600
    },
    {
      "epoch": 0.1688818565400844,
      "grad_norm": 0.5269106030464172,
      "learning_rate": 0.0015,
      "loss": 1.8265,
      "step": 1601
    },
    {
      "epoch": 0.1689873417721519,
      "grad_norm": 0.6660743951797485,
      "learning_rate": 0.0015,
      "loss": 1.896,
      "step": 1602
    },
    {
      "epoch": 0.1690928270042194,
      "grad_norm": 0.6672782897949219,
      "learning_rate": 0.0015,
      "loss": 1.8653,
      "step": 1603
    },
    {
      "epoch": 0.16919831223628692,
      "grad_norm": 0.5369535684585571,
      "learning_rate": 0.0015,
      "loss": 1.8526,
      "step": 1604
    },
    {
      "epoch": 0.16930379746835442,
      "grad_norm": 0.5175344944000244,
      "learning_rate": 0.0015,
      "loss": 1.8336,
      "step": 1605
    },
    {
      "epoch": 0.16940928270042194,
      "grad_norm": 0.5831400156021118,
      "learning_rate": 0.0015,
      "loss": 1.875,
      "step": 1606
    },
    {
      "epoch": 0.16951476793248946,
      "grad_norm": 0.5787636637687683,
      "learning_rate": 0.0015,
      "loss": 1.8797,
      "step": 1607
    },
    {
      "epoch": 0.16962025316455695,
      "grad_norm": 0.531754195690155,
      "learning_rate": 0.0015,
      "loss": 1.8332,
      "step": 1608
    },
    {
      "epoch": 0.16972573839662447,
      "grad_norm": 0.5120327472686768,
      "learning_rate": 0.0015,
      "loss": 1.8604,
      "step": 1609
    },
    {
      "epoch": 0.169831223628692,
      "grad_norm": 0.5185548067092896,
      "learning_rate": 0.0015,
      "loss": 1.8681,
      "step": 1610
    },
    {
      "epoch": 0.16993670886075948,
      "grad_norm": 0.4890459477901459,
      "learning_rate": 0.0015,
      "loss": 1.8058,
      "step": 1611
    },
    {
      "epoch": 0.170042194092827,
      "grad_norm": 0.5376722812652588,
      "learning_rate": 0.0015,
      "loss": 1.7789,
      "step": 1612
    },
    {
      "epoch": 0.17014767932489452,
      "grad_norm": 0.7383502125740051,
      "learning_rate": 0.0015,
      "loss": 1.8277,
      "step": 1613
    },
    {
      "epoch": 0.17025316455696202,
      "grad_norm": 0.736123263835907,
      "learning_rate": 0.0015,
      "loss": 1.8573,
      "step": 1614
    },
    {
      "epoch": 0.17035864978902954,
      "grad_norm": 0.5057313442230225,
      "learning_rate": 0.0015,
      "loss": 1.8295,
      "step": 1615
    },
    {
      "epoch": 0.17046413502109706,
      "grad_norm": 0.5502669215202332,
      "learning_rate": 0.0015,
      "loss": 1.8571,
      "step": 1616
    },
    {
      "epoch": 0.17056962025316455,
      "grad_norm": 0.6407621502876282,
      "learning_rate": 0.0015,
      "loss": 1.8458,
      "step": 1617
    },
    {
      "epoch": 0.17067510548523207,
      "grad_norm": 0.5297473669052124,
      "learning_rate": 0.0015,
      "loss": 1.8257,
      "step": 1618
    },
    {
      "epoch": 0.1707805907172996,
      "grad_norm": 0.4920409023761749,
      "learning_rate": 0.0015,
      "loss": 1.8458,
      "step": 1619
    },
    {
      "epoch": 0.17088607594936708,
      "grad_norm": 0.6494439840316772,
      "learning_rate": 0.0015,
      "loss": 1.8494,
      "step": 1620
    },
    {
      "epoch": 0.1709915611814346,
      "grad_norm": 0.5731796622276306,
      "learning_rate": 0.0015,
      "loss": 1.8435,
      "step": 1621
    },
    {
      "epoch": 0.1710970464135021,
      "grad_norm": 0.5125641226768494,
      "learning_rate": 0.0015,
      "loss": 1.8897,
      "step": 1622
    },
    {
      "epoch": 0.17120253164556962,
      "grad_norm": 0.5825533270835876,
      "learning_rate": 0.0015,
      "loss": 1.8596,
      "step": 1623
    },
    {
      "epoch": 0.17130801687763714,
      "grad_norm": 0.7571307420730591,
      "learning_rate": 0.0015,
      "loss": 1.8213,
      "step": 1624
    },
    {
      "epoch": 0.17141350210970463,
      "grad_norm": 0.6041271686553955,
      "learning_rate": 0.0015,
      "loss": 1.8801,
      "step": 1625
    },
    {
      "epoch": 0.17151898734177215,
      "grad_norm": 0.5118398070335388,
      "learning_rate": 0.0015,
      "loss": 1.8093,
      "step": 1626
    },
    {
      "epoch": 0.17162447257383967,
      "grad_norm": 0.579782247543335,
      "learning_rate": 0.0015,
      "loss": 1.83,
      "step": 1627
    },
    {
      "epoch": 0.17172995780590716,
      "grad_norm": 0.5675089955329895,
      "learning_rate": 0.0015,
      "loss": 1.8402,
      "step": 1628
    },
    {
      "epoch": 0.17183544303797468,
      "grad_norm": 0.7444484829902649,
      "learning_rate": 0.0015,
      "loss": 1.8657,
      "step": 1629
    },
    {
      "epoch": 0.1719409282700422,
      "grad_norm": 0.628352165222168,
      "learning_rate": 0.0015,
      "loss": 1.8231,
      "step": 1630
    },
    {
      "epoch": 0.1720464135021097,
      "grad_norm": 0.5874509811401367,
      "learning_rate": 0.0015,
      "loss": 1.8561,
      "step": 1631
    },
    {
      "epoch": 0.17215189873417722,
      "grad_norm": 0.5177331566810608,
      "learning_rate": 0.0015,
      "loss": 1.8524,
      "step": 1632
    },
    {
      "epoch": 0.17225738396624474,
      "grad_norm": 0.544594943523407,
      "learning_rate": 0.0015,
      "loss": 1.8433,
      "step": 1633
    },
    {
      "epoch": 0.17236286919831223,
      "grad_norm": 0.5621534585952759,
      "learning_rate": 0.0015,
      "loss": 1.8499,
      "step": 1634
    },
    {
      "epoch": 0.17246835443037975,
      "grad_norm": 0.4710772931575775,
      "learning_rate": 0.0015,
      "loss": 1.8341,
      "step": 1635
    },
    {
      "epoch": 0.17257383966244727,
      "grad_norm": 0.49431484937667847,
      "learning_rate": 0.0015,
      "loss": 1.8405,
      "step": 1636
    },
    {
      "epoch": 0.17267932489451476,
      "grad_norm": 0.5371549129486084,
      "learning_rate": 0.0015,
      "loss": 1.851,
      "step": 1637
    },
    {
      "epoch": 0.17278481012658228,
      "grad_norm": 0.6518941521644592,
      "learning_rate": 0.0015,
      "loss": 1.8644,
      "step": 1638
    },
    {
      "epoch": 0.17289029535864978,
      "grad_norm": 0.6416102051734924,
      "learning_rate": 0.0015,
      "loss": 1.8576,
      "step": 1639
    },
    {
      "epoch": 0.1729957805907173,
      "grad_norm": 0.5830026865005493,
      "learning_rate": 0.0015,
      "loss": 1.8284,
      "step": 1640
    },
    {
      "epoch": 0.17310126582278482,
      "grad_norm": 0.5949551463127136,
      "learning_rate": 0.0015,
      "loss": 1.8203,
      "step": 1641
    },
    {
      "epoch": 0.1732067510548523,
      "grad_norm": 0.6380580067634583,
      "learning_rate": 0.0015,
      "loss": 1.8122,
      "step": 1642
    },
    {
      "epoch": 0.17331223628691983,
      "grad_norm": 0.5648717880249023,
      "learning_rate": 0.0015,
      "loss": 1.8289,
      "step": 1643
    },
    {
      "epoch": 0.17341772151898735,
      "grad_norm": 0.6136492490768433,
      "learning_rate": 0.0015,
      "loss": 1.8305,
      "step": 1644
    },
    {
      "epoch": 0.17352320675105484,
      "grad_norm": 0.7076066136360168,
      "learning_rate": 0.0015,
      "loss": 1.8576,
      "step": 1645
    },
    {
      "epoch": 0.17362869198312236,
      "grad_norm": 0.7656069993972778,
      "learning_rate": 0.0015,
      "loss": 1.8514,
      "step": 1646
    },
    {
      "epoch": 0.17373417721518988,
      "grad_norm": 0.5621904730796814,
      "learning_rate": 0.0015,
      "loss": 1.8444,
      "step": 1647
    },
    {
      "epoch": 0.17383966244725738,
      "grad_norm": 0.5006690621376038,
      "learning_rate": 0.0015,
      "loss": 1.8273,
      "step": 1648
    },
    {
      "epoch": 0.1739451476793249,
      "grad_norm": 0.6614305377006531,
      "learning_rate": 0.0015,
      "loss": 1.8813,
      "step": 1649
    },
    {
      "epoch": 0.17405063291139242,
      "grad_norm": 0.6870092749595642,
      "learning_rate": 0.0015,
      "loss": 1.8497,
      "step": 1650
    },
    {
      "epoch": 0.1741561181434599,
      "grad_norm": 0.5306625962257385,
      "learning_rate": 0.0015,
      "loss": 1.8506,
      "step": 1651
    },
    {
      "epoch": 0.17426160337552743,
      "grad_norm": 0.5352807641029358,
      "learning_rate": 0.0015,
      "loss": 1.8216,
      "step": 1652
    },
    {
      "epoch": 0.17436708860759495,
      "grad_norm": 0.6174004077911377,
      "learning_rate": 0.0015,
      "loss": 1.8136,
      "step": 1653
    },
    {
      "epoch": 0.17447257383966244,
      "grad_norm": 0.5463118553161621,
      "learning_rate": 0.0015,
      "loss": 1.8617,
      "step": 1654
    },
    {
      "epoch": 0.17457805907172996,
      "grad_norm": 0.45304176211357117,
      "learning_rate": 0.0015,
      "loss": 1.8408,
      "step": 1655
    },
    {
      "epoch": 0.17468354430379746,
      "grad_norm": 0.4786616265773773,
      "learning_rate": 0.0015,
      "loss": 1.8728,
      "step": 1656
    },
    {
      "epoch": 0.17478902953586498,
      "grad_norm": 0.541914165019989,
      "learning_rate": 0.0015,
      "loss": 1.8261,
      "step": 1657
    },
    {
      "epoch": 0.1748945147679325,
      "grad_norm": 0.5445770621299744,
      "learning_rate": 0.0015,
      "loss": 1.8269,
      "step": 1658
    },
    {
      "epoch": 0.175,
      "grad_norm": 0.5142346620559692,
      "learning_rate": 0.0015,
      "loss": 1.8341,
      "step": 1659
    },
    {
      "epoch": 0.1751054852320675,
      "grad_norm": 0.5110147595405579,
      "learning_rate": 0.0015,
      "loss": 1.8259,
      "step": 1660
    },
    {
      "epoch": 0.17521097046413503,
      "grad_norm": 0.5237147212028503,
      "learning_rate": 0.0015,
      "loss": 1.8776,
      "step": 1661
    },
    {
      "epoch": 0.17531645569620252,
      "grad_norm": 0.6443389058113098,
      "learning_rate": 0.0015,
      "loss": 1.836,
      "step": 1662
    },
    {
      "epoch": 0.17542194092827004,
      "grad_norm": 0.7561408877372742,
      "learning_rate": 0.0015,
      "loss": 1.8373,
      "step": 1663
    },
    {
      "epoch": 0.17552742616033756,
      "grad_norm": 0.4879377782344818,
      "learning_rate": 0.0015,
      "loss": 1.8345,
      "step": 1664
    },
    {
      "epoch": 0.17563291139240506,
      "grad_norm": 0.7401023507118225,
      "learning_rate": 0.0015,
      "loss": 1.8542,
      "step": 1665
    },
    {
      "epoch": 0.17573839662447258,
      "grad_norm": 0.6402983665466309,
      "learning_rate": 0.0015,
      "loss": 1.8797,
      "step": 1666
    },
    {
      "epoch": 0.1758438818565401,
      "grad_norm": 0.5913488268852234,
      "learning_rate": 0.0015,
      "loss": 1.8581,
      "step": 1667
    },
    {
      "epoch": 0.1759493670886076,
      "grad_norm": 0.7639641761779785,
      "learning_rate": 0.0015,
      "loss": 1.871,
      "step": 1668
    },
    {
      "epoch": 0.1760548523206751,
      "grad_norm": 0.5228761434555054,
      "learning_rate": 0.0015,
      "loss": 1.8232,
      "step": 1669
    },
    {
      "epoch": 0.17616033755274263,
      "grad_norm": 0.6810857653617859,
      "learning_rate": 0.0015,
      "loss": 1.8015,
      "step": 1670
    },
    {
      "epoch": 0.17626582278481012,
      "grad_norm": 0.6449247598648071,
      "learning_rate": 0.0015,
      "loss": 1.8007,
      "step": 1671
    },
    {
      "epoch": 0.17637130801687764,
      "grad_norm": 0.5456721782684326,
      "learning_rate": 0.0015,
      "loss": 1.8397,
      "step": 1672
    },
    {
      "epoch": 0.17647679324894514,
      "grad_norm": 0.48542457818984985,
      "learning_rate": 0.0015,
      "loss": 1.8497,
      "step": 1673
    },
    {
      "epoch": 0.17658227848101266,
      "grad_norm": 0.5676390528678894,
      "learning_rate": 0.0015,
      "loss": 1.8531,
      "step": 1674
    },
    {
      "epoch": 0.17668776371308018,
      "grad_norm": 0.6341606974601746,
      "learning_rate": 0.0015,
      "loss": 1.8329,
      "step": 1675
    },
    {
      "epoch": 0.17679324894514767,
      "grad_norm": 0.5365669131278992,
      "learning_rate": 0.0015,
      "loss": 1.8516,
      "step": 1676
    },
    {
      "epoch": 0.1768987341772152,
      "grad_norm": 0.5370458960533142,
      "learning_rate": 0.0015,
      "loss": 1.7982,
      "step": 1677
    },
    {
      "epoch": 0.1770042194092827,
      "grad_norm": 0.7546199560165405,
      "learning_rate": 0.0015,
      "loss": 1.8696,
      "step": 1678
    },
    {
      "epoch": 0.1771097046413502,
      "grad_norm": 0.5257035493850708,
      "learning_rate": 0.0015,
      "loss": 1.8103,
      "step": 1679
    },
    {
      "epoch": 0.17721518987341772,
      "grad_norm": 0.7526178359985352,
      "learning_rate": 0.0015,
      "loss": 1.8424,
      "step": 1680
    },
    {
      "epoch": 0.17732067510548524,
      "grad_norm": 0.9093194007873535,
      "learning_rate": 0.0015,
      "loss": 1.8335,
      "step": 1681
    },
    {
      "epoch": 0.17742616033755274,
      "grad_norm": 0.6237613558769226,
      "learning_rate": 0.0015,
      "loss": 1.8225,
      "step": 1682
    },
    {
      "epoch": 0.17753164556962026,
      "grad_norm": 0.685141921043396,
      "learning_rate": 0.0015,
      "loss": 1.8651,
      "step": 1683
    },
    {
      "epoch": 0.17763713080168778,
      "grad_norm": 0.9615616202354431,
      "learning_rate": 0.0015,
      "loss": 1.8715,
      "step": 1684
    },
    {
      "epoch": 0.17774261603375527,
      "grad_norm": 0.6962200403213501,
      "learning_rate": 0.0015,
      "loss": 1.7927,
      "step": 1685
    },
    {
      "epoch": 0.1778481012658228,
      "grad_norm": 0.503422200679779,
      "learning_rate": 0.0015,
      "loss": 1.7943,
      "step": 1686
    },
    {
      "epoch": 0.17795358649789028,
      "grad_norm": 0.6377720236778259,
      "learning_rate": 0.0015,
      "loss": 1.847,
      "step": 1687
    },
    {
      "epoch": 0.1780590717299578,
      "grad_norm": 0.6209604144096375,
      "learning_rate": 0.0015,
      "loss": 1.7904,
      "step": 1688
    },
    {
      "epoch": 0.17816455696202532,
      "grad_norm": 0.5179665088653564,
      "learning_rate": 0.0015,
      "loss": 1.8506,
      "step": 1689
    },
    {
      "epoch": 0.17827004219409281,
      "grad_norm": 0.5614203810691833,
      "learning_rate": 0.0015,
      "loss": 1.8255,
      "step": 1690
    },
    {
      "epoch": 0.17837552742616034,
      "grad_norm": 0.6475895643234253,
      "learning_rate": 0.0015,
      "loss": 1.8837,
      "step": 1691
    },
    {
      "epoch": 0.17848101265822786,
      "grad_norm": 0.5590607523918152,
      "learning_rate": 0.0015,
      "loss": 1.8446,
      "step": 1692
    },
    {
      "epoch": 0.17858649789029535,
      "grad_norm": 0.5380192399024963,
      "learning_rate": 0.0015,
      "loss": 1.8111,
      "step": 1693
    },
    {
      "epoch": 0.17869198312236287,
      "grad_norm": 0.7930208444595337,
      "learning_rate": 0.0015,
      "loss": 1.84,
      "step": 1694
    },
    {
      "epoch": 0.1787974683544304,
      "grad_norm": 0.9309415817260742,
      "learning_rate": 0.0015,
      "loss": 1.8248,
      "step": 1695
    },
    {
      "epoch": 0.17890295358649788,
      "grad_norm": 0.6583831310272217,
      "learning_rate": 0.0015,
      "loss": 1.8256,
      "step": 1696
    },
    {
      "epoch": 0.1790084388185654,
      "grad_norm": 0.7181501984596252,
      "learning_rate": 0.0015,
      "loss": 1.8295,
      "step": 1697
    },
    {
      "epoch": 0.17911392405063292,
      "grad_norm": 0.9588510990142822,
      "learning_rate": 0.0015,
      "loss": 1.8566,
      "step": 1698
    },
    {
      "epoch": 0.17921940928270041,
      "grad_norm": 0.74287348985672,
      "learning_rate": 0.0015,
      "loss": 1.9011,
      "step": 1699
    },
    {
      "epoch": 0.17932489451476794,
      "grad_norm": 0.5708943605422974,
      "learning_rate": 0.0015,
      "loss": 1.8581,
      "step": 1700
    },
    {
      "epoch": 0.17943037974683546,
      "grad_norm": 0.849864661693573,
      "learning_rate": 0.0015,
      "loss": 1.8403,
      "step": 1701
    },
    {
      "epoch": 0.17953586497890295,
      "grad_norm": 0.6671258211135864,
      "learning_rate": 0.0015,
      "loss": 1.8395,
      "step": 1702
    },
    {
      "epoch": 0.17964135021097047,
      "grad_norm": 0.6216587424278259,
      "learning_rate": 0.0015,
      "loss": 1.7977,
      "step": 1703
    },
    {
      "epoch": 0.17974683544303796,
      "grad_norm": 0.8394911289215088,
      "learning_rate": 0.0015,
      "loss": 1.8265,
      "step": 1704
    },
    {
      "epoch": 0.17985232067510548,
      "grad_norm": 0.4909205138683319,
      "learning_rate": 0.0015,
      "loss": 1.8343,
      "step": 1705
    },
    {
      "epoch": 0.179957805907173,
      "grad_norm": 0.6571671366691589,
      "learning_rate": 0.0015,
      "loss": 1.8563,
      "step": 1706
    },
    {
      "epoch": 0.1800632911392405,
      "grad_norm": 0.6637552380561829,
      "learning_rate": 0.0015,
      "loss": 1.8051,
      "step": 1707
    },
    {
      "epoch": 0.18016877637130801,
      "grad_norm": 0.47139400243759155,
      "learning_rate": 0.0015,
      "loss": 1.8009,
      "step": 1708
    },
    {
      "epoch": 0.18027426160337554,
      "grad_norm": 0.6116742491722107,
      "learning_rate": 0.0015,
      "loss": 1.8116,
      "step": 1709
    },
    {
      "epoch": 0.18037974683544303,
      "grad_norm": 0.5905157923698425,
      "learning_rate": 0.0015,
      "loss": 1.8604,
      "step": 1710
    },
    {
      "epoch": 0.18048523206751055,
      "grad_norm": 0.502013087272644,
      "learning_rate": 0.0015,
      "loss": 1.8423,
      "step": 1711
    },
    {
      "epoch": 0.18059071729957807,
      "grad_norm": 0.5473625659942627,
      "learning_rate": 0.0015,
      "loss": 1.8591,
      "step": 1712
    },
    {
      "epoch": 0.18069620253164556,
      "grad_norm": 0.5390632748603821,
      "learning_rate": 0.0015,
      "loss": 1.8207,
      "step": 1713
    },
    {
      "epoch": 0.18080168776371308,
      "grad_norm": 0.5800744891166687,
      "learning_rate": 0.0015,
      "loss": 1.8051,
      "step": 1714
    },
    {
      "epoch": 0.1809071729957806,
      "grad_norm": 0.518036425113678,
      "learning_rate": 0.0015,
      "loss": 1.8016,
      "step": 1715
    },
    {
      "epoch": 0.1810126582278481,
      "grad_norm": 0.5341516733169556,
      "learning_rate": 0.0015,
      "loss": 1.8275,
      "step": 1716
    },
    {
      "epoch": 0.18111814345991561,
      "grad_norm": 0.5151551961898804,
      "learning_rate": 0.0015,
      "loss": 1.7892,
      "step": 1717
    },
    {
      "epoch": 0.18122362869198314,
      "grad_norm": 0.5301831364631653,
      "learning_rate": 0.0015,
      "loss": 1.8195,
      "step": 1718
    },
    {
      "epoch": 0.18132911392405063,
      "grad_norm": 0.4552363157272339,
      "learning_rate": 0.0015,
      "loss": 1.8206,
      "step": 1719
    },
    {
      "epoch": 0.18143459915611815,
      "grad_norm": 0.496009886264801,
      "learning_rate": 0.0015,
      "loss": 1.8484,
      "step": 1720
    },
    {
      "epoch": 0.18154008438818564,
      "grad_norm": 0.476816326379776,
      "learning_rate": 0.0015,
      "loss": 1.813,
      "step": 1721
    },
    {
      "epoch": 0.18164556962025316,
      "grad_norm": 0.48803362250328064,
      "learning_rate": 0.0015,
      "loss": 1.8112,
      "step": 1722
    },
    {
      "epoch": 0.18175105485232068,
      "grad_norm": 0.5525384545326233,
      "learning_rate": 0.0015,
      "loss": 1.811,
      "step": 1723
    },
    {
      "epoch": 0.18185654008438817,
      "grad_norm": 0.46394020318984985,
      "learning_rate": 0.0015,
      "loss": 1.808,
      "step": 1724
    },
    {
      "epoch": 0.1819620253164557,
      "grad_norm": 0.5969376564025879,
      "learning_rate": 0.0015,
      "loss": 1.8106,
      "step": 1725
    },
    {
      "epoch": 0.18206751054852321,
      "grad_norm": 0.5168143510818481,
      "learning_rate": 0.0015,
      "loss": 1.8467,
      "step": 1726
    },
    {
      "epoch": 0.1821729957805907,
      "grad_norm": 0.5381571650505066,
      "learning_rate": 0.0015,
      "loss": 1.8176,
      "step": 1727
    },
    {
      "epoch": 0.18227848101265823,
      "grad_norm": 0.46947160363197327,
      "learning_rate": 0.0015,
      "loss": 1.8212,
      "step": 1728
    },
    {
      "epoch": 0.18238396624472575,
      "grad_norm": 0.5031561851501465,
      "learning_rate": 0.0015,
      "loss": 1.8408,
      "step": 1729
    },
    {
      "epoch": 0.18248945147679324,
      "grad_norm": 0.5438616871833801,
      "learning_rate": 0.0015,
      "loss": 1.8651,
      "step": 1730
    },
    {
      "epoch": 0.18259493670886076,
      "grad_norm": 0.47203943133354187,
      "learning_rate": 0.0015,
      "loss": 1.8139,
      "step": 1731
    },
    {
      "epoch": 0.18270042194092828,
      "grad_norm": 0.5111636519432068,
      "learning_rate": 0.0015,
      "loss": 1.8216,
      "step": 1732
    },
    {
      "epoch": 0.18280590717299577,
      "grad_norm": 0.5975373387336731,
      "learning_rate": 0.0015,
      "loss": 1.8136,
      "step": 1733
    },
    {
      "epoch": 0.1829113924050633,
      "grad_norm": 0.5054951906204224,
      "learning_rate": 0.0015,
      "loss": 1.831,
      "step": 1734
    },
    {
      "epoch": 0.18301687763713081,
      "grad_norm": 0.5159053206443787,
      "learning_rate": 0.0015,
      "loss": 1.8149,
      "step": 1735
    },
    {
      "epoch": 0.1831223628691983,
      "grad_norm": 0.5666469931602478,
      "learning_rate": 0.0015,
      "loss": 1.8544,
      "step": 1736
    },
    {
      "epoch": 0.18322784810126583,
      "grad_norm": 0.6056023836135864,
      "learning_rate": 0.0015,
      "loss": 1.8151,
      "step": 1737
    },
    {
      "epoch": 0.18333333333333332,
      "grad_norm": 0.5668106079101562,
      "learning_rate": 0.0015,
      "loss": 1.8343,
      "step": 1738
    },
    {
      "epoch": 0.18343881856540084,
      "grad_norm": 0.5655359029769897,
      "learning_rate": 0.0015,
      "loss": 1.8585,
      "step": 1739
    },
    {
      "epoch": 0.18354430379746836,
      "grad_norm": 0.8024944067001343,
      "learning_rate": 0.0015,
      "loss": 1.8497,
      "step": 1740
    },
    {
      "epoch": 0.18364978902953585,
      "grad_norm": 0.6638942956924438,
      "learning_rate": 0.0015,
      "loss": 1.856,
      "step": 1741
    },
    {
      "epoch": 0.18375527426160337,
      "grad_norm": 0.5905331969261169,
      "learning_rate": 0.0015,
      "loss": 1.8435,
      "step": 1742
    },
    {
      "epoch": 0.1838607594936709,
      "grad_norm": 0.7127625942230225,
      "learning_rate": 0.0015,
      "loss": 1.8459,
      "step": 1743
    },
    {
      "epoch": 0.1839662447257384,
      "grad_norm": 0.5033014416694641,
      "learning_rate": 0.0015,
      "loss": 1.8483,
      "step": 1744
    },
    {
      "epoch": 0.1840717299578059,
      "grad_norm": 0.703793466091156,
      "learning_rate": 0.0015,
      "loss": 1.8413,
      "step": 1745
    },
    {
      "epoch": 0.18417721518987343,
      "grad_norm": 0.73264479637146,
      "learning_rate": 0.0015,
      "loss": 1.8548,
      "step": 1746
    },
    {
      "epoch": 0.18428270042194092,
      "grad_norm": 0.5461260080337524,
      "learning_rate": 0.0015,
      "loss": 1.875,
      "step": 1747
    },
    {
      "epoch": 0.18438818565400844,
      "grad_norm": 0.5849325656890869,
      "learning_rate": 0.0015,
      "loss": 1.8169,
      "step": 1748
    },
    {
      "epoch": 0.18449367088607596,
      "grad_norm": 0.5817967653274536,
      "learning_rate": 0.0015,
      "loss": 1.8647,
      "step": 1749
    },
    {
      "epoch": 0.18459915611814345,
      "grad_norm": 0.6045406460762024,
      "learning_rate": 0.0015,
      "loss": 1.8692,
      "step": 1750
    },
    {
      "epoch": 0.18470464135021097,
      "grad_norm": 0.5886181592941284,
      "learning_rate": 0.0015,
      "loss": 1.8272,
      "step": 1751
    },
    {
      "epoch": 0.1848101265822785,
      "grad_norm": 0.6443178653717041,
      "learning_rate": 0.0015,
      "loss": 1.8452,
      "step": 1752
    },
    {
      "epoch": 0.184915611814346,
      "grad_norm": 0.7146679759025574,
      "learning_rate": 0.0015,
      "loss": 1.8506,
      "step": 1753
    },
    {
      "epoch": 0.1850210970464135,
      "grad_norm": 0.49639758467674255,
      "learning_rate": 0.0015,
      "loss": 1.8355,
      "step": 1754
    },
    {
      "epoch": 0.185126582278481,
      "grad_norm": 0.6765944361686707,
      "learning_rate": 0.0015,
      "loss": 1.8424,
      "step": 1755
    },
    {
      "epoch": 0.18523206751054852,
      "grad_norm": 0.6170516014099121,
      "learning_rate": 0.0015,
      "loss": 1.8067,
      "step": 1756
    },
    {
      "epoch": 0.18533755274261604,
      "grad_norm": 0.5510251522064209,
      "learning_rate": 0.0015,
      "loss": 1.8241,
      "step": 1757
    },
    {
      "epoch": 0.18544303797468353,
      "grad_norm": 0.549027681350708,
      "learning_rate": 0.0015,
      "loss": 1.8427,
      "step": 1758
    },
    {
      "epoch": 0.18554852320675105,
      "grad_norm": 0.6877909898757935,
      "learning_rate": 0.0015,
      "loss": 1.8346,
      "step": 1759
    },
    {
      "epoch": 0.18565400843881857,
      "grad_norm": 0.643034815788269,
      "learning_rate": 0.0015,
      "loss": 1.7842,
      "step": 1760
    },
    {
      "epoch": 0.18575949367088607,
      "grad_norm": 0.48639509081840515,
      "learning_rate": 0.0015,
      "loss": 1.8209,
      "step": 1761
    },
    {
      "epoch": 0.1858649789029536,
      "grad_norm": 0.5269736647605896,
      "learning_rate": 0.0015,
      "loss": 1.8492,
      "step": 1762
    },
    {
      "epoch": 0.1859704641350211,
      "grad_norm": 0.560742199420929,
      "learning_rate": 0.0015,
      "loss": 1.8237,
      "step": 1763
    },
    {
      "epoch": 0.1860759493670886,
      "grad_norm": 0.5695181488990784,
      "learning_rate": 0.0015,
      "loss": 1.8118,
      "step": 1764
    },
    {
      "epoch": 0.18618143459915612,
      "grad_norm": 0.6077991127967834,
      "learning_rate": 0.0015,
      "loss": 1.8292,
      "step": 1765
    },
    {
      "epoch": 0.18628691983122364,
      "grad_norm": 0.5074968338012695,
      "learning_rate": 0.0015,
      "loss": 1.8301,
      "step": 1766
    },
    {
      "epoch": 0.18639240506329113,
      "grad_norm": 0.7066789269447327,
      "learning_rate": 0.0015,
      "loss": 1.8509,
      "step": 1767
    },
    {
      "epoch": 0.18649789029535865,
      "grad_norm": 0.6739769577980042,
      "learning_rate": 0.0015,
      "loss": 1.803,
      "step": 1768
    },
    {
      "epoch": 0.18660337552742617,
      "grad_norm": 0.47891584038734436,
      "learning_rate": 0.0015,
      "loss": 1.8413,
      "step": 1769
    },
    {
      "epoch": 0.18670886075949367,
      "grad_norm": 0.6155842542648315,
      "learning_rate": 0.0015,
      "loss": 1.8387,
      "step": 1770
    },
    {
      "epoch": 0.1868143459915612,
      "grad_norm": 0.562042236328125,
      "learning_rate": 0.0015,
      "loss": 1.8459,
      "step": 1771
    },
    {
      "epoch": 0.18691983122362868,
      "grad_norm": 0.5393577218055725,
      "learning_rate": 0.0015,
      "loss": 1.8357,
      "step": 1772
    },
    {
      "epoch": 0.1870253164556962,
      "grad_norm": 0.7965392470359802,
      "learning_rate": 0.0015,
      "loss": 1.8372,
      "step": 1773
    },
    {
      "epoch": 0.18713080168776372,
      "grad_norm": 0.7083988189697266,
      "learning_rate": 0.0015,
      "loss": 1.8761,
      "step": 1774
    },
    {
      "epoch": 0.1872362869198312,
      "grad_norm": 0.5049605369567871,
      "learning_rate": 0.0015,
      "loss": 1.878,
      "step": 1775
    },
    {
      "epoch": 0.18734177215189873,
      "grad_norm": 0.6535817384719849,
      "learning_rate": 0.0015,
      "loss": 1.839,
      "step": 1776
    },
    {
      "epoch": 0.18744725738396625,
      "grad_norm": 0.7916276454925537,
      "learning_rate": 0.0015,
      "loss": 1.8187,
      "step": 1777
    },
    {
      "epoch": 0.18755274261603375,
      "grad_norm": 0.5668764114379883,
      "learning_rate": 0.0015,
      "loss": 1.7739,
      "step": 1778
    },
    {
      "epoch": 0.18765822784810127,
      "grad_norm": 0.5987294316291809,
      "learning_rate": 0.0015,
      "loss": 1.8262,
      "step": 1779
    },
    {
      "epoch": 0.1877637130801688,
      "grad_norm": 0.7916189432144165,
      "learning_rate": 0.0015,
      "loss": 1.8266,
      "step": 1780
    },
    {
      "epoch": 0.18786919831223628,
      "grad_norm": 0.5684725642204285,
      "learning_rate": 0.0015,
      "loss": 1.8218,
      "step": 1781
    },
    {
      "epoch": 0.1879746835443038,
      "grad_norm": 0.6714436411857605,
      "learning_rate": 0.0015,
      "loss": 1.8894,
      "step": 1782
    },
    {
      "epoch": 0.18808016877637132,
      "grad_norm": 0.8307836651802063,
      "learning_rate": 0.0015,
      "loss": 1.8434,
      "step": 1783
    },
    {
      "epoch": 0.1881856540084388,
      "grad_norm": 0.5791143178939819,
      "learning_rate": 0.0015,
      "loss": 1.8103,
      "step": 1784
    },
    {
      "epoch": 0.18829113924050633,
      "grad_norm": 0.7713437676429749,
      "learning_rate": 0.0015,
      "loss": 1.8484,
      "step": 1785
    },
    {
      "epoch": 0.18839662447257383,
      "grad_norm": 1.103920578956604,
      "learning_rate": 0.0015,
      "loss": 1.8143,
      "step": 1786
    },
    {
      "epoch": 0.18850210970464135,
      "grad_norm": 0.5065297484397888,
      "learning_rate": 0.0015,
      "loss": 1.8272,
      "step": 1787
    },
    {
      "epoch": 0.18860759493670887,
      "grad_norm": 0.7157902717590332,
      "learning_rate": 0.0015,
      "loss": 1.8429,
      "step": 1788
    },
    {
      "epoch": 0.18871308016877636,
      "grad_norm": 0.5488754510879517,
      "learning_rate": 0.0015,
      "loss": 1.8247,
      "step": 1789
    },
    {
      "epoch": 0.18881856540084388,
      "grad_norm": 0.6051275730133057,
      "learning_rate": 0.0015,
      "loss": 1.8171,
      "step": 1790
    },
    {
      "epoch": 0.1889240506329114,
      "grad_norm": 0.744580864906311,
      "learning_rate": 0.0015,
      "loss": 1.799,
      "step": 1791
    },
    {
      "epoch": 0.1890295358649789,
      "grad_norm": 0.5383637547492981,
      "learning_rate": 0.0015,
      "loss": 1.8214,
      "step": 1792
    },
    {
      "epoch": 0.1891350210970464,
      "grad_norm": 0.6169219017028809,
      "learning_rate": 0.0015,
      "loss": 1.8502,
      "step": 1793
    },
    {
      "epoch": 0.18924050632911393,
      "grad_norm": 0.7362203598022461,
      "learning_rate": 0.0015,
      "loss": 1.891,
      "step": 1794
    },
    {
      "epoch": 0.18934599156118143,
      "grad_norm": 0.5578073859214783,
      "learning_rate": 0.0015,
      "loss": 1.8143,
      "step": 1795
    },
    {
      "epoch": 0.18945147679324895,
      "grad_norm": 0.5064363479614258,
      "learning_rate": 0.0015,
      "loss": 1.8513,
      "step": 1796
    },
    {
      "epoch": 0.18955696202531647,
      "grad_norm": 0.5139697194099426,
      "learning_rate": 0.0015,
      "loss": 1.81,
      "step": 1797
    },
    {
      "epoch": 0.18966244725738396,
      "grad_norm": 0.5032205581665039,
      "learning_rate": 0.0015,
      "loss": 1.8164,
      "step": 1798
    },
    {
      "epoch": 0.18976793248945148,
      "grad_norm": 0.5216019153594971,
      "learning_rate": 0.0015,
      "loss": 1.8131,
      "step": 1799
    },
    {
      "epoch": 0.189873417721519,
      "grad_norm": 0.5179572701454163,
      "learning_rate": 0.0015,
      "loss": 1.812,
      "step": 1800
    },
    {
      "epoch": 0.1899789029535865,
      "grad_norm": 0.46445566415786743,
      "learning_rate": 0.0015,
      "loss": 1.7668,
      "step": 1801
    },
    {
      "epoch": 0.190084388185654,
      "grad_norm": 0.536352276802063,
      "learning_rate": 0.0015,
      "loss": 1.8004,
      "step": 1802
    },
    {
      "epoch": 0.1901898734177215,
      "grad_norm": 0.5057159066200256,
      "learning_rate": 0.0015,
      "loss": 1.8296,
      "step": 1803
    },
    {
      "epoch": 0.19029535864978903,
      "grad_norm": 0.4822067320346832,
      "learning_rate": 0.0015,
      "loss": 1.8135,
      "step": 1804
    },
    {
      "epoch": 0.19040084388185655,
      "grad_norm": 0.634695827960968,
      "learning_rate": 0.0015,
      "loss": 1.8144,
      "step": 1805
    },
    {
      "epoch": 0.19050632911392404,
      "grad_norm": 0.4706762433052063,
      "learning_rate": 0.0015,
      "loss": 1.8149,
      "step": 1806
    },
    {
      "epoch": 0.19061181434599156,
      "grad_norm": 0.5567672848701477,
      "learning_rate": 0.0015,
      "loss": 1.8273,
      "step": 1807
    },
    {
      "epoch": 0.19071729957805908,
      "grad_norm": 0.4880026876926422,
      "learning_rate": 0.0015,
      "loss": 1.809,
      "step": 1808
    },
    {
      "epoch": 0.19082278481012657,
      "grad_norm": 0.5511531829833984,
      "learning_rate": 0.0015,
      "loss": 1.8503,
      "step": 1809
    },
    {
      "epoch": 0.1909282700421941,
      "grad_norm": 0.5229374170303345,
      "learning_rate": 0.0015,
      "loss": 1.7905,
      "step": 1810
    },
    {
      "epoch": 0.1910337552742616,
      "grad_norm": 0.6054973006248474,
      "learning_rate": 0.0015,
      "loss": 1.7705,
      "step": 1811
    },
    {
      "epoch": 0.1911392405063291,
      "grad_norm": 0.6326051354408264,
      "learning_rate": 0.0015,
      "loss": 1.823,
      "step": 1812
    },
    {
      "epoch": 0.19124472573839663,
      "grad_norm": 0.4894816279411316,
      "learning_rate": 0.0015,
      "loss": 1.8092,
      "step": 1813
    },
    {
      "epoch": 0.19135021097046415,
      "grad_norm": 0.5103054642677307,
      "learning_rate": 0.0015,
      "loss": 1.7862,
      "step": 1814
    },
    {
      "epoch": 0.19145569620253164,
      "grad_norm": 0.5828525424003601,
      "learning_rate": 0.0015,
      "loss": 1.799,
      "step": 1815
    },
    {
      "epoch": 0.19156118143459916,
      "grad_norm": 0.5314319133758545,
      "learning_rate": 0.0015,
      "loss": 1.824,
      "step": 1816
    },
    {
      "epoch": 0.19166666666666668,
      "grad_norm": 0.4891503155231476,
      "learning_rate": 0.0015,
      "loss": 1.7893,
      "step": 1817
    },
    {
      "epoch": 0.19177215189873417,
      "grad_norm": 0.5227344036102295,
      "learning_rate": 0.0015,
      "loss": 1.8017,
      "step": 1818
    },
    {
      "epoch": 0.1918776371308017,
      "grad_norm": 0.5064538717269897,
      "learning_rate": 0.0015,
      "loss": 1.8351,
      "step": 1819
    },
    {
      "epoch": 0.19198312236286919,
      "grad_norm": 0.4842202663421631,
      "learning_rate": 0.0015,
      "loss": 1.8762,
      "step": 1820
    },
    {
      "epoch": 0.1920886075949367,
      "grad_norm": 0.5408952832221985,
      "learning_rate": 0.0015,
      "loss": 1.8131,
      "step": 1821
    },
    {
      "epoch": 0.19219409282700423,
      "grad_norm": 0.559117317199707,
      "learning_rate": 0.0015,
      "loss": 1.8419,
      "step": 1822
    },
    {
      "epoch": 0.19229957805907172,
      "grad_norm": 0.6070255041122437,
      "learning_rate": 0.0015,
      "loss": 1.8325,
      "step": 1823
    },
    {
      "epoch": 0.19240506329113924,
      "grad_norm": 0.5837768912315369,
      "learning_rate": 0.0015,
      "loss": 1.84,
      "step": 1824
    },
    {
      "epoch": 0.19251054852320676,
      "grad_norm": 0.574431300163269,
      "learning_rate": 0.0015,
      "loss": 1.8667,
      "step": 1825
    },
    {
      "epoch": 0.19261603375527425,
      "grad_norm": 0.5465744137763977,
      "learning_rate": 0.0015,
      "loss": 1.8116,
      "step": 1826
    },
    {
      "epoch": 0.19272151898734177,
      "grad_norm": 0.4835458993911743,
      "learning_rate": 0.0015,
      "loss": 1.8011,
      "step": 1827
    },
    {
      "epoch": 0.1928270042194093,
      "grad_norm": 0.536300003528595,
      "learning_rate": 0.0015,
      "loss": 1.7582,
      "step": 1828
    },
    {
      "epoch": 0.19293248945147679,
      "grad_norm": 0.5421085953712463,
      "learning_rate": 0.0015,
      "loss": 1.796,
      "step": 1829
    },
    {
      "epoch": 0.1930379746835443,
      "grad_norm": 0.5275554060935974,
      "learning_rate": 0.0015,
      "loss": 1.7876,
      "step": 1830
    },
    {
      "epoch": 0.19314345991561183,
      "grad_norm": 0.536766529083252,
      "learning_rate": 0.0015,
      "loss": 1.7966,
      "step": 1831
    },
    {
      "epoch": 0.19324894514767932,
      "grad_norm": 0.4939540922641754,
      "learning_rate": 0.0015,
      "loss": 1.8043,
      "step": 1832
    },
    {
      "epoch": 0.19335443037974684,
      "grad_norm": 0.5534696578979492,
      "learning_rate": 0.0015,
      "loss": 1.8456,
      "step": 1833
    },
    {
      "epoch": 0.19345991561181436,
      "grad_norm": 0.6097571849822998,
      "learning_rate": 0.0015,
      "loss": 1.8403,
      "step": 1834
    },
    {
      "epoch": 0.19356540084388185,
      "grad_norm": 0.6294739842414856,
      "learning_rate": 0.0015,
      "loss": 1.8057,
      "step": 1835
    },
    {
      "epoch": 0.19367088607594937,
      "grad_norm": 0.6127732396125793,
      "learning_rate": 0.0015,
      "loss": 1.8297,
      "step": 1836
    },
    {
      "epoch": 0.19377637130801686,
      "grad_norm": 0.5328330993652344,
      "learning_rate": 0.0015,
      "loss": 1.7824,
      "step": 1837
    },
    {
      "epoch": 0.19388185654008439,
      "grad_norm": 0.5999840497970581,
      "learning_rate": 0.0015,
      "loss": 1.8088,
      "step": 1838
    },
    {
      "epoch": 0.1939873417721519,
      "grad_norm": 0.6274462342262268,
      "learning_rate": 0.0015,
      "loss": 1.822,
      "step": 1839
    },
    {
      "epoch": 0.1940928270042194,
      "grad_norm": 0.5135725736618042,
      "learning_rate": 0.0015,
      "loss": 1.809,
      "step": 1840
    },
    {
      "epoch": 0.19419831223628692,
      "grad_norm": 0.5076367259025574,
      "learning_rate": 0.0015,
      "loss": 1.8304,
      "step": 1841
    },
    {
      "epoch": 0.19430379746835444,
      "grad_norm": 0.5924019813537598,
      "learning_rate": 0.0015,
      "loss": 1.8179,
      "step": 1842
    },
    {
      "epoch": 0.19440928270042193,
      "grad_norm": 0.5452865362167358,
      "learning_rate": 0.0015,
      "loss": 1.7876,
      "step": 1843
    },
    {
      "epoch": 0.19451476793248945,
      "grad_norm": 0.6155842542648315,
      "learning_rate": 0.0015,
      "loss": 1.8497,
      "step": 1844
    },
    {
      "epoch": 0.19462025316455697,
      "grad_norm": 0.5350075960159302,
      "learning_rate": 0.0015,
      "loss": 1.8067,
      "step": 1845
    },
    {
      "epoch": 0.19472573839662446,
      "grad_norm": 0.5192350149154663,
      "learning_rate": 0.0015,
      "loss": 1.8387,
      "step": 1846
    },
    {
      "epoch": 0.19483122362869199,
      "grad_norm": 0.5149152874946594,
      "learning_rate": 0.0015,
      "loss": 1.83,
      "step": 1847
    },
    {
      "epoch": 0.1949367088607595,
      "grad_norm": 0.5416075587272644,
      "learning_rate": 0.0015,
      "loss": 1.7897,
      "step": 1848
    },
    {
      "epoch": 0.195042194092827,
      "grad_norm": 0.5153766870498657,
      "learning_rate": 0.0015,
      "loss": 1.7679,
      "step": 1849
    },
    {
      "epoch": 0.19514767932489452,
      "grad_norm": 0.6838843822479248,
      "learning_rate": 0.0015,
      "loss": 1.8395,
      "step": 1850
    },
    {
      "epoch": 0.19525316455696204,
      "grad_norm": 0.5588085055351257,
      "learning_rate": 0.0015,
      "loss": 1.7849,
      "step": 1851
    },
    {
      "epoch": 0.19535864978902953,
      "grad_norm": 0.5469839572906494,
      "learning_rate": 0.0015,
      "loss": 1.8287,
      "step": 1852
    },
    {
      "epoch": 0.19546413502109705,
      "grad_norm": 0.6076549887657166,
      "learning_rate": 0.0015,
      "loss": 1.8027,
      "step": 1853
    },
    {
      "epoch": 0.19556962025316454,
      "grad_norm": 0.6546722054481506,
      "learning_rate": 0.0015,
      "loss": 1.8364,
      "step": 1854
    },
    {
      "epoch": 0.19567510548523206,
      "grad_norm": 0.5049344897270203,
      "learning_rate": 0.0015,
      "loss": 1.8074,
      "step": 1855
    },
    {
      "epoch": 0.19578059071729959,
      "grad_norm": 0.6427181959152222,
      "learning_rate": 0.0015,
      "loss": 1.8203,
      "step": 1856
    },
    {
      "epoch": 0.19588607594936708,
      "grad_norm": 0.6976577639579773,
      "learning_rate": 0.0015,
      "loss": 1.8131,
      "step": 1857
    },
    {
      "epoch": 0.1959915611814346,
      "grad_norm": 0.5904081463813782,
      "learning_rate": 0.0015,
      "loss": 1.8154,
      "step": 1858
    },
    {
      "epoch": 0.19609704641350212,
      "grad_norm": 0.5384019017219543,
      "learning_rate": 0.0015,
      "loss": 1.8107,
      "step": 1859
    },
    {
      "epoch": 0.1962025316455696,
      "grad_norm": 0.7040228843688965,
      "learning_rate": 0.0015,
      "loss": 1.8424,
      "step": 1860
    },
    {
      "epoch": 0.19630801687763713,
      "grad_norm": 0.5308371186256409,
      "learning_rate": 0.0015,
      "loss": 1.8195,
      "step": 1861
    },
    {
      "epoch": 0.19641350210970465,
      "grad_norm": 0.5134651064872742,
      "learning_rate": 0.0015,
      "loss": 1.8195,
      "step": 1862
    },
    {
      "epoch": 0.19651898734177214,
      "grad_norm": 0.5103256702423096,
      "learning_rate": 0.0015,
      "loss": 1.8766,
      "step": 1863
    },
    {
      "epoch": 0.19662447257383966,
      "grad_norm": 0.4865470230579376,
      "learning_rate": 0.0015,
      "loss": 1.8013,
      "step": 1864
    },
    {
      "epoch": 0.19672995780590719,
      "grad_norm": 0.49293550848960876,
      "learning_rate": 0.0015,
      "loss": 1.8039,
      "step": 1865
    },
    {
      "epoch": 0.19683544303797468,
      "grad_norm": 0.546439528465271,
      "learning_rate": 0.0015,
      "loss": 1.7816,
      "step": 1866
    },
    {
      "epoch": 0.1969409282700422,
      "grad_norm": 0.5053555965423584,
      "learning_rate": 0.0015,
      "loss": 1.83,
      "step": 1867
    },
    {
      "epoch": 0.19704641350210972,
      "grad_norm": 0.5129899978637695,
      "learning_rate": 0.0015,
      "loss": 1.7929,
      "step": 1868
    },
    {
      "epoch": 0.1971518987341772,
      "grad_norm": 0.4739459455013275,
      "learning_rate": 0.0015,
      "loss": 1.8354,
      "step": 1869
    },
    {
      "epoch": 0.19725738396624473,
      "grad_norm": 0.5697641372680664,
      "learning_rate": 0.0015,
      "loss": 1.8476,
      "step": 1870
    },
    {
      "epoch": 0.19736286919831222,
      "grad_norm": 0.6657871603965759,
      "learning_rate": 0.0015,
      "loss": 1.781,
      "step": 1871
    },
    {
      "epoch": 0.19746835443037974,
      "grad_norm": 0.5394650101661682,
      "learning_rate": 0.0015,
      "loss": 1.7904,
      "step": 1872
    },
    {
      "epoch": 0.19757383966244726,
      "grad_norm": 0.5282880067825317,
      "learning_rate": 0.0015,
      "loss": 1.7813,
      "step": 1873
    },
    {
      "epoch": 0.19767932489451476,
      "grad_norm": 0.6685124039649963,
      "learning_rate": 0.0015,
      "loss": 1.8038,
      "step": 1874
    },
    {
      "epoch": 0.19778481012658228,
      "grad_norm": 0.6586697101593018,
      "learning_rate": 0.0015,
      "loss": 1.8199,
      "step": 1875
    },
    {
      "epoch": 0.1978902953586498,
      "grad_norm": 0.6609626412391663,
      "learning_rate": 0.0015,
      "loss": 1.8417,
      "step": 1876
    },
    {
      "epoch": 0.1979957805907173,
      "grad_norm": 0.6019270420074463,
      "learning_rate": 0.0015,
      "loss": 1.8683,
      "step": 1877
    },
    {
      "epoch": 0.1981012658227848,
      "grad_norm": 0.7429397702217102,
      "learning_rate": 0.0015,
      "loss": 1.7826,
      "step": 1878
    },
    {
      "epoch": 0.19820675105485233,
      "grad_norm": 0.948248028755188,
      "learning_rate": 0.0015,
      "loss": 1.7684,
      "step": 1879
    },
    {
      "epoch": 0.19831223628691982,
      "grad_norm": 0.49379172921180725,
      "learning_rate": 0.0015,
      "loss": 1.8243,
      "step": 1880
    },
    {
      "epoch": 0.19841772151898734,
      "grad_norm": 0.7900415062904358,
      "learning_rate": 0.0015,
      "loss": 1.8172,
      "step": 1881
    },
    {
      "epoch": 0.19852320675105486,
      "grad_norm": 0.8166360855102539,
      "learning_rate": 0.0015,
      "loss": 1.7975,
      "step": 1882
    },
    {
      "epoch": 0.19862869198312236,
      "grad_norm": 0.5131245255470276,
      "learning_rate": 0.0015,
      "loss": 1.8088,
      "step": 1883
    },
    {
      "epoch": 0.19873417721518988,
      "grad_norm": 0.4987246096134186,
      "learning_rate": 0.0015,
      "loss": 1.8229,
      "step": 1884
    },
    {
      "epoch": 0.19883966244725737,
      "grad_norm": 0.466266006231308,
      "learning_rate": 0.0015,
      "loss": 1.7958,
      "step": 1885
    },
    {
      "epoch": 0.1989451476793249,
      "grad_norm": 0.578522801399231,
      "learning_rate": 0.0015,
      "loss": 1.8149,
      "step": 1886
    },
    {
      "epoch": 0.1990506329113924,
      "grad_norm": 0.46351176500320435,
      "learning_rate": 0.0015,
      "loss": 1.8114,
      "step": 1887
    },
    {
      "epoch": 0.1991561181434599,
      "grad_norm": 0.6146172285079956,
      "learning_rate": 0.0015,
      "loss": 1.8381,
      "step": 1888
    },
    {
      "epoch": 0.19926160337552742,
      "grad_norm": 0.7563574910163879,
      "learning_rate": 0.0015,
      "loss": 1.803,
      "step": 1889
    },
    {
      "epoch": 0.19936708860759494,
      "grad_norm": 0.5614977478981018,
      "learning_rate": 0.0015,
      "loss": 1.8064,
      "step": 1890
    },
    {
      "epoch": 0.19947257383966244,
      "grad_norm": 0.6315422058105469,
      "learning_rate": 0.0015,
      "loss": 1.8257,
      "step": 1891
    },
    {
      "epoch": 0.19957805907172996,
      "grad_norm": 0.8877629637718201,
      "learning_rate": 0.0015,
      "loss": 1.7949,
      "step": 1892
    },
    {
      "epoch": 0.19968354430379748,
      "grad_norm": 0.6193661093711853,
      "learning_rate": 0.0015,
      "loss": 1.7596,
      "step": 1893
    },
    {
      "epoch": 0.19978902953586497,
      "grad_norm": 0.5928061008453369,
      "learning_rate": 0.0015,
      "loss": 1.8014,
      "step": 1894
    },
    {
      "epoch": 0.1998945147679325,
      "grad_norm": 0.7133049964904785,
      "learning_rate": 0.0015,
      "loss": 1.8423,
      "step": 1895
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48946404457092285,
      "learning_rate": 0.0015,
      "loss": 1.8044,
      "step": 1896
    },
    {
      "epoch": 0.2001054852320675,
      "grad_norm": 0.689228355884552,
      "learning_rate": 0.0015,
      "loss": 1.8168,
      "step": 1897
    },
    {
      "epoch": 0.20021097046413502,
      "grad_norm": 0.7096043825149536,
      "learning_rate": 0.0015,
      "loss": 1.827,
      "step": 1898
    },
    {
      "epoch": 0.20031645569620254,
      "grad_norm": 0.5263689756393433,
      "learning_rate": 0.0015,
      "loss": 1.805,
      "step": 1899
    },
    {
      "epoch": 0.20042194092827004,
      "grad_norm": 0.8081044554710388,
      "learning_rate": 0.0015,
      "loss": 1.8287,
      "step": 1900
    },
    {
      "epoch": 0.20052742616033756,
      "grad_norm": 0.7103652358055115,
      "learning_rate": 0.0015,
      "loss": 1.8024,
      "step": 1901
    },
    {
      "epoch": 0.20063291139240505,
      "grad_norm": 0.8252784013748169,
      "learning_rate": 0.0015,
      "loss": 1.7685,
      "step": 1902
    },
    {
      "epoch": 0.20073839662447257,
      "grad_norm": 0.9588136076927185,
      "learning_rate": 0.0015,
      "loss": 1.8297,
      "step": 1903
    },
    {
      "epoch": 0.2008438818565401,
      "grad_norm": 0.7772409319877625,
      "learning_rate": 0.0015,
      "loss": 1.8036,
      "step": 1904
    },
    {
      "epoch": 0.20094936708860758,
      "grad_norm": 0.4961509704589844,
      "learning_rate": 0.0015,
      "loss": 1.7533,
      "step": 1905
    },
    {
      "epoch": 0.2010548523206751,
      "grad_norm": 0.691749632358551,
      "learning_rate": 0.0015,
      "loss": 1.8402,
      "step": 1906
    },
    {
      "epoch": 0.20116033755274262,
      "grad_norm": 0.6328140497207642,
      "learning_rate": 0.0015,
      "loss": 1.8068,
      "step": 1907
    },
    {
      "epoch": 0.20126582278481012,
      "grad_norm": 0.5572474598884583,
      "learning_rate": 0.0015,
      "loss": 1.8157,
      "step": 1908
    },
    {
      "epoch": 0.20137130801687764,
      "grad_norm": 0.6494549512863159,
      "learning_rate": 0.0015,
      "loss": 1.8577,
      "step": 1909
    },
    {
      "epoch": 0.20147679324894516,
      "grad_norm": 0.607793390750885,
      "learning_rate": 0.0015,
      "loss": 1.8236,
      "step": 1910
    },
    {
      "epoch": 0.20158227848101265,
      "grad_norm": 0.6155970096588135,
      "learning_rate": 0.0015,
      "loss": 1.8284,
      "step": 1911
    },
    {
      "epoch": 0.20168776371308017,
      "grad_norm": 0.5968335866928101,
      "learning_rate": 0.0015,
      "loss": 1.7965,
      "step": 1912
    },
    {
      "epoch": 0.2017932489451477,
      "grad_norm": 0.5001022815704346,
      "learning_rate": 0.0015,
      "loss": 1.7886,
      "step": 1913
    },
    {
      "epoch": 0.20189873417721518,
      "grad_norm": 0.7552065253257751,
      "learning_rate": 0.0015,
      "loss": 1.814,
      "step": 1914
    },
    {
      "epoch": 0.2020042194092827,
      "grad_norm": 0.7346456050872803,
      "learning_rate": 0.0015,
      "loss": 1.7885,
      "step": 1915
    },
    {
      "epoch": 0.20210970464135022,
      "grad_norm": 0.5578684210777283,
      "learning_rate": 0.0015,
      "loss": 1.8142,
      "step": 1916
    },
    {
      "epoch": 0.20221518987341772,
      "grad_norm": 0.6246179342269897,
      "learning_rate": 0.0015,
      "loss": 1.7962,
      "step": 1917
    },
    {
      "epoch": 0.20232067510548524,
      "grad_norm": 0.6429421305656433,
      "learning_rate": 0.0015,
      "loss": 1.8151,
      "step": 1918
    },
    {
      "epoch": 0.20242616033755273,
      "grad_norm": 0.5428091883659363,
      "learning_rate": 0.0015,
      "loss": 1.8491,
      "step": 1919
    },
    {
      "epoch": 0.20253164556962025,
      "grad_norm": 0.5293623805046082,
      "learning_rate": 0.0015,
      "loss": 1.7877,
      "step": 1920
    },
    {
      "epoch": 0.20263713080168777,
      "grad_norm": 0.5695270299911499,
      "learning_rate": 0.0015,
      "loss": 1.826,
      "step": 1921
    },
    {
      "epoch": 0.20274261603375526,
      "grad_norm": 0.46148911118507385,
      "learning_rate": 0.0015,
      "loss": 1.7684,
      "step": 1922
    },
    {
      "epoch": 0.20284810126582278,
      "grad_norm": 0.8151350021362305,
      "learning_rate": 0.0015,
      "loss": 1.8195,
      "step": 1923
    },
    {
      "epoch": 0.2029535864978903,
      "grad_norm": 0.9036180973052979,
      "learning_rate": 0.0015,
      "loss": 1.7905,
      "step": 1924
    },
    {
      "epoch": 0.2030590717299578,
      "grad_norm": 0.5296486616134644,
      "learning_rate": 0.0015,
      "loss": 1.8419,
      "step": 1925
    },
    {
      "epoch": 0.20316455696202532,
      "grad_norm": 0.5247792601585388,
      "learning_rate": 0.0015,
      "loss": 1.7909,
      "step": 1926
    },
    {
      "epoch": 0.20327004219409284,
      "grad_norm": 0.5949251651763916,
      "learning_rate": 0.0015,
      "loss": 1.7899,
      "step": 1927
    },
    {
      "epoch": 0.20337552742616033,
      "grad_norm": 0.4845055043697357,
      "learning_rate": 0.0015,
      "loss": 1.8417,
      "step": 1928
    },
    {
      "epoch": 0.20348101265822785,
      "grad_norm": 0.8005715608596802,
      "learning_rate": 0.0015,
      "loss": 1.7822,
      "step": 1929
    },
    {
      "epoch": 0.20358649789029537,
      "grad_norm": 0.8493461012840271,
      "learning_rate": 0.0015,
      "loss": 1.8313,
      "step": 1930
    },
    {
      "epoch": 0.20369198312236286,
      "grad_norm": 0.47339266538619995,
      "learning_rate": 0.0015,
      "loss": 1.7568,
      "step": 1931
    },
    {
      "epoch": 0.20379746835443038,
      "grad_norm": 0.6098864674568176,
      "learning_rate": 0.0015,
      "loss": 1.8162,
      "step": 1932
    },
    {
      "epoch": 0.2039029535864979,
      "grad_norm": 0.5633969902992249,
      "learning_rate": 0.0015,
      "loss": 1.8032,
      "step": 1933
    },
    {
      "epoch": 0.2040084388185654,
      "grad_norm": 0.4892830550670624,
      "learning_rate": 0.0015,
      "loss": 1.8083,
      "step": 1934
    },
    {
      "epoch": 0.20411392405063292,
      "grad_norm": 0.5428746938705444,
      "learning_rate": 0.0015,
      "loss": 1.7833,
      "step": 1935
    },
    {
      "epoch": 0.2042194092827004,
      "grad_norm": 0.5698093175888062,
      "learning_rate": 0.0015,
      "loss": 1.8085,
      "step": 1936
    },
    {
      "epoch": 0.20432489451476793,
      "grad_norm": 0.5401638746261597,
      "learning_rate": 0.0015,
      "loss": 1.8205,
      "step": 1937
    },
    {
      "epoch": 0.20443037974683545,
      "grad_norm": 0.5090903639793396,
      "learning_rate": 0.0015,
      "loss": 1.8161,
      "step": 1938
    },
    {
      "epoch": 0.20453586497890294,
      "grad_norm": 0.6209476590156555,
      "learning_rate": 0.0015,
      "loss": 1.8176,
      "step": 1939
    },
    {
      "epoch": 0.20464135021097046,
      "grad_norm": 0.483394593000412,
      "learning_rate": 0.0015,
      "loss": 1.7667,
      "step": 1940
    },
    {
      "epoch": 0.20474683544303798,
      "grad_norm": 0.5928177237510681,
      "learning_rate": 0.0015,
      "loss": 1.813,
      "step": 1941
    },
    {
      "epoch": 0.20485232067510548,
      "grad_norm": 0.5569455623626709,
      "learning_rate": 0.0015,
      "loss": 1.8175,
      "step": 1942
    },
    {
      "epoch": 0.204957805907173,
      "grad_norm": 0.5346285104751587,
      "learning_rate": 0.0015,
      "loss": 1.7918,
      "step": 1943
    },
    {
      "epoch": 0.20506329113924052,
      "grad_norm": 0.5668536424636841,
      "learning_rate": 0.0015,
      "loss": 1.7953,
      "step": 1944
    },
    {
      "epoch": 0.205168776371308,
      "grad_norm": 0.6100139021873474,
      "learning_rate": 0.0015,
      "loss": 1.8392,
      "step": 1945
    },
    {
      "epoch": 0.20527426160337553,
      "grad_norm": 0.5450618267059326,
      "learning_rate": 0.0015,
      "loss": 1.765,
      "step": 1946
    },
    {
      "epoch": 0.20537974683544305,
      "grad_norm": 0.5417037010192871,
      "learning_rate": 0.0015,
      "loss": 1.8286,
      "step": 1947
    },
    {
      "epoch": 0.20548523206751054,
      "grad_norm": 0.5623211860656738,
      "learning_rate": 0.0015,
      "loss": 1.7838,
      "step": 1948
    },
    {
      "epoch": 0.20559071729957806,
      "grad_norm": 0.5735540986061096,
      "learning_rate": 0.0015,
      "loss": 1.7815,
      "step": 1949
    },
    {
      "epoch": 0.20569620253164558,
      "grad_norm": 0.514814019203186,
      "learning_rate": 0.0015,
      "loss": 1.7698,
      "step": 1950
    },
    {
      "epoch": 0.20580168776371308,
      "grad_norm": 0.532285213470459,
      "learning_rate": 0.0015,
      "loss": 1.8375,
      "step": 1951
    },
    {
      "epoch": 0.2059071729957806,
      "grad_norm": 0.6268038749694824,
      "learning_rate": 0.0015,
      "loss": 1.8088,
      "step": 1952
    },
    {
      "epoch": 0.2060126582278481,
      "grad_norm": 0.5530931353569031,
      "learning_rate": 0.0015,
      "loss": 1.834,
      "step": 1953
    },
    {
      "epoch": 0.2061181434599156,
      "grad_norm": 0.6239747405052185,
      "learning_rate": 0.0015,
      "loss": 1.7687,
      "step": 1954
    },
    {
      "epoch": 0.20622362869198313,
      "grad_norm": 0.6208072900772095,
      "learning_rate": 0.0015,
      "loss": 1.8158,
      "step": 1955
    },
    {
      "epoch": 0.20632911392405062,
      "grad_norm": 0.561712384223938,
      "learning_rate": 0.0015,
      "loss": 1.7592,
      "step": 1956
    },
    {
      "epoch": 0.20643459915611814,
      "grad_norm": 0.5547567009925842,
      "learning_rate": 0.0015,
      "loss": 1.7936,
      "step": 1957
    },
    {
      "epoch": 0.20654008438818566,
      "grad_norm": 0.5211175680160522,
      "learning_rate": 0.0015,
      "loss": 1.8057,
      "step": 1958
    },
    {
      "epoch": 0.20664556962025316,
      "grad_norm": 0.5201579928398132,
      "learning_rate": 0.0015,
      "loss": 1.8237,
      "step": 1959
    },
    {
      "epoch": 0.20675105485232068,
      "grad_norm": 0.49631425738334656,
      "learning_rate": 0.0015,
      "loss": 1.7696,
      "step": 1960
    },
    {
      "epoch": 0.2068565400843882,
      "grad_norm": 0.4951164126396179,
      "learning_rate": 0.0015,
      "loss": 1.8308,
      "step": 1961
    },
    {
      "epoch": 0.2069620253164557,
      "grad_norm": 0.5647163987159729,
      "learning_rate": 0.0015,
      "loss": 1.8411,
      "step": 1962
    },
    {
      "epoch": 0.2070675105485232,
      "grad_norm": 0.5356828570365906,
      "learning_rate": 0.0015,
      "loss": 1.8246,
      "step": 1963
    },
    {
      "epoch": 0.20717299578059073,
      "grad_norm": 0.5040904879570007,
      "learning_rate": 0.0015,
      "loss": 1.7794,
      "step": 1964
    },
    {
      "epoch": 0.20727848101265822,
      "grad_norm": 0.6009175777435303,
      "learning_rate": 0.0015,
      "loss": 1.8102,
      "step": 1965
    },
    {
      "epoch": 0.20738396624472574,
      "grad_norm": 0.583840548992157,
      "learning_rate": 0.0015,
      "loss": 1.7967,
      "step": 1966
    },
    {
      "epoch": 0.20748945147679324,
      "grad_norm": 0.5271764993667603,
      "learning_rate": 0.0015,
      "loss": 1.8245,
      "step": 1967
    },
    {
      "epoch": 0.20759493670886076,
      "grad_norm": 0.6596837043762207,
      "learning_rate": 0.0015,
      "loss": 1.8085,
      "step": 1968
    },
    {
      "epoch": 0.20770042194092828,
      "grad_norm": 0.7611307501792908,
      "learning_rate": 0.0015,
      "loss": 1.8126,
      "step": 1969
    },
    {
      "epoch": 0.20780590717299577,
      "grad_norm": 0.512626051902771,
      "learning_rate": 0.0015,
      "loss": 1.7983,
      "step": 1970
    },
    {
      "epoch": 0.2079113924050633,
      "grad_norm": 0.7491683959960938,
      "learning_rate": 0.0015,
      "loss": 1.8134,
      "step": 1971
    },
    {
      "epoch": 0.2080168776371308,
      "grad_norm": 1.0004591941833496,
      "learning_rate": 0.0015,
      "loss": 1.7722,
      "step": 1972
    },
    {
      "epoch": 0.2081223628691983,
      "grad_norm": 0.5658147931098938,
      "learning_rate": 0.0015,
      "loss": 1.8139,
      "step": 1973
    },
    {
      "epoch": 0.20822784810126582,
      "grad_norm": 0.8615539073944092,
      "learning_rate": 0.0015,
      "loss": 1.7961,
      "step": 1974
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 1.200351595878601,
      "learning_rate": 0.0015,
      "loss": 1.8216,
      "step": 1975
    },
    {
      "epoch": 0.20843881856540084,
      "grad_norm": 0.48392531275749207,
      "learning_rate": 0.0015,
      "loss": 1.7886,
      "step": 1976
    },
    {
      "epoch": 0.20854430379746836,
      "grad_norm": 0.9284475445747375,
      "learning_rate": 0.0015,
      "loss": 1.8182,
      "step": 1977
    },
    {
      "epoch": 0.20864978902953588,
      "grad_norm": 0.7154855728149414,
      "learning_rate": 0.0015,
      "loss": 1.792,
      "step": 1978
    },
    {
      "epoch": 0.20875527426160337,
      "grad_norm": 0.7586033940315247,
      "learning_rate": 0.0015,
      "loss": 1.7763,
      "step": 1979
    },
    {
      "epoch": 0.2088607594936709,
      "grad_norm": 1.0334913730621338,
      "learning_rate": 0.0015,
      "loss": 1.8212,
      "step": 1980
    },
    {
      "epoch": 0.2089662447257384,
      "grad_norm": 0.6829190254211426,
      "learning_rate": 0.0015,
      "loss": 1.7854,
      "step": 1981
    },
    {
      "epoch": 0.2090717299578059,
      "grad_norm": 0.7945849299430847,
      "learning_rate": 0.0015,
      "loss": 1.7529,
      "step": 1982
    },
    {
      "epoch": 0.20917721518987342,
      "grad_norm": 0.7287125587463379,
      "learning_rate": 0.0015,
      "loss": 1.8047,
      "step": 1983
    },
    {
      "epoch": 0.20928270042194091,
      "grad_norm": 0.6630305051803589,
      "learning_rate": 0.0015,
      "loss": 1.8365,
      "step": 1984
    },
    {
      "epoch": 0.20938818565400844,
      "grad_norm": 0.7076170444488525,
      "learning_rate": 0.0015,
      "loss": 1.7945,
      "step": 1985
    },
    {
      "epoch": 0.20949367088607596,
      "grad_norm": 0.6776531338691711,
      "learning_rate": 0.0015,
      "loss": 1.7725,
      "step": 1986
    },
    {
      "epoch": 0.20959915611814345,
      "grad_norm": 0.7501946091651917,
      "learning_rate": 0.0015,
      "loss": 1.762,
      "step": 1987
    },
    {
      "epoch": 0.20970464135021097,
      "grad_norm": 0.8988157510757446,
      "learning_rate": 0.0015,
      "loss": 1.8142,
      "step": 1988
    },
    {
      "epoch": 0.2098101265822785,
      "grad_norm": 0.5500444769859314,
      "learning_rate": 0.0015,
      "loss": 1.8292,
      "step": 1989
    },
    {
      "epoch": 0.20991561181434598,
      "grad_norm": 0.7037571668624878,
      "learning_rate": 0.0015,
      "loss": 1.8082,
      "step": 1990
    },
    {
      "epoch": 0.2100210970464135,
      "grad_norm": 0.6246312856674194,
      "learning_rate": 0.0015,
      "loss": 1.8202,
      "step": 1991
    },
    {
      "epoch": 0.21012658227848102,
      "grad_norm": 0.5556793808937073,
      "learning_rate": 0.0015,
      "loss": 1.7824,
      "step": 1992
    },
    {
      "epoch": 0.21023206751054851,
      "grad_norm": 0.7253715991973877,
      "learning_rate": 0.0015,
      "loss": 1.8209,
      "step": 1993
    },
    {
      "epoch": 0.21033755274261604,
      "grad_norm": 0.5947061777114868,
      "learning_rate": 0.0015,
      "loss": 1.7625,
      "step": 1994
    },
    {
      "epoch": 0.21044303797468356,
      "grad_norm": 0.5559309124946594,
      "learning_rate": 0.0015,
      "loss": 1.8327,
      "step": 1995
    },
    {
      "epoch": 0.21054852320675105,
      "grad_norm": 0.6597854495048523,
      "learning_rate": 0.0015,
      "loss": 1.7723,
      "step": 1996
    },
    {
      "epoch": 0.21065400843881857,
      "grad_norm": 0.5748874545097351,
      "learning_rate": 0.0015,
      "loss": 1.7907,
      "step": 1997
    },
    {
      "epoch": 0.2107594936708861,
      "grad_norm": 0.4824632406234741,
      "learning_rate": 0.0015,
      "loss": 1.7729,
      "step": 1998
    },
    {
      "epoch": 0.21086497890295358,
      "grad_norm": 0.5499720573425293,
      "learning_rate": 0.0015,
      "loss": 1.7715,
      "step": 1999
    },
    {
      "epoch": 0.2109704641350211,
      "grad_norm": 0.47200867533683777,
      "learning_rate": 0.0015,
      "loss": 1.8024,
      "step": 2000
    },
    {
      "epoch": 0.2110759493670886,
      "grad_norm": 0.52181476354599,
      "learning_rate": 0.0015,
      "loss": 1.8389,
      "step": 2001
    },
    {
      "epoch": 0.21118143459915611,
      "grad_norm": 0.5757996439933777,
      "learning_rate": 0.0015,
      "loss": 1.7929,
      "step": 2002
    },
    {
      "epoch": 0.21128691983122364,
      "grad_norm": 0.4950844645500183,
      "learning_rate": 0.0015,
      "loss": 1.7456,
      "step": 2003
    },
    {
      "epoch": 0.21139240506329113,
      "grad_norm": 0.47291117906570435,
      "learning_rate": 0.0015,
      "loss": 1.7939,
      "step": 2004
    },
    {
      "epoch": 0.21149789029535865,
      "grad_norm": 0.4911331832408905,
      "learning_rate": 0.0015,
      "loss": 1.8218,
      "step": 2005
    },
    {
      "epoch": 0.21160337552742617,
      "grad_norm": 0.477486789226532,
      "learning_rate": 0.0015,
      "loss": 1.8038,
      "step": 2006
    },
    {
      "epoch": 0.21170886075949366,
      "grad_norm": 0.4663199484348297,
      "learning_rate": 0.0015,
      "loss": 1.7897,
      "step": 2007
    },
    {
      "epoch": 0.21181434599156118,
      "grad_norm": 0.521081268787384,
      "learning_rate": 0.0015,
      "loss": 1.8358,
      "step": 2008
    },
    {
      "epoch": 0.2119198312236287,
      "grad_norm": 0.49341341853141785,
      "learning_rate": 0.0015,
      "loss": 1.7782,
      "step": 2009
    },
    {
      "epoch": 0.2120253164556962,
      "grad_norm": 0.5227444171905518,
      "learning_rate": 0.0015,
      "loss": 1.7784,
      "step": 2010
    },
    {
      "epoch": 0.21213080168776371,
      "grad_norm": 0.47457006573677063,
      "learning_rate": 0.0015,
      "loss": 1.8092,
      "step": 2011
    },
    {
      "epoch": 0.21223628691983124,
      "grad_norm": 0.4924984276294708,
      "learning_rate": 0.0015,
      "loss": 1.8039,
      "step": 2012
    },
    {
      "epoch": 0.21234177215189873,
      "grad_norm": 0.4714333713054657,
      "learning_rate": 0.0015,
      "loss": 1.8099,
      "step": 2013
    },
    {
      "epoch": 0.21244725738396625,
      "grad_norm": 0.45190441608428955,
      "learning_rate": 0.0015,
      "loss": 1.7932,
      "step": 2014
    },
    {
      "epoch": 0.21255274261603377,
      "grad_norm": 0.46433931589126587,
      "learning_rate": 0.0015,
      "loss": 1.8013,
      "step": 2015
    },
    {
      "epoch": 0.21265822784810126,
      "grad_norm": 0.5200649499893188,
      "learning_rate": 0.0015,
      "loss": 1.8317,
      "step": 2016
    },
    {
      "epoch": 0.21276371308016878,
      "grad_norm": 0.5095267295837402,
      "learning_rate": 0.0015,
      "loss": 1.7953,
      "step": 2017
    },
    {
      "epoch": 0.21286919831223627,
      "grad_norm": 0.4929715096950531,
      "learning_rate": 0.0015,
      "loss": 1.7978,
      "step": 2018
    },
    {
      "epoch": 0.2129746835443038,
      "grad_norm": 0.5923967957496643,
      "learning_rate": 0.0015,
      "loss": 1.8137,
      "step": 2019
    },
    {
      "epoch": 0.21308016877637131,
      "grad_norm": 0.4702209234237671,
      "learning_rate": 0.0015,
      "loss": 1.7838,
      "step": 2020
    },
    {
      "epoch": 0.2131856540084388,
      "grad_norm": 0.6560384035110474,
      "learning_rate": 0.0015,
      "loss": 1.7878,
      "step": 2021
    },
    {
      "epoch": 0.21329113924050633,
      "grad_norm": 0.6585400700569153,
      "learning_rate": 0.0015,
      "loss": 1.7798,
      "step": 2022
    },
    {
      "epoch": 0.21339662447257385,
      "grad_norm": 0.5050377249717712,
      "learning_rate": 0.0015,
      "loss": 1.8481,
      "step": 2023
    },
    {
      "epoch": 0.21350210970464134,
      "grad_norm": 0.7235761880874634,
      "learning_rate": 0.0015,
      "loss": 1.826,
      "step": 2024
    },
    {
      "epoch": 0.21360759493670886,
      "grad_norm": 0.5125359892845154,
      "learning_rate": 0.0015,
      "loss": 1.7541,
      "step": 2025
    },
    {
      "epoch": 0.21371308016877638,
      "grad_norm": 0.6349862217903137,
      "learning_rate": 0.0015,
      "loss": 1.7685,
      "step": 2026
    },
    {
      "epoch": 0.21381856540084387,
      "grad_norm": 0.559128999710083,
      "learning_rate": 0.0015,
      "loss": 1.7753,
      "step": 2027
    },
    {
      "epoch": 0.2139240506329114,
      "grad_norm": 0.5841044187545776,
      "learning_rate": 0.0015,
      "loss": 1.8158,
      "step": 2028
    },
    {
      "epoch": 0.21402953586497891,
      "grad_norm": 0.5483068823814392,
      "learning_rate": 0.0015,
      "loss": 1.8183,
      "step": 2029
    },
    {
      "epoch": 0.2141350210970464,
      "grad_norm": 0.5353325605392456,
      "learning_rate": 0.0015,
      "loss": 1.7804,
      "step": 2030
    },
    {
      "epoch": 0.21424050632911393,
      "grad_norm": 0.5862396955490112,
      "learning_rate": 0.0015,
      "loss": 1.8114,
      "step": 2031
    },
    {
      "epoch": 0.21434599156118145,
      "grad_norm": 0.612694501876831,
      "learning_rate": 0.0015,
      "loss": 1.7935,
      "step": 2032
    },
    {
      "epoch": 0.21445147679324894,
      "grad_norm": 0.4950452744960785,
      "learning_rate": 0.0015,
      "loss": 1.7931,
      "step": 2033
    },
    {
      "epoch": 0.21455696202531646,
      "grad_norm": 0.6351242065429688,
      "learning_rate": 0.0015,
      "loss": 1.7857,
      "step": 2034
    },
    {
      "epoch": 0.21466244725738395,
      "grad_norm": 0.4815472364425659,
      "learning_rate": 0.0015,
      "loss": 1.7829,
      "step": 2035
    },
    {
      "epoch": 0.21476793248945147,
      "grad_norm": 0.5411445498466492,
      "learning_rate": 0.0015,
      "loss": 1.8062,
      "step": 2036
    },
    {
      "epoch": 0.214873417721519,
      "grad_norm": 0.5612822771072388,
      "learning_rate": 0.0015,
      "loss": 1.7741,
      "step": 2037
    },
    {
      "epoch": 0.2149789029535865,
      "grad_norm": 0.5131843686103821,
      "learning_rate": 0.0015,
      "loss": 1.7879,
      "step": 2038
    },
    {
      "epoch": 0.215084388185654,
      "grad_norm": 0.5584938526153564,
      "learning_rate": 0.0015,
      "loss": 1.7966,
      "step": 2039
    },
    {
      "epoch": 0.21518987341772153,
      "grad_norm": 0.5455635190010071,
      "learning_rate": 0.0015,
      "loss": 1.8202,
      "step": 2040
    },
    {
      "epoch": 0.21529535864978902,
      "grad_norm": 0.5857892036437988,
      "learning_rate": 0.0015,
      "loss": 1.7788,
      "step": 2041
    },
    {
      "epoch": 0.21540084388185654,
      "grad_norm": 0.6014600396156311,
      "learning_rate": 0.0015,
      "loss": 1.7754,
      "step": 2042
    },
    {
      "epoch": 0.21550632911392406,
      "grad_norm": 0.5224863886833191,
      "learning_rate": 0.0015,
      "loss": 1.8038,
      "step": 2043
    },
    {
      "epoch": 0.21561181434599155,
      "grad_norm": 0.604583740234375,
      "learning_rate": 0.0015,
      "loss": 1.8325,
      "step": 2044
    },
    {
      "epoch": 0.21571729957805907,
      "grad_norm": 0.523348331451416,
      "learning_rate": 0.0015,
      "loss": 1.7784,
      "step": 2045
    },
    {
      "epoch": 0.2158227848101266,
      "grad_norm": 0.6126733422279358,
      "learning_rate": 0.0015,
      "loss": 1.8076,
      "step": 2046
    },
    {
      "epoch": 0.2159282700421941,
      "grad_norm": 0.553733766078949,
      "learning_rate": 0.0015,
      "loss": 1.7776,
      "step": 2047
    },
    {
      "epoch": 0.2160337552742616,
      "grad_norm": 0.6064372062683105,
      "learning_rate": 0.0015,
      "loss": 1.7918,
      "step": 2048
    },
    {
      "epoch": 0.21613924050632913,
      "grad_norm": 0.5542830228805542,
      "learning_rate": 0.0015,
      "loss": 1.8039,
      "step": 2049
    },
    {
      "epoch": 0.21624472573839662,
      "grad_norm": 0.49567410349845886,
      "learning_rate": 0.0015,
      "loss": 1.7975,
      "step": 2050
    },
    {
      "epoch": 0.21635021097046414,
      "grad_norm": 0.5664834976196289,
      "learning_rate": 0.0015,
      "loss": 1.8117,
      "step": 2051
    },
    {
      "epoch": 0.21645569620253163,
      "grad_norm": 0.5432770252227783,
      "learning_rate": 0.0015,
      "loss": 1.8184,
      "step": 2052
    },
    {
      "epoch": 0.21656118143459915,
      "grad_norm": 0.5299774408340454,
      "learning_rate": 0.0015,
      "loss": 1.7653,
      "step": 2053
    },
    {
      "epoch": 0.21666666666666667,
      "grad_norm": 0.6001642346382141,
      "learning_rate": 0.0015,
      "loss": 1.7823,
      "step": 2054
    },
    {
      "epoch": 0.21677215189873417,
      "grad_norm": 0.4917055666446686,
      "learning_rate": 0.0015,
      "loss": 1.785,
      "step": 2055
    },
    {
      "epoch": 0.2168776371308017,
      "grad_norm": 0.5083092451095581,
      "learning_rate": 0.0015,
      "loss": 1.7962,
      "step": 2056
    },
    {
      "epoch": 0.2169831223628692,
      "grad_norm": 0.5252932906150818,
      "learning_rate": 0.0015,
      "loss": 1.7852,
      "step": 2057
    },
    {
      "epoch": 0.2170886075949367,
      "grad_norm": 0.4780125021934509,
      "learning_rate": 0.0015,
      "loss": 1.7602,
      "step": 2058
    },
    {
      "epoch": 0.21719409282700422,
      "grad_norm": 0.617774248123169,
      "learning_rate": 0.0015,
      "loss": 1.7632,
      "step": 2059
    },
    {
      "epoch": 0.21729957805907174,
      "grad_norm": 0.6348524689674377,
      "learning_rate": 0.0015,
      "loss": 1.7597,
      "step": 2060
    },
    {
      "epoch": 0.21740506329113923,
      "grad_norm": 0.48544228076934814,
      "learning_rate": 0.0015,
      "loss": 1.7768,
      "step": 2061
    },
    {
      "epoch": 0.21751054852320675,
      "grad_norm": 0.4656197130680084,
      "learning_rate": 0.0015,
      "loss": 1.809,
      "step": 2062
    },
    {
      "epoch": 0.21761603375527427,
      "grad_norm": 0.518939197063446,
      "learning_rate": 0.0015,
      "loss": 1.7651,
      "step": 2063
    },
    {
      "epoch": 0.21772151898734177,
      "grad_norm": 0.6058339476585388,
      "learning_rate": 0.0015,
      "loss": 1.7675,
      "step": 2064
    },
    {
      "epoch": 0.2178270042194093,
      "grad_norm": 0.5695094466209412,
      "learning_rate": 0.0015,
      "loss": 1.7576,
      "step": 2065
    },
    {
      "epoch": 0.21793248945147678,
      "grad_norm": 0.5090291500091553,
      "learning_rate": 0.0015,
      "loss": 1.8055,
      "step": 2066
    },
    {
      "epoch": 0.2180379746835443,
      "grad_norm": 0.5104354619979858,
      "learning_rate": 0.0015,
      "loss": 1.7603,
      "step": 2067
    },
    {
      "epoch": 0.21814345991561182,
      "grad_norm": 0.5981147885322571,
      "learning_rate": 0.0015,
      "loss": 1.8037,
      "step": 2068
    },
    {
      "epoch": 0.2182489451476793,
      "grad_norm": 0.7407425045967102,
      "learning_rate": 0.0015,
      "loss": 1.8131,
      "step": 2069
    },
    {
      "epoch": 0.21835443037974683,
      "grad_norm": 0.5820272564888,
      "learning_rate": 0.0015,
      "loss": 1.7915,
      "step": 2070
    },
    {
      "epoch": 0.21845991561181435,
      "grad_norm": 0.7423871755599976,
      "learning_rate": 0.0015,
      "loss": 1.8166,
      "step": 2071
    },
    {
      "epoch": 0.21856540084388185,
      "grad_norm": 1.164667010307312,
      "learning_rate": 0.0015,
      "loss": 1.7936,
      "step": 2072
    },
    {
      "epoch": 0.21867088607594937,
      "grad_norm": 0.551436185836792,
      "learning_rate": 0.0015,
      "loss": 1.8044,
      "step": 2073
    },
    {
      "epoch": 0.2187763713080169,
      "grad_norm": 0.8940706253051758,
      "learning_rate": 0.0015,
      "loss": 1.7778,
      "step": 2074
    },
    {
      "epoch": 0.21888185654008438,
      "grad_norm": 0.8622190356254578,
      "learning_rate": 0.0015,
      "loss": 1.7781,
      "step": 2075
    },
    {
      "epoch": 0.2189873417721519,
      "grad_norm": 0.4920498728752136,
      "learning_rate": 0.0015,
      "loss": 1.7471,
      "step": 2076
    },
    {
      "epoch": 0.21909282700421942,
      "grad_norm": 0.8537492752075195,
      "learning_rate": 0.0015,
      "loss": 1.8177,
      "step": 2077
    },
    {
      "epoch": 0.2191983122362869,
      "grad_norm": 0.6676235198974609,
      "learning_rate": 0.0015,
      "loss": 1.7695,
      "step": 2078
    },
    {
      "epoch": 0.21930379746835443,
      "grad_norm": 0.5924880504608154,
      "learning_rate": 0.0015,
      "loss": 1.8098,
      "step": 2079
    },
    {
      "epoch": 0.21940928270042195,
      "grad_norm": 0.8442460298538208,
      "learning_rate": 0.0015,
      "loss": 1.8183,
      "step": 2080
    },
    {
      "epoch": 0.21951476793248945,
      "grad_norm": 0.5594147443771362,
      "learning_rate": 0.0015,
      "loss": 1.791,
      "step": 2081
    },
    {
      "epoch": 0.21962025316455697,
      "grad_norm": 0.814310610294342,
      "learning_rate": 0.0015,
      "loss": 1.7805,
      "step": 2082
    },
    {
      "epoch": 0.21972573839662446,
      "grad_norm": 0.929280161857605,
      "learning_rate": 0.0015,
      "loss": 1.7676,
      "step": 2083
    },
    {
      "epoch": 0.21983122362869198,
      "grad_norm": 0.482134610414505,
      "learning_rate": 0.0015,
      "loss": 1.7829,
      "step": 2084
    },
    {
      "epoch": 0.2199367088607595,
      "grad_norm": 0.916218638420105,
      "learning_rate": 0.0015,
      "loss": 1.797,
      "step": 2085
    },
    {
      "epoch": 0.220042194092827,
      "grad_norm": 0.8021676540374756,
      "learning_rate": 0.0015,
      "loss": 1.7477,
      "step": 2086
    },
    {
      "epoch": 0.2201476793248945,
      "grad_norm": 0.5192430019378662,
      "learning_rate": 0.0015,
      "loss": 1.8255,
      "step": 2087
    },
    {
      "epoch": 0.22025316455696203,
      "grad_norm": 0.8338963985443115,
      "learning_rate": 0.0015,
      "loss": 1.7566,
      "step": 2088
    },
    {
      "epoch": 0.22035864978902953,
      "grad_norm": 0.5037552118301392,
      "learning_rate": 0.0015,
      "loss": 1.7922,
      "step": 2089
    },
    {
      "epoch": 0.22046413502109705,
      "grad_norm": 0.7627525329589844,
      "learning_rate": 0.0015,
      "loss": 1.8104,
      "step": 2090
    },
    {
      "epoch": 0.22056962025316457,
      "grad_norm": 0.7417036294937134,
      "learning_rate": 0.0015,
      "loss": 1.7891,
      "step": 2091
    },
    {
      "epoch": 0.22067510548523206,
      "grad_norm": 0.5279183387756348,
      "learning_rate": 0.0015,
      "loss": 1.7959,
      "step": 2092
    },
    {
      "epoch": 0.22078059071729958,
      "grad_norm": 0.9817603826522827,
      "learning_rate": 0.0015,
      "loss": 1.7903,
      "step": 2093
    },
    {
      "epoch": 0.2208860759493671,
      "grad_norm": 0.7074859142303467,
      "learning_rate": 0.0015,
      "loss": 1.79,
      "step": 2094
    },
    {
      "epoch": 0.2209915611814346,
      "grad_norm": 0.677912175655365,
      "learning_rate": 0.0015,
      "loss": 1.734,
      "step": 2095
    },
    {
      "epoch": 0.2210970464135021,
      "grad_norm": 0.8714148998260498,
      "learning_rate": 0.0015,
      "loss": 1.7989,
      "step": 2096
    },
    {
      "epoch": 0.22120253164556963,
      "grad_norm": 0.5216149091720581,
      "learning_rate": 0.0015,
      "loss": 1.8042,
      "step": 2097
    },
    {
      "epoch": 0.22130801687763713,
      "grad_norm": 0.8973119854927063,
      "learning_rate": 0.0015,
      "loss": 1.8231,
      "step": 2098
    },
    {
      "epoch": 0.22141350210970465,
      "grad_norm": 0.5994203090667725,
      "learning_rate": 0.0015,
      "loss": 1.8379,
      "step": 2099
    },
    {
      "epoch": 0.22151898734177214,
      "grad_norm": 0.7807937264442444,
      "learning_rate": 0.0015,
      "loss": 1.7876,
      "step": 2100
    },
    {
      "epoch": 0.22162447257383966,
      "grad_norm": 0.7964973449707031,
      "learning_rate": 0.0015,
      "loss": 1.7758,
      "step": 2101
    },
    {
      "epoch": 0.22172995780590718,
      "grad_norm": 0.49382224678993225,
      "learning_rate": 0.0015,
      "loss": 1.8076,
      "step": 2102
    },
    {
      "epoch": 0.22183544303797467,
      "grad_norm": 0.7006746530532837,
      "learning_rate": 0.0015,
      "loss": 1.7815,
      "step": 2103
    },
    {
      "epoch": 0.2219409282700422,
      "grad_norm": 0.4756909906864166,
      "learning_rate": 0.0015,
      "loss": 1.7763,
      "step": 2104
    },
    {
      "epoch": 0.2220464135021097,
      "grad_norm": 0.6178300976753235,
      "learning_rate": 0.0015,
      "loss": 1.7967,
      "step": 2105
    },
    {
      "epoch": 0.2221518987341772,
      "grad_norm": 0.6046226024627686,
      "learning_rate": 0.0015,
      "loss": 1.7589,
      "step": 2106
    },
    {
      "epoch": 0.22225738396624473,
      "grad_norm": 0.4923652410507202,
      "learning_rate": 0.0015,
      "loss": 1.7768,
      "step": 2107
    },
    {
      "epoch": 0.22236286919831225,
      "grad_norm": 0.49180859327316284,
      "learning_rate": 0.0015,
      "loss": 1.7653,
      "step": 2108
    },
    {
      "epoch": 0.22246835443037974,
      "grad_norm": 0.49905455112457275,
      "learning_rate": 0.0015,
      "loss": 1.767,
      "step": 2109
    },
    {
      "epoch": 0.22257383966244726,
      "grad_norm": 0.5526487231254578,
      "learning_rate": 0.0015,
      "loss": 1.7682,
      "step": 2110
    },
    {
      "epoch": 0.22267932489451478,
      "grad_norm": 0.5231235027313232,
      "learning_rate": 0.0015,
      "loss": 1.7421,
      "step": 2111
    },
    {
      "epoch": 0.22278481012658227,
      "grad_norm": 0.4582445025444031,
      "learning_rate": 0.0015,
      "loss": 1.774,
      "step": 2112
    },
    {
      "epoch": 0.2228902953586498,
      "grad_norm": 0.45971113443374634,
      "learning_rate": 0.0015,
      "loss": 1.8047,
      "step": 2113
    },
    {
      "epoch": 0.2229957805907173,
      "grad_norm": 0.4648434519767761,
      "learning_rate": 0.0015,
      "loss": 1.7784,
      "step": 2114
    },
    {
      "epoch": 0.2231012658227848,
      "grad_norm": 0.46904227137565613,
      "learning_rate": 0.0015,
      "loss": 1.7714,
      "step": 2115
    },
    {
      "epoch": 0.22320675105485233,
      "grad_norm": 0.5289586186408997,
      "learning_rate": 0.0015,
      "loss": 1.7727,
      "step": 2116
    },
    {
      "epoch": 0.22331223628691982,
      "grad_norm": 0.5065954327583313,
      "learning_rate": 0.0015,
      "loss": 1.7892,
      "step": 2117
    },
    {
      "epoch": 0.22341772151898734,
      "grad_norm": 0.45888662338256836,
      "learning_rate": 0.0015,
      "loss": 1.7675,
      "step": 2118
    },
    {
      "epoch": 0.22352320675105486,
      "grad_norm": 0.4652993977069855,
      "learning_rate": 0.0015,
      "loss": 1.7902,
      "step": 2119
    },
    {
      "epoch": 0.22362869198312235,
      "grad_norm": 0.48693880438804626,
      "learning_rate": 0.0015,
      "loss": 1.7642,
      "step": 2120
    },
    {
      "epoch": 0.22373417721518987,
      "grad_norm": 0.6744024157524109,
      "learning_rate": 0.0015,
      "loss": 1.7821,
      "step": 2121
    },
    {
      "epoch": 0.2238396624472574,
      "grad_norm": 0.592976987361908,
      "learning_rate": 0.0015,
      "loss": 1.7468,
      "step": 2122
    },
    {
      "epoch": 0.22394514767932489,
      "grad_norm": 0.49479156732559204,
      "learning_rate": 0.0015,
      "loss": 1.8064,
      "step": 2123
    },
    {
      "epoch": 0.2240506329113924,
      "grad_norm": 0.7077398300170898,
      "learning_rate": 0.0015,
      "loss": 1.7812,
      "step": 2124
    },
    {
      "epoch": 0.22415611814345993,
      "grad_norm": 0.5500556826591492,
      "learning_rate": 0.0015,
      "loss": 1.7569,
      "step": 2125
    },
    {
      "epoch": 0.22426160337552742,
      "grad_norm": 0.5560785531997681,
      "learning_rate": 0.0015,
      "loss": 1.7717,
      "step": 2126
    },
    {
      "epoch": 0.22436708860759494,
      "grad_norm": 0.844450056552887,
      "learning_rate": 0.0015,
      "loss": 1.8187,
      "step": 2127
    },
    {
      "epoch": 0.22447257383966246,
      "grad_norm": 0.5791806578636169,
      "learning_rate": 0.0015,
      "loss": 1.7607,
      "step": 2128
    },
    {
      "epoch": 0.22457805907172995,
      "grad_norm": 0.5502035617828369,
      "learning_rate": 0.0015,
      "loss": 1.799,
      "step": 2129
    },
    {
      "epoch": 0.22468354430379747,
      "grad_norm": 0.5226722359657288,
      "learning_rate": 0.0015,
      "loss": 1.7848,
      "step": 2130
    },
    {
      "epoch": 0.224789029535865,
      "grad_norm": 0.5256772637367249,
      "learning_rate": 0.0015,
      "loss": 1.7574,
      "step": 2131
    },
    {
      "epoch": 0.22489451476793249,
      "grad_norm": 0.5230455994606018,
      "learning_rate": 0.0015,
      "loss": 1.7299,
      "step": 2132
    },
    {
      "epoch": 0.225,
      "grad_norm": 0.484761506319046,
      "learning_rate": 0.0015,
      "loss": 1.7566,
      "step": 2133
    },
    {
      "epoch": 0.2251054852320675,
      "grad_norm": 0.5470814108848572,
      "learning_rate": 0.0015,
      "loss": 1.8039,
      "step": 2134
    },
    {
      "epoch": 0.22521097046413502,
      "grad_norm": 0.5252546668052673,
      "learning_rate": 0.0015,
      "loss": 1.7829,
      "step": 2135
    },
    {
      "epoch": 0.22531645569620254,
      "grad_norm": 0.49324169754981995,
      "learning_rate": 0.0015,
      "loss": 1.8049,
      "step": 2136
    },
    {
      "epoch": 0.22542194092827003,
      "grad_norm": 0.5957626104354858,
      "learning_rate": 0.0015,
      "loss": 1.7664,
      "step": 2137
    },
    {
      "epoch": 0.22552742616033755,
      "grad_norm": 0.5057148337364197,
      "learning_rate": 0.0015,
      "loss": 1.7721,
      "step": 2138
    },
    {
      "epoch": 0.22563291139240507,
      "grad_norm": 0.594565749168396,
      "learning_rate": 0.0015,
      "loss": 1.8047,
      "step": 2139
    },
    {
      "epoch": 0.22573839662447256,
      "grad_norm": 0.57353276014328,
      "learning_rate": 0.0015,
      "loss": 1.7995,
      "step": 2140
    },
    {
      "epoch": 0.22584388185654009,
      "grad_norm": 0.5522382259368896,
      "learning_rate": 0.0015,
      "loss": 1.7697,
      "step": 2141
    },
    {
      "epoch": 0.2259493670886076,
      "grad_norm": 0.5503726601600647,
      "learning_rate": 0.0015,
      "loss": 1.7394,
      "step": 2142
    },
    {
      "epoch": 0.2260548523206751,
      "grad_norm": 0.5022009611129761,
      "learning_rate": 0.0015,
      "loss": 1.7725,
      "step": 2143
    },
    {
      "epoch": 0.22616033755274262,
      "grad_norm": 0.5625883936882019,
      "learning_rate": 0.0015,
      "loss": 1.8358,
      "step": 2144
    },
    {
      "epoch": 0.22626582278481014,
      "grad_norm": 0.5966932773590088,
      "learning_rate": 0.0015,
      "loss": 1.7974,
      "step": 2145
    },
    {
      "epoch": 0.22637130801687763,
      "grad_norm": 0.6162047982215881,
      "learning_rate": 0.0015,
      "loss": 1.7807,
      "step": 2146
    },
    {
      "epoch": 0.22647679324894515,
      "grad_norm": 0.4755229651927948,
      "learning_rate": 0.0015,
      "loss": 1.8087,
      "step": 2147
    },
    {
      "epoch": 0.22658227848101264,
      "grad_norm": 0.6164548397064209,
      "learning_rate": 0.0015,
      "loss": 1.806,
      "step": 2148
    },
    {
      "epoch": 0.22668776371308016,
      "grad_norm": 0.6168848872184753,
      "learning_rate": 0.0015,
      "loss": 1.7675,
      "step": 2149
    },
    {
      "epoch": 0.22679324894514769,
      "grad_norm": 0.5575670003890991,
      "learning_rate": 0.0015,
      "loss": 1.7559,
      "step": 2150
    },
    {
      "epoch": 0.22689873417721518,
      "grad_norm": 0.87114018201828,
      "learning_rate": 0.0015,
      "loss": 1.8342,
      "step": 2151
    },
    {
      "epoch": 0.2270042194092827,
      "grad_norm": 0.6856433749198914,
      "learning_rate": 0.0015,
      "loss": 1.8385,
      "step": 2152
    },
    {
      "epoch": 0.22710970464135022,
      "grad_norm": 0.5108639597892761,
      "learning_rate": 0.0015,
      "loss": 1.7644,
      "step": 2153
    },
    {
      "epoch": 0.2272151898734177,
      "grad_norm": 0.6017791032791138,
      "learning_rate": 0.0015,
      "loss": 1.7978,
      "step": 2154
    },
    {
      "epoch": 0.22732067510548523,
      "grad_norm": 0.48644721508026123,
      "learning_rate": 0.0015,
      "loss": 1.7952,
      "step": 2155
    },
    {
      "epoch": 0.22742616033755275,
      "grad_norm": 0.5356946587562561,
      "learning_rate": 0.0015,
      "loss": 1.7715,
      "step": 2156
    },
    {
      "epoch": 0.22753164556962024,
      "grad_norm": 0.5389983654022217,
      "learning_rate": 0.0015,
      "loss": 1.7918,
      "step": 2157
    },
    {
      "epoch": 0.22763713080168776,
      "grad_norm": 0.6026822924613953,
      "learning_rate": 0.0015,
      "loss": 1.7516,
      "step": 2158
    },
    {
      "epoch": 0.22774261603375529,
      "grad_norm": 0.5365722179412842,
      "learning_rate": 0.0015,
      "loss": 1.8183,
      "step": 2159
    },
    {
      "epoch": 0.22784810126582278,
      "grad_norm": 0.5630393624305725,
      "learning_rate": 0.0015,
      "loss": 1.7838,
      "step": 2160
    },
    {
      "epoch": 0.2279535864978903,
      "grad_norm": 0.6574785709381104,
      "learning_rate": 0.0015,
      "loss": 1.7782,
      "step": 2161
    },
    {
      "epoch": 0.22805907172995782,
      "grad_norm": 0.4817398488521576,
      "learning_rate": 0.0015,
      "loss": 1.7533,
      "step": 2162
    },
    {
      "epoch": 0.2281645569620253,
      "grad_norm": 0.5800243616104126,
      "learning_rate": 0.0015,
      "loss": 1.78,
      "step": 2163
    },
    {
      "epoch": 0.22827004219409283,
      "grad_norm": 0.6244105100631714,
      "learning_rate": 0.0015,
      "loss": 1.7533,
      "step": 2164
    },
    {
      "epoch": 0.22837552742616032,
      "grad_norm": 0.5163666009902954,
      "learning_rate": 0.0015,
      "loss": 1.7302,
      "step": 2165
    },
    {
      "epoch": 0.22848101265822784,
      "grad_norm": 0.5679610371589661,
      "learning_rate": 0.0015,
      "loss": 1.8157,
      "step": 2166
    },
    {
      "epoch": 0.22858649789029536,
      "grad_norm": 0.5949824452400208,
      "learning_rate": 0.0015,
      "loss": 1.8137,
      "step": 2167
    },
    {
      "epoch": 0.22869198312236286,
      "grad_norm": 0.5223953723907471,
      "learning_rate": 0.0015,
      "loss": 1.8106,
      "step": 2168
    },
    {
      "epoch": 0.22879746835443038,
      "grad_norm": 0.4686530530452728,
      "learning_rate": 0.0015,
      "loss": 1.7649,
      "step": 2169
    },
    {
      "epoch": 0.2289029535864979,
      "grad_norm": 0.5529415607452393,
      "learning_rate": 0.0015,
      "loss": 1.7757,
      "step": 2170
    },
    {
      "epoch": 0.2290084388185654,
      "grad_norm": 0.46849730610847473,
      "learning_rate": 0.0015,
      "loss": 1.7642,
      "step": 2171
    },
    {
      "epoch": 0.2291139240506329,
      "grad_norm": 0.596246063709259,
      "learning_rate": 0.0015,
      "loss": 1.7505,
      "step": 2172
    },
    {
      "epoch": 0.22921940928270043,
      "grad_norm": 0.563389241695404,
      "learning_rate": 0.0015,
      "loss": 1.7421,
      "step": 2173
    },
    {
      "epoch": 0.22932489451476792,
      "grad_norm": 0.5011084675788879,
      "learning_rate": 0.0015,
      "loss": 1.7505,
      "step": 2174
    },
    {
      "epoch": 0.22943037974683544,
      "grad_norm": 0.6441606879234314,
      "learning_rate": 0.0015,
      "loss": 1.7535,
      "step": 2175
    },
    {
      "epoch": 0.22953586497890296,
      "grad_norm": 0.5723125338554382,
      "learning_rate": 0.0015,
      "loss": 1.7804,
      "step": 2176
    },
    {
      "epoch": 0.22964135021097046,
      "grad_norm": 0.48091256618499756,
      "learning_rate": 0.0015,
      "loss": 1.7455,
      "step": 2177
    },
    {
      "epoch": 0.22974683544303798,
      "grad_norm": 0.5418523550033569,
      "learning_rate": 0.0015,
      "loss": 1.782,
      "step": 2178
    },
    {
      "epoch": 0.2298523206751055,
      "grad_norm": 0.5442428588867188,
      "learning_rate": 0.0015,
      "loss": 1.7752,
      "step": 2179
    },
    {
      "epoch": 0.229957805907173,
      "grad_norm": 0.48983144760131836,
      "learning_rate": 0.0015,
      "loss": 1.7709,
      "step": 2180
    },
    {
      "epoch": 0.2300632911392405,
      "grad_norm": 0.5266924500465393,
      "learning_rate": 0.0015,
      "loss": 1.8014,
      "step": 2181
    },
    {
      "epoch": 0.230168776371308,
      "grad_norm": 0.4973575472831726,
      "learning_rate": 0.0015,
      "loss": 1.7677,
      "step": 2182
    },
    {
      "epoch": 0.23027426160337552,
      "grad_norm": 0.5498241782188416,
      "learning_rate": 0.0015,
      "loss": 1.7918,
      "step": 2183
    },
    {
      "epoch": 0.23037974683544304,
      "grad_norm": 0.4875961244106293,
      "learning_rate": 0.0015,
      "loss": 1.7767,
      "step": 2184
    },
    {
      "epoch": 0.23048523206751054,
      "grad_norm": 0.6175986528396606,
      "learning_rate": 0.0015,
      "loss": 1.7595,
      "step": 2185
    },
    {
      "epoch": 0.23059071729957806,
      "grad_norm": 0.5438140034675598,
      "learning_rate": 0.0015,
      "loss": 1.7966,
      "step": 2186
    },
    {
      "epoch": 0.23069620253164558,
      "grad_norm": 0.48663565516471863,
      "learning_rate": 0.0015,
      "loss": 1.7493,
      "step": 2187
    },
    {
      "epoch": 0.23080168776371307,
      "grad_norm": 0.6219778656959534,
      "learning_rate": 0.0015,
      "loss": 1.7771,
      "step": 2188
    },
    {
      "epoch": 0.2309071729957806,
      "grad_norm": 0.48322415351867676,
      "learning_rate": 0.0015,
      "loss": 1.7713,
      "step": 2189
    },
    {
      "epoch": 0.2310126582278481,
      "grad_norm": 0.6042963266372681,
      "learning_rate": 0.0015,
      "loss": 1.7944,
      "step": 2190
    },
    {
      "epoch": 0.2311181434599156,
      "grad_norm": 0.6675837635993958,
      "learning_rate": 0.0015,
      "loss": 1.778,
      "step": 2191
    },
    {
      "epoch": 0.23122362869198312,
      "grad_norm": 0.5233267545700073,
      "learning_rate": 0.0015,
      "loss": 1.7748,
      "step": 2192
    },
    {
      "epoch": 0.23132911392405064,
      "grad_norm": 0.6182083487510681,
      "learning_rate": 0.0015,
      "loss": 1.8242,
      "step": 2193
    },
    {
      "epoch": 0.23143459915611814,
      "grad_norm": 0.5622542500495911,
      "learning_rate": 0.0015,
      "loss": 1.8129,
      "step": 2194
    },
    {
      "epoch": 0.23154008438818566,
      "grad_norm": 0.5527053475379944,
      "learning_rate": 0.0015,
      "loss": 1.7541,
      "step": 2195
    },
    {
      "epoch": 0.23164556962025318,
      "grad_norm": 0.5042627453804016,
      "learning_rate": 0.0015,
      "loss": 1.7409,
      "step": 2196
    },
    {
      "epoch": 0.23175105485232067,
      "grad_norm": 0.4917895793914795,
      "learning_rate": 0.0015,
      "loss": 1.7587,
      "step": 2197
    },
    {
      "epoch": 0.2318565400843882,
      "grad_norm": 0.5107989311218262,
      "learning_rate": 0.0015,
      "loss": 1.7495,
      "step": 2198
    },
    {
      "epoch": 0.23196202531645568,
      "grad_norm": 0.4925711750984192,
      "learning_rate": 0.0015,
      "loss": 1.751,
      "step": 2199
    },
    {
      "epoch": 0.2320675105485232,
      "grad_norm": 0.4993196427822113,
      "learning_rate": 0.0015,
      "loss": 1.8107,
      "step": 2200
    },
    {
      "epoch": 0.23217299578059072,
      "grad_norm": 0.470471054315567,
      "learning_rate": 0.0015,
      "loss": 1.7697,
      "step": 2201
    },
    {
      "epoch": 0.23227848101265822,
      "grad_norm": 0.4779837727546692,
      "learning_rate": 0.0015,
      "loss": 1.775,
      "step": 2202
    },
    {
      "epoch": 0.23238396624472574,
      "grad_norm": 0.4769928753376007,
      "learning_rate": 0.0015,
      "loss": 1.7647,
      "step": 2203
    },
    {
      "epoch": 0.23248945147679326,
      "grad_norm": 0.4866911768913269,
      "learning_rate": 0.0015,
      "loss": 1.7544,
      "step": 2204
    },
    {
      "epoch": 0.23259493670886075,
      "grad_norm": 0.5387329459190369,
      "learning_rate": 0.0015,
      "loss": 1.7114,
      "step": 2205
    },
    {
      "epoch": 0.23270042194092827,
      "grad_norm": 0.4828197956085205,
      "learning_rate": 0.0015,
      "loss": 1.8261,
      "step": 2206
    },
    {
      "epoch": 0.2328059071729958,
      "grad_norm": 0.6584717035293579,
      "learning_rate": 0.0015,
      "loss": 1.761,
      "step": 2207
    },
    {
      "epoch": 0.23291139240506328,
      "grad_norm": 0.5035545229911804,
      "learning_rate": 0.0015,
      "loss": 1.7768,
      "step": 2208
    },
    {
      "epoch": 0.2330168776371308,
      "grad_norm": 0.4784407615661621,
      "learning_rate": 0.0015,
      "loss": 1.7438,
      "step": 2209
    },
    {
      "epoch": 0.23312236286919832,
      "grad_norm": 0.5389178991317749,
      "learning_rate": 0.0015,
      "loss": 1.7768,
      "step": 2210
    },
    {
      "epoch": 0.23322784810126582,
      "grad_norm": 0.45946386456489563,
      "learning_rate": 0.0015,
      "loss": 1.7652,
      "step": 2211
    },
    {
      "epoch": 0.23333333333333334,
      "grad_norm": 0.6016244292259216,
      "learning_rate": 0.0015,
      "loss": 1.773,
      "step": 2212
    },
    {
      "epoch": 0.23343881856540086,
      "grad_norm": 0.48807260394096375,
      "learning_rate": 0.0015,
      "loss": 1.7706,
      "step": 2213
    },
    {
      "epoch": 0.23354430379746835,
      "grad_norm": 0.49960052967071533,
      "learning_rate": 0.0015,
      "loss": 1.7778,
      "step": 2214
    },
    {
      "epoch": 0.23364978902953587,
      "grad_norm": 0.5041942000389099,
      "learning_rate": 0.0015,
      "loss": 1.8037,
      "step": 2215
    },
    {
      "epoch": 0.23375527426160336,
      "grad_norm": 0.504001259803772,
      "learning_rate": 0.0015,
      "loss": 1.7668,
      "step": 2216
    },
    {
      "epoch": 0.23386075949367088,
      "grad_norm": 0.4873487055301666,
      "learning_rate": 0.0015,
      "loss": 1.7755,
      "step": 2217
    },
    {
      "epoch": 0.2339662447257384,
      "grad_norm": 0.5332985520362854,
      "learning_rate": 0.0015,
      "loss": 1.7913,
      "step": 2218
    },
    {
      "epoch": 0.2340717299578059,
      "grad_norm": 0.5267864465713501,
      "learning_rate": 0.0015,
      "loss": 1.7552,
      "step": 2219
    },
    {
      "epoch": 0.23417721518987342,
      "grad_norm": 0.4540298879146576,
      "learning_rate": 0.0015,
      "loss": 1.8023,
      "step": 2220
    },
    {
      "epoch": 0.23428270042194094,
      "grad_norm": 0.502090334892273,
      "learning_rate": 0.0015,
      "loss": 1.7619,
      "step": 2221
    },
    {
      "epoch": 0.23438818565400843,
      "grad_norm": 0.48418280482292175,
      "learning_rate": 0.0015,
      "loss": 1.7422,
      "step": 2222
    },
    {
      "epoch": 0.23449367088607595,
      "grad_norm": 0.5112397074699402,
      "learning_rate": 0.0015,
      "loss": 1.7564,
      "step": 2223
    },
    {
      "epoch": 0.23459915611814347,
      "grad_norm": 0.546126127243042,
      "learning_rate": 0.0015,
      "loss": 1.7984,
      "step": 2224
    },
    {
      "epoch": 0.23470464135021096,
      "grad_norm": 0.46486884355545044,
      "learning_rate": 0.0015,
      "loss": 1.7654,
      "step": 2225
    },
    {
      "epoch": 0.23481012658227848,
      "grad_norm": 0.5794968008995056,
      "learning_rate": 0.0015,
      "loss": 1.7965,
      "step": 2226
    },
    {
      "epoch": 0.234915611814346,
      "grad_norm": 0.6269369125366211,
      "learning_rate": 0.0015,
      "loss": 1.7652,
      "step": 2227
    },
    {
      "epoch": 0.2350210970464135,
      "grad_norm": 0.4783438444137573,
      "learning_rate": 0.0015,
      "loss": 1.7749,
      "step": 2228
    },
    {
      "epoch": 0.23512658227848102,
      "grad_norm": 0.5994933247566223,
      "learning_rate": 0.0015,
      "loss": 1.7969,
      "step": 2229
    },
    {
      "epoch": 0.23523206751054854,
      "grad_norm": 0.5959358215332031,
      "learning_rate": 0.0015,
      "loss": 1.7622,
      "step": 2230
    },
    {
      "epoch": 0.23533755274261603,
      "grad_norm": 0.47580185532569885,
      "learning_rate": 0.0015,
      "loss": 1.7595,
      "step": 2231
    },
    {
      "epoch": 0.23544303797468355,
      "grad_norm": 0.7106600403785706,
      "learning_rate": 0.0015,
      "loss": 1.7833,
      "step": 2232
    },
    {
      "epoch": 0.23554852320675104,
      "grad_norm": 0.7072519659996033,
      "learning_rate": 0.0015,
      "loss": 1.7867,
      "step": 2233
    },
    {
      "epoch": 0.23565400843881856,
      "grad_norm": 0.554775059223175,
      "learning_rate": 0.0015,
      "loss": 1.7611,
      "step": 2234
    },
    {
      "epoch": 0.23575949367088608,
      "grad_norm": 0.627525806427002,
      "learning_rate": 0.0015,
      "loss": 1.7524,
      "step": 2235
    },
    {
      "epoch": 0.23586497890295358,
      "grad_norm": 0.5165814757347107,
      "learning_rate": 0.0015,
      "loss": 1.7617,
      "step": 2236
    },
    {
      "epoch": 0.2359704641350211,
      "grad_norm": 0.5884402990341187,
      "learning_rate": 0.0015,
      "loss": 1.7329,
      "step": 2237
    },
    {
      "epoch": 0.23607594936708862,
      "grad_norm": 0.466094970703125,
      "learning_rate": 0.0015,
      "loss": 1.8022,
      "step": 2238
    },
    {
      "epoch": 0.2361814345991561,
      "grad_norm": 0.5659448504447937,
      "learning_rate": 0.0015,
      "loss": 1.7602,
      "step": 2239
    },
    {
      "epoch": 0.23628691983122363,
      "grad_norm": 0.5581298470497131,
      "learning_rate": 0.0015,
      "loss": 1.7634,
      "step": 2240
    },
    {
      "epoch": 0.23639240506329115,
      "grad_norm": 0.47687047719955444,
      "learning_rate": 0.0015,
      "loss": 1.7445,
      "step": 2241
    },
    {
      "epoch": 0.23649789029535864,
      "grad_norm": 0.6534562706947327,
      "learning_rate": 0.0015,
      "loss": 1.7454,
      "step": 2242
    },
    {
      "epoch": 0.23660337552742616,
      "grad_norm": 0.7514973282814026,
      "learning_rate": 0.0015,
      "loss": 1.748,
      "step": 2243
    },
    {
      "epoch": 0.23670886075949368,
      "grad_norm": 0.8936958312988281,
      "learning_rate": 0.0015,
      "loss": 1.7634,
      "step": 2244
    },
    {
      "epoch": 0.23681434599156118,
      "grad_norm": 0.7259150147438049,
      "learning_rate": 0.0015,
      "loss": 1.7698,
      "step": 2245
    },
    {
      "epoch": 0.2369198312236287,
      "grad_norm": 0.5334154963493347,
      "learning_rate": 0.0015,
      "loss": 1.8148,
      "step": 2246
    },
    {
      "epoch": 0.2370253164556962,
      "grad_norm": 0.9084137082099915,
      "learning_rate": 0.0015,
      "loss": 1.7314,
      "step": 2247
    },
    {
      "epoch": 0.2371308016877637,
      "grad_norm": 0.6602224111557007,
      "learning_rate": 0.0015,
      "loss": 1.7956,
      "step": 2248
    },
    {
      "epoch": 0.23723628691983123,
      "grad_norm": 0.6410700082778931,
      "learning_rate": 0.0015,
      "loss": 1.7958,
      "step": 2249
    },
    {
      "epoch": 0.23734177215189872,
      "grad_norm": 0.8773201704025269,
      "learning_rate": 0.0015,
      "loss": 1.768,
      "step": 2250
    },
    {
      "epoch": 0.23744725738396624,
      "grad_norm": 0.8360086679458618,
      "learning_rate": 0.0015,
      "loss": 1.7672,
      "step": 2251
    },
    {
      "epoch": 0.23755274261603376,
      "grad_norm": 0.4706181287765503,
      "learning_rate": 0.0015,
      "loss": 1.7657,
      "step": 2252
    },
    {
      "epoch": 0.23765822784810126,
      "grad_norm": 0.8980696201324463,
      "learning_rate": 0.0015,
      "loss": 1.7531,
      "step": 2253
    },
    {
      "epoch": 0.23776371308016878,
      "grad_norm": 0.9459096789360046,
      "learning_rate": 0.0015,
      "loss": 1.7694,
      "step": 2254
    },
    {
      "epoch": 0.2378691983122363,
      "grad_norm": 0.4759543836116791,
      "learning_rate": 0.0015,
      "loss": 1.7568,
      "step": 2255
    },
    {
      "epoch": 0.2379746835443038,
      "grad_norm": 0.8913567662239075,
      "learning_rate": 0.0015,
      "loss": 1.7587,
      "step": 2256
    },
    {
      "epoch": 0.2380801687763713,
      "grad_norm": 0.6646949648857117,
      "learning_rate": 0.0015,
      "loss": 1.7306,
      "step": 2257
    },
    {
      "epoch": 0.23818565400843883,
      "grad_norm": 0.6183149814605713,
      "learning_rate": 0.0015,
      "loss": 1.7644,
      "step": 2258
    },
    {
      "epoch": 0.23829113924050632,
      "grad_norm": 0.6488416194915771,
      "learning_rate": 0.0015,
      "loss": 1.7854,
      "step": 2259
    },
    {
      "epoch": 0.23839662447257384,
      "grad_norm": 0.5634511709213257,
      "learning_rate": 0.0015,
      "loss": 1.7799,
      "step": 2260
    },
    {
      "epoch": 0.23850210970464136,
      "grad_norm": 0.6052711009979248,
      "learning_rate": 0.0015,
      "loss": 1.7648,
      "step": 2261
    },
    {
      "epoch": 0.23860759493670886,
      "grad_norm": 0.5160919427871704,
      "learning_rate": 0.0015,
      "loss": 1.7572,
      "step": 2262
    },
    {
      "epoch": 0.23871308016877638,
      "grad_norm": 0.6135463714599609,
      "learning_rate": 0.0015,
      "loss": 1.7876,
      "step": 2263
    },
    {
      "epoch": 0.23881856540084387,
      "grad_norm": 0.5171403884887695,
      "learning_rate": 0.0015,
      "loss": 1.8167,
      "step": 2264
    },
    {
      "epoch": 0.2389240506329114,
      "grad_norm": 0.6402329802513123,
      "learning_rate": 0.0015,
      "loss": 1.828,
      "step": 2265
    },
    {
      "epoch": 0.2390295358649789,
      "grad_norm": 0.48066070675849915,
      "learning_rate": 0.0015,
      "loss": 1.7841,
      "step": 2266
    },
    {
      "epoch": 0.2391350210970464,
      "grad_norm": 0.6777234673500061,
      "learning_rate": 0.0015,
      "loss": 1.8165,
      "step": 2267
    },
    {
      "epoch": 0.23924050632911392,
      "grad_norm": 0.5749910473823547,
      "learning_rate": 0.0015,
      "loss": 1.7853,
      "step": 2268
    },
    {
      "epoch": 0.23934599156118144,
      "grad_norm": 0.5601595044136047,
      "learning_rate": 0.0015,
      "loss": 1.7819,
      "step": 2269
    },
    {
      "epoch": 0.23945147679324894,
      "grad_norm": 0.5099558234214783,
      "learning_rate": 0.0015,
      "loss": 1.7881,
      "step": 2270
    },
    {
      "epoch": 0.23955696202531646,
      "grad_norm": 0.5596107840538025,
      "learning_rate": 0.0015,
      "loss": 1.7902,
      "step": 2271
    },
    {
      "epoch": 0.23966244725738398,
      "grad_norm": 0.5314821004867554,
      "learning_rate": 0.0015,
      "loss": 1.7548,
      "step": 2272
    },
    {
      "epoch": 0.23976793248945147,
      "grad_norm": 0.4766457974910736,
      "learning_rate": 0.0015,
      "loss": 1.7468,
      "step": 2273
    },
    {
      "epoch": 0.239873417721519,
      "grad_norm": 0.6159982681274414,
      "learning_rate": 0.0015,
      "loss": 1.7946,
      "step": 2274
    },
    {
      "epoch": 0.2399789029535865,
      "grad_norm": 0.66802579164505,
      "learning_rate": 0.0015,
      "loss": 1.7656,
      "step": 2275
    },
    {
      "epoch": 0.240084388185654,
      "grad_norm": 0.5005107522010803,
      "learning_rate": 0.0015,
      "loss": 1.7842,
      "step": 2276
    },
    {
      "epoch": 0.24018987341772152,
      "grad_norm": 0.5707549452781677,
      "learning_rate": 0.0015,
      "loss": 1.7734,
      "step": 2277
    },
    {
      "epoch": 0.24029535864978904,
      "grad_norm": 0.5253913402557373,
      "learning_rate": 0.0015,
      "loss": 1.7752,
      "step": 2278
    },
    {
      "epoch": 0.24040084388185654,
      "grad_norm": 0.53165602684021,
      "learning_rate": 0.0015,
      "loss": 1.7891,
      "step": 2279
    },
    {
      "epoch": 0.24050632911392406,
      "grad_norm": 0.7151433825492859,
      "learning_rate": 0.0015,
      "loss": 1.7872,
      "step": 2280
    },
    {
      "epoch": 0.24061181434599155,
      "grad_norm": 0.5916571021080017,
      "learning_rate": 0.0015,
      "loss": 1.7706,
      "step": 2281
    },
    {
      "epoch": 0.24071729957805907,
      "grad_norm": 0.6928505897521973,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 2282
    },
    {
      "epoch": 0.2408227848101266,
      "grad_norm": 1.0352296829223633,
      "learning_rate": 0.0015,
      "loss": 1.7642,
      "step": 2283
    },
    {
      "epoch": 0.24092827004219408,
      "grad_norm": 0.7083351612091064,
      "learning_rate": 0.0015,
      "loss": 1.7254,
      "step": 2284
    },
    {
      "epoch": 0.2410337552742616,
      "grad_norm": 0.529478907585144,
      "learning_rate": 0.0015,
      "loss": 1.7556,
      "step": 2285
    },
    {
      "epoch": 0.24113924050632912,
      "grad_norm": 0.8601495623588562,
      "learning_rate": 0.0015,
      "loss": 1.7973,
      "step": 2286
    },
    {
      "epoch": 0.24124472573839661,
      "grad_norm": 0.5362940430641174,
      "learning_rate": 0.0015,
      "loss": 1.7757,
      "step": 2287
    },
    {
      "epoch": 0.24135021097046414,
      "grad_norm": 0.6998580098152161,
      "learning_rate": 0.0015,
      "loss": 1.7864,
      "step": 2288
    },
    {
      "epoch": 0.24145569620253166,
      "grad_norm": 0.8627341985702515,
      "learning_rate": 0.0015,
      "loss": 1.7583,
      "step": 2289
    },
    {
      "epoch": 0.24156118143459915,
      "grad_norm": 0.49089959263801575,
      "learning_rate": 0.0015,
      "loss": 1.7727,
      "step": 2290
    },
    {
      "epoch": 0.24166666666666667,
      "grad_norm": 0.8630306720733643,
      "learning_rate": 0.0015,
      "loss": 1.7806,
      "step": 2291
    },
    {
      "epoch": 0.2417721518987342,
      "grad_norm": 0.5493969917297363,
      "learning_rate": 0.0015,
      "loss": 1.7151,
      "step": 2292
    },
    {
      "epoch": 0.24187763713080168,
      "grad_norm": 0.7692028284072876,
      "learning_rate": 0.0015,
      "loss": 1.7705,
      "step": 2293
    },
    {
      "epoch": 0.2419831223628692,
      "grad_norm": 0.8354348540306091,
      "learning_rate": 0.0015,
      "loss": 1.7715,
      "step": 2294
    },
    {
      "epoch": 0.24208860759493672,
      "grad_norm": 0.47700226306915283,
      "learning_rate": 0.0015,
      "loss": 1.7695,
      "step": 2295
    },
    {
      "epoch": 0.24219409282700421,
      "grad_norm": 0.6921583414077759,
      "learning_rate": 0.0015,
      "loss": 1.7743,
      "step": 2296
    },
    {
      "epoch": 0.24229957805907174,
      "grad_norm": 0.5307833552360535,
      "learning_rate": 0.0015,
      "loss": 1.8149,
      "step": 2297
    },
    {
      "epoch": 0.24240506329113923,
      "grad_norm": 0.6120598912239075,
      "learning_rate": 0.0015,
      "loss": 1.7587,
      "step": 2298
    },
    {
      "epoch": 0.24251054852320675,
      "grad_norm": 0.6797193884849548,
      "learning_rate": 0.0015,
      "loss": 1.7552,
      "step": 2299
    },
    {
      "epoch": 0.24261603375527427,
      "grad_norm": 0.5515388250350952,
      "learning_rate": 0.0015,
      "loss": 1.7317,
      "step": 2300
    },
    {
      "epoch": 0.24272151898734176,
      "grad_norm": 0.8552504181861877,
      "learning_rate": 0.0015,
      "loss": 1.7891,
      "step": 2301
    },
    {
      "epoch": 0.24282700421940928,
      "grad_norm": 0.5781500339508057,
      "learning_rate": 0.0015,
      "loss": 1.7358,
      "step": 2302
    },
    {
      "epoch": 0.2429324894514768,
      "grad_norm": 0.7933359146118164,
      "learning_rate": 0.0015,
      "loss": 1.7919,
      "step": 2303
    },
    {
      "epoch": 0.2430379746835443,
      "grad_norm": 0.8131921291351318,
      "learning_rate": 0.0015,
      "loss": 1.7787,
      "step": 2304
    },
    {
      "epoch": 0.24314345991561181,
      "grad_norm": 0.5366730690002441,
      "learning_rate": 0.0015,
      "loss": 1.7492,
      "step": 2305
    },
    {
      "epoch": 0.24324894514767934,
      "grad_norm": 0.9996156096458435,
      "learning_rate": 0.0015,
      "loss": 1.7902,
      "step": 2306
    },
    {
      "epoch": 0.24335443037974683,
      "grad_norm": 0.7723273634910583,
      "learning_rate": 0.0015,
      "loss": 1.7579,
      "step": 2307
    },
    {
      "epoch": 0.24345991561181435,
      "grad_norm": 0.4938789904117584,
      "learning_rate": 0.0015,
      "loss": 1.7272,
      "step": 2308
    },
    {
      "epoch": 0.24356540084388187,
      "grad_norm": 0.5969580411911011,
      "learning_rate": 0.0015,
      "loss": 1.7757,
      "step": 2309
    },
    {
      "epoch": 0.24367088607594936,
      "grad_norm": 0.5582490563392639,
      "learning_rate": 0.0015,
      "loss": 1.7795,
      "step": 2310
    },
    {
      "epoch": 0.24377637130801688,
      "grad_norm": 0.4658015966415405,
      "learning_rate": 0.0015,
      "loss": 1.7685,
      "step": 2311
    },
    {
      "epoch": 0.2438818565400844,
      "grad_norm": 0.5754499435424805,
      "learning_rate": 0.0015,
      "loss": 1.7894,
      "step": 2312
    },
    {
      "epoch": 0.2439873417721519,
      "grad_norm": 0.5328736901283264,
      "learning_rate": 0.0015,
      "loss": 1.8219,
      "step": 2313
    },
    {
      "epoch": 0.24409282700421941,
      "grad_norm": 0.48569199442863464,
      "learning_rate": 0.0015,
      "loss": 1.7684,
      "step": 2314
    },
    {
      "epoch": 0.2441983122362869,
      "grad_norm": 0.47076699137687683,
      "learning_rate": 0.0015,
      "loss": 1.7653,
      "step": 2315
    },
    {
      "epoch": 0.24430379746835443,
      "grad_norm": 0.5177969932556152,
      "learning_rate": 0.0015,
      "loss": 1.8044,
      "step": 2316
    },
    {
      "epoch": 0.24440928270042195,
      "grad_norm": 0.5015473961830139,
      "learning_rate": 0.0015,
      "loss": 1.7133,
      "step": 2317
    },
    {
      "epoch": 0.24451476793248944,
      "grad_norm": 0.6228369474411011,
      "learning_rate": 0.0015,
      "loss": 1.7599,
      "step": 2318
    },
    {
      "epoch": 0.24462025316455696,
      "grad_norm": 0.4836505651473999,
      "learning_rate": 0.0015,
      "loss": 1.7856,
      "step": 2319
    },
    {
      "epoch": 0.24472573839662448,
      "grad_norm": 0.5013377666473389,
      "learning_rate": 0.0015,
      "loss": 1.7263,
      "step": 2320
    },
    {
      "epoch": 0.24483122362869197,
      "grad_norm": 0.5119166970252991,
      "learning_rate": 0.0015,
      "loss": 1.7297,
      "step": 2321
    },
    {
      "epoch": 0.2449367088607595,
      "grad_norm": 0.5447758436203003,
      "learning_rate": 0.0015,
      "loss": 1.784,
      "step": 2322
    },
    {
      "epoch": 0.24504219409282701,
      "grad_norm": 0.4415464997291565,
      "learning_rate": 0.0015,
      "loss": 1.7608,
      "step": 2323
    },
    {
      "epoch": 0.2451476793248945,
      "grad_norm": 0.5140018463134766,
      "learning_rate": 0.0015,
      "loss": 1.7631,
      "step": 2324
    },
    {
      "epoch": 0.24525316455696203,
      "grad_norm": 0.513386607170105,
      "learning_rate": 0.0015,
      "loss": 1.7412,
      "step": 2325
    },
    {
      "epoch": 0.24535864978902955,
      "grad_norm": 0.48736950755119324,
      "learning_rate": 0.0015,
      "loss": 1.7737,
      "step": 2326
    },
    {
      "epoch": 0.24546413502109704,
      "grad_norm": 0.4886453151702881,
      "learning_rate": 0.0015,
      "loss": 1.7501,
      "step": 2327
    },
    {
      "epoch": 0.24556962025316456,
      "grad_norm": 0.5579635500907898,
      "learning_rate": 0.0015,
      "loss": 1.7509,
      "step": 2328
    },
    {
      "epoch": 0.24567510548523205,
      "grad_norm": 0.5681976675987244,
      "learning_rate": 0.0015,
      "loss": 1.752,
      "step": 2329
    },
    {
      "epoch": 0.24578059071729957,
      "grad_norm": 0.4890390932559967,
      "learning_rate": 0.0015,
      "loss": 1.7307,
      "step": 2330
    },
    {
      "epoch": 0.2458860759493671,
      "grad_norm": 0.5141686797142029,
      "learning_rate": 0.0015,
      "loss": 1.7177,
      "step": 2331
    },
    {
      "epoch": 0.2459915611814346,
      "grad_norm": 0.5592617392539978,
      "learning_rate": 0.0015,
      "loss": 1.7774,
      "step": 2332
    },
    {
      "epoch": 0.2460970464135021,
      "grad_norm": 0.45796963572502136,
      "learning_rate": 0.0015,
      "loss": 1.7619,
      "step": 2333
    },
    {
      "epoch": 0.24620253164556963,
      "grad_norm": 0.4724443256855011,
      "learning_rate": 0.0015,
      "loss": 1.78,
      "step": 2334
    },
    {
      "epoch": 0.24630801687763712,
      "grad_norm": 0.5037837028503418,
      "learning_rate": 0.0015,
      "loss": 1.7774,
      "step": 2335
    },
    {
      "epoch": 0.24641350210970464,
      "grad_norm": 0.5889279246330261,
      "learning_rate": 0.0015,
      "loss": 1.769,
      "step": 2336
    },
    {
      "epoch": 0.24651898734177216,
      "grad_norm": 0.515529990196228,
      "learning_rate": 0.0015,
      "loss": 1.7033,
      "step": 2337
    },
    {
      "epoch": 0.24662447257383965,
      "grad_norm": 0.6168153882026672,
      "learning_rate": 0.0015,
      "loss": 1.743,
      "step": 2338
    },
    {
      "epoch": 0.24672995780590717,
      "grad_norm": 0.5972260236740112,
      "learning_rate": 0.0015,
      "loss": 1.7623,
      "step": 2339
    },
    {
      "epoch": 0.2468354430379747,
      "grad_norm": 0.5272727608680725,
      "learning_rate": 0.0015,
      "loss": 1.7522,
      "step": 2340
    },
    {
      "epoch": 0.2469409282700422,
      "grad_norm": 0.614957869052887,
      "learning_rate": 0.0015,
      "loss": 1.7773,
      "step": 2341
    },
    {
      "epoch": 0.2470464135021097,
      "grad_norm": 0.6469771862030029,
      "learning_rate": 0.0015,
      "loss": 1.7594,
      "step": 2342
    },
    {
      "epoch": 0.24715189873417723,
      "grad_norm": 0.5531512498855591,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 2343
    },
    {
      "epoch": 0.24725738396624472,
      "grad_norm": 0.6211775541305542,
      "learning_rate": 0.0015,
      "loss": 1.7691,
      "step": 2344
    },
    {
      "epoch": 0.24736286919831224,
      "grad_norm": 0.5498647689819336,
      "learning_rate": 0.0015,
      "loss": 1.7529,
      "step": 2345
    },
    {
      "epoch": 0.24746835443037973,
      "grad_norm": 0.7096125483512878,
      "learning_rate": 0.0015,
      "loss": 1.767,
      "step": 2346
    },
    {
      "epoch": 0.24757383966244725,
      "grad_norm": 0.7859359383583069,
      "learning_rate": 0.0015,
      "loss": 1.7357,
      "step": 2347
    },
    {
      "epoch": 0.24767932489451477,
      "grad_norm": 0.6023251414299011,
      "learning_rate": 0.0015,
      "loss": 1.7318,
      "step": 2348
    },
    {
      "epoch": 0.24778481012658227,
      "grad_norm": 0.6840221881866455,
      "learning_rate": 0.0015,
      "loss": 1.7355,
      "step": 2349
    },
    {
      "epoch": 0.2478902953586498,
      "grad_norm": 0.7530645132064819,
      "learning_rate": 0.0015,
      "loss": 1.7842,
      "step": 2350
    },
    {
      "epoch": 0.2479957805907173,
      "grad_norm": 0.5974234342575073,
      "learning_rate": 0.0015,
      "loss": 1.7545,
      "step": 2351
    },
    {
      "epoch": 0.2481012658227848,
      "grad_norm": 0.5100545287132263,
      "learning_rate": 0.0015,
      "loss": 1.7849,
      "step": 2352
    },
    {
      "epoch": 0.24820675105485232,
      "grad_norm": 0.6527138948440552,
      "learning_rate": 0.0015,
      "loss": 1.7419,
      "step": 2353
    },
    {
      "epoch": 0.24831223628691984,
      "grad_norm": 0.5429184436798096,
      "learning_rate": 0.0015,
      "loss": 1.773,
      "step": 2354
    },
    {
      "epoch": 0.24841772151898733,
      "grad_norm": 0.7330167889595032,
      "learning_rate": 0.0015,
      "loss": 1.7893,
      "step": 2355
    },
    {
      "epoch": 0.24852320675105485,
      "grad_norm": 0.6700617074966431,
      "learning_rate": 0.0015,
      "loss": 1.7491,
      "step": 2356
    },
    {
      "epoch": 0.24862869198312237,
      "grad_norm": 0.521098256111145,
      "learning_rate": 0.0015,
      "loss": 1.7625,
      "step": 2357
    },
    {
      "epoch": 0.24873417721518987,
      "grad_norm": 0.6005222797393799,
      "learning_rate": 0.0015,
      "loss": 1.7922,
      "step": 2358
    },
    {
      "epoch": 0.2488396624472574,
      "grad_norm": 0.5163782835006714,
      "learning_rate": 0.0015,
      "loss": 1.7432,
      "step": 2359
    },
    {
      "epoch": 0.2489451476793249,
      "grad_norm": 0.5147345662117004,
      "learning_rate": 0.0015,
      "loss": 1.7407,
      "step": 2360
    },
    {
      "epoch": 0.2490506329113924,
      "grad_norm": 0.4879380762577057,
      "learning_rate": 0.0015,
      "loss": 1.7367,
      "step": 2361
    },
    {
      "epoch": 0.24915611814345992,
      "grad_norm": 0.46628686785697937,
      "learning_rate": 0.0015,
      "loss": 1.7659,
      "step": 2362
    },
    {
      "epoch": 0.2492616033755274,
      "grad_norm": 0.49309617280960083,
      "learning_rate": 0.0015,
      "loss": 1.7663,
      "step": 2363
    },
    {
      "epoch": 0.24936708860759493,
      "grad_norm": 0.4331234097480774,
      "learning_rate": 0.0015,
      "loss": 1.747,
      "step": 2364
    },
    {
      "epoch": 0.24947257383966245,
      "grad_norm": 0.49875301122665405,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 2365
    },
    {
      "epoch": 0.24957805907172995,
      "grad_norm": 0.46396973729133606,
      "learning_rate": 0.0015,
      "loss": 1.7514,
      "step": 2366
    },
    {
      "epoch": 0.24968354430379747,
      "grad_norm": 0.44407719373703003,
      "learning_rate": 0.0015,
      "loss": 1.7757,
      "step": 2367
    },
    {
      "epoch": 0.249789029535865,
      "grad_norm": 0.48293107748031616,
      "learning_rate": 0.0015,
      "loss": 1.7652,
      "step": 2368
    },
    {
      "epoch": 0.24989451476793248,
      "grad_norm": 0.5071579217910767,
      "learning_rate": 0.0015,
      "loss": 1.7642,
      "step": 2369
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4867854714393616,
      "learning_rate": 0.0015,
      "loss": 1.7662,
      "step": 2370
    },
    {
      "epoch": 0.2501054852320675,
      "grad_norm": 0.46139228343963623,
      "learning_rate": 0.0015,
      "loss": 1.756,
      "step": 2371
    },
    {
      "epoch": 0.25021097046413504,
      "grad_norm": 0.5459284782409668,
      "learning_rate": 0.0015,
      "loss": 1.7577,
      "step": 2372
    },
    {
      "epoch": 0.25031645569620253,
      "grad_norm": 0.642245888710022,
      "learning_rate": 0.0015,
      "loss": 1.7512,
      "step": 2373
    },
    {
      "epoch": 0.25042194092827,
      "grad_norm": 0.6148761510848999,
      "learning_rate": 0.0015,
      "loss": 1.7565,
      "step": 2374
    },
    {
      "epoch": 0.2505274261603376,
      "grad_norm": 0.49759599566459656,
      "learning_rate": 0.0015,
      "loss": 1.7444,
      "step": 2375
    },
    {
      "epoch": 0.25063291139240507,
      "grad_norm": 0.668725848197937,
      "learning_rate": 0.0015,
      "loss": 1.8011,
      "step": 2376
    },
    {
      "epoch": 0.25073839662447256,
      "grad_norm": 0.4881817698478699,
      "learning_rate": 0.0015,
      "loss": 1.719,
      "step": 2377
    },
    {
      "epoch": 0.2508438818565401,
      "grad_norm": 0.5557053685188293,
      "learning_rate": 0.0015,
      "loss": 1.7453,
      "step": 2378
    },
    {
      "epoch": 0.2509493670886076,
      "grad_norm": 0.5239472985267639,
      "learning_rate": 0.0015,
      "loss": 1.7905,
      "step": 2379
    },
    {
      "epoch": 0.2510548523206751,
      "grad_norm": 0.49819645285606384,
      "learning_rate": 0.0015,
      "loss": 1.7447,
      "step": 2380
    },
    {
      "epoch": 0.25116033755274264,
      "grad_norm": 0.6239401698112488,
      "learning_rate": 0.0015,
      "loss": 1.769,
      "step": 2381
    },
    {
      "epoch": 0.25126582278481013,
      "grad_norm": 0.7374022603034973,
      "learning_rate": 0.0015,
      "loss": 1.7528,
      "step": 2382
    },
    {
      "epoch": 0.2513713080168776,
      "grad_norm": 0.578967809677124,
      "learning_rate": 0.0015,
      "loss": 1.7624,
      "step": 2383
    },
    {
      "epoch": 0.2514767932489452,
      "grad_norm": 0.5678188800811768,
      "learning_rate": 0.0015,
      "loss": 1.7281,
      "step": 2384
    },
    {
      "epoch": 0.25158227848101267,
      "grad_norm": 0.7271468043327332,
      "learning_rate": 0.0015,
      "loss": 1.7775,
      "step": 2385
    },
    {
      "epoch": 0.25168776371308016,
      "grad_norm": 0.5477023720741272,
      "learning_rate": 0.0015,
      "loss": 1.7672,
      "step": 2386
    },
    {
      "epoch": 0.25179324894514765,
      "grad_norm": 0.5434300303459167,
      "learning_rate": 0.0015,
      "loss": 1.8252,
      "step": 2387
    },
    {
      "epoch": 0.2518987341772152,
      "grad_norm": 0.6164018511772156,
      "learning_rate": 0.0015,
      "loss": 1.7287,
      "step": 2388
    },
    {
      "epoch": 0.2520042194092827,
      "grad_norm": 0.45354118943214417,
      "learning_rate": 0.0015,
      "loss": 1.7271,
      "step": 2389
    },
    {
      "epoch": 0.2521097046413502,
      "grad_norm": 0.554826557636261,
      "learning_rate": 0.0015,
      "loss": 1.7546,
      "step": 2390
    },
    {
      "epoch": 0.25221518987341773,
      "grad_norm": 0.5292742252349854,
      "learning_rate": 0.0015,
      "loss": 1.7347,
      "step": 2391
    },
    {
      "epoch": 0.2523206751054852,
      "grad_norm": 0.6176003813743591,
      "learning_rate": 0.0015,
      "loss": 1.7249,
      "step": 2392
    },
    {
      "epoch": 0.2524261603375527,
      "grad_norm": 0.8826800584793091,
      "learning_rate": 0.0015,
      "loss": 1.738,
      "step": 2393
    },
    {
      "epoch": 0.25253164556962027,
      "grad_norm": 0.7090798020362854,
      "learning_rate": 0.0015,
      "loss": 1.7608,
      "step": 2394
    },
    {
      "epoch": 0.25263713080168776,
      "grad_norm": 0.5037247538566589,
      "learning_rate": 0.0015,
      "loss": 1.7066,
      "step": 2395
    },
    {
      "epoch": 0.25274261603375525,
      "grad_norm": 0.7920147180557251,
      "learning_rate": 0.0015,
      "loss": 1.7855,
      "step": 2396
    },
    {
      "epoch": 0.2528481012658228,
      "grad_norm": 0.6410861015319824,
      "learning_rate": 0.0015,
      "loss": 1.8146,
      "step": 2397
    },
    {
      "epoch": 0.2529535864978903,
      "grad_norm": 0.5408570766448975,
      "learning_rate": 0.0015,
      "loss": 1.7941,
      "step": 2398
    },
    {
      "epoch": 0.2530590717299578,
      "grad_norm": 0.8117817640304565,
      "learning_rate": 0.0015,
      "loss": 1.754,
      "step": 2399
    },
    {
      "epoch": 0.25316455696202533,
      "grad_norm": 0.6268106698989868,
      "learning_rate": 0.0015,
      "loss": 1.7755,
      "step": 2400
    },
    {
      "epoch": 0.2532700421940928,
      "grad_norm": 0.553735077381134,
      "learning_rate": 0.0015,
      "loss": 1.7573,
      "step": 2401
    },
    {
      "epoch": 0.2533755274261603,
      "grad_norm": 0.8653258085250854,
      "learning_rate": 0.0015,
      "loss": 1.7728,
      "step": 2402
    },
    {
      "epoch": 0.25348101265822787,
      "grad_norm": 0.6993894577026367,
      "learning_rate": 0.0015,
      "loss": 1.784,
      "step": 2403
    },
    {
      "epoch": 0.25358649789029536,
      "grad_norm": 0.5147871971130371,
      "learning_rate": 0.0015,
      "loss": 1.7225,
      "step": 2404
    },
    {
      "epoch": 0.25369198312236285,
      "grad_norm": 0.6098518967628479,
      "learning_rate": 0.0015,
      "loss": 1.7917,
      "step": 2405
    },
    {
      "epoch": 0.2537974683544304,
      "grad_norm": 0.5055338144302368,
      "learning_rate": 0.0015,
      "loss": 1.7499,
      "step": 2406
    },
    {
      "epoch": 0.2539029535864979,
      "grad_norm": 0.5311747193336487,
      "learning_rate": 0.0015,
      "loss": 1.783,
      "step": 2407
    },
    {
      "epoch": 0.2540084388185654,
      "grad_norm": 0.6478826403617859,
      "learning_rate": 0.0015,
      "loss": 1.7385,
      "step": 2408
    },
    {
      "epoch": 0.25411392405063293,
      "grad_norm": 0.5538125038146973,
      "learning_rate": 0.0015,
      "loss": 1.7245,
      "step": 2409
    },
    {
      "epoch": 0.2542194092827004,
      "grad_norm": 0.4873404800891876,
      "learning_rate": 0.0015,
      "loss": 1.7841,
      "step": 2410
    },
    {
      "epoch": 0.2543248945147679,
      "grad_norm": 0.4269186854362488,
      "learning_rate": 0.0015,
      "loss": 1.7877,
      "step": 2411
    },
    {
      "epoch": 0.25443037974683547,
      "grad_norm": 0.49043551087379456,
      "learning_rate": 0.0015,
      "loss": 1.7437,
      "step": 2412
    },
    {
      "epoch": 0.25453586497890296,
      "grad_norm": 0.4809432327747345,
      "learning_rate": 0.0015,
      "loss": 1.7624,
      "step": 2413
    },
    {
      "epoch": 0.25464135021097045,
      "grad_norm": 0.4802286922931671,
      "learning_rate": 0.0015,
      "loss": 1.7227,
      "step": 2414
    },
    {
      "epoch": 0.254746835443038,
      "grad_norm": 0.5694488883018494,
      "learning_rate": 0.0015,
      "loss": 1.7404,
      "step": 2415
    },
    {
      "epoch": 0.2548523206751055,
      "grad_norm": 0.7195572853088379,
      "learning_rate": 0.0015,
      "loss": 1.7608,
      "step": 2416
    },
    {
      "epoch": 0.254957805907173,
      "grad_norm": 0.5387224555015564,
      "learning_rate": 0.0015,
      "loss": 1.7206,
      "step": 2417
    },
    {
      "epoch": 0.25506329113924053,
      "grad_norm": 0.5167665481567383,
      "learning_rate": 0.0015,
      "loss": 1.7602,
      "step": 2418
    },
    {
      "epoch": 0.255168776371308,
      "grad_norm": 0.6237902641296387,
      "learning_rate": 0.0015,
      "loss": 1.7328,
      "step": 2419
    },
    {
      "epoch": 0.2552742616033755,
      "grad_norm": 0.5012637972831726,
      "learning_rate": 0.0015,
      "loss": 1.7853,
      "step": 2420
    },
    {
      "epoch": 0.255379746835443,
      "grad_norm": 0.6789900660514832,
      "learning_rate": 0.0015,
      "loss": 1.7353,
      "step": 2421
    },
    {
      "epoch": 0.25548523206751056,
      "grad_norm": 0.5136539936065674,
      "learning_rate": 0.0015,
      "loss": 1.828,
      "step": 2422
    },
    {
      "epoch": 0.25559071729957805,
      "grad_norm": 0.607568085193634,
      "learning_rate": 0.0015,
      "loss": 1.7954,
      "step": 2423
    },
    {
      "epoch": 0.25569620253164554,
      "grad_norm": 0.5693743824958801,
      "learning_rate": 0.0015,
      "loss": 1.7383,
      "step": 2424
    },
    {
      "epoch": 0.2558016877637131,
      "grad_norm": 0.5048482418060303,
      "learning_rate": 0.0015,
      "loss": 1.7209,
      "step": 2425
    },
    {
      "epoch": 0.2559071729957806,
      "grad_norm": 0.6488950252532959,
      "learning_rate": 0.0015,
      "loss": 1.7569,
      "step": 2426
    },
    {
      "epoch": 0.2560126582278481,
      "grad_norm": 0.5138487815856934,
      "learning_rate": 0.0015,
      "loss": 1.7779,
      "step": 2427
    },
    {
      "epoch": 0.2561181434599156,
      "grad_norm": 0.5535092949867249,
      "learning_rate": 0.0015,
      "loss": 1.7229,
      "step": 2428
    },
    {
      "epoch": 0.2562236286919831,
      "grad_norm": 0.5026729702949524,
      "learning_rate": 0.0015,
      "loss": 1.7278,
      "step": 2429
    },
    {
      "epoch": 0.2563291139240506,
      "grad_norm": 0.5470440983772278,
      "learning_rate": 0.0015,
      "loss": 1.7285,
      "step": 2430
    },
    {
      "epoch": 0.25643459915611816,
      "grad_norm": 0.5697625279426575,
      "learning_rate": 0.0015,
      "loss": 1.7707,
      "step": 2431
    },
    {
      "epoch": 0.25654008438818565,
      "grad_norm": 0.45551273226737976,
      "learning_rate": 0.0015,
      "loss": 1.784,
      "step": 2432
    },
    {
      "epoch": 0.25664556962025314,
      "grad_norm": 0.5044599771499634,
      "learning_rate": 0.0015,
      "loss": 1.7985,
      "step": 2433
    },
    {
      "epoch": 0.2567510548523207,
      "grad_norm": 0.507034420967102,
      "learning_rate": 0.0015,
      "loss": 1.7076,
      "step": 2434
    },
    {
      "epoch": 0.2568565400843882,
      "grad_norm": 0.6209434270858765,
      "learning_rate": 0.0015,
      "loss": 1.7289,
      "step": 2435
    },
    {
      "epoch": 0.2569620253164557,
      "grad_norm": 0.6941367387771606,
      "learning_rate": 0.0015,
      "loss": 1.7307,
      "step": 2436
    },
    {
      "epoch": 0.2570675105485232,
      "grad_norm": 0.5281949043273926,
      "learning_rate": 0.0015,
      "loss": 1.7658,
      "step": 2437
    },
    {
      "epoch": 0.2571729957805907,
      "grad_norm": 0.5222294330596924,
      "learning_rate": 0.0015,
      "loss": 1.7592,
      "step": 2438
    },
    {
      "epoch": 0.2572784810126582,
      "grad_norm": 0.5669064521789551,
      "learning_rate": 0.0015,
      "loss": 1.7521,
      "step": 2439
    },
    {
      "epoch": 0.25738396624472576,
      "grad_norm": 0.4953993558883667,
      "learning_rate": 0.0015,
      "loss": 1.7841,
      "step": 2440
    },
    {
      "epoch": 0.25748945147679325,
      "grad_norm": 0.5113043785095215,
      "learning_rate": 0.0015,
      "loss": 1.7606,
      "step": 2441
    },
    {
      "epoch": 0.25759493670886074,
      "grad_norm": 0.4993591606616974,
      "learning_rate": 0.0015,
      "loss": 1.7429,
      "step": 2442
    },
    {
      "epoch": 0.2577004219409283,
      "grad_norm": 0.5303695201873779,
      "learning_rate": 0.0015,
      "loss": 1.7589,
      "step": 2443
    },
    {
      "epoch": 0.2578059071729958,
      "grad_norm": 0.5011569261550903,
      "learning_rate": 0.0015,
      "loss": 1.7346,
      "step": 2444
    },
    {
      "epoch": 0.2579113924050633,
      "grad_norm": 0.49535638093948364,
      "learning_rate": 0.0015,
      "loss": 1.7893,
      "step": 2445
    },
    {
      "epoch": 0.2580168776371308,
      "grad_norm": 0.5035144686698914,
      "learning_rate": 0.0015,
      "loss": 1.7236,
      "step": 2446
    },
    {
      "epoch": 0.2581223628691983,
      "grad_norm": 0.47144025564193726,
      "learning_rate": 0.0015,
      "loss": 1.7437,
      "step": 2447
    },
    {
      "epoch": 0.2582278481012658,
      "grad_norm": 0.49981555342674255,
      "learning_rate": 0.0015,
      "loss": 1.7408,
      "step": 2448
    },
    {
      "epoch": 0.25833333333333336,
      "grad_norm": 0.5020222067832947,
      "learning_rate": 0.0015,
      "loss": 1.732,
      "step": 2449
    },
    {
      "epoch": 0.25843881856540085,
      "grad_norm": 0.5577045679092407,
      "learning_rate": 0.0015,
      "loss": 1.7257,
      "step": 2450
    },
    {
      "epoch": 0.25854430379746834,
      "grad_norm": 0.4685960114002228,
      "learning_rate": 0.0015,
      "loss": 1.7441,
      "step": 2451
    },
    {
      "epoch": 0.2586497890295359,
      "grad_norm": 0.644061267375946,
      "learning_rate": 0.0015,
      "loss": 1.7921,
      "step": 2452
    },
    {
      "epoch": 0.2587552742616034,
      "grad_norm": 0.5561867952346802,
      "learning_rate": 0.0015,
      "loss": 1.7458,
      "step": 2453
    },
    {
      "epoch": 0.2588607594936709,
      "grad_norm": 0.5068477988243103,
      "learning_rate": 0.0015,
      "loss": 1.7944,
      "step": 2454
    },
    {
      "epoch": 0.25896624472573837,
      "grad_norm": 0.531140148639679,
      "learning_rate": 0.0015,
      "loss": 1.7259,
      "step": 2455
    },
    {
      "epoch": 0.2590717299578059,
      "grad_norm": 0.5053848624229431,
      "learning_rate": 0.0015,
      "loss": 1.7578,
      "step": 2456
    },
    {
      "epoch": 0.2591772151898734,
      "grad_norm": 0.5224964618682861,
      "learning_rate": 0.0015,
      "loss": 1.7489,
      "step": 2457
    },
    {
      "epoch": 0.2592827004219409,
      "grad_norm": 0.5178977251052856,
      "learning_rate": 0.0015,
      "loss": 1.7682,
      "step": 2458
    },
    {
      "epoch": 0.25938818565400845,
      "grad_norm": 0.5664845705032349,
      "learning_rate": 0.0015,
      "loss": 1.7788,
      "step": 2459
    },
    {
      "epoch": 0.25949367088607594,
      "grad_norm": 0.4939206838607788,
      "learning_rate": 0.0015,
      "loss": 1.7568,
      "step": 2460
    },
    {
      "epoch": 0.25959915611814344,
      "grad_norm": 0.46851345896720886,
      "learning_rate": 0.0015,
      "loss": 1.7087,
      "step": 2461
    },
    {
      "epoch": 0.259704641350211,
      "grad_norm": 0.4888216257095337,
      "learning_rate": 0.0015,
      "loss": 1.7576,
      "step": 2462
    },
    {
      "epoch": 0.2598101265822785,
      "grad_norm": 0.5279091000556946,
      "learning_rate": 0.0015,
      "loss": 1.7623,
      "step": 2463
    },
    {
      "epoch": 0.25991561181434597,
      "grad_norm": 0.4827437996864319,
      "learning_rate": 0.0015,
      "loss": 1.7188,
      "step": 2464
    },
    {
      "epoch": 0.2600210970464135,
      "grad_norm": 0.4542250633239746,
      "learning_rate": 0.0015,
      "loss": 1.7798,
      "step": 2465
    },
    {
      "epoch": 0.260126582278481,
      "grad_norm": 0.5125083923339844,
      "learning_rate": 0.0015,
      "loss": 1.732,
      "step": 2466
    },
    {
      "epoch": 0.2602320675105485,
      "grad_norm": 0.5349894762039185,
      "learning_rate": 0.0015,
      "loss": 1.7494,
      "step": 2467
    },
    {
      "epoch": 0.26033755274261605,
      "grad_norm": 0.5146206021308899,
      "learning_rate": 0.0015,
      "loss": 1.7601,
      "step": 2468
    },
    {
      "epoch": 0.26044303797468354,
      "grad_norm": 0.5680448412895203,
      "learning_rate": 0.0015,
      "loss": 1.769,
      "step": 2469
    },
    {
      "epoch": 0.26054852320675104,
      "grad_norm": 0.7838568091392517,
      "learning_rate": 0.0015,
      "loss": 1.7703,
      "step": 2470
    },
    {
      "epoch": 0.2606540084388186,
      "grad_norm": 0.5395078063011169,
      "learning_rate": 0.0015,
      "loss": 1.793,
      "step": 2471
    },
    {
      "epoch": 0.2607594936708861,
      "grad_norm": 0.5863775014877319,
      "learning_rate": 0.0015,
      "loss": 1.7372,
      "step": 2472
    },
    {
      "epoch": 0.26086497890295357,
      "grad_norm": 0.6507776379585266,
      "learning_rate": 0.0015,
      "loss": 1.7503,
      "step": 2473
    },
    {
      "epoch": 0.2609704641350211,
      "grad_norm": 0.6046692132949829,
      "learning_rate": 0.0015,
      "loss": 1.7513,
      "step": 2474
    },
    {
      "epoch": 0.2610759493670886,
      "grad_norm": 0.49888819456100464,
      "learning_rate": 0.0015,
      "loss": 1.7628,
      "step": 2475
    },
    {
      "epoch": 0.2611814345991561,
      "grad_norm": 0.5066530704498291,
      "learning_rate": 0.0015,
      "loss": 1.7337,
      "step": 2476
    },
    {
      "epoch": 0.26128691983122365,
      "grad_norm": 0.5808389782905579,
      "learning_rate": 0.0015,
      "loss": 1.7711,
      "step": 2477
    },
    {
      "epoch": 0.26139240506329114,
      "grad_norm": 0.5939879417419434,
      "learning_rate": 0.0015,
      "loss": 1.7491,
      "step": 2478
    },
    {
      "epoch": 0.26149789029535864,
      "grad_norm": 0.5914461612701416,
      "learning_rate": 0.0015,
      "loss": 1.7535,
      "step": 2479
    },
    {
      "epoch": 0.2616033755274262,
      "grad_norm": 0.5965650677680969,
      "learning_rate": 0.0015,
      "loss": 1.7548,
      "step": 2480
    },
    {
      "epoch": 0.2617088607594937,
      "grad_norm": 0.5369855165481567,
      "learning_rate": 0.0015,
      "loss": 1.7341,
      "step": 2481
    },
    {
      "epoch": 0.26181434599156117,
      "grad_norm": 0.5695091485977173,
      "learning_rate": 0.0015,
      "loss": 1.7267,
      "step": 2482
    },
    {
      "epoch": 0.2619198312236287,
      "grad_norm": 0.6204779744148254,
      "learning_rate": 0.0015,
      "loss": 1.7511,
      "step": 2483
    },
    {
      "epoch": 0.2620253164556962,
      "grad_norm": 0.5356783866882324,
      "learning_rate": 0.0015,
      "loss": 1.715,
      "step": 2484
    },
    {
      "epoch": 0.2621308016877637,
      "grad_norm": 0.57508784532547,
      "learning_rate": 0.0015,
      "loss": 1.7589,
      "step": 2485
    },
    {
      "epoch": 0.2622362869198312,
      "grad_norm": 0.5130199193954468,
      "learning_rate": 0.0015,
      "loss": 1.7382,
      "step": 2486
    },
    {
      "epoch": 0.26234177215189874,
      "grad_norm": 0.4947783946990967,
      "learning_rate": 0.0015,
      "loss": 1.742,
      "step": 2487
    },
    {
      "epoch": 0.26244725738396624,
      "grad_norm": 0.5992721319198608,
      "learning_rate": 0.0015,
      "loss": 1.7786,
      "step": 2488
    },
    {
      "epoch": 0.26255274261603373,
      "grad_norm": 0.5535210967063904,
      "learning_rate": 0.0015,
      "loss": 1.7184,
      "step": 2489
    },
    {
      "epoch": 0.2626582278481013,
      "grad_norm": 0.5110808610916138,
      "learning_rate": 0.0015,
      "loss": 1.7651,
      "step": 2490
    },
    {
      "epoch": 0.26276371308016877,
      "grad_norm": 0.5434951782226562,
      "learning_rate": 0.0015,
      "loss": 1.7297,
      "step": 2491
    },
    {
      "epoch": 0.26286919831223626,
      "grad_norm": 0.5436090230941772,
      "learning_rate": 0.0015,
      "loss": 1.7435,
      "step": 2492
    },
    {
      "epoch": 0.2629746835443038,
      "grad_norm": 0.5366165041923523,
      "learning_rate": 0.0015,
      "loss": 1.7876,
      "step": 2493
    },
    {
      "epoch": 0.2630801687763713,
      "grad_norm": 0.5321416854858398,
      "learning_rate": 0.0015,
      "loss": 1.738,
      "step": 2494
    },
    {
      "epoch": 0.2631856540084388,
      "grad_norm": 0.4772559404373169,
      "learning_rate": 0.0015,
      "loss": 1.7456,
      "step": 2495
    },
    {
      "epoch": 0.26329113924050634,
      "grad_norm": 0.5794147253036499,
      "learning_rate": 0.0015,
      "loss": 1.7415,
      "step": 2496
    },
    {
      "epoch": 0.26339662447257384,
      "grad_norm": 0.5084999203681946,
      "learning_rate": 0.0015,
      "loss": 1.7911,
      "step": 2497
    },
    {
      "epoch": 0.26350210970464133,
      "grad_norm": 0.5278586149215698,
      "learning_rate": 0.0015,
      "loss": 1.7408,
      "step": 2498
    },
    {
      "epoch": 0.2636075949367089,
      "grad_norm": 0.49334725737571716,
      "learning_rate": 0.0015,
      "loss": 1.7483,
      "step": 2499
    },
    {
      "epoch": 0.26371308016877637,
      "grad_norm": 0.5233896970748901,
      "learning_rate": 0.0015,
      "loss": 1.7296,
      "step": 2500
    },
    {
      "epoch": 0.26381856540084386,
      "grad_norm": 0.5626576542854309,
      "learning_rate": 0.0015,
      "loss": 1.7361,
      "step": 2501
    },
    {
      "epoch": 0.2639240506329114,
      "grad_norm": 0.48603805899620056,
      "learning_rate": 0.0015,
      "loss": 1.7742,
      "step": 2502
    },
    {
      "epoch": 0.2640295358649789,
      "grad_norm": 0.5147922039031982,
      "learning_rate": 0.0015,
      "loss": 1.7524,
      "step": 2503
    },
    {
      "epoch": 0.2641350210970464,
      "grad_norm": 0.4762532114982605,
      "learning_rate": 0.0015,
      "loss": 1.7223,
      "step": 2504
    },
    {
      "epoch": 0.26424050632911394,
      "grad_norm": 0.5490362048149109,
      "learning_rate": 0.0015,
      "loss": 1.7778,
      "step": 2505
    },
    {
      "epoch": 0.26434599156118144,
      "grad_norm": 0.534259557723999,
      "learning_rate": 0.0015,
      "loss": 1.7273,
      "step": 2506
    },
    {
      "epoch": 0.26445147679324893,
      "grad_norm": 0.5113989114761353,
      "learning_rate": 0.0015,
      "loss": 1.7562,
      "step": 2507
    },
    {
      "epoch": 0.2645569620253165,
      "grad_norm": 0.492646723985672,
      "learning_rate": 0.0015,
      "loss": 1.7309,
      "step": 2508
    },
    {
      "epoch": 0.26466244725738397,
      "grad_norm": 0.507503092288971,
      "learning_rate": 0.0015,
      "loss": 1.7302,
      "step": 2509
    },
    {
      "epoch": 0.26476793248945146,
      "grad_norm": 0.5104250311851501,
      "learning_rate": 0.0015,
      "loss": 1.7672,
      "step": 2510
    },
    {
      "epoch": 0.264873417721519,
      "grad_norm": 0.5170099139213562,
      "learning_rate": 0.0015,
      "loss": 1.7847,
      "step": 2511
    },
    {
      "epoch": 0.2649789029535865,
      "grad_norm": 0.4758244454860687,
      "learning_rate": 0.0015,
      "loss": 1.7553,
      "step": 2512
    },
    {
      "epoch": 0.265084388185654,
      "grad_norm": 0.503151535987854,
      "learning_rate": 0.0015,
      "loss": 1.7228,
      "step": 2513
    },
    {
      "epoch": 0.26518987341772154,
      "grad_norm": 0.47079765796661377,
      "learning_rate": 0.0015,
      "loss": 1.7403,
      "step": 2514
    },
    {
      "epoch": 0.26529535864978904,
      "grad_norm": 0.4670954644680023,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 2515
    },
    {
      "epoch": 0.26540084388185653,
      "grad_norm": 0.49063602089881897,
      "learning_rate": 0.0015,
      "loss": 1.7531,
      "step": 2516
    },
    {
      "epoch": 0.2655063291139241,
      "grad_norm": 0.5368266105651855,
      "learning_rate": 0.0015,
      "loss": 1.7577,
      "step": 2517
    },
    {
      "epoch": 0.26561181434599157,
      "grad_norm": 0.5149813890457153,
      "learning_rate": 0.0015,
      "loss": 1.763,
      "step": 2518
    },
    {
      "epoch": 0.26571729957805906,
      "grad_norm": 0.49429503083229065,
      "learning_rate": 0.0015,
      "loss": 1.7477,
      "step": 2519
    },
    {
      "epoch": 0.26582278481012656,
      "grad_norm": 0.5160220861434937,
      "learning_rate": 0.0015,
      "loss": 1.7603,
      "step": 2520
    },
    {
      "epoch": 0.2659282700421941,
      "grad_norm": 0.6020708680152893,
      "learning_rate": 0.0015,
      "loss": 1.7451,
      "step": 2521
    },
    {
      "epoch": 0.2660337552742616,
      "grad_norm": 0.6109582781791687,
      "learning_rate": 0.0015,
      "loss": 1.7396,
      "step": 2522
    },
    {
      "epoch": 0.2661392405063291,
      "grad_norm": 0.5616773366928101,
      "learning_rate": 0.0015,
      "loss": 1.7657,
      "step": 2523
    },
    {
      "epoch": 0.26624472573839664,
      "grad_norm": 0.4889385998249054,
      "learning_rate": 0.0015,
      "loss": 1.6985,
      "step": 2524
    },
    {
      "epoch": 0.26635021097046413,
      "grad_norm": 0.5477367639541626,
      "learning_rate": 0.0015,
      "loss": 1.7117,
      "step": 2525
    },
    {
      "epoch": 0.2664556962025316,
      "grad_norm": 0.5434495806694031,
      "learning_rate": 0.0015,
      "loss": 1.7252,
      "step": 2526
    },
    {
      "epoch": 0.26656118143459917,
      "grad_norm": 0.5032594203948975,
      "learning_rate": 0.0015,
      "loss": 1.7423,
      "step": 2527
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 0.6376218795776367,
      "learning_rate": 0.0015,
      "loss": 1.745,
      "step": 2528
    },
    {
      "epoch": 0.26677215189873416,
      "grad_norm": 0.740061342716217,
      "learning_rate": 0.0015,
      "loss": 1.8298,
      "step": 2529
    },
    {
      "epoch": 0.2668776371308017,
      "grad_norm": 0.7886686325073242,
      "learning_rate": 0.0015,
      "loss": 1.7371,
      "step": 2530
    },
    {
      "epoch": 0.2669831223628692,
      "grad_norm": 0.520111620426178,
      "learning_rate": 0.0015,
      "loss": 1.7384,
      "step": 2531
    },
    {
      "epoch": 0.2670886075949367,
      "grad_norm": 0.6698089241981506,
      "learning_rate": 0.0015,
      "loss": 1.7459,
      "step": 2532
    },
    {
      "epoch": 0.26719409282700424,
      "grad_norm": 0.8717212080955505,
      "learning_rate": 0.0015,
      "loss": 1.6923,
      "step": 2533
    },
    {
      "epoch": 0.26729957805907173,
      "grad_norm": 0.5864850282669067,
      "learning_rate": 0.0015,
      "loss": 1.7308,
      "step": 2534
    },
    {
      "epoch": 0.2674050632911392,
      "grad_norm": 0.4986039996147156,
      "learning_rate": 0.0015,
      "loss": 1.7879,
      "step": 2535
    },
    {
      "epoch": 0.26751054852320677,
      "grad_norm": 0.6276726126670837,
      "learning_rate": 0.0015,
      "loss": 1.7516,
      "step": 2536
    },
    {
      "epoch": 0.26761603375527426,
      "grad_norm": 0.5389881730079651,
      "learning_rate": 0.0015,
      "loss": 1.7396,
      "step": 2537
    },
    {
      "epoch": 0.26772151898734176,
      "grad_norm": 0.5643240213394165,
      "learning_rate": 0.0015,
      "loss": 1.7832,
      "step": 2538
    },
    {
      "epoch": 0.2678270042194093,
      "grad_norm": 0.6625216007232666,
      "learning_rate": 0.0015,
      "loss": 1.7545,
      "step": 2539
    },
    {
      "epoch": 0.2679324894514768,
      "grad_norm": 0.5147009491920471,
      "learning_rate": 0.0015,
      "loss": 1.7298,
      "step": 2540
    },
    {
      "epoch": 0.2680379746835443,
      "grad_norm": 0.6473565101623535,
      "learning_rate": 0.0015,
      "loss": 1.7371,
      "step": 2541
    },
    {
      "epoch": 0.26814345991561184,
      "grad_norm": 0.6117268204689026,
      "learning_rate": 0.0015,
      "loss": 1.7687,
      "step": 2542
    },
    {
      "epoch": 0.26824894514767933,
      "grad_norm": 0.4728650152683258,
      "learning_rate": 0.0015,
      "loss": 1.7349,
      "step": 2543
    },
    {
      "epoch": 0.2683544303797468,
      "grad_norm": 0.5104753375053406,
      "learning_rate": 0.0015,
      "loss": 1.7113,
      "step": 2544
    },
    {
      "epoch": 0.26845991561181437,
      "grad_norm": 0.6263878345489502,
      "learning_rate": 0.0015,
      "loss": 1.7494,
      "step": 2545
    },
    {
      "epoch": 0.26856540084388186,
      "grad_norm": 0.5564925074577332,
      "learning_rate": 0.0015,
      "loss": 1.7466,
      "step": 2546
    },
    {
      "epoch": 0.26867088607594936,
      "grad_norm": 0.48637768626213074,
      "learning_rate": 0.0015,
      "loss": 1.7278,
      "step": 2547
    },
    {
      "epoch": 0.2687763713080169,
      "grad_norm": 0.5296652913093567,
      "learning_rate": 0.0015,
      "loss": 1.7439,
      "step": 2548
    },
    {
      "epoch": 0.2688818565400844,
      "grad_norm": 0.5677223801612854,
      "learning_rate": 0.0015,
      "loss": 1.7483,
      "step": 2549
    },
    {
      "epoch": 0.2689873417721519,
      "grad_norm": 0.5286999344825745,
      "learning_rate": 0.0015,
      "loss": 1.7355,
      "step": 2550
    },
    {
      "epoch": 0.26909282700421944,
      "grad_norm": 0.4440421164035797,
      "learning_rate": 0.0015,
      "loss": 1.7615,
      "step": 2551
    },
    {
      "epoch": 0.26919831223628693,
      "grad_norm": 0.533702552318573,
      "learning_rate": 0.0015,
      "loss": 1.7746,
      "step": 2552
    },
    {
      "epoch": 0.2693037974683544,
      "grad_norm": 0.46381649374961853,
      "learning_rate": 0.0015,
      "loss": 1.7797,
      "step": 2553
    },
    {
      "epoch": 0.2694092827004219,
      "grad_norm": 0.5055655241012573,
      "learning_rate": 0.0015,
      "loss": 1.7355,
      "step": 2554
    },
    {
      "epoch": 0.26951476793248946,
      "grad_norm": 0.5017513036727905,
      "learning_rate": 0.0015,
      "loss": 1.748,
      "step": 2555
    },
    {
      "epoch": 0.26962025316455696,
      "grad_norm": 0.51686692237854,
      "learning_rate": 0.0015,
      "loss": 1.7609,
      "step": 2556
    },
    {
      "epoch": 0.26972573839662445,
      "grad_norm": 0.4669591188430786,
      "learning_rate": 0.0015,
      "loss": 1.7476,
      "step": 2557
    },
    {
      "epoch": 0.269831223628692,
      "grad_norm": 0.4892376661300659,
      "learning_rate": 0.0015,
      "loss": 1.7544,
      "step": 2558
    },
    {
      "epoch": 0.2699367088607595,
      "grad_norm": 0.5178463459014893,
      "learning_rate": 0.0015,
      "loss": 1.7458,
      "step": 2559
    },
    {
      "epoch": 0.270042194092827,
      "grad_norm": 0.5536075234413147,
      "learning_rate": 0.0015,
      "loss": 1.7082,
      "step": 2560
    },
    {
      "epoch": 0.27014767932489453,
      "grad_norm": 0.6125830411911011,
      "learning_rate": 0.0015,
      "loss": 1.7318,
      "step": 2561
    },
    {
      "epoch": 0.270253164556962,
      "grad_norm": 0.5590088963508606,
      "learning_rate": 0.0015,
      "loss": 1.7592,
      "step": 2562
    },
    {
      "epoch": 0.2703586497890295,
      "grad_norm": 0.4859699606895447,
      "learning_rate": 0.0015,
      "loss": 1.7826,
      "step": 2563
    },
    {
      "epoch": 0.27046413502109706,
      "grad_norm": 0.4657300114631653,
      "learning_rate": 0.0015,
      "loss": 1.7336,
      "step": 2564
    },
    {
      "epoch": 0.27056962025316456,
      "grad_norm": 0.4587562382221222,
      "learning_rate": 0.0015,
      "loss": 1.704,
      "step": 2565
    },
    {
      "epoch": 0.27067510548523205,
      "grad_norm": 0.5151911973953247,
      "learning_rate": 0.0015,
      "loss": 1.7224,
      "step": 2566
    },
    {
      "epoch": 0.2707805907172996,
      "grad_norm": 0.6008763313293457,
      "learning_rate": 0.0015,
      "loss": 1.764,
      "step": 2567
    },
    {
      "epoch": 0.2708860759493671,
      "grad_norm": 0.6141114234924316,
      "learning_rate": 0.0015,
      "loss": 1.7535,
      "step": 2568
    },
    {
      "epoch": 0.2709915611814346,
      "grad_norm": 0.5276148915290833,
      "learning_rate": 0.0015,
      "loss": 1.7049,
      "step": 2569
    },
    {
      "epoch": 0.27109704641350213,
      "grad_norm": 0.5054593682289124,
      "learning_rate": 0.0015,
      "loss": 1.7749,
      "step": 2570
    },
    {
      "epoch": 0.2712025316455696,
      "grad_norm": 0.5235294103622437,
      "learning_rate": 0.0015,
      "loss": 1.7161,
      "step": 2571
    },
    {
      "epoch": 0.2713080168776371,
      "grad_norm": 0.561466634273529,
      "learning_rate": 0.0015,
      "loss": 1.7458,
      "step": 2572
    },
    {
      "epoch": 0.27141350210970466,
      "grad_norm": 0.6114590167999268,
      "learning_rate": 0.0015,
      "loss": 1.7455,
      "step": 2573
    },
    {
      "epoch": 0.27151898734177216,
      "grad_norm": 0.5305688977241516,
      "learning_rate": 0.0015,
      "loss": 1.7409,
      "step": 2574
    },
    {
      "epoch": 0.27162447257383965,
      "grad_norm": 0.56999671459198,
      "learning_rate": 0.0015,
      "loss": 1.7445,
      "step": 2575
    },
    {
      "epoch": 0.2717299578059072,
      "grad_norm": 0.5928850173950195,
      "learning_rate": 0.0015,
      "loss": 1.7156,
      "step": 2576
    },
    {
      "epoch": 0.2718354430379747,
      "grad_norm": 0.5534878373146057,
      "learning_rate": 0.0015,
      "loss": 1.757,
      "step": 2577
    },
    {
      "epoch": 0.2719409282700422,
      "grad_norm": 0.4853445589542389,
      "learning_rate": 0.0015,
      "loss": 1.7935,
      "step": 2578
    },
    {
      "epoch": 0.27204641350210973,
      "grad_norm": 0.5856049060821533,
      "learning_rate": 0.0015,
      "loss": 1.7429,
      "step": 2579
    },
    {
      "epoch": 0.2721518987341772,
      "grad_norm": 0.5238052606582642,
      "learning_rate": 0.0015,
      "loss": 1.7442,
      "step": 2580
    },
    {
      "epoch": 0.2722573839662447,
      "grad_norm": 0.6355899572372437,
      "learning_rate": 0.0015,
      "loss": 1.7258,
      "step": 2581
    },
    {
      "epoch": 0.27236286919831226,
      "grad_norm": 0.880685567855835,
      "learning_rate": 0.0015,
      "loss": 1.7596,
      "step": 2582
    },
    {
      "epoch": 0.27246835443037976,
      "grad_norm": 0.7288837432861328,
      "learning_rate": 0.0015,
      "loss": 1.7502,
      "step": 2583
    },
    {
      "epoch": 0.27257383966244725,
      "grad_norm": 0.49775776267051697,
      "learning_rate": 0.0015,
      "loss": 1.7494,
      "step": 2584
    },
    {
      "epoch": 0.27267932489451474,
      "grad_norm": 0.899782657623291,
      "learning_rate": 0.0015,
      "loss": 1.774,
      "step": 2585
    },
    {
      "epoch": 0.2727848101265823,
      "grad_norm": 1.0974575281143188,
      "learning_rate": 0.0015,
      "loss": 1.736,
      "step": 2586
    },
    {
      "epoch": 0.2728902953586498,
      "grad_norm": 0.7350040078163147,
      "learning_rate": 0.0015,
      "loss": 1.7729,
      "step": 2587
    },
    {
      "epoch": 0.2729957805907173,
      "grad_norm": 0.5562779903411865,
      "learning_rate": 0.0015,
      "loss": 1.7598,
      "step": 2588
    },
    {
      "epoch": 0.2731012658227848,
      "grad_norm": 0.5916325449943542,
      "learning_rate": 0.0015,
      "loss": 1.7439,
      "step": 2589
    },
    {
      "epoch": 0.2732067510548523,
      "grad_norm": 0.5184524059295654,
      "learning_rate": 0.0015,
      "loss": 1.7488,
      "step": 2590
    },
    {
      "epoch": 0.2733122362869198,
      "grad_norm": 0.5797756314277649,
      "learning_rate": 0.0015,
      "loss": 1.7461,
      "step": 2591
    },
    {
      "epoch": 0.27341772151898736,
      "grad_norm": 0.6259637475013733,
      "learning_rate": 0.0015,
      "loss": 1.768,
      "step": 2592
    },
    {
      "epoch": 0.27352320675105485,
      "grad_norm": 0.5711591243743896,
      "learning_rate": 0.0015,
      "loss": 1.7512,
      "step": 2593
    },
    {
      "epoch": 0.27362869198312234,
      "grad_norm": 0.5064637660980225,
      "learning_rate": 0.0015,
      "loss": 1.7501,
      "step": 2594
    },
    {
      "epoch": 0.2737341772151899,
      "grad_norm": 0.5987383723258972,
      "learning_rate": 0.0015,
      "loss": 1.7777,
      "step": 2595
    },
    {
      "epoch": 0.2738396624472574,
      "grad_norm": 0.6491714119911194,
      "learning_rate": 0.0015,
      "loss": 1.7477,
      "step": 2596
    },
    {
      "epoch": 0.2739451476793249,
      "grad_norm": 0.5929031372070312,
      "learning_rate": 0.0015,
      "loss": 1.7369,
      "step": 2597
    },
    {
      "epoch": 0.2740506329113924,
      "grad_norm": 0.4909879267215729,
      "learning_rate": 0.0015,
      "loss": 1.7172,
      "step": 2598
    },
    {
      "epoch": 0.2741561181434599,
      "grad_norm": 0.6378136873245239,
      "learning_rate": 0.0015,
      "loss": 1.729,
      "step": 2599
    },
    {
      "epoch": 0.2742616033755274,
      "grad_norm": 0.598289430141449,
      "learning_rate": 0.0015,
      "loss": 1.776,
      "step": 2600
    },
    {
      "epoch": 0.27436708860759496,
      "grad_norm": 0.46960604190826416,
      "learning_rate": 0.0015,
      "loss": 1.7285,
      "step": 2601
    },
    {
      "epoch": 0.27447257383966245,
      "grad_norm": 0.7559182643890381,
      "learning_rate": 0.0015,
      "loss": 1.7455,
      "step": 2602
    },
    {
      "epoch": 0.27457805907172994,
      "grad_norm": 0.8152137398719788,
      "learning_rate": 0.0015,
      "loss": 1.6857,
      "step": 2603
    },
    {
      "epoch": 0.2746835443037975,
      "grad_norm": 0.5375908613204956,
      "learning_rate": 0.0015,
      "loss": 1.7668,
      "step": 2604
    },
    {
      "epoch": 0.274789029535865,
      "grad_norm": 0.4875588119029999,
      "learning_rate": 0.0015,
      "loss": 1.7754,
      "step": 2605
    },
    {
      "epoch": 0.2748945147679325,
      "grad_norm": 0.4983108341693878,
      "learning_rate": 0.0015,
      "loss": 1.7716,
      "step": 2606
    },
    {
      "epoch": 0.275,
      "grad_norm": 0.47570163011550903,
      "learning_rate": 0.0015,
      "loss": 1.7631,
      "step": 2607
    },
    {
      "epoch": 0.2751054852320675,
      "grad_norm": 0.5114532709121704,
      "learning_rate": 0.0015,
      "loss": 1.7378,
      "step": 2608
    },
    {
      "epoch": 0.275210970464135,
      "grad_norm": 0.5763033628463745,
      "learning_rate": 0.0015,
      "loss": 1.7628,
      "step": 2609
    },
    {
      "epoch": 0.27531645569620256,
      "grad_norm": 0.4745025336742401,
      "learning_rate": 0.0015,
      "loss": 1.7486,
      "step": 2610
    },
    {
      "epoch": 0.27542194092827005,
      "grad_norm": 0.6279363632202148,
      "learning_rate": 0.0015,
      "loss": 1.765,
      "step": 2611
    },
    {
      "epoch": 0.27552742616033754,
      "grad_norm": 0.7709202766418457,
      "learning_rate": 0.0015,
      "loss": 1.7587,
      "step": 2612
    },
    {
      "epoch": 0.2756329113924051,
      "grad_norm": 0.589849591255188,
      "learning_rate": 0.0015,
      "loss": 1.787,
      "step": 2613
    },
    {
      "epoch": 0.2757383966244726,
      "grad_norm": 0.6236584186553955,
      "learning_rate": 0.0015,
      "loss": 1.7393,
      "step": 2614
    },
    {
      "epoch": 0.2758438818565401,
      "grad_norm": 0.5918351411819458,
      "learning_rate": 0.0015,
      "loss": 1.7454,
      "step": 2615
    },
    {
      "epoch": 0.2759493670886076,
      "grad_norm": 0.5218271613121033,
      "learning_rate": 0.0015,
      "loss": 1.7274,
      "step": 2616
    },
    {
      "epoch": 0.2760548523206751,
      "grad_norm": 0.5378252267837524,
      "learning_rate": 0.0015,
      "loss": 1.7254,
      "step": 2617
    },
    {
      "epoch": 0.2761603375527426,
      "grad_norm": 0.4909058213233948,
      "learning_rate": 0.0015,
      "loss": 1.7187,
      "step": 2618
    },
    {
      "epoch": 0.2762658227848101,
      "grad_norm": 0.5239964127540588,
      "learning_rate": 0.0015,
      "loss": 1.7432,
      "step": 2619
    },
    {
      "epoch": 0.27637130801687765,
      "grad_norm": 0.5242708325386047,
      "learning_rate": 0.0015,
      "loss": 1.7262,
      "step": 2620
    },
    {
      "epoch": 0.27647679324894514,
      "grad_norm": 0.4703781008720398,
      "learning_rate": 0.0015,
      "loss": 1.7169,
      "step": 2621
    },
    {
      "epoch": 0.27658227848101263,
      "grad_norm": 0.6690729856491089,
      "learning_rate": 0.0015,
      "loss": 1.7061,
      "step": 2622
    },
    {
      "epoch": 0.2766877637130802,
      "grad_norm": 0.6361964344978333,
      "learning_rate": 0.0015,
      "loss": 1.7189,
      "step": 2623
    },
    {
      "epoch": 0.2767932489451477,
      "grad_norm": 0.520235002040863,
      "learning_rate": 0.0015,
      "loss": 1.7574,
      "step": 2624
    },
    {
      "epoch": 0.27689873417721517,
      "grad_norm": 0.6627557873725891,
      "learning_rate": 0.0015,
      "loss": 1.7287,
      "step": 2625
    },
    {
      "epoch": 0.2770042194092827,
      "grad_norm": 0.6653998494148254,
      "learning_rate": 0.0015,
      "loss": 1.7212,
      "step": 2626
    },
    {
      "epoch": 0.2771097046413502,
      "grad_norm": 0.49855709075927734,
      "learning_rate": 0.0015,
      "loss": 1.7185,
      "step": 2627
    },
    {
      "epoch": 0.2772151898734177,
      "grad_norm": 0.5100919008255005,
      "learning_rate": 0.0015,
      "loss": 1.7525,
      "step": 2628
    },
    {
      "epoch": 0.27732067510548525,
      "grad_norm": 0.5465761423110962,
      "learning_rate": 0.0015,
      "loss": 1.7255,
      "step": 2629
    },
    {
      "epoch": 0.27742616033755274,
      "grad_norm": 0.4805075228214264,
      "learning_rate": 0.0015,
      "loss": 1.7762,
      "step": 2630
    },
    {
      "epoch": 0.27753164556962023,
      "grad_norm": 0.5802726149559021,
      "learning_rate": 0.0015,
      "loss": 1.7474,
      "step": 2631
    },
    {
      "epoch": 0.2776371308016878,
      "grad_norm": 0.5042747259140015,
      "learning_rate": 0.0015,
      "loss": 1.7498,
      "step": 2632
    },
    {
      "epoch": 0.2777426160337553,
      "grad_norm": 0.526651918888092,
      "learning_rate": 0.0015,
      "loss": 1.7131,
      "step": 2633
    },
    {
      "epoch": 0.27784810126582277,
      "grad_norm": 0.6353644728660583,
      "learning_rate": 0.0015,
      "loss": 1.7585,
      "step": 2634
    },
    {
      "epoch": 0.2779535864978903,
      "grad_norm": 0.6921878457069397,
      "learning_rate": 0.0015,
      "loss": 1.7893,
      "step": 2635
    },
    {
      "epoch": 0.2780590717299578,
      "grad_norm": 0.45830094814300537,
      "learning_rate": 0.0015,
      "loss": 1.7361,
      "step": 2636
    },
    {
      "epoch": 0.2781645569620253,
      "grad_norm": 0.5728946924209595,
      "learning_rate": 0.0015,
      "loss": 1.7706,
      "step": 2637
    },
    {
      "epoch": 0.27827004219409285,
      "grad_norm": 0.5714189410209656,
      "learning_rate": 0.0015,
      "loss": 1.7181,
      "step": 2638
    },
    {
      "epoch": 0.27837552742616034,
      "grad_norm": 0.576870858669281,
      "learning_rate": 0.0015,
      "loss": 1.7343,
      "step": 2639
    },
    {
      "epoch": 0.27848101265822783,
      "grad_norm": 0.4691697657108307,
      "learning_rate": 0.0015,
      "loss": 1.7189,
      "step": 2640
    },
    {
      "epoch": 0.2785864978902954,
      "grad_norm": 0.46785059571266174,
      "learning_rate": 0.0015,
      "loss": 1.7545,
      "step": 2641
    },
    {
      "epoch": 0.2786919831223629,
      "grad_norm": 0.4791947305202484,
      "learning_rate": 0.0015,
      "loss": 1.7099,
      "step": 2642
    },
    {
      "epoch": 0.27879746835443037,
      "grad_norm": 0.4511741101741791,
      "learning_rate": 0.0015,
      "loss": 1.7102,
      "step": 2643
    },
    {
      "epoch": 0.2789029535864979,
      "grad_norm": 0.51612389087677,
      "learning_rate": 0.0015,
      "loss": 1.6999,
      "step": 2644
    },
    {
      "epoch": 0.2790084388185654,
      "grad_norm": 0.4879615902900696,
      "learning_rate": 0.0015,
      "loss": 1.7785,
      "step": 2645
    },
    {
      "epoch": 0.2791139240506329,
      "grad_norm": 0.5071017146110535,
      "learning_rate": 0.0015,
      "loss": 1.7351,
      "step": 2646
    },
    {
      "epoch": 0.27921940928270045,
      "grad_norm": 0.6026368141174316,
      "learning_rate": 0.0015,
      "loss": 1.7432,
      "step": 2647
    },
    {
      "epoch": 0.27932489451476794,
      "grad_norm": 0.5431324243545532,
      "learning_rate": 0.0015,
      "loss": 1.7542,
      "step": 2648
    },
    {
      "epoch": 0.27943037974683543,
      "grad_norm": 0.4874689280986786,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 2649
    },
    {
      "epoch": 0.2795358649789029,
      "grad_norm": 0.5022848844528198,
      "learning_rate": 0.0015,
      "loss": 1.747,
      "step": 2650
    },
    {
      "epoch": 0.2796413502109705,
      "grad_norm": 0.5504576563835144,
      "learning_rate": 0.0015,
      "loss": 1.7166,
      "step": 2651
    },
    {
      "epoch": 0.27974683544303797,
      "grad_norm": 0.528978705406189,
      "learning_rate": 0.0015,
      "loss": 1.7533,
      "step": 2652
    },
    {
      "epoch": 0.27985232067510546,
      "grad_norm": 0.4611363708972931,
      "learning_rate": 0.0015,
      "loss": 1.7609,
      "step": 2653
    },
    {
      "epoch": 0.279957805907173,
      "grad_norm": 0.4825553596019745,
      "learning_rate": 0.0015,
      "loss": 1.755,
      "step": 2654
    },
    {
      "epoch": 0.2800632911392405,
      "grad_norm": 0.5768417119979858,
      "learning_rate": 0.0015,
      "loss": 1.7338,
      "step": 2655
    },
    {
      "epoch": 0.280168776371308,
      "grad_norm": 0.48545992374420166,
      "learning_rate": 0.0015,
      "loss": 1.7299,
      "step": 2656
    },
    {
      "epoch": 0.28027426160337554,
      "grad_norm": 0.5208084583282471,
      "learning_rate": 0.0015,
      "loss": 1.7053,
      "step": 2657
    },
    {
      "epoch": 0.28037974683544303,
      "grad_norm": 0.610554575920105,
      "learning_rate": 0.0015,
      "loss": 1.6835,
      "step": 2658
    },
    {
      "epoch": 0.2804852320675105,
      "grad_norm": 0.5752939581871033,
      "learning_rate": 0.0015,
      "loss": 1.747,
      "step": 2659
    },
    {
      "epoch": 0.2805907172995781,
      "grad_norm": 0.46665632724761963,
      "learning_rate": 0.0015,
      "loss": 1.7333,
      "step": 2660
    },
    {
      "epoch": 0.28069620253164557,
      "grad_norm": 0.53729647397995,
      "learning_rate": 0.0015,
      "loss": 1.7549,
      "step": 2661
    },
    {
      "epoch": 0.28080168776371306,
      "grad_norm": 0.5107341408729553,
      "learning_rate": 0.0015,
      "loss": 1.7493,
      "step": 2662
    },
    {
      "epoch": 0.2809071729957806,
      "grad_norm": 0.4697841703891754,
      "learning_rate": 0.0015,
      "loss": 1.7258,
      "step": 2663
    },
    {
      "epoch": 0.2810126582278481,
      "grad_norm": 0.5092480182647705,
      "learning_rate": 0.0015,
      "loss": 1.724,
      "step": 2664
    },
    {
      "epoch": 0.2811181434599156,
      "grad_norm": 0.5099025964736938,
      "learning_rate": 0.0015,
      "loss": 1.7351,
      "step": 2665
    },
    {
      "epoch": 0.28122362869198314,
      "grad_norm": 0.5085151791572571,
      "learning_rate": 0.0015,
      "loss": 1.7288,
      "step": 2666
    },
    {
      "epoch": 0.28132911392405063,
      "grad_norm": 0.505859375,
      "learning_rate": 0.0015,
      "loss": 1.724,
      "step": 2667
    },
    {
      "epoch": 0.2814345991561181,
      "grad_norm": 0.5873626470565796,
      "learning_rate": 0.0015,
      "loss": 1.728,
      "step": 2668
    },
    {
      "epoch": 0.2815400843881857,
      "grad_norm": 0.49014726281166077,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 2669
    },
    {
      "epoch": 0.28164556962025317,
      "grad_norm": 0.5928011536598206,
      "learning_rate": 0.0015,
      "loss": 1.7303,
      "step": 2670
    },
    {
      "epoch": 0.28175105485232066,
      "grad_norm": 0.5059046745300293,
      "learning_rate": 0.0015,
      "loss": 1.7654,
      "step": 2671
    },
    {
      "epoch": 0.2818565400843882,
      "grad_norm": 0.5453937649726868,
      "learning_rate": 0.0015,
      "loss": 1.7406,
      "step": 2672
    },
    {
      "epoch": 0.2819620253164557,
      "grad_norm": 0.8845718502998352,
      "learning_rate": 0.0015,
      "loss": 1.781,
      "step": 2673
    },
    {
      "epoch": 0.2820675105485232,
      "grad_norm": 0.6845380663871765,
      "learning_rate": 0.0015,
      "loss": 1.7183,
      "step": 2674
    },
    {
      "epoch": 0.28217299578059074,
      "grad_norm": 0.6184827089309692,
      "learning_rate": 0.0015,
      "loss": 1.734,
      "step": 2675
    },
    {
      "epoch": 0.28227848101265823,
      "grad_norm": 1.1802616119384766,
      "learning_rate": 0.0015,
      "loss": 1.755,
      "step": 2676
    },
    {
      "epoch": 0.2823839662447257,
      "grad_norm": 0.759573757648468,
      "learning_rate": 0.0015,
      "loss": 1.712,
      "step": 2677
    },
    {
      "epoch": 0.2824894514767933,
      "grad_norm": 0.49365097284317017,
      "learning_rate": 0.0015,
      "loss": 1.7034,
      "step": 2678
    },
    {
      "epoch": 0.28259493670886077,
      "grad_norm": 0.6150100231170654,
      "learning_rate": 0.0015,
      "loss": 1.7474,
      "step": 2679
    },
    {
      "epoch": 0.28270042194092826,
      "grad_norm": 0.5978943705558777,
      "learning_rate": 0.0015,
      "loss": 1.7187,
      "step": 2680
    },
    {
      "epoch": 0.2828059071729958,
      "grad_norm": 0.5664392113685608,
      "learning_rate": 0.0015,
      "loss": 1.7102,
      "step": 2681
    },
    {
      "epoch": 0.2829113924050633,
      "grad_norm": 0.6059300303459167,
      "learning_rate": 0.0015,
      "loss": 1.6932,
      "step": 2682
    },
    {
      "epoch": 0.2830168776371308,
      "grad_norm": 0.5169327855110168,
      "learning_rate": 0.0015,
      "loss": 1.7585,
      "step": 2683
    },
    {
      "epoch": 0.2831223628691983,
      "grad_norm": 0.4633955955505371,
      "learning_rate": 0.0015,
      "loss": 1.7438,
      "step": 2684
    },
    {
      "epoch": 0.28322784810126583,
      "grad_norm": 0.5879747271537781,
      "learning_rate": 0.0015,
      "loss": 1.7128,
      "step": 2685
    },
    {
      "epoch": 0.2833333333333333,
      "grad_norm": 0.6198854446411133,
      "learning_rate": 0.0015,
      "loss": 1.7334,
      "step": 2686
    },
    {
      "epoch": 0.2834388185654008,
      "grad_norm": 0.526696503162384,
      "learning_rate": 0.0015,
      "loss": 1.7489,
      "step": 2687
    },
    {
      "epoch": 0.28354430379746837,
      "grad_norm": 0.5070342421531677,
      "learning_rate": 0.0015,
      "loss": 1.7274,
      "step": 2688
    },
    {
      "epoch": 0.28364978902953586,
      "grad_norm": 0.6639285087585449,
      "learning_rate": 0.0015,
      "loss": 1.7399,
      "step": 2689
    },
    {
      "epoch": 0.28375527426160335,
      "grad_norm": 0.6574095487594604,
      "learning_rate": 0.0015,
      "loss": 1.776,
      "step": 2690
    },
    {
      "epoch": 0.2838607594936709,
      "grad_norm": 0.5485591888427734,
      "learning_rate": 0.0015,
      "loss": 1.7485,
      "step": 2691
    },
    {
      "epoch": 0.2839662447257384,
      "grad_norm": 0.5603123903274536,
      "learning_rate": 0.0015,
      "loss": 1.7286,
      "step": 2692
    },
    {
      "epoch": 0.2840717299578059,
      "grad_norm": 0.5913242101669312,
      "learning_rate": 0.0015,
      "loss": 1.7524,
      "step": 2693
    },
    {
      "epoch": 0.28417721518987343,
      "grad_norm": 0.498909056186676,
      "learning_rate": 0.0015,
      "loss": 1.7077,
      "step": 2694
    },
    {
      "epoch": 0.2842827004219409,
      "grad_norm": 0.6316589713096619,
      "learning_rate": 0.0015,
      "loss": 1.7474,
      "step": 2695
    },
    {
      "epoch": 0.2843881856540084,
      "grad_norm": 0.7538981437683105,
      "learning_rate": 0.0015,
      "loss": 1.7706,
      "step": 2696
    },
    {
      "epoch": 0.28449367088607597,
      "grad_norm": 0.5047091841697693,
      "learning_rate": 0.0015,
      "loss": 1.7044,
      "step": 2697
    },
    {
      "epoch": 0.28459915611814346,
      "grad_norm": 0.6115390062332153,
      "learning_rate": 0.0015,
      "loss": 1.7308,
      "step": 2698
    },
    {
      "epoch": 0.28470464135021095,
      "grad_norm": 0.5837851762771606,
      "learning_rate": 0.0015,
      "loss": 1.7239,
      "step": 2699
    },
    {
      "epoch": 0.2848101265822785,
      "grad_norm": 0.5235541462898254,
      "learning_rate": 0.0015,
      "loss": 1.7002,
      "step": 2700
    },
    {
      "epoch": 0.284915611814346,
      "grad_norm": 0.8157408237457275,
      "learning_rate": 0.0015,
      "loss": 1.7714,
      "step": 2701
    },
    {
      "epoch": 0.2850210970464135,
      "grad_norm": 0.9191502928733826,
      "learning_rate": 0.0015,
      "loss": 1.7593,
      "step": 2702
    },
    {
      "epoch": 0.28512658227848103,
      "grad_norm": 0.6521652340888977,
      "learning_rate": 0.0015,
      "loss": 1.7492,
      "step": 2703
    },
    {
      "epoch": 0.2852320675105485,
      "grad_norm": 0.616782546043396,
      "learning_rate": 0.0015,
      "loss": 1.718,
      "step": 2704
    },
    {
      "epoch": 0.285337552742616,
      "grad_norm": 0.956119954586029,
      "learning_rate": 0.0015,
      "loss": 1.7578,
      "step": 2705
    },
    {
      "epoch": 0.28544303797468357,
      "grad_norm": 0.7189751267433167,
      "learning_rate": 0.0015,
      "loss": 1.7401,
      "step": 2706
    },
    {
      "epoch": 0.28554852320675106,
      "grad_norm": 0.7041029930114746,
      "learning_rate": 0.0015,
      "loss": 1.7295,
      "step": 2707
    },
    {
      "epoch": 0.28565400843881855,
      "grad_norm": 1.0130025148391724,
      "learning_rate": 0.0015,
      "loss": 1.7218,
      "step": 2708
    },
    {
      "epoch": 0.2857594936708861,
      "grad_norm": 0.5713807940483093,
      "learning_rate": 0.0015,
      "loss": 1.7234,
      "step": 2709
    },
    {
      "epoch": 0.2858649789029536,
      "grad_norm": 0.8522231578826904,
      "learning_rate": 0.0015,
      "loss": 1.7381,
      "step": 2710
    },
    {
      "epoch": 0.2859704641350211,
      "grad_norm": 0.9587810635566711,
      "learning_rate": 0.0015,
      "loss": 1.7452,
      "step": 2711
    },
    {
      "epoch": 0.28607594936708863,
      "grad_norm": 0.5355757474899292,
      "learning_rate": 0.0015,
      "loss": 1.7689,
      "step": 2712
    },
    {
      "epoch": 0.2861814345991561,
      "grad_norm": 0.9634062647819519,
      "learning_rate": 0.0015,
      "loss": 1.7289,
      "step": 2713
    },
    {
      "epoch": 0.2862869198312236,
      "grad_norm": 0.7838209867477417,
      "learning_rate": 0.0015,
      "loss": 1.7679,
      "step": 2714
    },
    {
      "epoch": 0.28639240506329117,
      "grad_norm": 0.5818883180618286,
      "learning_rate": 0.0015,
      "loss": 1.7443,
      "step": 2715
    },
    {
      "epoch": 0.28649789029535866,
      "grad_norm": 0.944904088973999,
      "learning_rate": 0.0015,
      "loss": 1.7506,
      "step": 2716
    },
    {
      "epoch": 0.28660337552742615,
      "grad_norm": 0.5879671573638916,
      "learning_rate": 0.0015,
      "loss": 1.7703,
      "step": 2717
    },
    {
      "epoch": 0.28670886075949364,
      "grad_norm": 0.6757920384407043,
      "learning_rate": 0.0015,
      "loss": 1.7026,
      "step": 2718
    },
    {
      "epoch": 0.2868143459915612,
      "grad_norm": 0.5808043479919434,
      "learning_rate": 0.0015,
      "loss": 1.6967,
      "step": 2719
    },
    {
      "epoch": 0.2869198312236287,
      "grad_norm": 0.5420714020729065,
      "learning_rate": 0.0015,
      "loss": 1.7191,
      "step": 2720
    },
    {
      "epoch": 0.2870253164556962,
      "grad_norm": 0.5985084176063538,
      "learning_rate": 0.0015,
      "loss": 1.7173,
      "step": 2721
    },
    {
      "epoch": 0.2871308016877637,
      "grad_norm": 0.4714157283306122,
      "learning_rate": 0.0015,
      "loss": 1.7065,
      "step": 2722
    },
    {
      "epoch": 0.2872362869198312,
      "grad_norm": 0.5194176435470581,
      "learning_rate": 0.0015,
      "loss": 1.7402,
      "step": 2723
    },
    {
      "epoch": 0.2873417721518987,
      "grad_norm": 0.5509008169174194,
      "learning_rate": 0.0015,
      "loss": 1.6874,
      "step": 2724
    },
    {
      "epoch": 0.28744725738396626,
      "grad_norm": 0.5198971629142761,
      "learning_rate": 0.0015,
      "loss": 1.7208,
      "step": 2725
    },
    {
      "epoch": 0.28755274261603375,
      "grad_norm": 0.595416784286499,
      "learning_rate": 0.0015,
      "loss": 1.7457,
      "step": 2726
    },
    {
      "epoch": 0.28765822784810124,
      "grad_norm": 0.6461116671562195,
      "learning_rate": 0.0015,
      "loss": 1.7234,
      "step": 2727
    },
    {
      "epoch": 0.2877637130801688,
      "grad_norm": 0.5399662256240845,
      "learning_rate": 0.0015,
      "loss": 1.6971,
      "step": 2728
    },
    {
      "epoch": 0.2878691983122363,
      "grad_norm": 0.5240736603736877,
      "learning_rate": 0.0015,
      "loss": 1.7337,
      "step": 2729
    },
    {
      "epoch": 0.2879746835443038,
      "grad_norm": 0.5210794806480408,
      "learning_rate": 0.0015,
      "loss": 1.727,
      "step": 2730
    },
    {
      "epoch": 0.2880801687763713,
      "grad_norm": 0.5225235223770142,
      "learning_rate": 0.0015,
      "loss": 1.7336,
      "step": 2731
    },
    {
      "epoch": 0.2881856540084388,
      "grad_norm": 0.4826267957687378,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 2732
    },
    {
      "epoch": 0.2882911392405063,
      "grad_norm": 0.5007849931716919,
      "learning_rate": 0.0015,
      "loss": 1.7366,
      "step": 2733
    },
    {
      "epoch": 0.28839662447257386,
      "grad_norm": 0.46511006355285645,
      "learning_rate": 0.0015,
      "loss": 1.708,
      "step": 2734
    },
    {
      "epoch": 0.28850210970464135,
      "grad_norm": 0.5794470906257629,
      "learning_rate": 0.0015,
      "loss": 1.7063,
      "step": 2735
    },
    {
      "epoch": 0.28860759493670884,
      "grad_norm": 0.5770504474639893,
      "learning_rate": 0.0015,
      "loss": 1.7044,
      "step": 2736
    },
    {
      "epoch": 0.2887130801687764,
      "grad_norm": 0.644075870513916,
      "learning_rate": 0.0015,
      "loss": 1.7161,
      "step": 2737
    },
    {
      "epoch": 0.2888185654008439,
      "grad_norm": 0.4943702518939972,
      "learning_rate": 0.0015,
      "loss": 1.7368,
      "step": 2738
    },
    {
      "epoch": 0.2889240506329114,
      "grad_norm": 0.8831140995025635,
      "learning_rate": 0.0015,
      "loss": 1.7638,
      "step": 2739
    },
    {
      "epoch": 0.2890295358649789,
      "grad_norm": 0.8674169182777405,
      "learning_rate": 0.0015,
      "loss": 1.7741,
      "step": 2740
    },
    {
      "epoch": 0.2891350210970464,
      "grad_norm": 0.4986017644405365,
      "learning_rate": 0.0015,
      "loss": 1.721,
      "step": 2741
    },
    {
      "epoch": 0.2892405063291139,
      "grad_norm": 0.6774818897247314,
      "learning_rate": 0.0015,
      "loss": 1.7467,
      "step": 2742
    },
    {
      "epoch": 0.28934599156118146,
      "grad_norm": 0.6801589131355286,
      "learning_rate": 0.0015,
      "loss": 1.7066,
      "step": 2743
    },
    {
      "epoch": 0.28945147679324895,
      "grad_norm": 0.48362305760383606,
      "learning_rate": 0.0015,
      "loss": 1.7435,
      "step": 2744
    },
    {
      "epoch": 0.28955696202531644,
      "grad_norm": 0.5517992377281189,
      "learning_rate": 0.0015,
      "loss": 1.7368,
      "step": 2745
    },
    {
      "epoch": 0.289662447257384,
      "grad_norm": 0.5449758172035217,
      "learning_rate": 0.0015,
      "loss": 1.75,
      "step": 2746
    },
    {
      "epoch": 0.2897679324894515,
      "grad_norm": 0.45939669013023376,
      "learning_rate": 0.0015,
      "loss": 1.7162,
      "step": 2747
    },
    {
      "epoch": 0.289873417721519,
      "grad_norm": 0.5620844960212708,
      "learning_rate": 0.0015,
      "loss": 1.7443,
      "step": 2748
    },
    {
      "epoch": 0.28997890295358647,
      "grad_norm": 0.5152401328086853,
      "learning_rate": 0.0015,
      "loss": 1.7468,
      "step": 2749
    },
    {
      "epoch": 0.290084388185654,
      "grad_norm": 0.5323880910873413,
      "learning_rate": 0.0015,
      "loss": 1.7283,
      "step": 2750
    },
    {
      "epoch": 0.2901898734177215,
      "grad_norm": 0.6154477596282959,
      "learning_rate": 0.0015,
      "loss": 1.7117,
      "step": 2751
    },
    {
      "epoch": 0.290295358649789,
      "grad_norm": 0.5759016275405884,
      "learning_rate": 0.0015,
      "loss": 1.7555,
      "step": 2752
    },
    {
      "epoch": 0.29040084388185655,
      "grad_norm": 0.6234085559844971,
      "learning_rate": 0.0015,
      "loss": 1.7218,
      "step": 2753
    },
    {
      "epoch": 0.29050632911392404,
      "grad_norm": 0.5297909379005432,
      "learning_rate": 0.0015,
      "loss": 1.7581,
      "step": 2754
    },
    {
      "epoch": 0.29061181434599154,
      "grad_norm": 0.5211373567581177,
      "learning_rate": 0.0015,
      "loss": 1.7784,
      "step": 2755
    },
    {
      "epoch": 0.2907172995780591,
      "grad_norm": 0.6669690012931824,
      "learning_rate": 0.0015,
      "loss": 1.7462,
      "step": 2756
    },
    {
      "epoch": 0.2908227848101266,
      "grad_norm": 0.47888511419296265,
      "learning_rate": 0.0015,
      "loss": 1.7186,
      "step": 2757
    },
    {
      "epoch": 0.29092827004219407,
      "grad_norm": 0.5764559507369995,
      "learning_rate": 0.0015,
      "loss": 1.7295,
      "step": 2758
    },
    {
      "epoch": 0.2910337552742616,
      "grad_norm": 0.4933163523674011,
      "learning_rate": 0.0015,
      "loss": 1.7892,
      "step": 2759
    },
    {
      "epoch": 0.2911392405063291,
      "grad_norm": 0.5667641162872314,
      "learning_rate": 0.0015,
      "loss": 1.7445,
      "step": 2760
    },
    {
      "epoch": 0.2912447257383966,
      "grad_norm": 0.537183940410614,
      "learning_rate": 0.0015,
      "loss": 1.7431,
      "step": 2761
    },
    {
      "epoch": 0.29135021097046415,
      "grad_norm": 0.5711116790771484,
      "learning_rate": 0.0015,
      "loss": 1.7607,
      "step": 2762
    },
    {
      "epoch": 0.29145569620253164,
      "grad_norm": 0.5116228461265564,
      "learning_rate": 0.0015,
      "loss": 1.763,
      "step": 2763
    },
    {
      "epoch": 0.29156118143459914,
      "grad_norm": 0.5435190796852112,
      "learning_rate": 0.0015,
      "loss": 1.6975,
      "step": 2764
    },
    {
      "epoch": 0.2916666666666667,
      "grad_norm": 0.6140767931938171,
      "learning_rate": 0.0015,
      "loss": 1.7492,
      "step": 2765
    },
    {
      "epoch": 0.2917721518987342,
      "grad_norm": 0.7195400595664978,
      "learning_rate": 0.0015,
      "loss": 1.7246,
      "step": 2766
    },
    {
      "epoch": 0.29187763713080167,
      "grad_norm": 0.4888113737106323,
      "learning_rate": 0.0015,
      "loss": 1.7272,
      "step": 2767
    },
    {
      "epoch": 0.2919831223628692,
      "grad_norm": 0.5987370014190674,
      "learning_rate": 0.0015,
      "loss": 1.7215,
      "step": 2768
    },
    {
      "epoch": 0.2920886075949367,
      "grad_norm": 0.6660099625587463,
      "learning_rate": 0.0015,
      "loss": 1.7405,
      "step": 2769
    },
    {
      "epoch": 0.2921940928270042,
      "grad_norm": 0.5677423477172852,
      "learning_rate": 0.0015,
      "loss": 1.7637,
      "step": 2770
    },
    {
      "epoch": 0.29229957805907175,
      "grad_norm": 0.5299758911132812,
      "learning_rate": 0.0015,
      "loss": 1.7331,
      "step": 2771
    },
    {
      "epoch": 0.29240506329113924,
      "grad_norm": 0.5895119309425354,
      "learning_rate": 0.0015,
      "loss": 1.7466,
      "step": 2772
    },
    {
      "epoch": 0.29251054852320674,
      "grad_norm": 0.673026978969574,
      "learning_rate": 0.0015,
      "loss": 1.7351,
      "step": 2773
    },
    {
      "epoch": 0.2926160337552743,
      "grad_norm": 0.48601093888282776,
      "learning_rate": 0.0015,
      "loss": 1.7307,
      "step": 2774
    },
    {
      "epoch": 0.2927215189873418,
      "grad_norm": 0.6198064684867859,
      "learning_rate": 0.0015,
      "loss": 1.7285,
      "step": 2775
    },
    {
      "epoch": 0.29282700421940927,
      "grad_norm": 0.6173463463783264,
      "learning_rate": 0.0015,
      "loss": 1.7415,
      "step": 2776
    },
    {
      "epoch": 0.2929324894514768,
      "grad_norm": 0.5128932595252991,
      "learning_rate": 0.0015,
      "loss": 1.7547,
      "step": 2777
    },
    {
      "epoch": 0.2930379746835443,
      "grad_norm": 0.4993072748184204,
      "learning_rate": 0.0015,
      "loss": 1.7219,
      "step": 2778
    },
    {
      "epoch": 0.2931434599156118,
      "grad_norm": 0.5432206988334656,
      "learning_rate": 0.0015,
      "loss": 1.7338,
      "step": 2779
    },
    {
      "epoch": 0.29324894514767935,
      "grad_norm": 0.45908355712890625,
      "learning_rate": 0.0015,
      "loss": 1.7145,
      "step": 2780
    },
    {
      "epoch": 0.29335443037974684,
      "grad_norm": 0.6190879940986633,
      "learning_rate": 0.0015,
      "loss": 1.7317,
      "step": 2781
    },
    {
      "epoch": 0.29345991561181434,
      "grad_norm": 0.5287309885025024,
      "learning_rate": 0.0015,
      "loss": 1.7292,
      "step": 2782
    },
    {
      "epoch": 0.29356540084388183,
      "grad_norm": 0.5456938743591309,
      "learning_rate": 0.0015,
      "loss": 1.7226,
      "step": 2783
    },
    {
      "epoch": 0.2936708860759494,
      "grad_norm": 0.6921320557594299,
      "learning_rate": 0.0015,
      "loss": 1.7396,
      "step": 2784
    },
    {
      "epoch": 0.29377637130801687,
      "grad_norm": 0.6134920120239258,
      "learning_rate": 0.0015,
      "loss": 1.7344,
      "step": 2785
    },
    {
      "epoch": 0.29388185654008436,
      "grad_norm": 0.5075782537460327,
      "learning_rate": 0.0015,
      "loss": 1.6991,
      "step": 2786
    },
    {
      "epoch": 0.2939873417721519,
      "grad_norm": 0.6686756610870361,
      "learning_rate": 0.0015,
      "loss": 1.7203,
      "step": 2787
    },
    {
      "epoch": 0.2940928270042194,
      "grad_norm": 0.6644595265388489,
      "learning_rate": 0.0015,
      "loss": 1.7523,
      "step": 2788
    },
    {
      "epoch": 0.2941983122362869,
      "grad_norm": 0.6102534532546997,
      "learning_rate": 0.0015,
      "loss": 1.7682,
      "step": 2789
    },
    {
      "epoch": 0.29430379746835444,
      "grad_norm": 0.5273255705833435,
      "learning_rate": 0.0015,
      "loss": 1.739,
      "step": 2790
    },
    {
      "epoch": 0.29440928270042194,
      "grad_norm": 0.6021966934204102,
      "learning_rate": 0.0015,
      "loss": 1.7521,
      "step": 2791
    },
    {
      "epoch": 0.29451476793248943,
      "grad_norm": 0.5897009968757629,
      "learning_rate": 0.0015,
      "loss": 1.6936,
      "step": 2792
    },
    {
      "epoch": 0.294620253164557,
      "grad_norm": 0.4723966419696808,
      "learning_rate": 0.0015,
      "loss": 1.7393,
      "step": 2793
    },
    {
      "epoch": 0.29472573839662447,
      "grad_norm": 0.6469066143035889,
      "learning_rate": 0.0015,
      "loss": 1.745,
      "step": 2794
    },
    {
      "epoch": 0.29483122362869196,
      "grad_norm": 0.6034923195838928,
      "learning_rate": 0.0015,
      "loss": 1.7117,
      "step": 2795
    },
    {
      "epoch": 0.2949367088607595,
      "grad_norm": 0.6067705750465393,
      "learning_rate": 0.0015,
      "loss": 1.7498,
      "step": 2796
    },
    {
      "epoch": 0.295042194092827,
      "grad_norm": 0.5731272101402283,
      "learning_rate": 0.0015,
      "loss": 1.7323,
      "step": 2797
    },
    {
      "epoch": 0.2951476793248945,
      "grad_norm": 0.5448641180992126,
      "learning_rate": 0.0015,
      "loss": 1.7512,
      "step": 2798
    },
    {
      "epoch": 0.29525316455696204,
      "grad_norm": 0.7797780632972717,
      "learning_rate": 0.0015,
      "loss": 1.7388,
      "step": 2799
    },
    {
      "epoch": 0.29535864978902954,
      "grad_norm": 0.6417586803436279,
      "learning_rate": 0.0015,
      "loss": 1.7027,
      "step": 2800
    },
    {
      "epoch": 0.29546413502109703,
      "grad_norm": 0.6263667941093445,
      "learning_rate": 0.0015,
      "loss": 1.7426,
      "step": 2801
    },
    {
      "epoch": 0.2955696202531646,
      "grad_norm": 0.738879382610321,
      "learning_rate": 0.0015,
      "loss": 1.7164,
      "step": 2802
    },
    {
      "epoch": 0.29567510548523207,
      "grad_norm": 0.5009618997573853,
      "learning_rate": 0.0015,
      "loss": 1.7453,
      "step": 2803
    },
    {
      "epoch": 0.29578059071729956,
      "grad_norm": 0.6718484163284302,
      "learning_rate": 0.0015,
      "loss": 1.7328,
      "step": 2804
    },
    {
      "epoch": 0.2958860759493671,
      "grad_norm": 0.6552501916885376,
      "learning_rate": 0.0015,
      "loss": 1.7161,
      "step": 2805
    },
    {
      "epoch": 0.2959915611814346,
      "grad_norm": 0.5674899816513062,
      "learning_rate": 0.0015,
      "loss": 1.7329,
      "step": 2806
    },
    {
      "epoch": 0.2960970464135021,
      "grad_norm": 0.7102161049842834,
      "learning_rate": 0.0015,
      "loss": 1.7392,
      "step": 2807
    },
    {
      "epoch": 0.29620253164556964,
      "grad_norm": 0.6125746965408325,
      "learning_rate": 0.0015,
      "loss": 1.7171,
      "step": 2808
    },
    {
      "epoch": 0.29630801687763714,
      "grad_norm": 0.5231237411499023,
      "learning_rate": 0.0015,
      "loss": 1.7608,
      "step": 2809
    },
    {
      "epoch": 0.29641350210970463,
      "grad_norm": 0.5651828050613403,
      "learning_rate": 0.0015,
      "loss": 1.7352,
      "step": 2810
    },
    {
      "epoch": 0.2965189873417722,
      "grad_norm": 0.6057689189910889,
      "learning_rate": 0.0015,
      "loss": 1.7257,
      "step": 2811
    },
    {
      "epoch": 0.29662447257383967,
      "grad_norm": 0.46668606996536255,
      "learning_rate": 0.0015,
      "loss": 1.7183,
      "step": 2812
    },
    {
      "epoch": 0.29672995780590716,
      "grad_norm": 0.5443451404571533,
      "learning_rate": 0.0015,
      "loss": 1.7253,
      "step": 2813
    },
    {
      "epoch": 0.2968354430379747,
      "grad_norm": 0.5878294110298157,
      "learning_rate": 0.0015,
      "loss": 1.7629,
      "step": 2814
    },
    {
      "epoch": 0.2969409282700422,
      "grad_norm": 0.6273327469825745,
      "learning_rate": 0.0015,
      "loss": 1.7293,
      "step": 2815
    },
    {
      "epoch": 0.2970464135021097,
      "grad_norm": 0.5364030003547668,
      "learning_rate": 0.0015,
      "loss": 1.6985,
      "step": 2816
    },
    {
      "epoch": 0.2971518987341772,
      "grad_norm": 0.6072271466255188,
      "learning_rate": 0.0015,
      "loss": 1.7349,
      "step": 2817
    },
    {
      "epoch": 0.29725738396624474,
      "grad_norm": 0.5207487344741821,
      "learning_rate": 0.0015,
      "loss": 1.7655,
      "step": 2818
    },
    {
      "epoch": 0.29736286919831223,
      "grad_norm": 0.5988492965698242,
      "learning_rate": 0.0015,
      "loss": 1.7504,
      "step": 2819
    },
    {
      "epoch": 0.2974683544303797,
      "grad_norm": 0.7192591428756714,
      "learning_rate": 0.0015,
      "loss": 1.7633,
      "step": 2820
    },
    {
      "epoch": 0.29757383966244727,
      "grad_norm": 0.5050835013389587,
      "learning_rate": 0.0015,
      "loss": 1.7105,
      "step": 2821
    },
    {
      "epoch": 0.29767932489451476,
      "grad_norm": 0.6090943813323975,
      "learning_rate": 0.0015,
      "loss": 1.7095,
      "step": 2822
    },
    {
      "epoch": 0.29778481012658226,
      "grad_norm": 0.6430670022964478,
      "learning_rate": 0.0015,
      "loss": 1.7344,
      "step": 2823
    },
    {
      "epoch": 0.2978902953586498,
      "grad_norm": 0.5029327869415283,
      "learning_rate": 0.0015,
      "loss": 1.7001,
      "step": 2824
    },
    {
      "epoch": 0.2979957805907173,
      "grad_norm": 0.7795506715774536,
      "learning_rate": 0.0015,
      "loss": 1.7041,
      "step": 2825
    },
    {
      "epoch": 0.2981012658227848,
      "grad_norm": 0.5343037247657776,
      "learning_rate": 0.0015,
      "loss": 1.7147,
      "step": 2826
    },
    {
      "epoch": 0.29820675105485234,
      "grad_norm": 0.8471195101737976,
      "learning_rate": 0.0015,
      "loss": 1.6798,
      "step": 2827
    },
    {
      "epoch": 0.29831223628691983,
      "grad_norm": 1.0902858972549438,
      "learning_rate": 0.0015,
      "loss": 1.7153,
      "step": 2828
    },
    {
      "epoch": 0.2984177215189873,
      "grad_norm": 0.5062287449836731,
      "learning_rate": 0.0015,
      "loss": 1.7136,
      "step": 2829
    },
    {
      "epoch": 0.29852320675105487,
      "grad_norm": 1.2234694957733154,
      "learning_rate": 0.0015,
      "loss": 1.7266,
      "step": 2830
    },
    {
      "epoch": 0.29862869198312236,
      "grad_norm": 0.6013873815536499,
      "learning_rate": 0.0015,
      "loss": 1.7032,
      "step": 2831
    },
    {
      "epoch": 0.29873417721518986,
      "grad_norm": 0.7629626393318176,
      "learning_rate": 0.0015,
      "loss": 1.7367,
      "step": 2832
    },
    {
      "epoch": 0.2988396624472574,
      "grad_norm": 0.7414044141769409,
      "learning_rate": 0.0015,
      "loss": 1.708,
      "step": 2833
    },
    {
      "epoch": 0.2989451476793249,
      "grad_norm": 0.5013367533683777,
      "learning_rate": 0.0015,
      "loss": 1.7726,
      "step": 2834
    },
    {
      "epoch": 0.2990506329113924,
      "grad_norm": 0.9379156231880188,
      "learning_rate": 0.0015,
      "loss": 1.7,
      "step": 2835
    },
    {
      "epoch": 0.29915611814345994,
      "grad_norm": 0.7603035569190979,
      "learning_rate": 0.0015,
      "loss": 1.6982,
      "step": 2836
    },
    {
      "epoch": 0.29926160337552743,
      "grad_norm": 0.6432632207870483,
      "learning_rate": 0.0015,
      "loss": 1.7146,
      "step": 2837
    },
    {
      "epoch": 0.2993670886075949,
      "grad_norm": 0.9125514030456543,
      "learning_rate": 0.0015,
      "loss": 1.6937,
      "step": 2838
    },
    {
      "epoch": 0.29947257383966247,
      "grad_norm": 0.4466330409049988,
      "learning_rate": 0.0015,
      "loss": 1.7059,
      "step": 2839
    },
    {
      "epoch": 0.29957805907172996,
      "grad_norm": 0.8298995494842529,
      "learning_rate": 0.0015,
      "loss": 1.7424,
      "step": 2840
    },
    {
      "epoch": 0.29968354430379746,
      "grad_norm": 0.7232890725135803,
      "learning_rate": 0.0015,
      "loss": 1.703,
      "step": 2841
    },
    {
      "epoch": 0.299789029535865,
      "grad_norm": 0.575537383556366,
      "learning_rate": 0.0015,
      "loss": 1.7472,
      "step": 2842
    },
    {
      "epoch": 0.2998945147679325,
      "grad_norm": 0.9341709613800049,
      "learning_rate": 0.0015,
      "loss": 1.7129,
      "step": 2843
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.513477623462677,
      "learning_rate": 0.0015,
      "loss": 1.7432,
      "step": 2844
    },
    {
      "epoch": 0.30010548523206754,
      "grad_norm": 0.7083409428596497,
      "learning_rate": 0.0015,
      "loss": 1.7367,
      "step": 2845
    },
    {
      "epoch": 0.30021097046413503,
      "grad_norm": 0.7591075301170349,
      "learning_rate": 0.0015,
      "loss": 1.7368,
      "step": 2846
    },
    {
      "epoch": 0.3003164556962025,
      "grad_norm": 0.494125097990036,
      "learning_rate": 0.0015,
      "loss": 1.7292,
      "step": 2847
    },
    {
      "epoch": 0.30042194092827,
      "grad_norm": 0.6274526715278625,
      "learning_rate": 0.0015,
      "loss": 1.6976,
      "step": 2848
    },
    {
      "epoch": 0.30052742616033756,
      "grad_norm": 0.6833388805389404,
      "learning_rate": 0.0015,
      "loss": 1.7105,
      "step": 2849
    },
    {
      "epoch": 0.30063291139240506,
      "grad_norm": 0.5163332223892212,
      "learning_rate": 0.0015,
      "loss": 1.7262,
      "step": 2850
    },
    {
      "epoch": 0.30073839662447255,
      "grad_norm": 0.7678579688072205,
      "learning_rate": 0.0015,
      "loss": 1.6895,
      "step": 2851
    },
    {
      "epoch": 0.3008438818565401,
      "grad_norm": 0.681102991104126,
      "learning_rate": 0.0015,
      "loss": 1.7337,
      "step": 2852
    },
    {
      "epoch": 0.3009493670886076,
      "grad_norm": 0.4950621724128723,
      "learning_rate": 0.0015,
      "loss": 1.7244,
      "step": 2853
    },
    {
      "epoch": 0.3010548523206751,
      "grad_norm": 0.6309622526168823,
      "learning_rate": 0.0015,
      "loss": 1.7224,
      "step": 2854
    },
    {
      "epoch": 0.30116033755274263,
      "grad_norm": 0.49587345123291016,
      "learning_rate": 0.0015,
      "loss": 1.7716,
      "step": 2855
    },
    {
      "epoch": 0.3012658227848101,
      "grad_norm": 0.6604416966438293,
      "learning_rate": 0.0015,
      "loss": 1.7472,
      "step": 2856
    },
    {
      "epoch": 0.3013713080168776,
      "grad_norm": 0.6719710826873779,
      "learning_rate": 0.0015,
      "loss": 1.7316,
      "step": 2857
    },
    {
      "epoch": 0.30147679324894516,
      "grad_norm": 0.5016152262687683,
      "learning_rate": 0.0015,
      "loss": 1.7132,
      "step": 2858
    },
    {
      "epoch": 0.30158227848101266,
      "grad_norm": 0.8282347321510315,
      "learning_rate": 0.0015,
      "loss": 1.705,
      "step": 2859
    },
    {
      "epoch": 0.30168776371308015,
      "grad_norm": 0.5938814878463745,
      "learning_rate": 0.0015,
      "loss": 1.7257,
      "step": 2860
    },
    {
      "epoch": 0.3017932489451477,
      "grad_norm": 0.5885493755340576,
      "learning_rate": 0.0015,
      "loss": 1.7371,
      "step": 2861
    },
    {
      "epoch": 0.3018987341772152,
      "grad_norm": 0.8027797937393188,
      "learning_rate": 0.0015,
      "loss": 1.7325,
      "step": 2862
    },
    {
      "epoch": 0.3020042194092827,
      "grad_norm": 0.4907360374927521,
      "learning_rate": 0.0015,
      "loss": 1.7129,
      "step": 2863
    },
    {
      "epoch": 0.30210970464135023,
      "grad_norm": 0.6164902448654175,
      "learning_rate": 0.0015,
      "loss": 1.6973,
      "step": 2864
    },
    {
      "epoch": 0.3022151898734177,
      "grad_norm": 0.6104419231414795,
      "learning_rate": 0.0015,
      "loss": 1.7314,
      "step": 2865
    },
    {
      "epoch": 0.3023206751054852,
      "grad_norm": 0.5501196384429932,
      "learning_rate": 0.0015,
      "loss": 1.7788,
      "step": 2866
    },
    {
      "epoch": 0.30242616033755276,
      "grad_norm": 0.7411067485809326,
      "learning_rate": 0.0015,
      "loss": 1.7342,
      "step": 2867
    },
    {
      "epoch": 0.30253164556962026,
      "grad_norm": 0.48683708906173706,
      "learning_rate": 0.0015,
      "loss": 1.7198,
      "step": 2868
    },
    {
      "epoch": 0.30263713080168775,
      "grad_norm": 0.6525079011917114,
      "learning_rate": 0.0015,
      "loss": 1.6716,
      "step": 2869
    },
    {
      "epoch": 0.3027426160337553,
      "grad_norm": 0.6791862845420837,
      "learning_rate": 0.0015,
      "loss": 1.7684,
      "step": 2870
    },
    {
      "epoch": 0.3028481012658228,
      "grad_norm": 0.4720754623413086,
      "learning_rate": 0.0015,
      "loss": 1.7144,
      "step": 2871
    },
    {
      "epoch": 0.3029535864978903,
      "grad_norm": 0.6469252109527588,
      "learning_rate": 0.0015,
      "loss": 1.726,
      "step": 2872
    },
    {
      "epoch": 0.30305907172995783,
      "grad_norm": 0.4780896306037903,
      "learning_rate": 0.0015,
      "loss": 1.7315,
      "step": 2873
    },
    {
      "epoch": 0.3031645569620253,
      "grad_norm": 0.737733006477356,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 2874
    },
    {
      "epoch": 0.3032700421940928,
      "grad_norm": 0.5257196426391602,
      "learning_rate": 0.0015,
      "loss": 1.7096,
      "step": 2875
    },
    {
      "epoch": 0.30337552742616036,
      "grad_norm": 0.6989250779151917,
      "learning_rate": 0.0015,
      "loss": 1.7032,
      "step": 2876
    },
    {
      "epoch": 0.30348101265822786,
      "grad_norm": 0.6388779878616333,
      "learning_rate": 0.0015,
      "loss": 1.708,
      "step": 2877
    },
    {
      "epoch": 0.30358649789029535,
      "grad_norm": 0.6642478704452515,
      "learning_rate": 0.0015,
      "loss": 1.7282,
      "step": 2878
    },
    {
      "epoch": 0.3036919831223629,
      "grad_norm": 0.8404462933540344,
      "learning_rate": 0.0015,
      "loss": 1.7249,
      "step": 2879
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 0.6265426278114319,
      "learning_rate": 0.0015,
      "loss": 1.7113,
      "step": 2880
    },
    {
      "epoch": 0.3039029535864979,
      "grad_norm": 0.5950288772583008,
      "learning_rate": 0.0015,
      "loss": 1.7172,
      "step": 2881
    },
    {
      "epoch": 0.3040084388185654,
      "grad_norm": 0.619999349117279,
      "learning_rate": 0.0015,
      "loss": 1.7168,
      "step": 2882
    },
    {
      "epoch": 0.3041139240506329,
      "grad_norm": 0.5390220284461975,
      "learning_rate": 0.0015,
      "loss": 1.68,
      "step": 2883
    },
    {
      "epoch": 0.3042194092827004,
      "grad_norm": 0.5932281613349915,
      "learning_rate": 0.0015,
      "loss": 1.7146,
      "step": 2884
    },
    {
      "epoch": 0.3043248945147679,
      "grad_norm": 0.5918823480606079,
      "learning_rate": 0.0015,
      "loss": 1.7275,
      "step": 2885
    },
    {
      "epoch": 0.30443037974683546,
      "grad_norm": 0.5813516974449158,
      "learning_rate": 0.0015,
      "loss": 1.7332,
      "step": 2886
    },
    {
      "epoch": 0.30453586497890295,
      "grad_norm": 0.6132948398590088,
      "learning_rate": 0.0015,
      "loss": 1.7404,
      "step": 2887
    },
    {
      "epoch": 0.30464135021097044,
      "grad_norm": 0.5324879288673401,
      "learning_rate": 0.0015,
      "loss": 1.6782,
      "step": 2888
    },
    {
      "epoch": 0.304746835443038,
      "grad_norm": 0.7246313095092773,
      "learning_rate": 0.0015,
      "loss": 1.7571,
      "step": 2889
    },
    {
      "epoch": 0.3048523206751055,
      "grad_norm": 0.5872277617454529,
      "learning_rate": 0.0015,
      "loss": 1.7026,
      "step": 2890
    },
    {
      "epoch": 0.304957805907173,
      "grad_norm": 0.48824891448020935,
      "learning_rate": 0.0015,
      "loss": 1.7329,
      "step": 2891
    },
    {
      "epoch": 0.3050632911392405,
      "grad_norm": 0.4681529402732849,
      "learning_rate": 0.0015,
      "loss": 1.7251,
      "step": 2892
    },
    {
      "epoch": 0.305168776371308,
      "grad_norm": 0.4926391839981079,
      "learning_rate": 0.0015,
      "loss": 1.7054,
      "step": 2893
    },
    {
      "epoch": 0.3052742616033755,
      "grad_norm": 0.5023272633552551,
      "learning_rate": 0.0015,
      "loss": 1.7257,
      "step": 2894
    },
    {
      "epoch": 0.30537974683544306,
      "grad_norm": 0.4859912693500519,
      "learning_rate": 0.0015,
      "loss": 1.715,
      "step": 2895
    },
    {
      "epoch": 0.30548523206751055,
      "grad_norm": 0.5448835492134094,
      "learning_rate": 0.0015,
      "loss": 1.7207,
      "step": 2896
    },
    {
      "epoch": 0.30559071729957804,
      "grad_norm": 0.5105636715888977,
      "learning_rate": 0.0015,
      "loss": 1.7434,
      "step": 2897
    },
    {
      "epoch": 0.3056962025316456,
      "grad_norm": 0.6418217420578003,
      "learning_rate": 0.0015,
      "loss": 1.7409,
      "step": 2898
    },
    {
      "epoch": 0.3058016877637131,
      "grad_norm": 0.6759374737739563,
      "learning_rate": 0.0015,
      "loss": 1.7238,
      "step": 2899
    },
    {
      "epoch": 0.3059071729957806,
      "grad_norm": 0.629976749420166,
      "learning_rate": 0.0015,
      "loss": 1.7007,
      "step": 2900
    },
    {
      "epoch": 0.3060126582278481,
      "grad_norm": 0.7432752847671509,
      "learning_rate": 0.0015,
      "loss": 1.7249,
      "step": 2901
    },
    {
      "epoch": 0.3061181434599156,
      "grad_norm": 0.5830813050270081,
      "learning_rate": 0.0015,
      "loss": 1.7463,
      "step": 2902
    },
    {
      "epoch": 0.3062236286919831,
      "grad_norm": 0.6720617413520813,
      "learning_rate": 0.0015,
      "loss": 1.7277,
      "step": 2903
    },
    {
      "epoch": 0.30632911392405066,
      "grad_norm": 0.7102701663970947,
      "learning_rate": 0.0015,
      "loss": 1.7351,
      "step": 2904
    },
    {
      "epoch": 0.30643459915611815,
      "grad_norm": 0.5022579431533813,
      "learning_rate": 0.0015,
      "loss": 1.7315,
      "step": 2905
    },
    {
      "epoch": 0.30654008438818564,
      "grad_norm": 0.5378018021583557,
      "learning_rate": 0.0015,
      "loss": 1.7517,
      "step": 2906
    },
    {
      "epoch": 0.3066455696202532,
      "grad_norm": 0.5608876943588257,
      "learning_rate": 0.0015,
      "loss": 1.6931,
      "step": 2907
    },
    {
      "epoch": 0.3067510548523207,
      "grad_norm": 0.4984451234340668,
      "learning_rate": 0.0015,
      "loss": 1.719,
      "step": 2908
    },
    {
      "epoch": 0.3068565400843882,
      "grad_norm": 0.5655184984207153,
      "learning_rate": 0.0015,
      "loss": 1.7065,
      "step": 2909
    },
    {
      "epoch": 0.3069620253164557,
      "grad_norm": 0.6536985039710999,
      "learning_rate": 0.0015,
      "loss": 1.6989,
      "step": 2910
    },
    {
      "epoch": 0.3070675105485232,
      "grad_norm": 0.459410160779953,
      "learning_rate": 0.0015,
      "loss": 1.6869,
      "step": 2911
    },
    {
      "epoch": 0.3071729957805907,
      "grad_norm": 0.6984475255012512,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 2912
    },
    {
      "epoch": 0.30727848101265826,
      "grad_norm": 0.6061699986457825,
      "learning_rate": 0.0015,
      "loss": 1.6939,
      "step": 2913
    },
    {
      "epoch": 0.30738396624472575,
      "grad_norm": 0.5774768590927124,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 2914
    },
    {
      "epoch": 0.30748945147679324,
      "grad_norm": 0.8641024231910706,
      "learning_rate": 0.0015,
      "loss": 1.7554,
      "step": 2915
    },
    {
      "epoch": 0.30759493670886073,
      "grad_norm": 0.5568138957023621,
      "learning_rate": 0.0015,
      "loss": 1.7343,
      "step": 2916
    },
    {
      "epoch": 0.3077004219409283,
      "grad_norm": 0.7313710451126099,
      "learning_rate": 0.0015,
      "loss": 1.6874,
      "step": 2917
    },
    {
      "epoch": 0.3078059071729958,
      "grad_norm": 0.853188693523407,
      "learning_rate": 0.0015,
      "loss": 1.7139,
      "step": 2918
    },
    {
      "epoch": 0.30791139240506327,
      "grad_norm": 0.49975845217704773,
      "learning_rate": 0.0015,
      "loss": 1.7565,
      "step": 2919
    },
    {
      "epoch": 0.3080168776371308,
      "grad_norm": 0.656856894493103,
      "learning_rate": 0.0015,
      "loss": 1.7059,
      "step": 2920
    },
    {
      "epoch": 0.3081223628691983,
      "grad_norm": 0.6693185567855835,
      "learning_rate": 0.0015,
      "loss": 1.679,
      "step": 2921
    },
    {
      "epoch": 0.3082278481012658,
      "grad_norm": 0.4762989580631256,
      "learning_rate": 0.0015,
      "loss": 1.7177,
      "step": 2922
    },
    {
      "epoch": 0.30833333333333335,
      "grad_norm": 0.489740252494812,
      "learning_rate": 0.0015,
      "loss": 1.7347,
      "step": 2923
    },
    {
      "epoch": 0.30843881856540084,
      "grad_norm": 0.5571384429931641,
      "learning_rate": 0.0015,
      "loss": 1.7074,
      "step": 2924
    },
    {
      "epoch": 0.30854430379746833,
      "grad_norm": 0.5981157422065735,
      "learning_rate": 0.0015,
      "loss": 1.7049,
      "step": 2925
    },
    {
      "epoch": 0.3086497890295359,
      "grad_norm": 0.5322088599205017,
      "learning_rate": 0.0015,
      "loss": 1.718,
      "step": 2926
    },
    {
      "epoch": 0.3087552742616034,
      "grad_norm": 0.4946162700653076,
      "learning_rate": 0.0015,
      "loss": 1.7362,
      "step": 2927
    },
    {
      "epoch": 0.30886075949367087,
      "grad_norm": 0.5483956336975098,
      "learning_rate": 0.0015,
      "loss": 1.7498,
      "step": 2928
    },
    {
      "epoch": 0.3089662447257384,
      "grad_norm": 0.574830174446106,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 2929
    },
    {
      "epoch": 0.3090717299578059,
      "grad_norm": 0.4502004384994507,
      "learning_rate": 0.0015,
      "loss": 1.6995,
      "step": 2930
    },
    {
      "epoch": 0.3091772151898734,
      "grad_norm": 0.5166926980018616,
      "learning_rate": 0.0015,
      "loss": 1.7415,
      "step": 2931
    },
    {
      "epoch": 0.30928270042194095,
      "grad_norm": 0.4438752830028534,
      "learning_rate": 0.0015,
      "loss": 1.7599,
      "step": 2932
    },
    {
      "epoch": 0.30938818565400844,
      "grad_norm": 0.5299575328826904,
      "learning_rate": 0.0015,
      "loss": 1.7177,
      "step": 2933
    },
    {
      "epoch": 0.30949367088607593,
      "grad_norm": 0.4622742831707001,
      "learning_rate": 0.0015,
      "loss": 1.7428,
      "step": 2934
    },
    {
      "epoch": 0.3095991561181435,
      "grad_norm": 0.5356643795967102,
      "learning_rate": 0.0015,
      "loss": 1.7258,
      "step": 2935
    },
    {
      "epoch": 0.309704641350211,
      "grad_norm": 0.6198638677597046,
      "learning_rate": 0.0015,
      "loss": 1.762,
      "step": 2936
    },
    {
      "epoch": 0.30981012658227847,
      "grad_norm": 0.4738200008869171,
      "learning_rate": 0.0015,
      "loss": 1.6758,
      "step": 2937
    },
    {
      "epoch": 0.309915611814346,
      "grad_norm": 0.5630542039871216,
      "learning_rate": 0.0015,
      "loss": 1.7026,
      "step": 2938
    },
    {
      "epoch": 0.3100210970464135,
      "grad_norm": 0.6402138471603394,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 2939
    },
    {
      "epoch": 0.310126582278481,
      "grad_norm": 0.6138157844543457,
      "learning_rate": 0.0015,
      "loss": 1.7062,
      "step": 2940
    },
    {
      "epoch": 0.31023206751054855,
      "grad_norm": 0.5457233190536499,
      "learning_rate": 0.0015,
      "loss": 1.7429,
      "step": 2941
    },
    {
      "epoch": 0.31033755274261604,
      "grad_norm": 0.5193164348602295,
      "learning_rate": 0.0015,
      "loss": 1.7104,
      "step": 2942
    },
    {
      "epoch": 0.31044303797468353,
      "grad_norm": 0.5176413059234619,
      "learning_rate": 0.0015,
      "loss": 1.7357,
      "step": 2943
    },
    {
      "epoch": 0.3105485232067511,
      "grad_norm": 0.5201972723007202,
      "learning_rate": 0.0015,
      "loss": 1.7581,
      "step": 2944
    },
    {
      "epoch": 0.3106540084388186,
      "grad_norm": 0.5292931795120239,
      "learning_rate": 0.0015,
      "loss": 1.7433,
      "step": 2945
    },
    {
      "epoch": 0.31075949367088607,
      "grad_norm": 0.5314915776252747,
      "learning_rate": 0.0015,
      "loss": 1.7061,
      "step": 2946
    },
    {
      "epoch": 0.31086497890295356,
      "grad_norm": 0.5484752655029297,
      "learning_rate": 0.0015,
      "loss": 1.7291,
      "step": 2947
    },
    {
      "epoch": 0.3109704641350211,
      "grad_norm": 0.47969773411750793,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 2948
    },
    {
      "epoch": 0.3110759493670886,
      "grad_norm": 0.528062641620636,
      "learning_rate": 0.0015,
      "loss": 1.7333,
      "step": 2949
    },
    {
      "epoch": 0.3111814345991561,
      "grad_norm": 0.4798565208911896,
      "learning_rate": 0.0015,
      "loss": 1.6886,
      "step": 2950
    },
    {
      "epoch": 0.31128691983122364,
      "grad_norm": 0.5226189494132996,
      "learning_rate": 0.0015,
      "loss": 1.6907,
      "step": 2951
    },
    {
      "epoch": 0.31139240506329113,
      "grad_norm": 0.698269248008728,
      "learning_rate": 0.0015,
      "loss": 1.7147,
      "step": 2952
    },
    {
      "epoch": 0.3114978902953586,
      "grad_norm": 0.7282271981239319,
      "learning_rate": 0.0015,
      "loss": 1.726,
      "step": 2953
    },
    {
      "epoch": 0.3116033755274262,
      "grad_norm": 0.4884694516658783,
      "learning_rate": 0.0015,
      "loss": 1.6855,
      "step": 2954
    },
    {
      "epoch": 0.31170886075949367,
      "grad_norm": 0.8627981543540955,
      "learning_rate": 0.0015,
      "loss": 1.7064,
      "step": 2955
    },
    {
      "epoch": 0.31181434599156116,
      "grad_norm": 0.6796111464500427,
      "learning_rate": 0.0015,
      "loss": 1.7161,
      "step": 2956
    },
    {
      "epoch": 0.3119198312236287,
      "grad_norm": 0.7278868556022644,
      "learning_rate": 0.0015,
      "loss": 1.732,
      "step": 2957
    },
    {
      "epoch": 0.3120253164556962,
      "grad_norm": 0.7733821272850037,
      "learning_rate": 0.0015,
      "loss": 1.6823,
      "step": 2958
    },
    {
      "epoch": 0.3121308016877637,
      "grad_norm": 0.4810079038143158,
      "learning_rate": 0.0015,
      "loss": 1.7187,
      "step": 2959
    },
    {
      "epoch": 0.31223628691983124,
      "grad_norm": 0.7000905275344849,
      "learning_rate": 0.0015,
      "loss": 1.7062,
      "step": 2960
    },
    {
      "epoch": 0.31234177215189873,
      "grad_norm": 0.5330955982208252,
      "learning_rate": 0.0015,
      "loss": 1.6838,
      "step": 2961
    },
    {
      "epoch": 0.3124472573839662,
      "grad_norm": 0.5643135905265808,
      "learning_rate": 0.0015,
      "loss": 1.7133,
      "step": 2962
    },
    {
      "epoch": 0.3125527426160338,
      "grad_norm": 0.49947667121887207,
      "learning_rate": 0.0015,
      "loss": 1.7098,
      "step": 2963
    },
    {
      "epoch": 0.31265822784810127,
      "grad_norm": 0.508200466632843,
      "learning_rate": 0.0015,
      "loss": 1.6731,
      "step": 2964
    },
    {
      "epoch": 0.31276371308016876,
      "grad_norm": 0.721143364906311,
      "learning_rate": 0.0015,
      "loss": 1.7061,
      "step": 2965
    },
    {
      "epoch": 0.3128691983122363,
      "grad_norm": 0.4859545826911926,
      "learning_rate": 0.0015,
      "loss": 1.7129,
      "step": 2966
    },
    {
      "epoch": 0.3129746835443038,
      "grad_norm": 0.651206910610199,
      "learning_rate": 0.0015,
      "loss": 1.7326,
      "step": 2967
    },
    {
      "epoch": 0.3130801687763713,
      "grad_norm": 0.6910227537155151,
      "learning_rate": 0.0015,
      "loss": 1.7347,
      "step": 2968
    },
    {
      "epoch": 0.31318565400843884,
      "grad_norm": 0.5312160849571228,
      "learning_rate": 0.0015,
      "loss": 1.6943,
      "step": 2969
    },
    {
      "epoch": 0.31329113924050633,
      "grad_norm": 0.6909522414207458,
      "learning_rate": 0.0015,
      "loss": 1.7064,
      "step": 2970
    },
    {
      "epoch": 0.3133966244725738,
      "grad_norm": 0.527764618396759,
      "learning_rate": 0.0015,
      "loss": 1.6689,
      "step": 2971
    },
    {
      "epoch": 0.3135021097046414,
      "grad_norm": 0.5629932284355164,
      "learning_rate": 0.0015,
      "loss": 1.6881,
      "step": 2972
    },
    {
      "epoch": 0.31360759493670887,
      "grad_norm": 0.5080920457839966,
      "learning_rate": 0.0015,
      "loss": 1.7148,
      "step": 2973
    },
    {
      "epoch": 0.31371308016877636,
      "grad_norm": 0.5579750537872314,
      "learning_rate": 0.0015,
      "loss": 1.7264,
      "step": 2974
    },
    {
      "epoch": 0.3138185654008439,
      "grad_norm": 0.6935030817985535,
      "learning_rate": 0.0015,
      "loss": 1.6895,
      "step": 2975
    },
    {
      "epoch": 0.3139240506329114,
      "grad_norm": 0.47138023376464844,
      "learning_rate": 0.0015,
      "loss": 1.7101,
      "step": 2976
    },
    {
      "epoch": 0.3140295358649789,
      "grad_norm": 0.5409736037254333,
      "learning_rate": 0.0015,
      "loss": 1.7181,
      "step": 2977
    },
    {
      "epoch": 0.31413502109704644,
      "grad_norm": 0.5488565564155579,
      "learning_rate": 0.0015,
      "loss": 1.7199,
      "step": 2978
    },
    {
      "epoch": 0.31424050632911393,
      "grad_norm": 0.5884570479393005,
      "learning_rate": 0.0015,
      "loss": 1.7214,
      "step": 2979
    },
    {
      "epoch": 0.3143459915611814,
      "grad_norm": 0.5556483268737793,
      "learning_rate": 0.0015,
      "loss": 1.7449,
      "step": 2980
    },
    {
      "epoch": 0.3144514767932489,
      "grad_norm": 0.5683891177177429,
      "learning_rate": 0.0015,
      "loss": 1.7487,
      "step": 2981
    },
    {
      "epoch": 0.31455696202531647,
      "grad_norm": 0.6940754652023315,
      "learning_rate": 0.0015,
      "loss": 1.6898,
      "step": 2982
    },
    {
      "epoch": 0.31466244725738396,
      "grad_norm": 0.5430545806884766,
      "learning_rate": 0.0015,
      "loss": 1.7081,
      "step": 2983
    },
    {
      "epoch": 0.31476793248945145,
      "grad_norm": 0.5181131958961487,
      "learning_rate": 0.0015,
      "loss": 1.7108,
      "step": 2984
    },
    {
      "epoch": 0.314873417721519,
      "grad_norm": 0.5396592020988464,
      "learning_rate": 0.0015,
      "loss": 1.7025,
      "step": 2985
    },
    {
      "epoch": 0.3149789029535865,
      "grad_norm": 0.45588329434394836,
      "learning_rate": 0.0015,
      "loss": 1.7474,
      "step": 2986
    },
    {
      "epoch": 0.315084388185654,
      "grad_norm": 0.5171492099761963,
      "learning_rate": 0.0015,
      "loss": 1.7475,
      "step": 2987
    },
    {
      "epoch": 0.31518987341772153,
      "grad_norm": 0.4931403398513794,
      "learning_rate": 0.0015,
      "loss": 1.7322,
      "step": 2988
    },
    {
      "epoch": 0.315295358649789,
      "grad_norm": 0.5297608375549316,
      "learning_rate": 0.0015,
      "loss": 1.7583,
      "step": 2989
    },
    {
      "epoch": 0.3154008438818565,
      "grad_norm": 0.4950697422027588,
      "learning_rate": 0.0015,
      "loss": 1.7387,
      "step": 2990
    },
    {
      "epoch": 0.31550632911392407,
      "grad_norm": 0.5757505893707275,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 2991
    },
    {
      "epoch": 0.31561181434599156,
      "grad_norm": 0.5489315390586853,
      "learning_rate": 0.0015,
      "loss": 1.6818,
      "step": 2992
    },
    {
      "epoch": 0.31571729957805905,
      "grad_norm": 0.5464658141136169,
      "learning_rate": 0.0015,
      "loss": 1.6652,
      "step": 2993
    },
    {
      "epoch": 0.3158227848101266,
      "grad_norm": 0.5630892515182495,
      "learning_rate": 0.0015,
      "loss": 1.7377,
      "step": 2994
    },
    {
      "epoch": 0.3159282700421941,
      "grad_norm": 0.6165306568145752,
      "learning_rate": 0.0015,
      "loss": 1.7545,
      "step": 2995
    },
    {
      "epoch": 0.3160337552742616,
      "grad_norm": 0.5498163104057312,
      "learning_rate": 0.0015,
      "loss": 1.6951,
      "step": 2996
    },
    {
      "epoch": 0.31613924050632913,
      "grad_norm": 0.5323984622955322,
      "learning_rate": 0.0015,
      "loss": 1.6957,
      "step": 2997
    },
    {
      "epoch": 0.3162447257383966,
      "grad_norm": 0.5057029128074646,
      "learning_rate": 0.0015,
      "loss": 1.7063,
      "step": 2998
    },
    {
      "epoch": 0.3163502109704641,
      "grad_norm": 0.4992408752441406,
      "learning_rate": 0.0015,
      "loss": 1.7293,
      "step": 2999
    },
    {
      "epoch": 0.31645569620253167,
      "grad_norm": 0.4843266010284424,
      "learning_rate": 0.0015,
      "loss": 1.7005,
      "step": 3000
    },
    {
      "epoch": 0.31656118143459916,
      "grad_norm": 0.5039722323417664,
      "learning_rate": 0.0015,
      "loss": 1.7098,
      "step": 3001
    },
    {
      "epoch": 0.31666666666666665,
      "grad_norm": 0.5445347428321838,
      "learning_rate": 0.0015,
      "loss": 1.693,
      "step": 3002
    },
    {
      "epoch": 0.3167721518987342,
      "grad_norm": 0.6046627759933472,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 3003
    },
    {
      "epoch": 0.3168776371308017,
      "grad_norm": 0.6104879379272461,
      "learning_rate": 0.0015,
      "loss": 1.7645,
      "step": 3004
    },
    {
      "epoch": 0.3169831223628692,
      "grad_norm": 0.5497164726257324,
      "learning_rate": 0.0015,
      "loss": 1.7295,
      "step": 3005
    },
    {
      "epoch": 0.31708860759493673,
      "grad_norm": 0.5327602624893188,
      "learning_rate": 0.0015,
      "loss": 1.7384,
      "step": 3006
    },
    {
      "epoch": 0.3171940928270042,
      "grad_norm": 0.6222022175788879,
      "learning_rate": 0.0015,
      "loss": 1.7203,
      "step": 3007
    },
    {
      "epoch": 0.3172995780590717,
      "grad_norm": 0.5037438273429871,
      "learning_rate": 0.0015,
      "loss": 1.7107,
      "step": 3008
    },
    {
      "epoch": 0.31740506329113927,
      "grad_norm": 0.6280902028083801,
      "learning_rate": 0.0015,
      "loss": 1.702,
      "step": 3009
    },
    {
      "epoch": 0.31751054852320676,
      "grad_norm": 0.7413647770881653,
      "learning_rate": 0.0015,
      "loss": 1.7166,
      "step": 3010
    },
    {
      "epoch": 0.31761603375527425,
      "grad_norm": 0.4961790144443512,
      "learning_rate": 0.0015,
      "loss": 1.6976,
      "step": 3011
    },
    {
      "epoch": 0.31772151898734174,
      "grad_norm": 0.7486870884895325,
      "learning_rate": 0.0015,
      "loss": 1.7016,
      "step": 3012
    },
    {
      "epoch": 0.3178270042194093,
      "grad_norm": 0.7110238075256348,
      "learning_rate": 0.0015,
      "loss": 1.6876,
      "step": 3013
    },
    {
      "epoch": 0.3179324894514768,
      "grad_norm": 0.5200576186180115,
      "learning_rate": 0.0015,
      "loss": 1.7166,
      "step": 3014
    },
    {
      "epoch": 0.3180379746835443,
      "grad_norm": 0.8104680180549622,
      "learning_rate": 0.0015,
      "loss": 1.7449,
      "step": 3015
    },
    {
      "epoch": 0.3181434599156118,
      "grad_norm": 0.7071999311447144,
      "learning_rate": 0.0015,
      "loss": 1.7138,
      "step": 3016
    },
    {
      "epoch": 0.3182489451476793,
      "grad_norm": 0.5642274022102356,
      "learning_rate": 0.0015,
      "loss": 1.7436,
      "step": 3017
    },
    {
      "epoch": 0.3183544303797468,
      "grad_norm": 0.8863946795463562,
      "learning_rate": 0.0015,
      "loss": 1.7472,
      "step": 3018
    },
    {
      "epoch": 0.31845991561181436,
      "grad_norm": 0.8234111070632935,
      "learning_rate": 0.0015,
      "loss": 1.7178,
      "step": 3019
    },
    {
      "epoch": 0.31856540084388185,
      "grad_norm": 0.6567312479019165,
      "learning_rate": 0.0015,
      "loss": 1.7014,
      "step": 3020
    },
    {
      "epoch": 0.31867088607594934,
      "grad_norm": 0.8978010416030884,
      "learning_rate": 0.0015,
      "loss": 1.6899,
      "step": 3021
    },
    {
      "epoch": 0.3187763713080169,
      "grad_norm": 0.7658066153526306,
      "learning_rate": 0.0015,
      "loss": 1.7148,
      "step": 3022
    },
    {
      "epoch": 0.3188818565400844,
      "grad_norm": 0.4774928390979767,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 3023
    },
    {
      "epoch": 0.3189873417721519,
      "grad_norm": 0.7254661321640015,
      "learning_rate": 0.0015,
      "loss": 1.7017,
      "step": 3024
    },
    {
      "epoch": 0.3190928270042194,
      "grad_norm": 0.536165714263916,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 3025
    },
    {
      "epoch": 0.3191983122362869,
      "grad_norm": 0.663199245929718,
      "learning_rate": 0.0015,
      "loss": 1.7304,
      "step": 3026
    },
    {
      "epoch": 0.3193037974683544,
      "grad_norm": 0.654486894607544,
      "learning_rate": 0.0015,
      "loss": 1.7107,
      "step": 3027
    },
    {
      "epoch": 0.31940928270042196,
      "grad_norm": 0.6329241991043091,
      "learning_rate": 0.0015,
      "loss": 1.7341,
      "step": 3028
    },
    {
      "epoch": 0.31951476793248945,
      "grad_norm": 0.5985587239265442,
      "learning_rate": 0.0015,
      "loss": 1.6952,
      "step": 3029
    },
    {
      "epoch": 0.31962025316455694,
      "grad_norm": 0.8229667544364929,
      "learning_rate": 0.0015,
      "loss": 1.747,
      "step": 3030
    },
    {
      "epoch": 0.3197257383966245,
      "grad_norm": 0.5880804657936096,
      "learning_rate": 0.0015,
      "loss": 1.7174,
      "step": 3031
    },
    {
      "epoch": 0.319831223628692,
      "grad_norm": 0.5924350023269653,
      "learning_rate": 0.0015,
      "loss": 1.7108,
      "step": 3032
    },
    {
      "epoch": 0.3199367088607595,
      "grad_norm": 0.5878428220748901,
      "learning_rate": 0.0015,
      "loss": 1.6887,
      "step": 3033
    },
    {
      "epoch": 0.320042194092827,
      "grad_norm": 0.5516447424888611,
      "learning_rate": 0.0015,
      "loss": 1.7428,
      "step": 3034
    },
    {
      "epoch": 0.3201476793248945,
      "grad_norm": 0.7194355130195618,
      "learning_rate": 0.0015,
      "loss": 1.7037,
      "step": 3035
    },
    {
      "epoch": 0.320253164556962,
      "grad_norm": 0.47816890478134155,
      "learning_rate": 0.0015,
      "loss": 1.7629,
      "step": 3036
    },
    {
      "epoch": 0.32035864978902956,
      "grad_norm": 0.6952998638153076,
      "learning_rate": 0.0015,
      "loss": 1.7469,
      "step": 3037
    },
    {
      "epoch": 0.32046413502109705,
      "grad_norm": 0.5314906239509583,
      "learning_rate": 0.0015,
      "loss": 1.6811,
      "step": 3038
    },
    {
      "epoch": 0.32056962025316454,
      "grad_norm": 0.5869095921516418,
      "learning_rate": 0.0015,
      "loss": 1.693,
      "step": 3039
    },
    {
      "epoch": 0.3206751054852321,
      "grad_norm": 0.5194697380065918,
      "learning_rate": 0.0015,
      "loss": 1.6861,
      "step": 3040
    },
    {
      "epoch": 0.3207805907172996,
      "grad_norm": 0.641189694404602,
      "learning_rate": 0.0015,
      "loss": 1.7105,
      "step": 3041
    },
    {
      "epoch": 0.3208860759493671,
      "grad_norm": 0.5106596946716309,
      "learning_rate": 0.0015,
      "loss": 1.7043,
      "step": 3042
    },
    {
      "epoch": 0.3209915611814346,
      "grad_norm": 0.6200544238090515,
      "learning_rate": 0.0015,
      "loss": 1.7395,
      "step": 3043
    },
    {
      "epoch": 0.3210970464135021,
      "grad_norm": 0.5583987832069397,
      "learning_rate": 0.0015,
      "loss": 1.7112,
      "step": 3044
    },
    {
      "epoch": 0.3212025316455696,
      "grad_norm": 0.5801132321357727,
      "learning_rate": 0.0015,
      "loss": 1.7095,
      "step": 3045
    },
    {
      "epoch": 0.3213080168776371,
      "grad_norm": 0.536475419998169,
      "learning_rate": 0.0015,
      "loss": 1.757,
      "step": 3046
    },
    {
      "epoch": 0.32141350210970465,
      "grad_norm": 0.5250360369682312,
      "learning_rate": 0.0015,
      "loss": 1.6634,
      "step": 3047
    },
    {
      "epoch": 0.32151898734177214,
      "grad_norm": 0.5058603882789612,
      "learning_rate": 0.0015,
      "loss": 1.7073,
      "step": 3048
    },
    {
      "epoch": 0.32162447257383964,
      "grad_norm": 0.5103098154067993,
      "learning_rate": 0.0015,
      "loss": 1.6794,
      "step": 3049
    },
    {
      "epoch": 0.3217299578059072,
      "grad_norm": 0.4832077622413635,
      "learning_rate": 0.0015,
      "loss": 1.7062,
      "step": 3050
    },
    {
      "epoch": 0.3218354430379747,
      "grad_norm": 0.5129528641700745,
      "learning_rate": 0.0015,
      "loss": 1.699,
      "step": 3051
    },
    {
      "epoch": 0.32194092827004217,
      "grad_norm": 0.5048354268074036,
      "learning_rate": 0.0015,
      "loss": 1.6927,
      "step": 3052
    },
    {
      "epoch": 0.3220464135021097,
      "grad_norm": 0.5547788143157959,
      "learning_rate": 0.0015,
      "loss": 1.7307,
      "step": 3053
    },
    {
      "epoch": 0.3221518987341772,
      "grad_norm": 0.5981379151344299,
      "learning_rate": 0.0015,
      "loss": 1.7147,
      "step": 3054
    },
    {
      "epoch": 0.3222573839662447,
      "grad_norm": 0.46103549003601074,
      "learning_rate": 0.0015,
      "loss": 1.7425,
      "step": 3055
    },
    {
      "epoch": 0.32236286919831225,
      "grad_norm": 0.601043701171875,
      "learning_rate": 0.0015,
      "loss": 1.7254,
      "step": 3056
    },
    {
      "epoch": 0.32246835443037974,
      "grad_norm": 0.5374560952186584,
      "learning_rate": 0.0015,
      "loss": 1.7188,
      "step": 3057
    },
    {
      "epoch": 0.32257383966244724,
      "grad_norm": 0.4482406973838806,
      "learning_rate": 0.0015,
      "loss": 1.7083,
      "step": 3058
    },
    {
      "epoch": 0.3226793248945148,
      "grad_norm": 0.5059371590614319,
      "learning_rate": 0.0015,
      "loss": 1.7242,
      "step": 3059
    },
    {
      "epoch": 0.3227848101265823,
      "grad_norm": 0.5071547031402588,
      "learning_rate": 0.0015,
      "loss": 1.7208,
      "step": 3060
    },
    {
      "epoch": 0.32289029535864977,
      "grad_norm": 0.46012645959854126,
      "learning_rate": 0.0015,
      "loss": 1.6941,
      "step": 3061
    },
    {
      "epoch": 0.3229957805907173,
      "grad_norm": 0.5413433909416199,
      "learning_rate": 0.0015,
      "loss": 1.6894,
      "step": 3062
    },
    {
      "epoch": 0.3231012658227848,
      "grad_norm": 0.5513799786567688,
      "learning_rate": 0.0015,
      "loss": 1.7154,
      "step": 3063
    },
    {
      "epoch": 0.3232067510548523,
      "grad_norm": 0.4934634864330292,
      "learning_rate": 0.0015,
      "loss": 1.7136,
      "step": 3064
    },
    {
      "epoch": 0.32331223628691985,
      "grad_norm": 0.49460723996162415,
      "learning_rate": 0.0015,
      "loss": 1.6926,
      "step": 3065
    },
    {
      "epoch": 0.32341772151898734,
      "grad_norm": 0.5192188024520874,
      "learning_rate": 0.0015,
      "loss": 1.707,
      "step": 3066
    },
    {
      "epoch": 0.32352320675105484,
      "grad_norm": 0.4758225381374359,
      "learning_rate": 0.0015,
      "loss": 1.7233,
      "step": 3067
    },
    {
      "epoch": 0.3236286919831224,
      "grad_norm": 0.4821212887763977,
      "learning_rate": 0.0015,
      "loss": 1.7053,
      "step": 3068
    },
    {
      "epoch": 0.3237341772151899,
      "grad_norm": 0.4905916154384613,
      "learning_rate": 0.0015,
      "loss": 1.7183,
      "step": 3069
    },
    {
      "epoch": 0.32383966244725737,
      "grad_norm": 0.5138663649559021,
      "learning_rate": 0.0015,
      "loss": 1.7293,
      "step": 3070
    },
    {
      "epoch": 0.3239451476793249,
      "grad_norm": 0.5315477848052979,
      "learning_rate": 0.0015,
      "loss": 1.7419,
      "step": 3071
    },
    {
      "epoch": 0.3240506329113924,
      "grad_norm": 0.4920712113380432,
      "learning_rate": 0.0015,
      "loss": 1.6559,
      "step": 3072
    },
    {
      "epoch": 0.3241561181434599,
      "grad_norm": 0.48032838106155396,
      "learning_rate": 0.0015,
      "loss": 1.7018,
      "step": 3073
    },
    {
      "epoch": 0.32426160337552745,
      "grad_norm": 0.5117155909538269,
      "learning_rate": 0.0015,
      "loss": 1.7267,
      "step": 3074
    },
    {
      "epoch": 0.32436708860759494,
      "grad_norm": 0.4488002359867096,
      "learning_rate": 0.0015,
      "loss": 1.6693,
      "step": 3075
    },
    {
      "epoch": 0.32447257383966244,
      "grad_norm": 0.48116782307624817,
      "learning_rate": 0.0015,
      "loss": 1.6683,
      "step": 3076
    },
    {
      "epoch": 0.32457805907173,
      "grad_norm": 0.5115820169448853,
      "learning_rate": 0.0015,
      "loss": 1.6754,
      "step": 3077
    },
    {
      "epoch": 0.3246835443037975,
      "grad_norm": 0.49160560965538025,
      "learning_rate": 0.0015,
      "loss": 1.6932,
      "step": 3078
    },
    {
      "epoch": 0.32478902953586497,
      "grad_norm": 0.5501735210418701,
      "learning_rate": 0.0015,
      "loss": 1.6965,
      "step": 3079
    },
    {
      "epoch": 0.32489451476793246,
      "grad_norm": 0.5930479764938354,
      "learning_rate": 0.0015,
      "loss": 1.749,
      "step": 3080
    },
    {
      "epoch": 0.325,
      "grad_norm": 0.4873003363609314,
      "learning_rate": 0.0015,
      "loss": 1.6475,
      "step": 3081
    },
    {
      "epoch": 0.3251054852320675,
      "grad_norm": 0.5444014072418213,
      "learning_rate": 0.0015,
      "loss": 1.7198,
      "step": 3082
    },
    {
      "epoch": 0.325210970464135,
      "grad_norm": 0.6106745600700378,
      "learning_rate": 0.0015,
      "loss": 1.7062,
      "step": 3083
    },
    {
      "epoch": 0.32531645569620254,
      "grad_norm": 0.45090463757514954,
      "learning_rate": 0.0015,
      "loss": 1.7129,
      "step": 3084
    },
    {
      "epoch": 0.32542194092827004,
      "grad_norm": 0.5389440655708313,
      "learning_rate": 0.0015,
      "loss": 1.7278,
      "step": 3085
    },
    {
      "epoch": 0.32552742616033753,
      "grad_norm": 0.5276144742965698,
      "learning_rate": 0.0015,
      "loss": 1.7077,
      "step": 3086
    },
    {
      "epoch": 0.3256329113924051,
      "grad_norm": 0.45780766010284424,
      "learning_rate": 0.0015,
      "loss": 1.761,
      "step": 3087
    },
    {
      "epoch": 0.32573839662447257,
      "grad_norm": 0.5008159279823303,
      "learning_rate": 0.0015,
      "loss": 1.6925,
      "step": 3088
    },
    {
      "epoch": 0.32584388185654006,
      "grad_norm": 0.49904295802116394,
      "learning_rate": 0.0015,
      "loss": 1.7068,
      "step": 3089
    },
    {
      "epoch": 0.3259493670886076,
      "grad_norm": 0.504729151725769,
      "learning_rate": 0.0015,
      "loss": 1.7373,
      "step": 3090
    },
    {
      "epoch": 0.3260548523206751,
      "grad_norm": 0.494392991065979,
      "learning_rate": 0.0015,
      "loss": 1.7291,
      "step": 3091
    },
    {
      "epoch": 0.3261603375527426,
      "grad_norm": 0.46707990765571594,
      "learning_rate": 0.0015,
      "loss": 1.6951,
      "step": 3092
    },
    {
      "epoch": 0.32626582278481014,
      "grad_norm": 0.4759694039821625,
      "learning_rate": 0.0015,
      "loss": 1.6798,
      "step": 3093
    },
    {
      "epoch": 0.32637130801687764,
      "grad_norm": 0.4642212986946106,
      "learning_rate": 0.0015,
      "loss": 1.6961,
      "step": 3094
    },
    {
      "epoch": 0.32647679324894513,
      "grad_norm": 0.47130927443504333,
      "learning_rate": 0.0015,
      "loss": 1.694,
      "step": 3095
    },
    {
      "epoch": 0.3265822784810127,
      "grad_norm": 0.5033254027366638,
      "learning_rate": 0.0015,
      "loss": 1.7039,
      "step": 3096
    },
    {
      "epoch": 0.32668776371308017,
      "grad_norm": 0.5418142080307007,
      "learning_rate": 0.0015,
      "loss": 1.6881,
      "step": 3097
    },
    {
      "epoch": 0.32679324894514766,
      "grad_norm": 0.4699496924877167,
      "learning_rate": 0.0015,
      "loss": 1.7271,
      "step": 3098
    },
    {
      "epoch": 0.3268987341772152,
      "grad_norm": 0.7430257797241211,
      "learning_rate": 0.0015,
      "loss": 1.6879,
      "step": 3099
    },
    {
      "epoch": 0.3270042194092827,
      "grad_norm": 0.8292415142059326,
      "learning_rate": 0.0015,
      "loss": 1.7252,
      "step": 3100
    },
    {
      "epoch": 0.3271097046413502,
      "grad_norm": 0.5245999097824097,
      "learning_rate": 0.0015,
      "loss": 1.7084,
      "step": 3101
    },
    {
      "epoch": 0.32721518987341774,
      "grad_norm": 0.5449711084365845,
      "learning_rate": 0.0015,
      "loss": 1.717,
      "step": 3102
    },
    {
      "epoch": 0.32732067510548524,
      "grad_norm": 0.6642726063728333,
      "learning_rate": 0.0015,
      "loss": 1.7238,
      "step": 3103
    },
    {
      "epoch": 0.32742616033755273,
      "grad_norm": 0.5721220970153809,
      "learning_rate": 0.0015,
      "loss": 1.69,
      "step": 3104
    },
    {
      "epoch": 0.3275316455696203,
      "grad_norm": 0.5738294124603271,
      "learning_rate": 0.0015,
      "loss": 1.6925,
      "step": 3105
    },
    {
      "epoch": 0.32763713080168777,
      "grad_norm": 0.5988900065422058,
      "learning_rate": 0.0015,
      "loss": 1.7273,
      "step": 3106
    },
    {
      "epoch": 0.32774261603375526,
      "grad_norm": 0.4937974214553833,
      "learning_rate": 0.0015,
      "loss": 1.6868,
      "step": 3107
    },
    {
      "epoch": 0.3278481012658228,
      "grad_norm": 0.684008002281189,
      "learning_rate": 0.0015,
      "loss": 1.6909,
      "step": 3108
    },
    {
      "epoch": 0.3279535864978903,
      "grad_norm": 0.5166898965835571,
      "learning_rate": 0.0015,
      "loss": 1.7227,
      "step": 3109
    },
    {
      "epoch": 0.3280590717299578,
      "grad_norm": 0.661108136177063,
      "learning_rate": 0.0015,
      "loss": 1.7195,
      "step": 3110
    },
    {
      "epoch": 0.3281645569620253,
      "grad_norm": 0.5461713075637817,
      "learning_rate": 0.0015,
      "loss": 1.7208,
      "step": 3111
    },
    {
      "epoch": 0.32827004219409284,
      "grad_norm": 0.5846986174583435,
      "learning_rate": 0.0015,
      "loss": 1.7103,
      "step": 3112
    },
    {
      "epoch": 0.32837552742616033,
      "grad_norm": 0.6091597676277161,
      "learning_rate": 0.0015,
      "loss": 1.7008,
      "step": 3113
    },
    {
      "epoch": 0.3284810126582278,
      "grad_norm": 0.6286957859992981,
      "learning_rate": 0.0015,
      "loss": 1.7222,
      "step": 3114
    },
    {
      "epoch": 0.32858649789029537,
      "grad_norm": 0.6247303485870361,
      "learning_rate": 0.0015,
      "loss": 1.7073,
      "step": 3115
    },
    {
      "epoch": 0.32869198312236286,
      "grad_norm": 0.57159024477005,
      "learning_rate": 0.0015,
      "loss": 1.678,
      "step": 3116
    },
    {
      "epoch": 0.32879746835443036,
      "grad_norm": 0.735164999961853,
      "learning_rate": 0.0015,
      "loss": 1.7103,
      "step": 3117
    },
    {
      "epoch": 0.3289029535864979,
      "grad_norm": 0.5963243246078491,
      "learning_rate": 0.0015,
      "loss": 1.6868,
      "step": 3118
    },
    {
      "epoch": 0.3290084388185654,
      "grad_norm": 0.5325501561164856,
      "learning_rate": 0.0015,
      "loss": 1.7046,
      "step": 3119
    },
    {
      "epoch": 0.3291139240506329,
      "grad_norm": 0.7457340955734253,
      "learning_rate": 0.0015,
      "loss": 1.7651,
      "step": 3120
    },
    {
      "epoch": 0.32921940928270044,
      "grad_norm": 0.6895079016685486,
      "learning_rate": 0.0015,
      "loss": 1.7074,
      "step": 3121
    },
    {
      "epoch": 0.32932489451476793,
      "grad_norm": 0.5428807735443115,
      "learning_rate": 0.0015,
      "loss": 1.689,
      "step": 3122
    },
    {
      "epoch": 0.3294303797468354,
      "grad_norm": 0.5404313802719116,
      "learning_rate": 0.0015,
      "loss": 1.7589,
      "step": 3123
    },
    {
      "epoch": 0.32953586497890297,
      "grad_norm": 0.5746237635612488,
      "learning_rate": 0.0015,
      "loss": 1.686,
      "step": 3124
    },
    {
      "epoch": 0.32964135021097046,
      "grad_norm": 0.49167463183403015,
      "learning_rate": 0.0015,
      "loss": 1.743,
      "step": 3125
    },
    {
      "epoch": 0.32974683544303796,
      "grad_norm": 0.7660045623779297,
      "learning_rate": 0.0015,
      "loss": 1.7277,
      "step": 3126
    },
    {
      "epoch": 0.3298523206751055,
      "grad_norm": 0.9002141356468201,
      "learning_rate": 0.0015,
      "loss": 1.7226,
      "step": 3127
    },
    {
      "epoch": 0.329957805907173,
      "grad_norm": 0.4946906268596649,
      "learning_rate": 0.0015,
      "loss": 1.7029,
      "step": 3128
    },
    {
      "epoch": 0.3300632911392405,
      "grad_norm": 0.8130706548690796,
      "learning_rate": 0.0015,
      "loss": 1.7452,
      "step": 3129
    },
    {
      "epoch": 0.33016877637130804,
      "grad_norm": 0.6559513211250305,
      "learning_rate": 0.0015,
      "loss": 1.7304,
      "step": 3130
    },
    {
      "epoch": 0.33027426160337553,
      "grad_norm": 0.5682690143585205,
      "learning_rate": 0.0015,
      "loss": 1.6972,
      "step": 3131
    },
    {
      "epoch": 0.330379746835443,
      "grad_norm": 0.88380366563797,
      "learning_rate": 0.0015,
      "loss": 1.7255,
      "step": 3132
    },
    {
      "epoch": 0.33048523206751057,
      "grad_norm": 0.8223167061805725,
      "learning_rate": 0.0015,
      "loss": 1.6579,
      "step": 3133
    },
    {
      "epoch": 0.33059071729957806,
      "grad_norm": 0.46287649869918823,
      "learning_rate": 0.0015,
      "loss": 1.734,
      "step": 3134
    },
    {
      "epoch": 0.33069620253164556,
      "grad_norm": 0.6548208594322205,
      "learning_rate": 0.0015,
      "loss": 1.6651,
      "step": 3135
    },
    {
      "epoch": 0.3308016877637131,
      "grad_norm": 0.6544517874717712,
      "learning_rate": 0.0015,
      "loss": 1.6888,
      "step": 3136
    },
    {
      "epoch": 0.3309071729957806,
      "grad_norm": 0.4662328064441681,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 3137
    },
    {
      "epoch": 0.3310126582278481,
      "grad_norm": 0.6554684042930603,
      "learning_rate": 0.0015,
      "loss": 1.7382,
      "step": 3138
    },
    {
      "epoch": 0.33111814345991564,
      "grad_norm": 0.7112213373184204,
      "learning_rate": 0.0015,
      "loss": 1.7187,
      "step": 3139
    },
    {
      "epoch": 0.33122362869198313,
      "grad_norm": 0.6192386150360107,
      "learning_rate": 0.0015,
      "loss": 1.6896,
      "step": 3140
    },
    {
      "epoch": 0.3313291139240506,
      "grad_norm": 0.5745393633842468,
      "learning_rate": 0.0015,
      "loss": 1.7242,
      "step": 3141
    },
    {
      "epoch": 0.33143459915611817,
      "grad_norm": 0.5276301503181458,
      "learning_rate": 0.0015,
      "loss": 1.6827,
      "step": 3142
    },
    {
      "epoch": 0.33154008438818566,
      "grad_norm": 0.49275004863739014,
      "learning_rate": 0.0015,
      "loss": 1.7192,
      "step": 3143
    },
    {
      "epoch": 0.33164556962025316,
      "grad_norm": 0.5142596364021301,
      "learning_rate": 0.0015,
      "loss": 1.7245,
      "step": 3144
    },
    {
      "epoch": 0.33175105485232065,
      "grad_norm": 0.4448336362838745,
      "learning_rate": 0.0015,
      "loss": 1.7213,
      "step": 3145
    },
    {
      "epoch": 0.3318565400843882,
      "grad_norm": 0.554429829120636,
      "learning_rate": 0.0015,
      "loss": 1.7104,
      "step": 3146
    },
    {
      "epoch": 0.3319620253164557,
      "grad_norm": 0.515129804611206,
      "learning_rate": 0.0015,
      "loss": 1.6915,
      "step": 3147
    },
    {
      "epoch": 0.3320675105485232,
      "grad_norm": 0.5072916746139526,
      "learning_rate": 0.0015,
      "loss": 1.6949,
      "step": 3148
    },
    {
      "epoch": 0.33217299578059073,
      "grad_norm": 0.5899186134338379,
      "learning_rate": 0.0015,
      "loss": 1.6962,
      "step": 3149
    },
    {
      "epoch": 0.3322784810126582,
      "grad_norm": 0.4840143620967865,
      "learning_rate": 0.0015,
      "loss": 1.6973,
      "step": 3150
    },
    {
      "epoch": 0.3323839662447257,
      "grad_norm": 0.6333123445510864,
      "learning_rate": 0.0015,
      "loss": 1.7048,
      "step": 3151
    },
    {
      "epoch": 0.33248945147679326,
      "grad_norm": 0.6574004292488098,
      "learning_rate": 0.0015,
      "loss": 1.6943,
      "step": 3152
    },
    {
      "epoch": 0.33259493670886076,
      "grad_norm": 0.5177522301673889,
      "learning_rate": 0.0015,
      "loss": 1.7207,
      "step": 3153
    },
    {
      "epoch": 0.33270042194092825,
      "grad_norm": 0.5760917663574219,
      "learning_rate": 0.0015,
      "loss": 1.7131,
      "step": 3154
    },
    {
      "epoch": 0.3328059071729958,
      "grad_norm": 0.5342174172401428,
      "learning_rate": 0.0015,
      "loss": 1.7097,
      "step": 3155
    },
    {
      "epoch": 0.3329113924050633,
      "grad_norm": 0.5675168037414551,
      "learning_rate": 0.0015,
      "loss": 1.7096,
      "step": 3156
    },
    {
      "epoch": 0.3330168776371308,
      "grad_norm": 0.6600531339645386,
      "learning_rate": 0.0015,
      "loss": 1.7023,
      "step": 3157
    },
    {
      "epoch": 0.33312236286919833,
      "grad_norm": 0.5559056401252747,
      "learning_rate": 0.0015,
      "loss": 1.6869,
      "step": 3158
    },
    {
      "epoch": 0.3332278481012658,
      "grad_norm": 0.57248455286026,
      "learning_rate": 0.0015,
      "loss": 1.7187,
      "step": 3159
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.5283870697021484,
      "learning_rate": 0.0015,
      "loss": 1.7089,
      "step": 3160
    },
    {
      "epoch": 0.33343881856540086,
      "grad_norm": 0.4742368459701538,
      "learning_rate": 0.0015,
      "loss": 1.7136,
      "step": 3161
    },
    {
      "epoch": 0.33354430379746836,
      "grad_norm": 0.7028859853744507,
      "learning_rate": 0.0015,
      "loss": 1.707,
      "step": 3162
    },
    {
      "epoch": 0.33364978902953585,
      "grad_norm": 0.7126950621604919,
      "learning_rate": 0.0015,
      "loss": 1.7071,
      "step": 3163
    },
    {
      "epoch": 0.3337552742616034,
      "grad_norm": 0.4972867965698242,
      "learning_rate": 0.0015,
      "loss": 1.716,
      "step": 3164
    },
    {
      "epoch": 0.3338607594936709,
      "grad_norm": 0.7651951909065247,
      "learning_rate": 0.0015,
      "loss": 1.7067,
      "step": 3165
    },
    {
      "epoch": 0.3339662447257384,
      "grad_norm": 0.5141854882240295,
      "learning_rate": 0.0015,
      "loss": 1.7433,
      "step": 3166
    },
    {
      "epoch": 0.33407172995780593,
      "grad_norm": 0.6878163814544678,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 3167
    },
    {
      "epoch": 0.3341772151898734,
      "grad_norm": 0.7308759093284607,
      "learning_rate": 0.0015,
      "loss": 1.7179,
      "step": 3168
    },
    {
      "epoch": 0.3342827004219409,
      "grad_norm": 0.5225549340248108,
      "learning_rate": 0.0015,
      "loss": 1.7305,
      "step": 3169
    },
    {
      "epoch": 0.33438818565400846,
      "grad_norm": 0.5995153784751892,
      "learning_rate": 0.0015,
      "loss": 1.6682,
      "step": 3170
    },
    {
      "epoch": 0.33449367088607596,
      "grad_norm": 0.5042096972465515,
      "learning_rate": 0.0015,
      "loss": 1.6979,
      "step": 3171
    },
    {
      "epoch": 0.33459915611814345,
      "grad_norm": 0.5545108914375305,
      "learning_rate": 0.0015,
      "loss": 1.6876,
      "step": 3172
    },
    {
      "epoch": 0.334704641350211,
      "grad_norm": 0.7023755311965942,
      "learning_rate": 0.0015,
      "loss": 1.7053,
      "step": 3173
    },
    {
      "epoch": 0.3348101265822785,
      "grad_norm": 0.7239559888839722,
      "learning_rate": 0.0015,
      "loss": 1.7178,
      "step": 3174
    },
    {
      "epoch": 0.334915611814346,
      "grad_norm": 0.6107579469680786,
      "learning_rate": 0.0015,
      "loss": 1.7103,
      "step": 3175
    },
    {
      "epoch": 0.33502109704641353,
      "grad_norm": 0.4597165584564209,
      "learning_rate": 0.0015,
      "loss": 1.7002,
      "step": 3176
    },
    {
      "epoch": 0.335126582278481,
      "grad_norm": 0.5891026258468628,
      "learning_rate": 0.0015,
      "loss": 1.7052,
      "step": 3177
    },
    {
      "epoch": 0.3352320675105485,
      "grad_norm": 0.47593066096305847,
      "learning_rate": 0.0015,
      "loss": 1.6954,
      "step": 3178
    },
    {
      "epoch": 0.335337552742616,
      "grad_norm": 0.5646541118621826,
      "learning_rate": 0.0015,
      "loss": 1.6659,
      "step": 3179
    },
    {
      "epoch": 0.33544303797468356,
      "grad_norm": 0.503830075263977,
      "learning_rate": 0.0015,
      "loss": 1.7446,
      "step": 3180
    },
    {
      "epoch": 0.33554852320675105,
      "grad_norm": 0.631014883518219,
      "learning_rate": 0.0015,
      "loss": 1.6698,
      "step": 3181
    },
    {
      "epoch": 0.33565400843881854,
      "grad_norm": 0.7992089986801147,
      "learning_rate": 0.0015,
      "loss": 1.6867,
      "step": 3182
    },
    {
      "epoch": 0.3357594936708861,
      "grad_norm": 0.6834596991539001,
      "learning_rate": 0.0015,
      "loss": 1.7043,
      "step": 3183
    },
    {
      "epoch": 0.3358649789029536,
      "grad_norm": 0.46089988946914673,
      "learning_rate": 0.0015,
      "loss": 1.7046,
      "step": 3184
    },
    {
      "epoch": 0.3359704641350211,
      "grad_norm": 0.6685697436332703,
      "learning_rate": 0.0015,
      "loss": 1.7151,
      "step": 3185
    },
    {
      "epoch": 0.3360759493670886,
      "grad_norm": 0.6084441542625427,
      "learning_rate": 0.0015,
      "loss": 1.7049,
      "step": 3186
    },
    {
      "epoch": 0.3361814345991561,
      "grad_norm": 0.45125362277030945,
      "learning_rate": 0.0015,
      "loss": 1.7253,
      "step": 3187
    },
    {
      "epoch": 0.3362869198312236,
      "grad_norm": 0.5677869319915771,
      "learning_rate": 0.0015,
      "loss": 1.708,
      "step": 3188
    },
    {
      "epoch": 0.33639240506329116,
      "grad_norm": 0.671842098236084,
      "learning_rate": 0.0015,
      "loss": 1.6861,
      "step": 3189
    },
    {
      "epoch": 0.33649789029535865,
      "grad_norm": 0.5831478238105774,
      "learning_rate": 0.0015,
      "loss": 1.734,
      "step": 3190
    },
    {
      "epoch": 0.33660337552742614,
      "grad_norm": 0.4758757948875427,
      "learning_rate": 0.0015,
      "loss": 1.6978,
      "step": 3191
    },
    {
      "epoch": 0.3367088607594937,
      "grad_norm": 0.5622855424880981,
      "learning_rate": 0.0015,
      "loss": 1.6779,
      "step": 3192
    },
    {
      "epoch": 0.3368143459915612,
      "grad_norm": 0.5027768015861511,
      "learning_rate": 0.0015,
      "loss": 1.7407,
      "step": 3193
    },
    {
      "epoch": 0.3369198312236287,
      "grad_norm": 0.5022798776626587,
      "learning_rate": 0.0015,
      "loss": 1.7238,
      "step": 3194
    },
    {
      "epoch": 0.3370253164556962,
      "grad_norm": 0.4664076864719391,
      "learning_rate": 0.0015,
      "loss": 1.6678,
      "step": 3195
    },
    {
      "epoch": 0.3371308016877637,
      "grad_norm": 0.4845595061779022,
      "learning_rate": 0.0015,
      "loss": 1.7356,
      "step": 3196
    },
    {
      "epoch": 0.3372362869198312,
      "grad_norm": 0.5015625357627869,
      "learning_rate": 0.0015,
      "loss": 1.6471,
      "step": 3197
    },
    {
      "epoch": 0.33734177215189876,
      "grad_norm": 0.5129271745681763,
      "learning_rate": 0.0015,
      "loss": 1.7026,
      "step": 3198
    },
    {
      "epoch": 0.33744725738396625,
      "grad_norm": 0.5271135568618774,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 3199
    },
    {
      "epoch": 0.33755274261603374,
      "grad_norm": 0.503288209438324,
      "learning_rate": 0.0015,
      "loss": 1.707,
      "step": 3200
    },
    {
      "epoch": 0.3376582278481013,
      "grad_norm": 0.45782169699668884,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 3201
    },
    {
      "epoch": 0.3377637130801688,
      "grad_norm": 0.5114582777023315,
      "learning_rate": 0.0015,
      "loss": 1.737,
      "step": 3202
    },
    {
      "epoch": 0.3378691983122363,
      "grad_norm": 0.5901846289634705,
      "learning_rate": 0.0015,
      "loss": 1.7242,
      "step": 3203
    },
    {
      "epoch": 0.3379746835443038,
      "grad_norm": 0.672936737537384,
      "learning_rate": 0.0015,
      "loss": 1.7104,
      "step": 3204
    },
    {
      "epoch": 0.3380801687763713,
      "grad_norm": 0.5681142210960388,
      "learning_rate": 0.0015,
      "loss": 1.7256,
      "step": 3205
    },
    {
      "epoch": 0.3381856540084388,
      "grad_norm": 0.5117339491844177,
      "learning_rate": 0.0015,
      "loss": 1.7215,
      "step": 3206
    },
    {
      "epoch": 0.33829113924050636,
      "grad_norm": 0.5524427890777588,
      "learning_rate": 0.0015,
      "loss": 1.7258,
      "step": 3207
    },
    {
      "epoch": 0.33839662447257385,
      "grad_norm": 0.5113054513931274,
      "learning_rate": 0.0015,
      "loss": 1.6759,
      "step": 3208
    },
    {
      "epoch": 0.33850210970464134,
      "grad_norm": 0.4930611848831177,
      "learning_rate": 0.0015,
      "loss": 1.6929,
      "step": 3209
    },
    {
      "epoch": 0.33860759493670883,
      "grad_norm": 0.5216551423072815,
      "learning_rate": 0.0015,
      "loss": 1.7071,
      "step": 3210
    },
    {
      "epoch": 0.3387130801687764,
      "grad_norm": 0.5543256402015686,
      "learning_rate": 0.0015,
      "loss": 1.6885,
      "step": 3211
    },
    {
      "epoch": 0.3388185654008439,
      "grad_norm": 0.5151799321174622,
      "learning_rate": 0.0015,
      "loss": 1.684,
      "step": 3212
    },
    {
      "epoch": 0.33892405063291137,
      "grad_norm": 0.49063053727149963,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 3213
    },
    {
      "epoch": 0.3390295358649789,
      "grad_norm": 0.5205222368240356,
      "learning_rate": 0.0015,
      "loss": 1.733,
      "step": 3214
    },
    {
      "epoch": 0.3391350210970464,
      "grad_norm": 0.5935371518135071,
      "learning_rate": 0.0015,
      "loss": 1.7125,
      "step": 3215
    },
    {
      "epoch": 0.3392405063291139,
      "grad_norm": 0.6582415103912354,
      "learning_rate": 0.0015,
      "loss": 1.7074,
      "step": 3216
    },
    {
      "epoch": 0.33934599156118145,
      "grad_norm": 0.5578335523605347,
      "learning_rate": 0.0015,
      "loss": 1.6992,
      "step": 3217
    },
    {
      "epoch": 0.33945147679324894,
      "grad_norm": 0.5155923366546631,
      "learning_rate": 0.0015,
      "loss": 1.7109,
      "step": 3218
    },
    {
      "epoch": 0.33955696202531643,
      "grad_norm": 0.6253606677055359,
      "learning_rate": 0.0015,
      "loss": 1.6796,
      "step": 3219
    },
    {
      "epoch": 0.339662447257384,
      "grad_norm": 0.5452165603637695,
      "learning_rate": 0.0015,
      "loss": 1.7119,
      "step": 3220
    },
    {
      "epoch": 0.3397679324894515,
      "grad_norm": 0.4783080220222473,
      "learning_rate": 0.0015,
      "loss": 1.7015,
      "step": 3221
    },
    {
      "epoch": 0.33987341772151897,
      "grad_norm": 0.605499804019928,
      "learning_rate": 0.0015,
      "loss": 1.7164,
      "step": 3222
    },
    {
      "epoch": 0.3399789029535865,
      "grad_norm": 0.6417616009712219,
      "learning_rate": 0.0015,
      "loss": 1.7096,
      "step": 3223
    },
    {
      "epoch": 0.340084388185654,
      "grad_norm": 0.49299174547195435,
      "learning_rate": 0.0015,
      "loss": 1.6775,
      "step": 3224
    },
    {
      "epoch": 0.3401898734177215,
      "grad_norm": 0.6858759522438049,
      "learning_rate": 0.0015,
      "loss": 1.7109,
      "step": 3225
    },
    {
      "epoch": 0.34029535864978905,
      "grad_norm": 0.7292691469192505,
      "learning_rate": 0.0015,
      "loss": 1.6802,
      "step": 3226
    },
    {
      "epoch": 0.34040084388185654,
      "grad_norm": 0.5831139087677002,
      "learning_rate": 0.0015,
      "loss": 1.6743,
      "step": 3227
    },
    {
      "epoch": 0.34050632911392403,
      "grad_norm": 0.5639909505844116,
      "learning_rate": 0.0015,
      "loss": 1.6706,
      "step": 3228
    },
    {
      "epoch": 0.3406118143459916,
      "grad_norm": 0.7858826518058777,
      "learning_rate": 0.0015,
      "loss": 1.7151,
      "step": 3229
    },
    {
      "epoch": 0.3407172995780591,
      "grad_norm": 0.4706067442893982,
      "learning_rate": 0.0015,
      "loss": 1.7146,
      "step": 3230
    },
    {
      "epoch": 0.34082278481012657,
      "grad_norm": 0.8369712233543396,
      "learning_rate": 0.0015,
      "loss": 1.686,
      "step": 3231
    },
    {
      "epoch": 0.3409282700421941,
      "grad_norm": 0.716052234172821,
      "learning_rate": 0.0015,
      "loss": 1.6772,
      "step": 3232
    },
    {
      "epoch": 0.3410337552742616,
      "grad_norm": 0.5431948900222778,
      "learning_rate": 0.0015,
      "loss": 1.6752,
      "step": 3233
    },
    {
      "epoch": 0.3411392405063291,
      "grad_norm": 0.8990529775619507,
      "learning_rate": 0.0015,
      "loss": 1.6745,
      "step": 3234
    },
    {
      "epoch": 0.34124472573839665,
      "grad_norm": 0.6069479584693909,
      "learning_rate": 0.0015,
      "loss": 1.6627,
      "step": 3235
    },
    {
      "epoch": 0.34135021097046414,
      "grad_norm": 0.6533997654914856,
      "learning_rate": 0.0015,
      "loss": 1.7082,
      "step": 3236
    },
    {
      "epoch": 0.34145569620253163,
      "grad_norm": 0.7731319665908813,
      "learning_rate": 0.0015,
      "loss": 1.6739,
      "step": 3237
    },
    {
      "epoch": 0.3415611814345992,
      "grad_norm": 0.7820937037467957,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 3238
    },
    {
      "epoch": 0.3416666666666667,
      "grad_norm": 0.5627657771110535,
      "learning_rate": 0.0015,
      "loss": 1.663,
      "step": 3239
    },
    {
      "epoch": 0.34177215189873417,
      "grad_norm": 0.5278221964836121,
      "learning_rate": 0.0015,
      "loss": 1.7004,
      "step": 3240
    },
    {
      "epoch": 0.3418776371308017,
      "grad_norm": 0.5488499402999878,
      "learning_rate": 0.0015,
      "loss": 1.6825,
      "step": 3241
    },
    {
      "epoch": 0.3419831223628692,
      "grad_norm": 0.524059534072876,
      "learning_rate": 0.0015,
      "loss": 1.7194,
      "step": 3242
    },
    {
      "epoch": 0.3420886075949367,
      "grad_norm": 0.4945333003997803,
      "learning_rate": 0.0015,
      "loss": 1.6782,
      "step": 3243
    },
    {
      "epoch": 0.3421940928270042,
      "grad_norm": 0.5472710728645325,
      "learning_rate": 0.0015,
      "loss": 1.7152,
      "step": 3244
    },
    {
      "epoch": 0.34229957805907174,
      "grad_norm": 0.46813690662384033,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3245
    },
    {
      "epoch": 0.34240506329113923,
      "grad_norm": 0.5513242483139038,
      "learning_rate": 0.0015,
      "loss": 1.6979,
      "step": 3246
    },
    {
      "epoch": 0.3425105485232067,
      "grad_norm": 0.4633464217185974,
      "learning_rate": 0.0015,
      "loss": 1.7162,
      "step": 3247
    },
    {
      "epoch": 0.3426160337552743,
      "grad_norm": 0.5948613286018372,
      "learning_rate": 0.0015,
      "loss": 1.7221,
      "step": 3248
    },
    {
      "epoch": 0.34272151898734177,
      "grad_norm": 0.6489377617835999,
      "learning_rate": 0.0015,
      "loss": 1.7034,
      "step": 3249
    },
    {
      "epoch": 0.34282700421940926,
      "grad_norm": 0.5473290085792542,
      "learning_rate": 0.0015,
      "loss": 1.7181,
      "step": 3250
    },
    {
      "epoch": 0.3429324894514768,
      "grad_norm": 0.6085308194160461,
      "learning_rate": 0.0015,
      "loss": 1.6865,
      "step": 3251
    },
    {
      "epoch": 0.3430379746835443,
      "grad_norm": 0.8947362899780273,
      "learning_rate": 0.0015,
      "loss": 1.7276,
      "step": 3252
    },
    {
      "epoch": 0.3431434599156118,
      "grad_norm": 0.6903753876686096,
      "learning_rate": 0.0015,
      "loss": 1.7199,
      "step": 3253
    },
    {
      "epoch": 0.34324894514767934,
      "grad_norm": 0.5013677477836609,
      "learning_rate": 0.0015,
      "loss": 1.6789,
      "step": 3254
    },
    {
      "epoch": 0.34335443037974683,
      "grad_norm": 0.690724790096283,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 3255
    },
    {
      "epoch": 0.3434599156118143,
      "grad_norm": 0.5765038132667542,
      "learning_rate": 0.0015,
      "loss": 1.6566,
      "step": 3256
    },
    {
      "epoch": 0.3435654008438819,
      "grad_norm": 0.5136321187019348,
      "learning_rate": 0.0015,
      "loss": 1.7373,
      "step": 3257
    },
    {
      "epoch": 0.34367088607594937,
      "grad_norm": 0.5360373854637146,
      "learning_rate": 0.0015,
      "loss": 1.6834,
      "step": 3258
    },
    {
      "epoch": 0.34377637130801686,
      "grad_norm": 0.4985573887825012,
      "learning_rate": 0.0015,
      "loss": 1.693,
      "step": 3259
    },
    {
      "epoch": 0.3438818565400844,
      "grad_norm": 0.5356287360191345,
      "learning_rate": 0.0015,
      "loss": 1.6777,
      "step": 3260
    },
    {
      "epoch": 0.3439873417721519,
      "grad_norm": 0.6050174832344055,
      "learning_rate": 0.0015,
      "loss": 1.7266,
      "step": 3261
    },
    {
      "epoch": 0.3440928270042194,
      "grad_norm": 0.6644989848136902,
      "learning_rate": 0.0015,
      "loss": 1.7146,
      "step": 3262
    },
    {
      "epoch": 0.34419831223628694,
      "grad_norm": 0.6384313106536865,
      "learning_rate": 0.0015,
      "loss": 1.7211,
      "step": 3263
    },
    {
      "epoch": 0.34430379746835443,
      "grad_norm": 0.6091856360435486,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 3264
    },
    {
      "epoch": 0.3444092827004219,
      "grad_norm": 0.6349808573722839,
      "learning_rate": 0.0015,
      "loss": 1.6829,
      "step": 3265
    },
    {
      "epoch": 0.3445147679324895,
      "grad_norm": 0.5130767226219177,
      "learning_rate": 0.0015,
      "loss": 1.7253,
      "step": 3266
    },
    {
      "epoch": 0.34462025316455697,
      "grad_norm": 0.6537332534790039,
      "learning_rate": 0.0015,
      "loss": 1.7367,
      "step": 3267
    },
    {
      "epoch": 0.34472573839662446,
      "grad_norm": 0.7101198434829712,
      "learning_rate": 0.0015,
      "loss": 1.7207,
      "step": 3268
    },
    {
      "epoch": 0.344831223628692,
      "grad_norm": 0.49545493721961975,
      "learning_rate": 0.0015,
      "loss": 1.7137,
      "step": 3269
    },
    {
      "epoch": 0.3449367088607595,
      "grad_norm": 0.646725058555603,
      "learning_rate": 0.0015,
      "loss": 1.7131,
      "step": 3270
    },
    {
      "epoch": 0.345042194092827,
      "grad_norm": 0.7167512774467468,
      "learning_rate": 0.0015,
      "loss": 1.6758,
      "step": 3271
    },
    {
      "epoch": 0.34514767932489454,
      "grad_norm": 0.6485745310783386,
      "learning_rate": 0.0015,
      "loss": 1.7517,
      "step": 3272
    },
    {
      "epoch": 0.34525316455696203,
      "grad_norm": 0.5508519411087036,
      "learning_rate": 0.0015,
      "loss": 1.7139,
      "step": 3273
    },
    {
      "epoch": 0.3453586497890295,
      "grad_norm": 0.7126834988594055,
      "learning_rate": 0.0015,
      "loss": 1.7087,
      "step": 3274
    },
    {
      "epoch": 0.3454641350210971,
      "grad_norm": 0.6241485476493835,
      "learning_rate": 0.0015,
      "loss": 1.7243,
      "step": 3275
    },
    {
      "epoch": 0.34556962025316457,
      "grad_norm": 0.55812007188797,
      "learning_rate": 0.0015,
      "loss": 1.7246,
      "step": 3276
    },
    {
      "epoch": 0.34567510548523206,
      "grad_norm": 0.5232017636299133,
      "learning_rate": 0.0015,
      "loss": 1.7105,
      "step": 3277
    },
    {
      "epoch": 0.34578059071729955,
      "grad_norm": 0.4928762912750244,
      "learning_rate": 0.0015,
      "loss": 1.6624,
      "step": 3278
    },
    {
      "epoch": 0.3458860759493671,
      "grad_norm": 0.5147992372512817,
      "learning_rate": 0.0015,
      "loss": 1.7166,
      "step": 3279
    },
    {
      "epoch": 0.3459915611814346,
      "grad_norm": 0.6979365348815918,
      "learning_rate": 0.0015,
      "loss": 1.7093,
      "step": 3280
    },
    {
      "epoch": 0.3460970464135021,
      "grad_norm": 0.5939792394638062,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 3281
    },
    {
      "epoch": 0.34620253164556963,
      "grad_norm": 0.5686169266700745,
      "learning_rate": 0.0015,
      "loss": 1.7079,
      "step": 3282
    },
    {
      "epoch": 0.3463080168776371,
      "grad_norm": 0.7128338813781738,
      "learning_rate": 0.0015,
      "loss": 1.6631,
      "step": 3283
    },
    {
      "epoch": 0.3464135021097046,
      "grad_norm": 0.6922389268875122,
      "learning_rate": 0.0015,
      "loss": 1.713,
      "step": 3284
    },
    {
      "epoch": 0.34651898734177217,
      "grad_norm": 0.5139843821525574,
      "learning_rate": 0.0015,
      "loss": 1.6965,
      "step": 3285
    },
    {
      "epoch": 0.34662447257383966,
      "grad_norm": 0.5343854427337646,
      "learning_rate": 0.0015,
      "loss": 1.721,
      "step": 3286
    },
    {
      "epoch": 0.34672995780590715,
      "grad_norm": 0.5553687810897827,
      "learning_rate": 0.0015,
      "loss": 1.7585,
      "step": 3287
    },
    {
      "epoch": 0.3468354430379747,
      "grad_norm": 0.504325270652771,
      "learning_rate": 0.0015,
      "loss": 1.6627,
      "step": 3288
    },
    {
      "epoch": 0.3469409282700422,
      "grad_norm": 0.5255725979804993,
      "learning_rate": 0.0015,
      "loss": 1.6705,
      "step": 3289
    },
    {
      "epoch": 0.3470464135021097,
      "grad_norm": 0.507466733455658,
      "learning_rate": 0.0015,
      "loss": 1.6276,
      "step": 3290
    },
    {
      "epoch": 0.34715189873417723,
      "grad_norm": 0.5654557943344116,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 3291
    },
    {
      "epoch": 0.3472573839662447,
      "grad_norm": 0.671724259853363,
      "learning_rate": 0.0015,
      "loss": 1.6957,
      "step": 3292
    },
    {
      "epoch": 0.3473628691983122,
      "grad_norm": 0.4625714421272278,
      "learning_rate": 0.0015,
      "loss": 1.6956,
      "step": 3293
    },
    {
      "epoch": 0.34746835443037977,
      "grad_norm": 0.6441212296485901,
      "learning_rate": 0.0015,
      "loss": 1.697,
      "step": 3294
    },
    {
      "epoch": 0.34757383966244726,
      "grad_norm": 0.6679786443710327,
      "learning_rate": 0.0015,
      "loss": 1.688,
      "step": 3295
    },
    {
      "epoch": 0.34767932489451475,
      "grad_norm": 0.4963196814060211,
      "learning_rate": 0.0015,
      "loss": 1.7361,
      "step": 3296
    },
    {
      "epoch": 0.3477848101265823,
      "grad_norm": 0.5643613338470459,
      "learning_rate": 0.0015,
      "loss": 1.7034,
      "step": 3297
    },
    {
      "epoch": 0.3478902953586498,
      "grad_norm": 0.6231514811515808,
      "learning_rate": 0.0015,
      "loss": 1.688,
      "step": 3298
    },
    {
      "epoch": 0.3479957805907173,
      "grad_norm": 0.5145208835601807,
      "learning_rate": 0.0015,
      "loss": 1.7219,
      "step": 3299
    },
    {
      "epoch": 0.34810126582278483,
      "grad_norm": 0.5588470697402954,
      "learning_rate": 0.0015,
      "loss": 1.7082,
      "step": 3300
    },
    {
      "epoch": 0.3482067510548523,
      "grad_norm": 0.661913275718689,
      "learning_rate": 0.0015,
      "loss": 1.7018,
      "step": 3301
    },
    {
      "epoch": 0.3483122362869198,
      "grad_norm": 0.551858127117157,
      "learning_rate": 0.0015,
      "loss": 1.6866,
      "step": 3302
    },
    {
      "epoch": 0.34841772151898737,
      "grad_norm": 0.5006975531578064,
      "learning_rate": 0.0015,
      "loss": 1.7312,
      "step": 3303
    },
    {
      "epoch": 0.34852320675105486,
      "grad_norm": 0.5418669581413269,
      "learning_rate": 0.0015,
      "loss": 1.7068,
      "step": 3304
    },
    {
      "epoch": 0.34862869198312235,
      "grad_norm": 0.6015976071357727,
      "learning_rate": 0.0015,
      "loss": 1.7178,
      "step": 3305
    },
    {
      "epoch": 0.3487341772151899,
      "grad_norm": 0.6234529614448547,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 3306
    },
    {
      "epoch": 0.3488396624472574,
      "grad_norm": 0.5737875699996948,
      "learning_rate": 0.0015,
      "loss": 1.6975,
      "step": 3307
    },
    {
      "epoch": 0.3489451476793249,
      "grad_norm": 0.44024980068206787,
      "learning_rate": 0.0015,
      "loss": 1.7066,
      "step": 3308
    },
    {
      "epoch": 0.3490506329113924,
      "grad_norm": 0.5740571022033691,
      "learning_rate": 0.0015,
      "loss": 1.7012,
      "step": 3309
    },
    {
      "epoch": 0.3491561181434599,
      "grad_norm": 0.5544969439506531,
      "learning_rate": 0.0015,
      "loss": 1.6758,
      "step": 3310
    },
    {
      "epoch": 0.3492616033755274,
      "grad_norm": 0.46073102951049805,
      "learning_rate": 0.0015,
      "loss": 1.6946,
      "step": 3311
    },
    {
      "epoch": 0.3493670886075949,
      "grad_norm": 0.4999867081642151,
      "learning_rate": 0.0015,
      "loss": 1.7307,
      "step": 3312
    },
    {
      "epoch": 0.34947257383966246,
      "grad_norm": 0.5623162984848022,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 3313
    },
    {
      "epoch": 0.34957805907172995,
      "grad_norm": 0.5500815510749817,
      "learning_rate": 0.0015,
      "loss": 1.7309,
      "step": 3314
    },
    {
      "epoch": 0.34968354430379744,
      "grad_norm": 0.5207936763763428,
      "learning_rate": 0.0015,
      "loss": 1.6843,
      "step": 3315
    },
    {
      "epoch": 0.349789029535865,
      "grad_norm": 0.5976893305778503,
      "learning_rate": 0.0015,
      "loss": 1.6713,
      "step": 3316
    },
    {
      "epoch": 0.3498945147679325,
      "grad_norm": 0.5292049050331116,
      "learning_rate": 0.0015,
      "loss": 1.7348,
      "step": 3317
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.48040205240249634,
      "learning_rate": 0.0015,
      "loss": 1.7415,
      "step": 3318
    },
    {
      "epoch": 0.3501054852320675,
      "grad_norm": 0.5762725472450256,
      "learning_rate": 0.0015,
      "loss": 1.6996,
      "step": 3319
    },
    {
      "epoch": 0.350210970464135,
      "grad_norm": 0.4599584639072418,
      "learning_rate": 0.0015,
      "loss": 1.7012,
      "step": 3320
    },
    {
      "epoch": 0.3503164556962025,
      "grad_norm": 0.5592414736747742,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 3321
    },
    {
      "epoch": 0.35042194092827006,
      "grad_norm": 0.6016160845756531,
      "learning_rate": 0.0015,
      "loss": 1.6848,
      "step": 3322
    },
    {
      "epoch": 0.35052742616033755,
      "grad_norm": 0.5023660659790039,
      "learning_rate": 0.0015,
      "loss": 1.7341,
      "step": 3323
    },
    {
      "epoch": 0.35063291139240504,
      "grad_norm": 0.5222054123878479,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3324
    },
    {
      "epoch": 0.3507383966244726,
      "grad_norm": 0.5310649275779724,
      "learning_rate": 0.0015,
      "loss": 1.7295,
      "step": 3325
    },
    {
      "epoch": 0.3508438818565401,
      "grad_norm": 0.5232838988304138,
      "learning_rate": 0.0015,
      "loss": 1.6732,
      "step": 3326
    },
    {
      "epoch": 0.3509493670886076,
      "grad_norm": 0.511332631111145,
      "learning_rate": 0.0015,
      "loss": 1.7079,
      "step": 3327
    },
    {
      "epoch": 0.3510548523206751,
      "grad_norm": 0.5557478666305542,
      "learning_rate": 0.0015,
      "loss": 1.7198,
      "step": 3328
    },
    {
      "epoch": 0.3511603375527426,
      "grad_norm": 0.5167666673660278,
      "learning_rate": 0.0015,
      "loss": 1.7036,
      "step": 3329
    },
    {
      "epoch": 0.3512658227848101,
      "grad_norm": 0.5409457087516785,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 3330
    },
    {
      "epoch": 0.35137130801687766,
      "grad_norm": 0.6356794834136963,
      "learning_rate": 0.0015,
      "loss": 1.6929,
      "step": 3331
    },
    {
      "epoch": 0.35147679324894515,
      "grad_norm": 0.5135475397109985,
      "learning_rate": 0.0015,
      "loss": 1.6845,
      "step": 3332
    },
    {
      "epoch": 0.35158227848101264,
      "grad_norm": 0.5805373191833496,
      "learning_rate": 0.0015,
      "loss": 1.7524,
      "step": 3333
    },
    {
      "epoch": 0.3516877637130802,
      "grad_norm": 0.865698516368866,
      "learning_rate": 0.0015,
      "loss": 1.6803,
      "step": 3334
    },
    {
      "epoch": 0.3517932489451477,
      "grad_norm": 0.718709409236908,
      "learning_rate": 0.0015,
      "loss": 1.6771,
      "step": 3335
    },
    {
      "epoch": 0.3518987341772152,
      "grad_norm": 0.5176778435707092,
      "learning_rate": 0.0015,
      "loss": 1.6953,
      "step": 3336
    },
    {
      "epoch": 0.3520042194092827,
      "grad_norm": 0.835837721824646,
      "learning_rate": 0.0015,
      "loss": 1.698,
      "step": 3337
    },
    {
      "epoch": 0.3521097046413502,
      "grad_norm": 0.7923001646995544,
      "learning_rate": 0.0015,
      "loss": 1.7018,
      "step": 3338
    },
    {
      "epoch": 0.3522151898734177,
      "grad_norm": 0.6018409729003906,
      "learning_rate": 0.0015,
      "loss": 1.7262,
      "step": 3339
    },
    {
      "epoch": 0.35232067510548526,
      "grad_norm": 0.5894249677658081,
      "learning_rate": 0.0015,
      "loss": 1.678,
      "step": 3340
    },
    {
      "epoch": 0.35242616033755275,
      "grad_norm": 0.6654856204986572,
      "learning_rate": 0.0015,
      "loss": 1.661,
      "step": 3341
    },
    {
      "epoch": 0.35253164556962024,
      "grad_norm": 0.5093351006507874,
      "learning_rate": 0.0015,
      "loss": 1.7185,
      "step": 3342
    },
    {
      "epoch": 0.35263713080168774,
      "grad_norm": 0.6728792786598206,
      "learning_rate": 0.0015,
      "loss": 1.6778,
      "step": 3343
    },
    {
      "epoch": 0.3527426160337553,
      "grad_norm": 0.5834736227989197,
      "learning_rate": 0.0015,
      "loss": 1.7246,
      "step": 3344
    },
    {
      "epoch": 0.3528481012658228,
      "grad_norm": 0.5098950266838074,
      "learning_rate": 0.0015,
      "loss": 1.7201,
      "step": 3345
    },
    {
      "epoch": 0.35295358649789027,
      "grad_norm": 0.6730625033378601,
      "learning_rate": 0.0015,
      "loss": 1.6871,
      "step": 3346
    },
    {
      "epoch": 0.3530590717299578,
      "grad_norm": 0.6545631289482117,
      "learning_rate": 0.0015,
      "loss": 1.705,
      "step": 3347
    },
    {
      "epoch": 0.3531645569620253,
      "grad_norm": 0.5896464586257935,
      "learning_rate": 0.0015,
      "loss": 1.6806,
      "step": 3348
    },
    {
      "epoch": 0.3532700421940928,
      "grad_norm": 0.5974704027175903,
      "learning_rate": 0.0015,
      "loss": 1.6507,
      "step": 3349
    },
    {
      "epoch": 0.35337552742616035,
      "grad_norm": 0.7370885014533997,
      "learning_rate": 0.0015,
      "loss": 1.6856,
      "step": 3350
    },
    {
      "epoch": 0.35348101265822784,
      "grad_norm": 0.651846706867218,
      "learning_rate": 0.0015,
      "loss": 1.6806,
      "step": 3351
    },
    {
      "epoch": 0.35358649789029534,
      "grad_norm": 0.5507676601409912,
      "learning_rate": 0.0015,
      "loss": 1.7234,
      "step": 3352
    },
    {
      "epoch": 0.3536919831223629,
      "grad_norm": 0.8470252752304077,
      "learning_rate": 0.0015,
      "loss": 1.6935,
      "step": 3353
    },
    {
      "epoch": 0.3537974683544304,
      "grad_norm": 0.6926529407501221,
      "learning_rate": 0.0015,
      "loss": 1.6617,
      "step": 3354
    },
    {
      "epoch": 0.35390295358649787,
      "grad_norm": 0.5039480924606323,
      "learning_rate": 0.0015,
      "loss": 1.7087,
      "step": 3355
    },
    {
      "epoch": 0.3540084388185654,
      "grad_norm": 0.6330252885818481,
      "learning_rate": 0.0015,
      "loss": 1.6756,
      "step": 3356
    },
    {
      "epoch": 0.3541139240506329,
      "grad_norm": 0.6043148040771484,
      "learning_rate": 0.0015,
      "loss": 1.6967,
      "step": 3357
    },
    {
      "epoch": 0.3542194092827004,
      "grad_norm": 0.5623552203178406,
      "learning_rate": 0.0015,
      "loss": 1.6649,
      "step": 3358
    },
    {
      "epoch": 0.35432489451476795,
      "grad_norm": 0.633874237537384,
      "learning_rate": 0.0015,
      "loss": 1.6943,
      "step": 3359
    },
    {
      "epoch": 0.35443037974683544,
      "grad_norm": 0.6607420444488525,
      "learning_rate": 0.0015,
      "loss": 1.715,
      "step": 3360
    },
    {
      "epoch": 0.35453586497890294,
      "grad_norm": 0.5188843011856079,
      "learning_rate": 0.0015,
      "loss": 1.7042,
      "step": 3361
    },
    {
      "epoch": 0.3546413502109705,
      "grad_norm": 0.6697191596031189,
      "learning_rate": 0.0015,
      "loss": 1.7113,
      "step": 3362
    },
    {
      "epoch": 0.354746835443038,
      "grad_norm": 0.6414597630500793,
      "learning_rate": 0.0015,
      "loss": 1.7045,
      "step": 3363
    },
    {
      "epoch": 0.35485232067510547,
      "grad_norm": 0.49116820096969604,
      "learning_rate": 0.0015,
      "loss": 1.6996,
      "step": 3364
    },
    {
      "epoch": 0.354957805907173,
      "grad_norm": 0.5321905016899109,
      "learning_rate": 0.0015,
      "loss": 1.6555,
      "step": 3365
    },
    {
      "epoch": 0.3550632911392405,
      "grad_norm": 0.587928831577301,
      "learning_rate": 0.0015,
      "loss": 1.69,
      "step": 3366
    },
    {
      "epoch": 0.355168776371308,
      "grad_norm": 0.46752479672431946,
      "learning_rate": 0.0015,
      "loss": 1.7064,
      "step": 3367
    },
    {
      "epoch": 0.35527426160337555,
      "grad_norm": 0.5310870409011841,
      "learning_rate": 0.0015,
      "loss": 1.7015,
      "step": 3368
    },
    {
      "epoch": 0.35537974683544304,
      "grad_norm": 0.47283926606178284,
      "learning_rate": 0.0015,
      "loss": 1.725,
      "step": 3369
    },
    {
      "epoch": 0.35548523206751054,
      "grad_norm": 0.48126205801963806,
      "learning_rate": 0.0015,
      "loss": 1.6881,
      "step": 3370
    },
    {
      "epoch": 0.3555907172995781,
      "grad_norm": 0.49770766496658325,
      "learning_rate": 0.0015,
      "loss": 1.6998,
      "step": 3371
    },
    {
      "epoch": 0.3556962025316456,
      "grad_norm": 0.6166954040527344,
      "learning_rate": 0.0015,
      "loss": 1.6695,
      "step": 3372
    },
    {
      "epoch": 0.35580168776371307,
      "grad_norm": 0.5064780712127686,
      "learning_rate": 0.0015,
      "loss": 1.6667,
      "step": 3373
    },
    {
      "epoch": 0.35590717299578056,
      "grad_norm": 0.559788703918457,
      "learning_rate": 0.0015,
      "loss": 1.7025,
      "step": 3374
    },
    {
      "epoch": 0.3560126582278481,
      "grad_norm": 0.5952900648117065,
      "learning_rate": 0.0015,
      "loss": 1.714,
      "step": 3375
    },
    {
      "epoch": 0.3561181434599156,
      "grad_norm": 0.6047338843345642,
      "learning_rate": 0.0015,
      "loss": 1.7156,
      "step": 3376
    },
    {
      "epoch": 0.3562236286919831,
      "grad_norm": 0.6075417995452881,
      "learning_rate": 0.0015,
      "loss": 1.6837,
      "step": 3377
    },
    {
      "epoch": 0.35632911392405064,
      "grad_norm": 0.5179601907730103,
      "learning_rate": 0.0015,
      "loss": 1.6903,
      "step": 3378
    },
    {
      "epoch": 0.35643459915611814,
      "grad_norm": 0.6053937673568726,
      "learning_rate": 0.0015,
      "loss": 1.7,
      "step": 3379
    },
    {
      "epoch": 0.35654008438818563,
      "grad_norm": 0.6172323822975159,
      "learning_rate": 0.0015,
      "loss": 1.6984,
      "step": 3380
    },
    {
      "epoch": 0.3566455696202532,
      "grad_norm": 0.5589747428894043,
      "learning_rate": 0.0015,
      "loss": 1.6824,
      "step": 3381
    },
    {
      "epoch": 0.35675105485232067,
      "grad_norm": 0.6385650038719177,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 3382
    },
    {
      "epoch": 0.35685654008438816,
      "grad_norm": 0.5532436370849609,
      "learning_rate": 0.0015,
      "loss": 1.7325,
      "step": 3383
    },
    {
      "epoch": 0.3569620253164557,
      "grad_norm": 0.6474589705467224,
      "learning_rate": 0.0015,
      "loss": 1.7054,
      "step": 3384
    },
    {
      "epoch": 0.3570675105485232,
      "grad_norm": 0.6483468413352966,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 3385
    },
    {
      "epoch": 0.3571729957805907,
      "grad_norm": 0.4812987148761749,
      "learning_rate": 0.0015,
      "loss": 1.6583,
      "step": 3386
    },
    {
      "epoch": 0.35727848101265824,
      "grad_norm": 0.6307127475738525,
      "learning_rate": 0.0015,
      "loss": 1.7284,
      "step": 3387
    },
    {
      "epoch": 0.35738396624472574,
      "grad_norm": 0.6140430569648743,
      "learning_rate": 0.0015,
      "loss": 1.714,
      "step": 3388
    },
    {
      "epoch": 0.35748945147679323,
      "grad_norm": 0.5917832255363464,
      "learning_rate": 0.0015,
      "loss": 1.7245,
      "step": 3389
    },
    {
      "epoch": 0.3575949367088608,
      "grad_norm": 0.5120094418525696,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 3390
    },
    {
      "epoch": 0.35770042194092827,
      "grad_norm": 0.5447616577148438,
      "learning_rate": 0.0015,
      "loss": 1.6904,
      "step": 3391
    },
    {
      "epoch": 0.35780590717299576,
      "grad_norm": 0.6497238278388977,
      "learning_rate": 0.0015,
      "loss": 1.6673,
      "step": 3392
    },
    {
      "epoch": 0.3579113924050633,
      "grad_norm": 0.526225745677948,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 3393
    },
    {
      "epoch": 0.3580168776371308,
      "grad_norm": 0.5980810523033142,
      "learning_rate": 0.0015,
      "loss": 1.6607,
      "step": 3394
    },
    {
      "epoch": 0.3581223628691983,
      "grad_norm": 0.6649575233459473,
      "learning_rate": 0.0015,
      "loss": 1.7108,
      "step": 3395
    },
    {
      "epoch": 0.35822784810126584,
      "grad_norm": 0.4878542125225067,
      "learning_rate": 0.0015,
      "loss": 1.7025,
      "step": 3396
    },
    {
      "epoch": 0.35833333333333334,
      "grad_norm": 0.6462936401367188,
      "learning_rate": 0.0015,
      "loss": 1.6887,
      "step": 3397
    },
    {
      "epoch": 0.35843881856540083,
      "grad_norm": 0.6519804000854492,
      "learning_rate": 0.0015,
      "loss": 1.6975,
      "step": 3398
    },
    {
      "epoch": 0.3585443037974684,
      "grad_norm": 0.5274231433868408,
      "learning_rate": 0.0015,
      "loss": 1.7348,
      "step": 3399
    },
    {
      "epoch": 0.35864978902953587,
      "grad_norm": 0.5992415547370911,
      "learning_rate": 0.0015,
      "loss": 1.746,
      "step": 3400
    },
    {
      "epoch": 0.35875527426160336,
      "grad_norm": 0.6707643866539001,
      "learning_rate": 0.0015,
      "loss": 1.7174,
      "step": 3401
    },
    {
      "epoch": 0.3588607594936709,
      "grad_norm": 0.5592296719551086,
      "learning_rate": 0.0015,
      "loss": 1.6891,
      "step": 3402
    },
    {
      "epoch": 0.3589662447257384,
      "grad_norm": 0.8578243255615234,
      "learning_rate": 0.0015,
      "loss": 1.7337,
      "step": 3403
    },
    {
      "epoch": 0.3590717299578059,
      "grad_norm": 1.013196587562561,
      "learning_rate": 0.0015,
      "loss": 1.6993,
      "step": 3404
    },
    {
      "epoch": 0.35917721518987344,
      "grad_norm": 0.48511528968811035,
      "learning_rate": 0.0015,
      "loss": 1.6781,
      "step": 3405
    },
    {
      "epoch": 0.35928270042194094,
      "grad_norm": 0.7466984391212463,
      "learning_rate": 0.0015,
      "loss": 1.7116,
      "step": 3406
    },
    {
      "epoch": 0.35938818565400843,
      "grad_norm": 0.7377890944480896,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 3407
    },
    {
      "epoch": 0.3594936708860759,
      "grad_norm": 0.5139774084091187,
      "learning_rate": 0.0015,
      "loss": 1.6971,
      "step": 3408
    },
    {
      "epoch": 0.35959915611814347,
      "grad_norm": 0.5240907073020935,
      "learning_rate": 0.0015,
      "loss": 1.7014,
      "step": 3409
    },
    {
      "epoch": 0.35970464135021096,
      "grad_norm": 0.5521343350410461,
      "learning_rate": 0.0015,
      "loss": 1.6623,
      "step": 3410
    },
    {
      "epoch": 0.35981012658227846,
      "grad_norm": 0.5915968418121338,
      "learning_rate": 0.0015,
      "loss": 1.6941,
      "step": 3411
    },
    {
      "epoch": 0.359915611814346,
      "grad_norm": 0.4795050024986267,
      "learning_rate": 0.0015,
      "loss": 1.6764,
      "step": 3412
    },
    {
      "epoch": 0.3600210970464135,
      "grad_norm": 0.5866679549217224,
      "learning_rate": 0.0015,
      "loss": 1.688,
      "step": 3413
    },
    {
      "epoch": 0.360126582278481,
      "grad_norm": 0.5616958737373352,
      "learning_rate": 0.0015,
      "loss": 1.6571,
      "step": 3414
    },
    {
      "epoch": 0.36023206751054854,
      "grad_norm": 0.5041908025741577,
      "learning_rate": 0.0015,
      "loss": 1.654,
      "step": 3415
    },
    {
      "epoch": 0.36033755274261603,
      "grad_norm": 0.5769246220588684,
      "learning_rate": 0.0015,
      "loss": 1.6911,
      "step": 3416
    },
    {
      "epoch": 0.3604430379746835,
      "grad_norm": 0.4528326094150543,
      "learning_rate": 0.0015,
      "loss": 1.6819,
      "step": 3417
    },
    {
      "epoch": 0.36054852320675107,
      "grad_norm": 0.6060860753059387,
      "learning_rate": 0.0015,
      "loss": 1.6593,
      "step": 3418
    },
    {
      "epoch": 0.36065400843881856,
      "grad_norm": 0.6590308547019958,
      "learning_rate": 0.0015,
      "loss": 1.658,
      "step": 3419
    },
    {
      "epoch": 0.36075949367088606,
      "grad_norm": 0.4749763309955597,
      "learning_rate": 0.0015,
      "loss": 1.6813,
      "step": 3420
    },
    {
      "epoch": 0.3608649789029536,
      "grad_norm": 0.5621888637542725,
      "learning_rate": 0.0015,
      "loss": 1.7103,
      "step": 3421
    },
    {
      "epoch": 0.3609704641350211,
      "grad_norm": 0.5397945642471313,
      "learning_rate": 0.0015,
      "loss": 1.7349,
      "step": 3422
    },
    {
      "epoch": 0.3610759493670886,
      "grad_norm": 0.5255492925643921,
      "learning_rate": 0.0015,
      "loss": 1.69,
      "step": 3423
    },
    {
      "epoch": 0.36118143459915614,
      "grad_norm": 0.5748757123947144,
      "learning_rate": 0.0015,
      "loss": 1.7167,
      "step": 3424
    },
    {
      "epoch": 0.36128691983122363,
      "grad_norm": 0.5286357998847961,
      "learning_rate": 0.0015,
      "loss": 1.7132,
      "step": 3425
    },
    {
      "epoch": 0.3613924050632911,
      "grad_norm": 0.5291661620140076,
      "learning_rate": 0.0015,
      "loss": 1.666,
      "step": 3426
    },
    {
      "epoch": 0.36149789029535867,
      "grad_norm": 0.6045889854431152,
      "learning_rate": 0.0015,
      "loss": 1.7058,
      "step": 3427
    },
    {
      "epoch": 0.36160337552742616,
      "grad_norm": 0.5522410869598389,
      "learning_rate": 0.0015,
      "loss": 1.6802,
      "step": 3428
    },
    {
      "epoch": 0.36170886075949366,
      "grad_norm": 0.5047609806060791,
      "learning_rate": 0.0015,
      "loss": 1.6703,
      "step": 3429
    },
    {
      "epoch": 0.3618143459915612,
      "grad_norm": 0.6251084804534912,
      "learning_rate": 0.0015,
      "loss": 1.6601,
      "step": 3430
    },
    {
      "epoch": 0.3619198312236287,
      "grad_norm": 0.5223478674888611,
      "learning_rate": 0.0015,
      "loss": 1.6906,
      "step": 3431
    },
    {
      "epoch": 0.3620253164556962,
      "grad_norm": 0.5176872611045837,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 3432
    },
    {
      "epoch": 0.36213080168776374,
      "grad_norm": 0.648583710193634,
      "learning_rate": 0.0015,
      "loss": 1.6961,
      "step": 3433
    },
    {
      "epoch": 0.36223628691983123,
      "grad_norm": 0.5592969655990601,
      "learning_rate": 0.0015,
      "loss": 1.665,
      "step": 3434
    },
    {
      "epoch": 0.3623417721518987,
      "grad_norm": 0.5014697313308716,
      "learning_rate": 0.0015,
      "loss": 1.7031,
      "step": 3435
    },
    {
      "epoch": 0.36244725738396627,
      "grad_norm": 0.4879317879676819,
      "learning_rate": 0.0015,
      "loss": 1.6671,
      "step": 3436
    },
    {
      "epoch": 0.36255274261603376,
      "grad_norm": 0.5001959204673767,
      "learning_rate": 0.0015,
      "loss": 1.7284,
      "step": 3437
    },
    {
      "epoch": 0.36265822784810126,
      "grad_norm": 0.50409334897995,
      "learning_rate": 0.0015,
      "loss": 1.7124,
      "step": 3438
    },
    {
      "epoch": 0.3627637130801688,
      "grad_norm": 0.4925854504108429,
      "learning_rate": 0.0015,
      "loss": 1.6638,
      "step": 3439
    },
    {
      "epoch": 0.3628691983122363,
      "grad_norm": 0.5322001576423645,
      "learning_rate": 0.0015,
      "loss": 1.6791,
      "step": 3440
    },
    {
      "epoch": 0.3629746835443038,
      "grad_norm": 0.5502275824546814,
      "learning_rate": 0.0015,
      "loss": 1.6581,
      "step": 3441
    },
    {
      "epoch": 0.3630801687763713,
      "grad_norm": 0.5756074786186218,
      "learning_rate": 0.0015,
      "loss": 1.697,
      "step": 3442
    },
    {
      "epoch": 0.36318565400843883,
      "grad_norm": 0.5817195773124695,
      "learning_rate": 0.0015,
      "loss": 1.7084,
      "step": 3443
    },
    {
      "epoch": 0.3632911392405063,
      "grad_norm": 0.5023086667060852,
      "learning_rate": 0.0015,
      "loss": 1.6589,
      "step": 3444
    },
    {
      "epoch": 0.3633966244725738,
      "grad_norm": 0.5629333257675171,
      "learning_rate": 0.0015,
      "loss": 1.7129,
      "step": 3445
    },
    {
      "epoch": 0.36350210970464136,
      "grad_norm": 0.6314513087272644,
      "learning_rate": 0.0015,
      "loss": 1.7207,
      "step": 3446
    },
    {
      "epoch": 0.36360759493670886,
      "grad_norm": 0.46431276202201843,
      "learning_rate": 0.0015,
      "loss": 1.6628,
      "step": 3447
    },
    {
      "epoch": 0.36371308016877635,
      "grad_norm": 0.5817503333091736,
      "learning_rate": 0.0015,
      "loss": 1.6999,
      "step": 3448
    },
    {
      "epoch": 0.3638185654008439,
      "grad_norm": 0.5450338125228882,
      "learning_rate": 0.0015,
      "loss": 1.6854,
      "step": 3449
    },
    {
      "epoch": 0.3639240506329114,
      "grad_norm": 0.4929513931274414,
      "learning_rate": 0.0015,
      "loss": 1.6556,
      "step": 3450
    },
    {
      "epoch": 0.3640295358649789,
      "grad_norm": 0.5406349897384644,
      "learning_rate": 0.0015,
      "loss": 1.6832,
      "step": 3451
    },
    {
      "epoch": 0.36413502109704643,
      "grad_norm": 0.5617015361785889,
      "learning_rate": 0.0015,
      "loss": 1.6448,
      "step": 3452
    },
    {
      "epoch": 0.3642405063291139,
      "grad_norm": 0.4781974256038666,
      "learning_rate": 0.0015,
      "loss": 1.686,
      "step": 3453
    },
    {
      "epoch": 0.3643459915611814,
      "grad_norm": 0.46796074509620667,
      "learning_rate": 0.0015,
      "loss": 1.6588,
      "step": 3454
    },
    {
      "epoch": 0.36445147679324896,
      "grad_norm": 0.4808714687824249,
      "learning_rate": 0.0015,
      "loss": 1.7232,
      "step": 3455
    },
    {
      "epoch": 0.36455696202531646,
      "grad_norm": 0.508586585521698,
      "learning_rate": 0.0015,
      "loss": 1.6751,
      "step": 3456
    },
    {
      "epoch": 0.36466244725738395,
      "grad_norm": 0.5230977535247803,
      "learning_rate": 0.0015,
      "loss": 1.6868,
      "step": 3457
    },
    {
      "epoch": 0.3647679324894515,
      "grad_norm": 0.75013267993927,
      "learning_rate": 0.0015,
      "loss": 1.7005,
      "step": 3458
    },
    {
      "epoch": 0.364873417721519,
      "grad_norm": 0.889573872089386,
      "learning_rate": 0.0015,
      "loss": 1.6587,
      "step": 3459
    },
    {
      "epoch": 0.3649789029535865,
      "grad_norm": 0.5489333271980286,
      "learning_rate": 0.0015,
      "loss": 1.6915,
      "step": 3460
    },
    {
      "epoch": 0.36508438818565403,
      "grad_norm": 0.7285661697387695,
      "learning_rate": 0.0015,
      "loss": 1.7091,
      "step": 3461
    },
    {
      "epoch": 0.3651898734177215,
      "grad_norm": 0.6302562355995178,
      "learning_rate": 0.0015,
      "loss": 1.6871,
      "step": 3462
    },
    {
      "epoch": 0.365295358649789,
      "grad_norm": 0.6151975989341736,
      "learning_rate": 0.0015,
      "loss": 1.704,
      "step": 3463
    },
    {
      "epoch": 0.36540084388185656,
      "grad_norm": 0.749674916267395,
      "learning_rate": 0.0015,
      "loss": 1.7232,
      "step": 3464
    },
    {
      "epoch": 0.36550632911392406,
      "grad_norm": 0.6529875993728638,
      "learning_rate": 0.0015,
      "loss": 1.6926,
      "step": 3465
    },
    {
      "epoch": 0.36561181434599155,
      "grad_norm": 0.5716555714607239,
      "learning_rate": 0.0015,
      "loss": 1.6607,
      "step": 3466
    },
    {
      "epoch": 0.3657172995780591,
      "grad_norm": 0.5743492245674133,
      "learning_rate": 0.0015,
      "loss": 1.72,
      "step": 3467
    },
    {
      "epoch": 0.3658227848101266,
      "grad_norm": 0.5516340136528015,
      "learning_rate": 0.0015,
      "loss": 1.6534,
      "step": 3468
    },
    {
      "epoch": 0.3659282700421941,
      "grad_norm": 0.5176778435707092,
      "learning_rate": 0.0015,
      "loss": 1.7363,
      "step": 3469
    },
    {
      "epoch": 0.36603375527426163,
      "grad_norm": 0.7524895071983337,
      "learning_rate": 0.0015,
      "loss": 1.7043,
      "step": 3470
    },
    {
      "epoch": 0.3661392405063291,
      "grad_norm": 0.6692596673965454,
      "learning_rate": 0.0015,
      "loss": 1.6868,
      "step": 3471
    },
    {
      "epoch": 0.3662447257383966,
      "grad_norm": 0.5070008039474487,
      "learning_rate": 0.0015,
      "loss": 1.7232,
      "step": 3472
    },
    {
      "epoch": 0.3663502109704641,
      "grad_norm": 0.5661128163337708,
      "learning_rate": 0.0015,
      "loss": 1.7427,
      "step": 3473
    },
    {
      "epoch": 0.36645569620253166,
      "grad_norm": 0.5610405802726746,
      "learning_rate": 0.0015,
      "loss": 1.7179,
      "step": 3474
    },
    {
      "epoch": 0.36656118143459915,
      "grad_norm": 0.5527447462081909,
      "learning_rate": 0.0015,
      "loss": 1.7357,
      "step": 3475
    },
    {
      "epoch": 0.36666666666666664,
      "grad_norm": 0.6391531229019165,
      "learning_rate": 0.0015,
      "loss": 1.6924,
      "step": 3476
    },
    {
      "epoch": 0.3667721518987342,
      "grad_norm": 0.5474950075149536,
      "learning_rate": 0.0015,
      "loss": 1.693,
      "step": 3477
    },
    {
      "epoch": 0.3668776371308017,
      "grad_norm": 0.5395394563674927,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 3478
    },
    {
      "epoch": 0.3669831223628692,
      "grad_norm": 0.7075352668762207,
      "learning_rate": 0.0015,
      "loss": 1.6958,
      "step": 3479
    },
    {
      "epoch": 0.3670886075949367,
      "grad_norm": 0.6539794206619263,
      "learning_rate": 0.0015,
      "loss": 1.6706,
      "step": 3480
    },
    {
      "epoch": 0.3671940928270042,
      "grad_norm": 0.4946688115596771,
      "learning_rate": 0.0015,
      "loss": 1.705,
      "step": 3481
    },
    {
      "epoch": 0.3672995780590717,
      "grad_norm": 0.7209545373916626,
      "learning_rate": 0.0015,
      "loss": 1.7183,
      "step": 3482
    },
    {
      "epoch": 0.36740506329113926,
      "grad_norm": 0.526837170124054,
      "learning_rate": 0.0015,
      "loss": 1.6724,
      "step": 3483
    },
    {
      "epoch": 0.36751054852320675,
      "grad_norm": 0.7148000001907349,
      "learning_rate": 0.0015,
      "loss": 1.7337,
      "step": 3484
    },
    {
      "epoch": 0.36761603375527424,
      "grad_norm": 0.7433557510375977,
      "learning_rate": 0.0015,
      "loss": 1.7034,
      "step": 3485
    },
    {
      "epoch": 0.3677215189873418,
      "grad_norm": 0.6516178250312805,
      "learning_rate": 0.0015,
      "loss": 1.6902,
      "step": 3486
    },
    {
      "epoch": 0.3678270042194093,
      "grad_norm": 0.6386226415634155,
      "learning_rate": 0.0015,
      "loss": 1.6812,
      "step": 3487
    },
    {
      "epoch": 0.3679324894514768,
      "grad_norm": 0.6712808012962341,
      "learning_rate": 0.0015,
      "loss": 1.6975,
      "step": 3488
    },
    {
      "epoch": 0.3680379746835443,
      "grad_norm": 0.5676007866859436,
      "learning_rate": 0.0015,
      "loss": 1.69,
      "step": 3489
    },
    {
      "epoch": 0.3681434599156118,
      "grad_norm": 0.6194564700126648,
      "learning_rate": 0.0015,
      "loss": 1.706,
      "step": 3490
    },
    {
      "epoch": 0.3682489451476793,
      "grad_norm": 0.4818286895751953,
      "learning_rate": 0.0015,
      "loss": 1.6859,
      "step": 3491
    },
    {
      "epoch": 0.36835443037974686,
      "grad_norm": 0.6806401610374451,
      "learning_rate": 0.0015,
      "loss": 1.6761,
      "step": 3492
    },
    {
      "epoch": 0.36845991561181435,
      "grad_norm": 0.6845285296440125,
      "learning_rate": 0.0015,
      "loss": 1.6628,
      "step": 3493
    },
    {
      "epoch": 0.36856540084388184,
      "grad_norm": 0.6205683350563049,
      "learning_rate": 0.0015,
      "loss": 1.7087,
      "step": 3494
    },
    {
      "epoch": 0.3686708860759494,
      "grad_norm": 0.8780078887939453,
      "learning_rate": 0.0015,
      "loss": 1.6959,
      "step": 3495
    },
    {
      "epoch": 0.3687763713080169,
      "grad_norm": 0.9626905918121338,
      "learning_rate": 0.0015,
      "loss": 1.6691,
      "step": 3496
    },
    {
      "epoch": 0.3688818565400844,
      "grad_norm": 0.5961605906486511,
      "learning_rate": 0.0015,
      "loss": 1.6969,
      "step": 3497
    },
    {
      "epoch": 0.3689873417721519,
      "grad_norm": 0.5036250948905945,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 3498
    },
    {
      "epoch": 0.3690928270042194,
      "grad_norm": 0.5939358472824097,
      "learning_rate": 0.0015,
      "loss": 1.6715,
      "step": 3499
    },
    {
      "epoch": 0.3691983122362869,
      "grad_norm": 0.5004081130027771,
      "learning_rate": 0.0015,
      "loss": 1.6406,
      "step": 3500
    },
    {
      "epoch": 0.36930379746835446,
      "grad_norm": 0.5770102739334106,
      "learning_rate": 0.0015,
      "loss": 1.669,
      "step": 3501
    },
    {
      "epoch": 0.36940928270042195,
      "grad_norm": 0.6931666135787964,
      "learning_rate": 0.0015,
      "loss": 1.7007,
      "step": 3502
    },
    {
      "epoch": 0.36951476793248944,
      "grad_norm": 0.48336613178253174,
      "learning_rate": 0.0015,
      "loss": 1.7153,
      "step": 3503
    },
    {
      "epoch": 0.369620253164557,
      "grad_norm": 0.6822316646575928,
      "learning_rate": 0.0015,
      "loss": 1.7156,
      "step": 3504
    },
    {
      "epoch": 0.3697257383966245,
      "grad_norm": 0.7500801682472229,
      "learning_rate": 0.0015,
      "loss": 1.6706,
      "step": 3505
    },
    {
      "epoch": 0.369831223628692,
      "grad_norm": 0.52363520860672,
      "learning_rate": 0.0015,
      "loss": 1.6952,
      "step": 3506
    },
    {
      "epoch": 0.36993670886075947,
      "grad_norm": 0.5796246528625488,
      "learning_rate": 0.0015,
      "loss": 1.7191,
      "step": 3507
    },
    {
      "epoch": 0.370042194092827,
      "grad_norm": 0.6043279767036438,
      "learning_rate": 0.0015,
      "loss": 1.6686,
      "step": 3508
    },
    {
      "epoch": 0.3701476793248945,
      "grad_norm": 0.5411955118179321,
      "learning_rate": 0.0015,
      "loss": 1.6904,
      "step": 3509
    },
    {
      "epoch": 0.370253164556962,
      "grad_norm": 0.6494709849357605,
      "learning_rate": 0.0015,
      "loss": 1.6949,
      "step": 3510
    },
    {
      "epoch": 0.37035864978902955,
      "grad_norm": 0.5679773092269897,
      "learning_rate": 0.0015,
      "loss": 1.6993,
      "step": 3511
    },
    {
      "epoch": 0.37046413502109704,
      "grad_norm": 0.5008822083473206,
      "learning_rate": 0.0015,
      "loss": 1.6691,
      "step": 3512
    },
    {
      "epoch": 0.37056962025316453,
      "grad_norm": 0.597727358341217,
      "learning_rate": 0.0015,
      "loss": 1.6763,
      "step": 3513
    },
    {
      "epoch": 0.3706751054852321,
      "grad_norm": 0.44443100690841675,
      "learning_rate": 0.0015,
      "loss": 1.6866,
      "step": 3514
    },
    {
      "epoch": 0.3707805907172996,
      "grad_norm": 0.6263906955718994,
      "learning_rate": 0.0015,
      "loss": 1.6779,
      "step": 3515
    },
    {
      "epoch": 0.37088607594936707,
      "grad_norm": 0.5585700273513794,
      "learning_rate": 0.0015,
      "loss": 1.72,
      "step": 3516
    },
    {
      "epoch": 0.3709915611814346,
      "grad_norm": 0.5889336466789246,
      "learning_rate": 0.0015,
      "loss": 1.7009,
      "step": 3517
    },
    {
      "epoch": 0.3710970464135021,
      "grad_norm": 0.6008414626121521,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 3518
    },
    {
      "epoch": 0.3712025316455696,
      "grad_norm": 0.5108413100242615,
      "learning_rate": 0.0015,
      "loss": 1.7061,
      "step": 3519
    },
    {
      "epoch": 0.37130801687763715,
      "grad_norm": 0.5178086757659912,
      "learning_rate": 0.0015,
      "loss": 1.6999,
      "step": 3520
    },
    {
      "epoch": 0.37141350210970464,
      "grad_norm": 0.5006667971611023,
      "learning_rate": 0.0015,
      "loss": 1.7104,
      "step": 3521
    },
    {
      "epoch": 0.37151898734177213,
      "grad_norm": 0.44895803928375244,
      "learning_rate": 0.0015,
      "loss": 1.6833,
      "step": 3522
    },
    {
      "epoch": 0.3716244725738397,
      "grad_norm": 0.4816954731941223,
      "learning_rate": 0.0015,
      "loss": 1.6781,
      "step": 3523
    },
    {
      "epoch": 0.3717299578059072,
      "grad_norm": 0.5086579918861389,
      "learning_rate": 0.0015,
      "loss": 1.6406,
      "step": 3524
    },
    {
      "epoch": 0.37183544303797467,
      "grad_norm": 0.5456579327583313,
      "learning_rate": 0.0015,
      "loss": 1.7272,
      "step": 3525
    },
    {
      "epoch": 0.3719409282700422,
      "grad_norm": 0.45087578892707825,
      "learning_rate": 0.0015,
      "loss": 1.7092,
      "step": 3526
    },
    {
      "epoch": 0.3720464135021097,
      "grad_norm": 0.46919184923171997,
      "learning_rate": 0.0015,
      "loss": 1.6826,
      "step": 3527
    },
    {
      "epoch": 0.3721518987341772,
      "grad_norm": 0.47783172130584717,
      "learning_rate": 0.0015,
      "loss": 1.7211,
      "step": 3528
    },
    {
      "epoch": 0.37225738396624475,
      "grad_norm": 0.4555913209915161,
      "learning_rate": 0.0015,
      "loss": 1.6597,
      "step": 3529
    },
    {
      "epoch": 0.37236286919831224,
      "grad_norm": 0.514130711555481,
      "learning_rate": 0.0015,
      "loss": 1.7586,
      "step": 3530
    },
    {
      "epoch": 0.37246835443037973,
      "grad_norm": 0.6121523380279541,
      "learning_rate": 0.0015,
      "loss": 1.7147,
      "step": 3531
    },
    {
      "epoch": 0.3725738396624473,
      "grad_norm": 0.6196393370628357,
      "learning_rate": 0.0015,
      "loss": 1.7315,
      "step": 3532
    },
    {
      "epoch": 0.3726793248945148,
      "grad_norm": 0.4390181303024292,
      "learning_rate": 0.0015,
      "loss": 1.668,
      "step": 3533
    },
    {
      "epoch": 0.37278481012658227,
      "grad_norm": 0.47705987095832825,
      "learning_rate": 0.0015,
      "loss": 1.7102,
      "step": 3534
    },
    {
      "epoch": 0.3728902953586498,
      "grad_norm": 0.4802907705307007,
      "learning_rate": 0.0015,
      "loss": 1.7155,
      "step": 3535
    },
    {
      "epoch": 0.3729957805907173,
      "grad_norm": 0.4534493386745453,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 3536
    },
    {
      "epoch": 0.3731012658227848,
      "grad_norm": 0.5185301899909973,
      "learning_rate": 0.0015,
      "loss": 1.66,
      "step": 3537
    },
    {
      "epoch": 0.37320675105485235,
      "grad_norm": 0.5581870675086975,
      "learning_rate": 0.0015,
      "loss": 1.7176,
      "step": 3538
    },
    {
      "epoch": 0.37331223628691984,
      "grad_norm": 0.5368343591690063,
      "learning_rate": 0.0015,
      "loss": 1.6821,
      "step": 3539
    },
    {
      "epoch": 0.37341772151898733,
      "grad_norm": 0.5538505911827087,
      "learning_rate": 0.0015,
      "loss": 1.676,
      "step": 3540
    },
    {
      "epoch": 0.3735232067510548,
      "grad_norm": 0.7850205302238464,
      "learning_rate": 0.0015,
      "loss": 1.6848,
      "step": 3541
    },
    {
      "epoch": 0.3736286919831224,
      "grad_norm": 0.8336942791938782,
      "learning_rate": 0.0015,
      "loss": 1.7212,
      "step": 3542
    },
    {
      "epoch": 0.37373417721518987,
      "grad_norm": 0.5784370303153992,
      "learning_rate": 0.0015,
      "loss": 1.676,
      "step": 3543
    },
    {
      "epoch": 0.37383966244725736,
      "grad_norm": 0.5390932559967041,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 3544
    },
    {
      "epoch": 0.3739451476793249,
      "grad_norm": 0.530036449432373,
      "learning_rate": 0.0015,
      "loss": 1.7075,
      "step": 3545
    },
    {
      "epoch": 0.3740506329113924,
      "grad_norm": 0.5599520802497864,
      "learning_rate": 0.0015,
      "loss": 1.6757,
      "step": 3546
    },
    {
      "epoch": 0.3741561181434599,
      "grad_norm": 0.5387759208679199,
      "learning_rate": 0.0015,
      "loss": 1.6474,
      "step": 3547
    },
    {
      "epoch": 0.37426160337552744,
      "grad_norm": 0.47868961095809937,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 3548
    },
    {
      "epoch": 0.37436708860759493,
      "grad_norm": 0.5268681049346924,
      "learning_rate": 0.0015,
      "loss": 1.6768,
      "step": 3549
    },
    {
      "epoch": 0.3744725738396624,
      "grad_norm": 0.5130245089530945,
      "learning_rate": 0.0015,
      "loss": 1.6945,
      "step": 3550
    },
    {
      "epoch": 0.37457805907173,
      "grad_norm": 0.5134006142616272,
      "learning_rate": 0.0015,
      "loss": 1.668,
      "step": 3551
    },
    {
      "epoch": 0.37468354430379747,
      "grad_norm": 0.4919327199459076,
      "learning_rate": 0.0015,
      "loss": 1.6785,
      "step": 3552
    },
    {
      "epoch": 0.37478902953586496,
      "grad_norm": 0.48739781975746155,
      "learning_rate": 0.0015,
      "loss": 1.6938,
      "step": 3553
    },
    {
      "epoch": 0.3748945147679325,
      "grad_norm": 0.48200276494026184,
      "learning_rate": 0.0015,
      "loss": 1.6828,
      "step": 3554
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.4728035628795624,
      "learning_rate": 0.0015,
      "loss": 1.6841,
      "step": 3555
    },
    {
      "epoch": 0.3751054852320675,
      "grad_norm": 0.5085388422012329,
      "learning_rate": 0.0015,
      "loss": 1.688,
      "step": 3556
    },
    {
      "epoch": 0.37521097046413504,
      "grad_norm": 0.44792282581329346,
      "learning_rate": 0.0015,
      "loss": 1.69,
      "step": 3557
    },
    {
      "epoch": 0.37531645569620253,
      "grad_norm": 0.4960862398147583,
      "learning_rate": 0.0015,
      "loss": 1.6586,
      "step": 3558
    },
    {
      "epoch": 0.37542194092827,
      "grad_norm": 0.47321751713752747,
      "learning_rate": 0.0015,
      "loss": 1.6827,
      "step": 3559
    },
    {
      "epoch": 0.3755274261603376,
      "grad_norm": 0.45882028341293335,
      "learning_rate": 0.0015,
      "loss": 1.6731,
      "step": 3560
    },
    {
      "epoch": 0.37563291139240507,
      "grad_norm": 0.4504358768463135,
      "learning_rate": 0.0015,
      "loss": 1.6895,
      "step": 3561
    },
    {
      "epoch": 0.37573839662447256,
      "grad_norm": 0.5343787670135498,
      "learning_rate": 0.0015,
      "loss": 1.6657,
      "step": 3562
    },
    {
      "epoch": 0.3758438818565401,
      "grad_norm": 0.4659075140953064,
      "learning_rate": 0.0015,
      "loss": 1.6703,
      "step": 3563
    },
    {
      "epoch": 0.3759493670886076,
      "grad_norm": 0.5344029664993286,
      "learning_rate": 0.0015,
      "loss": 1.6789,
      "step": 3564
    },
    {
      "epoch": 0.3760548523206751,
      "grad_norm": 0.5266735553741455,
      "learning_rate": 0.0015,
      "loss": 1.6796,
      "step": 3565
    },
    {
      "epoch": 0.37616033755274264,
      "grad_norm": 0.5037487149238586,
      "learning_rate": 0.0015,
      "loss": 1.6501,
      "step": 3566
    },
    {
      "epoch": 0.37626582278481013,
      "grad_norm": 0.56449294090271,
      "learning_rate": 0.0015,
      "loss": 1.6799,
      "step": 3567
    },
    {
      "epoch": 0.3763713080168776,
      "grad_norm": 0.5012770295143127,
      "learning_rate": 0.0015,
      "loss": 1.7164,
      "step": 3568
    },
    {
      "epoch": 0.3764767932489452,
      "grad_norm": 0.6296719312667847,
      "learning_rate": 0.0015,
      "loss": 1.6677,
      "step": 3569
    },
    {
      "epoch": 0.37658227848101267,
      "grad_norm": 0.7946227788925171,
      "learning_rate": 0.0015,
      "loss": 1.7011,
      "step": 3570
    },
    {
      "epoch": 0.37668776371308016,
      "grad_norm": 0.5402594208717346,
      "learning_rate": 0.0015,
      "loss": 1.6501,
      "step": 3571
    },
    {
      "epoch": 0.37679324894514765,
      "grad_norm": 0.6937465071678162,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 3572
    },
    {
      "epoch": 0.3768987341772152,
      "grad_norm": 0.9502573013305664,
      "learning_rate": 0.0015,
      "loss": 1.6942,
      "step": 3573
    },
    {
      "epoch": 0.3770042194092827,
      "grad_norm": 0.7560267448425293,
      "learning_rate": 0.0015,
      "loss": 1.7044,
      "step": 3574
    },
    {
      "epoch": 0.3771097046413502,
      "grad_norm": 0.6838465929031372,
      "learning_rate": 0.0015,
      "loss": 1.6969,
      "step": 3575
    },
    {
      "epoch": 0.37721518987341773,
      "grad_norm": 0.5273986458778381,
      "learning_rate": 0.0015,
      "loss": 1.6884,
      "step": 3576
    },
    {
      "epoch": 0.3773206751054852,
      "grad_norm": 0.6105398535728455,
      "learning_rate": 0.0015,
      "loss": 1.6734,
      "step": 3577
    },
    {
      "epoch": 0.3774261603375527,
      "grad_norm": 0.6233408451080322,
      "learning_rate": 0.0015,
      "loss": 1.683,
      "step": 3578
    },
    {
      "epoch": 0.37753164556962027,
      "grad_norm": 0.5808678865432739,
      "learning_rate": 0.0015,
      "loss": 1.6754,
      "step": 3579
    },
    {
      "epoch": 0.37763713080168776,
      "grad_norm": 0.8989353179931641,
      "learning_rate": 0.0015,
      "loss": 1.7109,
      "step": 3580
    },
    {
      "epoch": 0.37774261603375525,
      "grad_norm": 0.8094796538352966,
      "learning_rate": 0.0015,
      "loss": 1.7011,
      "step": 3581
    },
    {
      "epoch": 0.3778481012658228,
      "grad_norm": 0.5858950614929199,
      "learning_rate": 0.0015,
      "loss": 1.6706,
      "step": 3582
    },
    {
      "epoch": 0.3779535864978903,
      "grad_norm": 0.7963961362838745,
      "learning_rate": 0.0015,
      "loss": 1.6938,
      "step": 3583
    },
    {
      "epoch": 0.3780590717299578,
      "grad_norm": 0.8486836552619934,
      "learning_rate": 0.0015,
      "loss": 1.6775,
      "step": 3584
    },
    {
      "epoch": 0.37816455696202533,
      "grad_norm": 0.502236008644104,
      "learning_rate": 0.0015,
      "loss": 1.7269,
      "step": 3585
    },
    {
      "epoch": 0.3782700421940928,
      "grad_norm": 0.8007808327674866,
      "learning_rate": 0.0015,
      "loss": 1.7224,
      "step": 3586
    },
    {
      "epoch": 0.3783755274261603,
      "grad_norm": 0.6669566631317139,
      "learning_rate": 0.0015,
      "loss": 1.6765,
      "step": 3587
    },
    {
      "epoch": 0.37848101265822787,
      "grad_norm": 0.5253190398216248,
      "learning_rate": 0.0015,
      "loss": 1.6831,
      "step": 3588
    },
    {
      "epoch": 0.37858649789029536,
      "grad_norm": 0.6445272564888,
      "learning_rate": 0.0015,
      "loss": 1.727,
      "step": 3589
    },
    {
      "epoch": 0.37869198312236285,
      "grad_norm": 0.5694237947463989,
      "learning_rate": 0.0015,
      "loss": 1.7016,
      "step": 3590
    },
    {
      "epoch": 0.3787974683544304,
      "grad_norm": 0.5158687829971313,
      "learning_rate": 0.0015,
      "loss": 1.6597,
      "step": 3591
    },
    {
      "epoch": 0.3789029535864979,
      "grad_norm": 0.5416021347045898,
      "learning_rate": 0.0015,
      "loss": 1.6818,
      "step": 3592
    },
    {
      "epoch": 0.3790084388185654,
      "grad_norm": 0.5724902153015137,
      "learning_rate": 0.0015,
      "loss": 1.7238,
      "step": 3593
    },
    {
      "epoch": 0.37911392405063293,
      "grad_norm": 0.5864480137825012,
      "learning_rate": 0.0015,
      "loss": 1.6853,
      "step": 3594
    },
    {
      "epoch": 0.3792194092827004,
      "grad_norm": 0.516903281211853,
      "learning_rate": 0.0015,
      "loss": 1.6973,
      "step": 3595
    },
    {
      "epoch": 0.3793248945147679,
      "grad_norm": 0.6607723236083984,
      "learning_rate": 0.0015,
      "loss": 1.6425,
      "step": 3596
    },
    {
      "epoch": 0.37943037974683547,
      "grad_norm": 0.5852009654045105,
      "learning_rate": 0.0015,
      "loss": 1.6743,
      "step": 3597
    },
    {
      "epoch": 0.37953586497890296,
      "grad_norm": 0.5943871140480042,
      "learning_rate": 0.0015,
      "loss": 1.6878,
      "step": 3598
    },
    {
      "epoch": 0.37964135021097045,
      "grad_norm": 0.5274428129196167,
      "learning_rate": 0.0015,
      "loss": 1.661,
      "step": 3599
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 0.9265272617340088,
      "learning_rate": 0.0015,
      "loss": 1.6695,
      "step": 3600
    },
    {
      "epoch": 0.3798523206751055,
      "grad_norm": 0.7401799559593201,
      "learning_rate": 0.0015,
      "loss": 1.6856,
      "step": 3601
    },
    {
      "epoch": 0.379957805907173,
      "grad_norm": 0.5657160878181458,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 3602
    },
    {
      "epoch": 0.38006329113924053,
      "grad_norm": 0.5467101335525513,
      "learning_rate": 0.0015,
      "loss": 1.6838,
      "step": 3603
    },
    {
      "epoch": 0.380168776371308,
      "grad_norm": 0.6029835343360901,
      "learning_rate": 0.0015,
      "loss": 1.6596,
      "step": 3604
    },
    {
      "epoch": 0.3802742616033755,
      "grad_norm": 0.5620792508125305,
      "learning_rate": 0.0015,
      "loss": 1.7066,
      "step": 3605
    },
    {
      "epoch": 0.380379746835443,
      "grad_norm": 0.5603548288345337,
      "learning_rate": 0.0015,
      "loss": 1.6749,
      "step": 3606
    },
    {
      "epoch": 0.38048523206751056,
      "grad_norm": 0.6930073499679565,
      "learning_rate": 0.0015,
      "loss": 1.6731,
      "step": 3607
    },
    {
      "epoch": 0.38059071729957805,
      "grad_norm": 0.5464161038398743,
      "learning_rate": 0.0015,
      "loss": 1.7097,
      "step": 3608
    },
    {
      "epoch": 0.38069620253164554,
      "grad_norm": 0.5768519639968872,
      "learning_rate": 0.0015,
      "loss": 1.7014,
      "step": 3609
    },
    {
      "epoch": 0.3808016877637131,
      "grad_norm": 0.5359461307525635,
      "learning_rate": 0.0015,
      "loss": 1.6362,
      "step": 3610
    },
    {
      "epoch": 0.3809071729957806,
      "grad_norm": 0.6051581501960754,
      "learning_rate": 0.0015,
      "loss": 1.7067,
      "step": 3611
    },
    {
      "epoch": 0.3810126582278481,
      "grad_norm": 0.7465226054191589,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 3612
    },
    {
      "epoch": 0.3811181434599156,
      "grad_norm": 0.7165592908859253,
      "learning_rate": 0.0015,
      "loss": 1.6878,
      "step": 3613
    },
    {
      "epoch": 0.3812236286919831,
      "grad_norm": 0.6232821941375732,
      "learning_rate": 0.0015,
      "loss": 1.6578,
      "step": 3614
    },
    {
      "epoch": 0.3813291139240506,
      "grad_norm": 0.6930992603302002,
      "learning_rate": 0.0015,
      "loss": 1.6385,
      "step": 3615
    },
    {
      "epoch": 0.38143459915611816,
      "grad_norm": 0.6751329302787781,
      "learning_rate": 0.0015,
      "loss": 1.6862,
      "step": 3616
    },
    {
      "epoch": 0.38154008438818565,
      "grad_norm": 0.5513862371444702,
      "learning_rate": 0.0015,
      "loss": 1.7347,
      "step": 3617
    },
    {
      "epoch": 0.38164556962025314,
      "grad_norm": 0.689549446105957,
      "learning_rate": 0.0015,
      "loss": 1.7143,
      "step": 3618
    },
    {
      "epoch": 0.3817510548523207,
      "grad_norm": 0.6732983589172363,
      "learning_rate": 0.0015,
      "loss": 1.6811,
      "step": 3619
    },
    {
      "epoch": 0.3818565400843882,
      "grad_norm": 0.5490033030509949,
      "learning_rate": 0.0015,
      "loss": 1.648,
      "step": 3620
    },
    {
      "epoch": 0.3819620253164557,
      "grad_norm": 0.6244326233863831,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 3621
    },
    {
      "epoch": 0.3820675105485232,
      "grad_norm": 0.6083469986915588,
      "learning_rate": 0.0015,
      "loss": 1.6538,
      "step": 3622
    },
    {
      "epoch": 0.3821729957805907,
      "grad_norm": 0.4693094789981842,
      "learning_rate": 0.0015,
      "loss": 1.6975,
      "step": 3623
    },
    {
      "epoch": 0.3822784810126582,
      "grad_norm": 0.6091998219490051,
      "learning_rate": 0.0015,
      "loss": 1.6792,
      "step": 3624
    },
    {
      "epoch": 0.38238396624472576,
      "grad_norm": 0.4976237416267395,
      "learning_rate": 0.0015,
      "loss": 1.6932,
      "step": 3625
    },
    {
      "epoch": 0.38248945147679325,
      "grad_norm": 0.5420821905136108,
      "learning_rate": 0.0015,
      "loss": 1.6832,
      "step": 3626
    },
    {
      "epoch": 0.38259493670886074,
      "grad_norm": 0.5881454944610596,
      "learning_rate": 0.0015,
      "loss": 1.6845,
      "step": 3627
    },
    {
      "epoch": 0.3827004219409283,
      "grad_norm": 0.6723463535308838,
      "learning_rate": 0.0015,
      "loss": 1.6944,
      "step": 3628
    },
    {
      "epoch": 0.3828059071729958,
      "grad_norm": 0.4542931020259857,
      "learning_rate": 0.0015,
      "loss": 1.6569,
      "step": 3629
    },
    {
      "epoch": 0.3829113924050633,
      "grad_norm": 0.7008521556854248,
      "learning_rate": 0.0015,
      "loss": 1.6824,
      "step": 3630
    },
    {
      "epoch": 0.3830168776371308,
      "grad_norm": 0.8241995573043823,
      "learning_rate": 0.0015,
      "loss": 1.6934,
      "step": 3631
    },
    {
      "epoch": 0.3831223628691983,
      "grad_norm": 0.5964398980140686,
      "learning_rate": 0.0015,
      "loss": 1.672,
      "step": 3632
    },
    {
      "epoch": 0.3832278481012658,
      "grad_norm": 0.61642986536026,
      "learning_rate": 0.0015,
      "loss": 1.712,
      "step": 3633
    },
    {
      "epoch": 0.38333333333333336,
      "grad_norm": 0.7122082710266113,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 3634
    },
    {
      "epoch": 0.38343881856540085,
      "grad_norm": 0.4951452314853668,
      "learning_rate": 0.0015,
      "loss": 1.6825,
      "step": 3635
    },
    {
      "epoch": 0.38354430379746834,
      "grad_norm": 0.7421321272850037,
      "learning_rate": 0.0015,
      "loss": 1.667,
      "step": 3636
    },
    {
      "epoch": 0.3836497890295359,
      "grad_norm": 0.8305424451828003,
      "learning_rate": 0.0015,
      "loss": 1.7546,
      "step": 3637
    },
    {
      "epoch": 0.3837552742616034,
      "grad_norm": 0.5369387269020081,
      "learning_rate": 0.0015,
      "loss": 1.6798,
      "step": 3638
    },
    {
      "epoch": 0.3838607594936709,
      "grad_norm": 1.0308220386505127,
      "learning_rate": 0.0015,
      "loss": 1.6897,
      "step": 3639
    },
    {
      "epoch": 0.38396624472573837,
      "grad_norm": 0.7028903365135193,
      "learning_rate": 0.0015,
      "loss": 1.6426,
      "step": 3640
    },
    {
      "epoch": 0.3840717299578059,
      "grad_norm": 0.6929757595062256,
      "learning_rate": 0.0015,
      "loss": 1.676,
      "step": 3641
    },
    {
      "epoch": 0.3841772151898734,
      "grad_norm": 0.8227697610855103,
      "learning_rate": 0.0015,
      "loss": 1.7256,
      "step": 3642
    },
    {
      "epoch": 0.3842827004219409,
      "grad_norm": 0.641666829586029,
      "learning_rate": 0.0015,
      "loss": 1.6802,
      "step": 3643
    },
    {
      "epoch": 0.38438818565400845,
      "grad_norm": 0.5322589874267578,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 3644
    },
    {
      "epoch": 0.38449367088607594,
      "grad_norm": 0.5577198266983032,
      "learning_rate": 0.0015,
      "loss": 1.6821,
      "step": 3645
    },
    {
      "epoch": 0.38459915611814344,
      "grad_norm": 0.6370434761047363,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 3646
    },
    {
      "epoch": 0.384704641350211,
      "grad_norm": 0.6994079351425171,
      "learning_rate": 0.0015,
      "loss": 1.6883,
      "step": 3647
    },
    {
      "epoch": 0.3848101265822785,
      "grad_norm": 0.5557304620742798,
      "learning_rate": 0.0015,
      "loss": 1.7202,
      "step": 3648
    },
    {
      "epoch": 0.38491561181434597,
      "grad_norm": 0.5214542150497437,
      "learning_rate": 0.0015,
      "loss": 1.7004,
      "step": 3649
    },
    {
      "epoch": 0.3850210970464135,
      "grad_norm": 0.6828441023826599,
      "learning_rate": 0.0015,
      "loss": 1.6893,
      "step": 3650
    },
    {
      "epoch": 0.385126582278481,
      "grad_norm": 0.5492691993713379,
      "learning_rate": 0.0015,
      "loss": 1.6958,
      "step": 3651
    },
    {
      "epoch": 0.3852320675105485,
      "grad_norm": 0.5482195019721985,
      "learning_rate": 0.0015,
      "loss": 1.6727,
      "step": 3652
    },
    {
      "epoch": 0.38533755274261605,
      "grad_norm": 0.7347174286842346,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 3653
    },
    {
      "epoch": 0.38544303797468354,
      "grad_norm": 0.5783145427703857,
      "learning_rate": 0.0015,
      "loss": 1.707,
      "step": 3654
    },
    {
      "epoch": 0.38554852320675104,
      "grad_norm": 0.5347406268119812,
      "learning_rate": 0.0015,
      "loss": 1.6933,
      "step": 3655
    },
    {
      "epoch": 0.3856540084388186,
      "grad_norm": 0.5907634496688843,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 3656
    },
    {
      "epoch": 0.3857594936708861,
      "grad_norm": 0.44710153341293335,
      "learning_rate": 0.0015,
      "loss": 1.6503,
      "step": 3657
    },
    {
      "epoch": 0.38586497890295357,
      "grad_norm": 0.5879906415939331,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 3658
    },
    {
      "epoch": 0.3859704641350211,
      "grad_norm": 0.5563458800315857,
      "learning_rate": 0.0015,
      "loss": 1.7,
      "step": 3659
    },
    {
      "epoch": 0.3860759493670886,
      "grad_norm": 0.5694467425346375,
      "learning_rate": 0.0015,
      "loss": 1.6745,
      "step": 3660
    },
    {
      "epoch": 0.3861814345991561,
      "grad_norm": 0.603987991809845,
      "learning_rate": 0.0015,
      "loss": 1.696,
      "step": 3661
    },
    {
      "epoch": 0.38628691983122365,
      "grad_norm": 0.5052303671836853,
      "learning_rate": 0.0015,
      "loss": 1.6646,
      "step": 3662
    },
    {
      "epoch": 0.38639240506329114,
      "grad_norm": 0.626789391040802,
      "learning_rate": 0.0015,
      "loss": 1.6805,
      "step": 3663
    },
    {
      "epoch": 0.38649789029535864,
      "grad_norm": 0.47768598794937134,
      "learning_rate": 0.0015,
      "loss": 1.6612,
      "step": 3664
    },
    {
      "epoch": 0.3866033755274262,
      "grad_norm": 0.6381677389144897,
      "learning_rate": 0.0015,
      "loss": 1.6644,
      "step": 3665
    },
    {
      "epoch": 0.3867088607594937,
      "grad_norm": 0.673236608505249,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3666
    },
    {
      "epoch": 0.38681434599156117,
      "grad_norm": 0.6323572993278503,
      "learning_rate": 0.0015,
      "loss": 1.6956,
      "step": 3667
    },
    {
      "epoch": 0.3869198312236287,
      "grad_norm": 0.5218010544776917,
      "learning_rate": 0.0015,
      "loss": 1.7002,
      "step": 3668
    },
    {
      "epoch": 0.3870253164556962,
      "grad_norm": 0.6393988728523254,
      "learning_rate": 0.0015,
      "loss": 1.667,
      "step": 3669
    },
    {
      "epoch": 0.3871308016877637,
      "grad_norm": 0.5353789925575256,
      "learning_rate": 0.0015,
      "loss": 1.6668,
      "step": 3670
    },
    {
      "epoch": 0.3872362869198312,
      "grad_norm": 0.5642716288566589,
      "learning_rate": 0.0015,
      "loss": 1.6861,
      "step": 3671
    },
    {
      "epoch": 0.38734177215189874,
      "grad_norm": 0.5615326762199402,
      "learning_rate": 0.0015,
      "loss": 1.6695,
      "step": 3672
    },
    {
      "epoch": 0.38744725738396624,
      "grad_norm": 0.5854267477989197,
      "learning_rate": 0.0015,
      "loss": 1.6752,
      "step": 3673
    },
    {
      "epoch": 0.38755274261603373,
      "grad_norm": 0.6541895270347595,
      "learning_rate": 0.0015,
      "loss": 1.6762,
      "step": 3674
    },
    {
      "epoch": 0.3876582278481013,
      "grad_norm": 0.5889910459518433,
      "learning_rate": 0.0015,
      "loss": 1.6572,
      "step": 3675
    },
    {
      "epoch": 0.38776371308016877,
      "grad_norm": 0.7357600331306458,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 3676
    },
    {
      "epoch": 0.38786919831223626,
      "grad_norm": 0.48305729031562805,
      "learning_rate": 0.0015,
      "loss": 1.692,
      "step": 3677
    },
    {
      "epoch": 0.3879746835443038,
      "grad_norm": 0.7296749353408813,
      "learning_rate": 0.0015,
      "loss": 1.7031,
      "step": 3678
    },
    {
      "epoch": 0.3880801687763713,
      "grad_norm": 0.4649110734462738,
      "learning_rate": 0.0015,
      "loss": 1.7132,
      "step": 3679
    },
    {
      "epoch": 0.3881856540084388,
      "grad_norm": 0.8049523830413818,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 3680
    },
    {
      "epoch": 0.38829113924050634,
      "grad_norm": 0.903431236743927,
      "learning_rate": 0.0015,
      "loss": 1.6914,
      "step": 3681
    },
    {
      "epoch": 0.38839662447257384,
      "grad_norm": 0.594926118850708,
      "learning_rate": 0.0015,
      "loss": 1.6904,
      "step": 3682
    },
    {
      "epoch": 0.38850210970464133,
      "grad_norm": 0.647506594657898,
      "learning_rate": 0.0015,
      "loss": 1.6956,
      "step": 3683
    },
    {
      "epoch": 0.3886075949367089,
      "grad_norm": 0.6687235832214355,
      "learning_rate": 0.0015,
      "loss": 1.7001,
      "step": 3684
    },
    {
      "epoch": 0.38871308016877637,
      "grad_norm": 0.5746626257896423,
      "learning_rate": 0.0015,
      "loss": 1.7193,
      "step": 3685
    },
    {
      "epoch": 0.38881856540084386,
      "grad_norm": 0.544668436050415,
      "learning_rate": 0.0015,
      "loss": 1.6703,
      "step": 3686
    },
    {
      "epoch": 0.3889240506329114,
      "grad_norm": 0.5631168484687805,
      "learning_rate": 0.0015,
      "loss": 1.6649,
      "step": 3687
    },
    {
      "epoch": 0.3890295358649789,
      "grad_norm": 0.6437835097312927,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 3688
    },
    {
      "epoch": 0.3891350210970464,
      "grad_norm": 0.6942353248596191,
      "learning_rate": 0.0015,
      "loss": 1.6989,
      "step": 3689
    },
    {
      "epoch": 0.38924050632911394,
      "grad_norm": 0.5834630727767944,
      "learning_rate": 0.0015,
      "loss": 1.6953,
      "step": 3690
    },
    {
      "epoch": 0.38934599156118144,
      "grad_norm": 0.6948363780975342,
      "learning_rate": 0.0015,
      "loss": 1.7541,
      "step": 3691
    },
    {
      "epoch": 0.38945147679324893,
      "grad_norm": 0.7779099345207214,
      "learning_rate": 0.0015,
      "loss": 1.683,
      "step": 3692
    },
    {
      "epoch": 0.3895569620253165,
      "grad_norm": 0.5741925239562988,
      "learning_rate": 0.0015,
      "loss": 1.6449,
      "step": 3693
    },
    {
      "epoch": 0.38966244725738397,
      "grad_norm": 0.6894943118095398,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 3694
    },
    {
      "epoch": 0.38976793248945146,
      "grad_norm": 0.5647335052490234,
      "learning_rate": 0.0015,
      "loss": 1.696,
      "step": 3695
    },
    {
      "epoch": 0.389873417721519,
      "grad_norm": 0.6527907252311707,
      "learning_rate": 0.0015,
      "loss": 1.6848,
      "step": 3696
    },
    {
      "epoch": 0.3899789029535865,
      "grad_norm": 0.5589083433151245,
      "learning_rate": 0.0015,
      "loss": 1.698,
      "step": 3697
    },
    {
      "epoch": 0.390084388185654,
      "grad_norm": 0.5453910827636719,
      "learning_rate": 0.0015,
      "loss": 1.7233,
      "step": 3698
    },
    {
      "epoch": 0.39018987341772154,
      "grad_norm": 0.563664972782135,
      "learning_rate": 0.0015,
      "loss": 1.6532,
      "step": 3699
    },
    {
      "epoch": 0.39029535864978904,
      "grad_norm": 0.46357178688049316,
      "learning_rate": 0.0015,
      "loss": 1.7509,
      "step": 3700
    },
    {
      "epoch": 0.39040084388185653,
      "grad_norm": 0.6345269083976746,
      "learning_rate": 0.0015,
      "loss": 1.696,
      "step": 3701
    },
    {
      "epoch": 0.3905063291139241,
      "grad_norm": 0.5798743367195129,
      "learning_rate": 0.0015,
      "loss": 1.6799,
      "step": 3702
    },
    {
      "epoch": 0.39061181434599157,
      "grad_norm": 0.5159527659416199,
      "learning_rate": 0.0015,
      "loss": 1.6489,
      "step": 3703
    },
    {
      "epoch": 0.39071729957805906,
      "grad_norm": 0.5321524739265442,
      "learning_rate": 0.0015,
      "loss": 1.6931,
      "step": 3704
    },
    {
      "epoch": 0.39082278481012656,
      "grad_norm": 0.4516725540161133,
      "learning_rate": 0.0015,
      "loss": 1.6685,
      "step": 3705
    },
    {
      "epoch": 0.3909282700421941,
      "grad_norm": 0.6135842800140381,
      "learning_rate": 0.0015,
      "loss": 1.7076,
      "step": 3706
    },
    {
      "epoch": 0.3910337552742616,
      "grad_norm": 0.53651362657547,
      "learning_rate": 0.0015,
      "loss": 1.7377,
      "step": 3707
    },
    {
      "epoch": 0.3911392405063291,
      "grad_norm": 0.4846319258213043,
      "learning_rate": 0.0015,
      "loss": 1.6698,
      "step": 3708
    },
    {
      "epoch": 0.39124472573839664,
      "grad_norm": 0.747732400894165,
      "learning_rate": 0.0015,
      "loss": 1.6904,
      "step": 3709
    },
    {
      "epoch": 0.39135021097046413,
      "grad_norm": 0.6844327449798584,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 3710
    },
    {
      "epoch": 0.3914556962025316,
      "grad_norm": 0.4895083010196686,
      "learning_rate": 0.0015,
      "loss": 1.6522,
      "step": 3711
    },
    {
      "epoch": 0.39156118143459917,
      "grad_norm": 0.5964961647987366,
      "learning_rate": 0.0015,
      "loss": 1.639,
      "step": 3712
    },
    {
      "epoch": 0.39166666666666666,
      "grad_norm": 0.5458452105522156,
      "learning_rate": 0.0015,
      "loss": 1.6464,
      "step": 3713
    },
    {
      "epoch": 0.39177215189873416,
      "grad_norm": 0.525438129901886,
      "learning_rate": 0.0015,
      "loss": 1.6764,
      "step": 3714
    },
    {
      "epoch": 0.3918776371308017,
      "grad_norm": 0.5726062059402466,
      "learning_rate": 0.0015,
      "loss": 1.6551,
      "step": 3715
    },
    {
      "epoch": 0.3919831223628692,
      "grad_norm": 0.5745319724082947,
      "learning_rate": 0.0015,
      "loss": 1.6547,
      "step": 3716
    },
    {
      "epoch": 0.3920886075949367,
      "grad_norm": 0.49845850467681885,
      "learning_rate": 0.0015,
      "loss": 1.6947,
      "step": 3717
    },
    {
      "epoch": 0.39219409282700424,
      "grad_norm": 0.5194739699363708,
      "learning_rate": 0.0015,
      "loss": 1.6184,
      "step": 3718
    },
    {
      "epoch": 0.39229957805907173,
      "grad_norm": 0.5362001657485962,
      "learning_rate": 0.0015,
      "loss": 1.6903,
      "step": 3719
    },
    {
      "epoch": 0.3924050632911392,
      "grad_norm": 0.4627896845340729,
      "learning_rate": 0.0015,
      "loss": 1.6628,
      "step": 3720
    },
    {
      "epoch": 0.39251054852320677,
      "grad_norm": 0.4402790665626526,
      "learning_rate": 0.0015,
      "loss": 1.6595,
      "step": 3721
    },
    {
      "epoch": 0.39261603375527426,
      "grad_norm": 0.532944917678833,
      "learning_rate": 0.0015,
      "loss": 1.6875,
      "step": 3722
    },
    {
      "epoch": 0.39272151898734176,
      "grad_norm": 0.47930651903152466,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 3723
    },
    {
      "epoch": 0.3928270042194093,
      "grad_norm": 0.5077054500579834,
      "learning_rate": 0.0015,
      "loss": 1.6914,
      "step": 3724
    },
    {
      "epoch": 0.3929324894514768,
      "grad_norm": 0.6024961471557617,
      "learning_rate": 0.0015,
      "loss": 1.6527,
      "step": 3725
    },
    {
      "epoch": 0.3930379746835443,
      "grad_norm": 0.5431039929389954,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 3726
    },
    {
      "epoch": 0.39314345991561184,
      "grad_norm": 0.5446413159370422,
      "learning_rate": 0.0015,
      "loss": 1.6777,
      "step": 3727
    },
    {
      "epoch": 0.39324894514767933,
      "grad_norm": 0.5875133275985718,
      "learning_rate": 0.0015,
      "loss": 1.6922,
      "step": 3728
    },
    {
      "epoch": 0.3933544303797468,
      "grad_norm": 0.4683308005332947,
      "learning_rate": 0.0015,
      "loss": 1.6927,
      "step": 3729
    },
    {
      "epoch": 0.39345991561181437,
      "grad_norm": 0.5920150876045227,
      "learning_rate": 0.0015,
      "loss": 1.6566,
      "step": 3730
    },
    {
      "epoch": 0.39356540084388186,
      "grad_norm": 0.5581077933311462,
      "learning_rate": 0.0015,
      "loss": 1.678,
      "step": 3731
    },
    {
      "epoch": 0.39367088607594936,
      "grad_norm": 0.4967675805091858,
      "learning_rate": 0.0015,
      "loss": 1.7229,
      "step": 3732
    },
    {
      "epoch": 0.3937763713080169,
      "grad_norm": 0.5351163744926453,
      "learning_rate": 0.0015,
      "loss": 1.6764,
      "step": 3733
    },
    {
      "epoch": 0.3938818565400844,
      "grad_norm": 0.5785093307495117,
      "learning_rate": 0.0015,
      "loss": 1.6846,
      "step": 3734
    },
    {
      "epoch": 0.3939873417721519,
      "grad_norm": 0.6192387342453003,
      "learning_rate": 0.0015,
      "loss": 1.698,
      "step": 3735
    },
    {
      "epoch": 0.39409282700421944,
      "grad_norm": 0.6738994121551514,
      "learning_rate": 0.0015,
      "loss": 1.6925,
      "step": 3736
    },
    {
      "epoch": 0.39419831223628693,
      "grad_norm": 0.48425158858299255,
      "learning_rate": 0.0015,
      "loss": 1.6502,
      "step": 3737
    },
    {
      "epoch": 0.3943037974683544,
      "grad_norm": 0.6813925504684448,
      "learning_rate": 0.0015,
      "loss": 1.6548,
      "step": 3738
    },
    {
      "epoch": 0.3944092827004219,
      "grad_norm": 0.6188768148422241,
      "learning_rate": 0.0015,
      "loss": 1.638,
      "step": 3739
    },
    {
      "epoch": 0.39451476793248946,
      "grad_norm": 0.5234917998313904,
      "learning_rate": 0.0015,
      "loss": 1.7169,
      "step": 3740
    },
    {
      "epoch": 0.39462025316455696,
      "grad_norm": 0.5968658924102783,
      "learning_rate": 0.0015,
      "loss": 1.6956,
      "step": 3741
    },
    {
      "epoch": 0.39472573839662445,
      "grad_norm": 0.5261411070823669,
      "learning_rate": 0.0015,
      "loss": 1.6959,
      "step": 3742
    },
    {
      "epoch": 0.394831223628692,
      "grad_norm": 0.46268901228904724,
      "learning_rate": 0.0015,
      "loss": 1.6725,
      "step": 3743
    },
    {
      "epoch": 0.3949367088607595,
      "grad_norm": 0.4687832295894623,
      "learning_rate": 0.0015,
      "loss": 1.7272,
      "step": 3744
    },
    {
      "epoch": 0.395042194092827,
      "grad_norm": 0.44519439339637756,
      "learning_rate": 0.0015,
      "loss": 1.6461,
      "step": 3745
    },
    {
      "epoch": 0.39514767932489453,
      "grad_norm": 0.5245404839515686,
      "learning_rate": 0.0015,
      "loss": 1.6872,
      "step": 3746
    },
    {
      "epoch": 0.395253164556962,
      "grad_norm": 0.6123425960540771,
      "learning_rate": 0.0015,
      "loss": 1.7192,
      "step": 3747
    },
    {
      "epoch": 0.3953586497890295,
      "grad_norm": 0.5783941149711609,
      "learning_rate": 0.0015,
      "loss": 1.6711,
      "step": 3748
    },
    {
      "epoch": 0.39546413502109706,
      "grad_norm": 0.5005989074707031,
      "learning_rate": 0.0015,
      "loss": 1.6433,
      "step": 3749
    },
    {
      "epoch": 0.39556962025316456,
      "grad_norm": 0.8036031723022461,
      "learning_rate": 0.0015,
      "loss": 1.6805,
      "step": 3750
    },
    {
      "epoch": 0.39567510548523205,
      "grad_norm": 0.5644263029098511,
      "learning_rate": 0.0015,
      "loss": 1.679,
      "step": 3751
    },
    {
      "epoch": 0.3957805907172996,
      "grad_norm": 0.6158437728881836,
      "learning_rate": 0.0015,
      "loss": 1.6959,
      "step": 3752
    },
    {
      "epoch": 0.3958860759493671,
      "grad_norm": 0.6980195641517639,
      "learning_rate": 0.0015,
      "loss": 1.6971,
      "step": 3753
    },
    {
      "epoch": 0.3959915611814346,
      "grad_norm": 0.5789486169815063,
      "learning_rate": 0.0015,
      "loss": 1.6834,
      "step": 3754
    },
    {
      "epoch": 0.39609704641350213,
      "grad_norm": 0.6175979971885681,
      "learning_rate": 0.0015,
      "loss": 1.6789,
      "step": 3755
    },
    {
      "epoch": 0.3962025316455696,
      "grad_norm": 0.771101176738739,
      "learning_rate": 0.0015,
      "loss": 1.6868,
      "step": 3756
    },
    {
      "epoch": 0.3963080168776371,
      "grad_norm": 0.6815974116325378,
      "learning_rate": 0.0015,
      "loss": 1.6821,
      "step": 3757
    },
    {
      "epoch": 0.39641350210970466,
      "grad_norm": 0.5077992677688599,
      "learning_rate": 0.0015,
      "loss": 1.6602,
      "step": 3758
    },
    {
      "epoch": 0.39651898734177216,
      "grad_norm": 0.5861387252807617,
      "learning_rate": 0.0015,
      "loss": 1.6371,
      "step": 3759
    },
    {
      "epoch": 0.39662447257383965,
      "grad_norm": 0.5313556790351868,
      "learning_rate": 0.0015,
      "loss": 1.6614,
      "step": 3760
    },
    {
      "epoch": 0.3967299578059072,
      "grad_norm": 0.7050797939300537,
      "learning_rate": 0.0015,
      "loss": 1.7039,
      "step": 3761
    },
    {
      "epoch": 0.3968354430379747,
      "grad_norm": 0.7246404886245728,
      "learning_rate": 0.0015,
      "loss": 1.6759,
      "step": 3762
    },
    {
      "epoch": 0.3969409282700422,
      "grad_norm": 0.49107906222343445,
      "learning_rate": 0.0015,
      "loss": 1.6749,
      "step": 3763
    },
    {
      "epoch": 0.39704641350210973,
      "grad_norm": 0.7503883838653564,
      "learning_rate": 0.0015,
      "loss": 1.6641,
      "step": 3764
    },
    {
      "epoch": 0.3971518987341772,
      "grad_norm": 0.5992181897163391,
      "learning_rate": 0.0015,
      "loss": 1.6818,
      "step": 3765
    },
    {
      "epoch": 0.3972573839662447,
      "grad_norm": 0.5650216937065125,
      "learning_rate": 0.0015,
      "loss": 1.6864,
      "step": 3766
    },
    {
      "epoch": 0.39736286919831226,
      "grad_norm": 0.726324737071991,
      "learning_rate": 0.0015,
      "loss": 1.7338,
      "step": 3767
    },
    {
      "epoch": 0.39746835443037976,
      "grad_norm": 0.5421167612075806,
      "learning_rate": 0.0015,
      "loss": 1.7374,
      "step": 3768
    },
    {
      "epoch": 0.39757383966244725,
      "grad_norm": 0.5234367847442627,
      "learning_rate": 0.0015,
      "loss": 1.6793,
      "step": 3769
    },
    {
      "epoch": 0.39767932489451474,
      "grad_norm": 0.7114390730857849,
      "learning_rate": 0.0015,
      "loss": 1.6803,
      "step": 3770
    },
    {
      "epoch": 0.3977848101265823,
      "grad_norm": 0.6242325305938721,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 3771
    },
    {
      "epoch": 0.3978902953586498,
      "grad_norm": 0.4838646650314331,
      "learning_rate": 0.0015,
      "loss": 1.6856,
      "step": 3772
    },
    {
      "epoch": 0.3979957805907173,
      "grad_norm": 0.6684880256652832,
      "learning_rate": 0.0015,
      "loss": 1.6957,
      "step": 3773
    },
    {
      "epoch": 0.3981012658227848,
      "grad_norm": 0.5313114523887634,
      "learning_rate": 0.0015,
      "loss": 1.7024,
      "step": 3774
    },
    {
      "epoch": 0.3982067510548523,
      "grad_norm": 0.4991389811038971,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 3775
    },
    {
      "epoch": 0.3983122362869198,
      "grad_norm": 0.5732834935188293,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 3776
    },
    {
      "epoch": 0.39841772151898736,
      "grad_norm": 0.4613068401813507,
      "learning_rate": 0.0015,
      "loss": 1.6583,
      "step": 3777
    },
    {
      "epoch": 0.39852320675105485,
      "grad_norm": 0.5855815410614014,
      "learning_rate": 0.0015,
      "loss": 1.6774,
      "step": 3778
    },
    {
      "epoch": 0.39862869198312234,
      "grad_norm": 0.6076484322547913,
      "learning_rate": 0.0015,
      "loss": 1.6883,
      "step": 3779
    },
    {
      "epoch": 0.3987341772151899,
      "grad_norm": 0.5026043057441711,
      "learning_rate": 0.0015,
      "loss": 1.6551,
      "step": 3780
    },
    {
      "epoch": 0.3988396624472574,
      "grad_norm": 0.6369363069534302,
      "learning_rate": 0.0015,
      "loss": 1.66,
      "step": 3781
    },
    {
      "epoch": 0.3989451476793249,
      "grad_norm": 0.6031306982040405,
      "learning_rate": 0.0015,
      "loss": 1.6576,
      "step": 3782
    },
    {
      "epoch": 0.3990506329113924,
      "grad_norm": 0.43691784143447876,
      "learning_rate": 0.0015,
      "loss": 1.7045,
      "step": 3783
    },
    {
      "epoch": 0.3991561181434599,
      "grad_norm": 0.6470429301261902,
      "learning_rate": 0.0015,
      "loss": 1.6453,
      "step": 3784
    },
    {
      "epoch": 0.3992616033755274,
      "grad_norm": 0.5600720643997192,
      "learning_rate": 0.0015,
      "loss": 1.7107,
      "step": 3785
    },
    {
      "epoch": 0.39936708860759496,
      "grad_norm": 0.4862491488456726,
      "learning_rate": 0.0015,
      "loss": 1.6978,
      "step": 3786
    },
    {
      "epoch": 0.39947257383966245,
      "grad_norm": 0.8172095417976379,
      "learning_rate": 0.0015,
      "loss": 1.6486,
      "step": 3787
    },
    {
      "epoch": 0.39957805907172994,
      "grad_norm": 0.7997248768806458,
      "learning_rate": 0.0015,
      "loss": 1.7115,
      "step": 3788
    },
    {
      "epoch": 0.3996835443037975,
      "grad_norm": 0.5401503443717957,
      "learning_rate": 0.0015,
      "loss": 1.672,
      "step": 3789
    },
    {
      "epoch": 0.399789029535865,
      "grad_norm": 0.47454023361206055,
      "learning_rate": 0.0015,
      "loss": 1.6829,
      "step": 3790
    },
    {
      "epoch": 0.3998945147679325,
      "grad_norm": 0.5431615710258484,
      "learning_rate": 0.0015,
      "loss": 1.6758,
      "step": 3791
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4792156517505646,
      "learning_rate": 0.0015,
      "loss": 1.6456,
      "step": 3792
    },
    {
      "epoch": 0.4001054852320675,
      "grad_norm": 0.5675703287124634,
      "learning_rate": 0.0015,
      "loss": 1.7152,
      "step": 3793
    },
    {
      "epoch": 0.400210970464135,
      "grad_norm": 0.5410193204879761,
      "learning_rate": 0.0015,
      "loss": 1.6924,
      "step": 3794
    },
    {
      "epoch": 0.40031645569620256,
      "grad_norm": 0.535546600818634,
      "learning_rate": 0.0015,
      "loss": 1.6979,
      "step": 3795
    },
    {
      "epoch": 0.40042194092827005,
      "grad_norm": 0.4831826090812683,
      "learning_rate": 0.0015,
      "loss": 1.7037,
      "step": 3796
    },
    {
      "epoch": 0.40052742616033754,
      "grad_norm": 0.5048959851264954,
      "learning_rate": 0.0015,
      "loss": 1.6748,
      "step": 3797
    },
    {
      "epoch": 0.4006329113924051,
      "grad_norm": 0.6044279932975769,
      "learning_rate": 0.0015,
      "loss": 1.7105,
      "step": 3798
    },
    {
      "epoch": 0.4007383966244726,
      "grad_norm": 0.5984317064285278,
      "learning_rate": 0.0015,
      "loss": 1.6391,
      "step": 3799
    },
    {
      "epoch": 0.4008438818565401,
      "grad_norm": 0.476684033870697,
      "learning_rate": 0.0015,
      "loss": 1.6675,
      "step": 3800
    },
    {
      "epoch": 0.4009493670886076,
      "grad_norm": 0.5303037166595459,
      "learning_rate": 0.0015,
      "loss": 1.6328,
      "step": 3801
    },
    {
      "epoch": 0.4010548523206751,
      "grad_norm": 0.6264271140098572,
      "learning_rate": 0.0015,
      "loss": 1.6645,
      "step": 3802
    },
    {
      "epoch": 0.4011603375527426,
      "grad_norm": 0.6265786290168762,
      "learning_rate": 0.0015,
      "loss": 1.7183,
      "step": 3803
    },
    {
      "epoch": 0.4012658227848101,
      "grad_norm": 0.6287268400192261,
      "learning_rate": 0.0015,
      "loss": 1.7058,
      "step": 3804
    },
    {
      "epoch": 0.40137130801687765,
      "grad_norm": 0.6235065460205078,
      "learning_rate": 0.0015,
      "loss": 1.6452,
      "step": 3805
    },
    {
      "epoch": 0.40147679324894514,
      "grad_norm": 0.523852527141571,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 3806
    },
    {
      "epoch": 0.40158227848101263,
      "grad_norm": 0.6052106022834778,
      "learning_rate": 0.0015,
      "loss": 1.6541,
      "step": 3807
    },
    {
      "epoch": 0.4016877637130802,
      "grad_norm": 0.6712532043457031,
      "learning_rate": 0.0015,
      "loss": 1.6944,
      "step": 3808
    },
    {
      "epoch": 0.4017932489451477,
      "grad_norm": 0.5733087658882141,
      "learning_rate": 0.0015,
      "loss": 1.7142,
      "step": 3809
    },
    {
      "epoch": 0.40189873417721517,
      "grad_norm": 0.6598334312438965,
      "learning_rate": 0.0015,
      "loss": 1.714,
      "step": 3810
    },
    {
      "epoch": 0.4020042194092827,
      "grad_norm": 0.7662893533706665,
      "learning_rate": 0.0015,
      "loss": 1.6801,
      "step": 3811
    },
    {
      "epoch": 0.4021097046413502,
      "grad_norm": 0.5452477335929871,
      "learning_rate": 0.0015,
      "loss": 1.6553,
      "step": 3812
    },
    {
      "epoch": 0.4022151898734177,
      "grad_norm": 0.6024952530860901,
      "learning_rate": 0.0015,
      "loss": 1.6529,
      "step": 3813
    },
    {
      "epoch": 0.40232067510548525,
      "grad_norm": 0.6055412888526917,
      "learning_rate": 0.0015,
      "loss": 1.6617,
      "step": 3814
    },
    {
      "epoch": 0.40242616033755274,
      "grad_norm": 0.6142078638076782,
      "learning_rate": 0.0015,
      "loss": 1.6929,
      "step": 3815
    },
    {
      "epoch": 0.40253164556962023,
      "grad_norm": 0.5741285681724548,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 3816
    },
    {
      "epoch": 0.4026371308016878,
      "grad_norm": 0.5197274088859558,
      "learning_rate": 0.0015,
      "loss": 1.6384,
      "step": 3817
    },
    {
      "epoch": 0.4027426160337553,
      "grad_norm": 0.5791266560554504,
      "learning_rate": 0.0015,
      "loss": 1.6669,
      "step": 3818
    },
    {
      "epoch": 0.40284810126582277,
      "grad_norm": 0.5819272994995117,
      "learning_rate": 0.0015,
      "loss": 1.6887,
      "step": 3819
    },
    {
      "epoch": 0.4029535864978903,
      "grad_norm": 0.5928919911384583,
      "learning_rate": 0.0015,
      "loss": 1.6404,
      "step": 3820
    },
    {
      "epoch": 0.4030590717299578,
      "grad_norm": 0.5293534994125366,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 3821
    },
    {
      "epoch": 0.4031645569620253,
      "grad_norm": 0.6402086019515991,
      "learning_rate": 0.0015,
      "loss": 1.6326,
      "step": 3822
    },
    {
      "epoch": 0.40327004219409285,
      "grad_norm": 0.575337827205658,
      "learning_rate": 0.0015,
      "loss": 1.6765,
      "step": 3823
    },
    {
      "epoch": 0.40337552742616034,
      "grad_norm": 0.5857341289520264,
      "learning_rate": 0.0015,
      "loss": 1.6779,
      "step": 3824
    },
    {
      "epoch": 0.40348101265822783,
      "grad_norm": 0.6004738211631775,
      "learning_rate": 0.0015,
      "loss": 1.6775,
      "step": 3825
    },
    {
      "epoch": 0.4035864978902954,
      "grad_norm": 0.5944442749023438,
      "learning_rate": 0.0015,
      "loss": 1.7099,
      "step": 3826
    },
    {
      "epoch": 0.4036919831223629,
      "grad_norm": 0.46930718421936035,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 3827
    },
    {
      "epoch": 0.40379746835443037,
      "grad_norm": 0.5892956256866455,
      "learning_rate": 0.0015,
      "loss": 1.6539,
      "step": 3828
    },
    {
      "epoch": 0.4039029535864979,
      "grad_norm": 0.50191730260849,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 3829
    },
    {
      "epoch": 0.4040084388185654,
      "grad_norm": 0.6760581731796265,
      "learning_rate": 0.0015,
      "loss": 1.6473,
      "step": 3830
    },
    {
      "epoch": 0.4041139240506329,
      "grad_norm": 0.5088383555412292,
      "learning_rate": 0.0015,
      "loss": 1.6884,
      "step": 3831
    },
    {
      "epoch": 0.40421940928270045,
      "grad_norm": 0.5728570818901062,
      "learning_rate": 0.0015,
      "loss": 1.7008,
      "step": 3832
    },
    {
      "epoch": 0.40432489451476794,
      "grad_norm": 0.5390267372131348,
      "learning_rate": 0.0015,
      "loss": 1.6782,
      "step": 3833
    },
    {
      "epoch": 0.40443037974683543,
      "grad_norm": 0.67784184217453,
      "learning_rate": 0.0015,
      "loss": 1.6698,
      "step": 3834
    },
    {
      "epoch": 0.4045358649789029,
      "grad_norm": 0.5902341604232788,
      "learning_rate": 0.0015,
      "loss": 1.6753,
      "step": 3835
    },
    {
      "epoch": 0.4046413502109705,
      "grad_norm": 0.6139640808105469,
      "learning_rate": 0.0015,
      "loss": 1.7032,
      "step": 3836
    },
    {
      "epoch": 0.40474683544303797,
      "grad_norm": 0.5352752804756165,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 3837
    },
    {
      "epoch": 0.40485232067510546,
      "grad_norm": 0.6846832036972046,
      "learning_rate": 0.0015,
      "loss": 1.6845,
      "step": 3838
    },
    {
      "epoch": 0.404957805907173,
      "grad_norm": 0.5652015805244446,
      "learning_rate": 0.0015,
      "loss": 1.7339,
      "step": 3839
    },
    {
      "epoch": 0.4050632911392405,
      "grad_norm": 0.6455852389335632,
      "learning_rate": 0.0015,
      "loss": 1.7196,
      "step": 3840
    },
    {
      "epoch": 0.405168776371308,
      "grad_norm": 0.5148798823356628,
      "learning_rate": 0.0015,
      "loss": 1.7008,
      "step": 3841
    },
    {
      "epoch": 0.40527426160337554,
      "grad_norm": 0.4962198734283447,
      "learning_rate": 0.0015,
      "loss": 1.6705,
      "step": 3842
    },
    {
      "epoch": 0.40537974683544303,
      "grad_norm": 0.47228044271469116,
      "learning_rate": 0.0015,
      "loss": 1.6685,
      "step": 3843
    },
    {
      "epoch": 0.4054852320675105,
      "grad_norm": 0.4935452342033386,
      "learning_rate": 0.0015,
      "loss": 1.6626,
      "step": 3844
    },
    {
      "epoch": 0.4055907172995781,
      "grad_norm": 0.4716067612171173,
      "learning_rate": 0.0015,
      "loss": 1.6946,
      "step": 3845
    },
    {
      "epoch": 0.40569620253164557,
      "grad_norm": 0.5209110975265503,
      "learning_rate": 0.0015,
      "loss": 1.6575,
      "step": 3846
    },
    {
      "epoch": 0.40580168776371306,
      "grad_norm": 0.46358731389045715,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 3847
    },
    {
      "epoch": 0.4059071729957806,
      "grad_norm": 0.4754665791988373,
      "learning_rate": 0.0015,
      "loss": 1.6636,
      "step": 3848
    },
    {
      "epoch": 0.4060126582278481,
      "grad_norm": 0.5017957091331482,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 3849
    },
    {
      "epoch": 0.4061181434599156,
      "grad_norm": 0.4882739782333374,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 3850
    },
    {
      "epoch": 0.40622362869198314,
      "grad_norm": 0.5001218318939209,
      "learning_rate": 0.0015,
      "loss": 1.728,
      "step": 3851
    },
    {
      "epoch": 0.40632911392405063,
      "grad_norm": 0.4525369107723236,
      "learning_rate": 0.0015,
      "loss": 1.6816,
      "step": 3852
    },
    {
      "epoch": 0.4064345991561181,
      "grad_norm": 0.47382867336273193,
      "learning_rate": 0.0015,
      "loss": 1.6877,
      "step": 3853
    },
    {
      "epoch": 0.4065400843881857,
      "grad_norm": 0.48720699548721313,
      "learning_rate": 0.0015,
      "loss": 1.689,
      "step": 3854
    },
    {
      "epoch": 0.40664556962025317,
      "grad_norm": 0.547744631767273,
      "learning_rate": 0.0015,
      "loss": 1.6695,
      "step": 3855
    },
    {
      "epoch": 0.40675105485232066,
      "grad_norm": 0.47330442070961,
      "learning_rate": 0.0015,
      "loss": 1.6891,
      "step": 3856
    },
    {
      "epoch": 0.4068565400843882,
      "grad_norm": 0.5014797449111938,
      "learning_rate": 0.0015,
      "loss": 1.6693,
      "step": 3857
    },
    {
      "epoch": 0.4069620253164557,
      "grad_norm": 0.5758227109909058,
      "learning_rate": 0.0015,
      "loss": 1.6735,
      "step": 3858
    },
    {
      "epoch": 0.4070675105485232,
      "grad_norm": 0.5283249616622925,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 3859
    },
    {
      "epoch": 0.40717299578059074,
      "grad_norm": 0.4935728907585144,
      "learning_rate": 0.0015,
      "loss": 1.6629,
      "step": 3860
    },
    {
      "epoch": 0.40727848101265823,
      "grad_norm": 0.6933180689811707,
      "learning_rate": 0.0015,
      "loss": 1.6698,
      "step": 3861
    },
    {
      "epoch": 0.4073839662447257,
      "grad_norm": 0.6887080073356628,
      "learning_rate": 0.0015,
      "loss": 1.6715,
      "step": 3862
    },
    {
      "epoch": 0.4074894514767933,
      "grad_norm": 0.5577717423439026,
      "learning_rate": 0.0015,
      "loss": 1.6984,
      "step": 3863
    },
    {
      "epoch": 0.40759493670886077,
      "grad_norm": 0.487842857837677,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 3864
    },
    {
      "epoch": 0.40770042194092826,
      "grad_norm": 0.5179023742675781,
      "learning_rate": 0.0015,
      "loss": 1.6469,
      "step": 3865
    },
    {
      "epoch": 0.4078059071729958,
      "grad_norm": 0.6370635628700256,
      "learning_rate": 0.0015,
      "loss": 1.7037,
      "step": 3866
    },
    {
      "epoch": 0.4079113924050633,
      "grad_norm": 0.6766543388366699,
      "learning_rate": 0.0015,
      "loss": 1.6504,
      "step": 3867
    },
    {
      "epoch": 0.4080168776371308,
      "grad_norm": 0.5281383991241455,
      "learning_rate": 0.0015,
      "loss": 1.6563,
      "step": 3868
    },
    {
      "epoch": 0.4081223628691983,
      "grad_norm": 0.5986506342887878,
      "learning_rate": 0.0015,
      "loss": 1.71,
      "step": 3869
    },
    {
      "epoch": 0.40822784810126583,
      "grad_norm": 0.7948133945465088,
      "learning_rate": 0.0015,
      "loss": 1.6859,
      "step": 3870
    },
    {
      "epoch": 0.4083333333333333,
      "grad_norm": 0.6055651903152466,
      "learning_rate": 0.0015,
      "loss": 1.6562,
      "step": 3871
    },
    {
      "epoch": 0.4084388185654008,
      "grad_norm": 0.6119669079780579,
      "learning_rate": 0.0015,
      "loss": 1.7158,
      "step": 3872
    },
    {
      "epoch": 0.40854430379746837,
      "grad_norm": 0.7874894142150879,
      "learning_rate": 0.0015,
      "loss": 1.6664,
      "step": 3873
    },
    {
      "epoch": 0.40864978902953586,
      "grad_norm": 0.45923927426338196,
      "learning_rate": 0.0015,
      "loss": 1.6798,
      "step": 3874
    },
    {
      "epoch": 0.40875527426160335,
      "grad_norm": 0.5760701298713684,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3875
    },
    {
      "epoch": 0.4088607594936709,
      "grad_norm": 0.5290331244468689,
      "learning_rate": 0.0015,
      "loss": 1.6343,
      "step": 3876
    },
    {
      "epoch": 0.4089662447257384,
      "grad_norm": 0.49807408452033997,
      "learning_rate": 0.0015,
      "loss": 1.6388,
      "step": 3877
    },
    {
      "epoch": 0.4090717299578059,
      "grad_norm": 0.6555059552192688,
      "learning_rate": 0.0015,
      "loss": 1.635,
      "step": 3878
    },
    {
      "epoch": 0.40917721518987343,
      "grad_norm": 0.6197727918624878,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3879
    },
    {
      "epoch": 0.4092827004219409,
      "grad_norm": 0.5036622881889343,
      "learning_rate": 0.0015,
      "loss": 1.6352,
      "step": 3880
    },
    {
      "epoch": 0.4093881856540084,
      "grad_norm": 0.9330415725708008,
      "learning_rate": 0.0015,
      "loss": 1.6869,
      "step": 3881
    },
    {
      "epoch": 0.40949367088607597,
      "grad_norm": 0.7329707741737366,
      "learning_rate": 0.0015,
      "loss": 1.6606,
      "step": 3882
    },
    {
      "epoch": 0.40959915611814346,
      "grad_norm": 0.6673707962036133,
      "learning_rate": 0.0015,
      "loss": 1.6589,
      "step": 3883
    },
    {
      "epoch": 0.40970464135021095,
      "grad_norm": 1.2548285722732544,
      "learning_rate": 0.0015,
      "loss": 1.6533,
      "step": 3884
    },
    {
      "epoch": 0.4098101265822785,
      "grad_norm": 0.49074581265449524,
      "learning_rate": 0.0015,
      "loss": 1.6531,
      "step": 3885
    },
    {
      "epoch": 0.409915611814346,
      "grad_norm": 1.0034308433532715,
      "learning_rate": 0.0015,
      "loss": 1.6957,
      "step": 3886
    },
    {
      "epoch": 0.4100210970464135,
      "grad_norm": 0.7492836713790894,
      "learning_rate": 0.0015,
      "loss": 1.6752,
      "step": 3887
    },
    {
      "epoch": 0.41012658227848103,
      "grad_norm": 0.6365775465965271,
      "learning_rate": 0.0015,
      "loss": 1.6898,
      "step": 3888
    },
    {
      "epoch": 0.4102320675105485,
      "grad_norm": 1.0673199892044067,
      "learning_rate": 0.0015,
      "loss": 1.6529,
      "step": 3889
    },
    {
      "epoch": 0.410337552742616,
      "grad_norm": 0.5610344409942627,
      "learning_rate": 0.0015,
      "loss": 1.6325,
      "step": 3890
    },
    {
      "epoch": 0.41044303797468357,
      "grad_norm": 0.8004913330078125,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 3891
    },
    {
      "epoch": 0.41054852320675106,
      "grad_norm": 0.7554417848587036,
      "learning_rate": 0.0015,
      "loss": 1.7075,
      "step": 3892
    },
    {
      "epoch": 0.41065400843881855,
      "grad_norm": 0.591577410697937,
      "learning_rate": 0.0015,
      "loss": 1.6813,
      "step": 3893
    },
    {
      "epoch": 0.4107594936708861,
      "grad_norm": 0.8795300722122192,
      "learning_rate": 0.0015,
      "loss": 1.6723,
      "step": 3894
    },
    {
      "epoch": 0.4108649789029536,
      "grad_norm": 0.5651753544807434,
      "learning_rate": 0.0015,
      "loss": 1.6678,
      "step": 3895
    },
    {
      "epoch": 0.4109704641350211,
      "grad_norm": 0.7733017802238464,
      "learning_rate": 0.0015,
      "loss": 1.6354,
      "step": 3896
    },
    {
      "epoch": 0.41107594936708863,
      "grad_norm": 0.5560656785964966,
      "learning_rate": 0.0015,
      "loss": 1.6528,
      "step": 3897
    },
    {
      "epoch": 0.4111814345991561,
      "grad_norm": 0.5901712775230408,
      "learning_rate": 0.0015,
      "loss": 1.661,
      "step": 3898
    },
    {
      "epoch": 0.4112869198312236,
      "grad_norm": 0.5299215316772461,
      "learning_rate": 0.0015,
      "loss": 1.6341,
      "step": 3899
    },
    {
      "epoch": 0.41139240506329117,
      "grad_norm": 0.7282924056053162,
      "learning_rate": 0.0015,
      "loss": 1.6906,
      "step": 3900
    },
    {
      "epoch": 0.41149789029535866,
      "grad_norm": 0.7884114384651184,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 3901
    },
    {
      "epoch": 0.41160337552742615,
      "grad_norm": 0.5577768087387085,
      "learning_rate": 0.0015,
      "loss": 1.7069,
      "step": 3902
    },
    {
      "epoch": 0.41170886075949364,
      "grad_norm": 0.7427865862846375,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 3903
    },
    {
      "epoch": 0.4118143459915612,
      "grad_norm": 0.5344990491867065,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 3904
    },
    {
      "epoch": 0.4119198312236287,
      "grad_norm": 0.6284624934196472,
      "learning_rate": 0.0015,
      "loss": 1.662,
      "step": 3905
    },
    {
      "epoch": 0.4120253164556962,
      "grad_norm": 0.5771468877792358,
      "learning_rate": 0.0015,
      "loss": 1.6791,
      "step": 3906
    },
    {
      "epoch": 0.4121308016877637,
      "grad_norm": 0.5664799809455872,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 3907
    },
    {
      "epoch": 0.4122362869198312,
      "grad_norm": 0.533612847328186,
      "learning_rate": 0.0015,
      "loss": 1.6675,
      "step": 3908
    },
    {
      "epoch": 0.4123417721518987,
      "grad_norm": 0.5474640727043152,
      "learning_rate": 0.0015,
      "loss": 1.6573,
      "step": 3909
    },
    {
      "epoch": 0.41244725738396626,
      "grad_norm": 0.5034437775611877,
      "learning_rate": 0.0015,
      "loss": 1.6944,
      "step": 3910
    },
    {
      "epoch": 0.41255274261603375,
      "grad_norm": 0.4837951064109802,
      "learning_rate": 0.0015,
      "loss": 1.6777,
      "step": 3911
    },
    {
      "epoch": 0.41265822784810124,
      "grad_norm": 0.5077068209648132,
      "learning_rate": 0.0015,
      "loss": 1.7078,
      "step": 3912
    },
    {
      "epoch": 0.4127637130801688,
      "grad_norm": 0.5174497365951538,
      "learning_rate": 0.0015,
      "loss": 1.6754,
      "step": 3913
    },
    {
      "epoch": 0.4128691983122363,
      "grad_norm": 0.547358512878418,
      "learning_rate": 0.0015,
      "loss": 1.6041,
      "step": 3914
    },
    {
      "epoch": 0.4129746835443038,
      "grad_norm": 0.5077178478240967,
      "learning_rate": 0.0015,
      "loss": 1.6516,
      "step": 3915
    },
    {
      "epoch": 0.4130801687763713,
      "grad_norm": 0.5167761445045471,
      "learning_rate": 0.0015,
      "loss": 1.6626,
      "step": 3916
    },
    {
      "epoch": 0.4131856540084388,
      "grad_norm": 0.6522733569145203,
      "learning_rate": 0.0015,
      "loss": 1.6408,
      "step": 3917
    },
    {
      "epoch": 0.4132911392405063,
      "grad_norm": 0.4896742105484009,
      "learning_rate": 0.0015,
      "loss": 1.6819,
      "step": 3918
    },
    {
      "epoch": 0.41339662447257386,
      "grad_norm": 0.6426261067390442,
      "learning_rate": 0.0015,
      "loss": 1.6478,
      "step": 3919
    },
    {
      "epoch": 0.41350210970464135,
      "grad_norm": 0.5804246068000793,
      "learning_rate": 0.0015,
      "loss": 1.6911,
      "step": 3920
    },
    {
      "epoch": 0.41360759493670884,
      "grad_norm": 0.4942731261253357,
      "learning_rate": 0.0015,
      "loss": 1.6418,
      "step": 3921
    },
    {
      "epoch": 0.4137130801687764,
      "grad_norm": 0.6210718750953674,
      "learning_rate": 0.0015,
      "loss": 1.7146,
      "step": 3922
    },
    {
      "epoch": 0.4138185654008439,
      "grad_norm": 0.5851173400878906,
      "learning_rate": 0.0015,
      "loss": 1.681,
      "step": 3923
    },
    {
      "epoch": 0.4139240506329114,
      "grad_norm": 0.502113401889801,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 3924
    },
    {
      "epoch": 0.4140295358649789,
      "grad_norm": 0.5490169525146484,
      "learning_rate": 0.0015,
      "loss": 1.6177,
      "step": 3925
    },
    {
      "epoch": 0.4141350210970464,
      "grad_norm": 0.5450034141540527,
      "learning_rate": 0.0015,
      "loss": 1.668,
      "step": 3926
    },
    {
      "epoch": 0.4142405063291139,
      "grad_norm": 0.6607049107551575,
      "learning_rate": 0.0015,
      "loss": 1.6594,
      "step": 3927
    },
    {
      "epoch": 0.41434599156118146,
      "grad_norm": 0.5202460885047913,
      "learning_rate": 0.0015,
      "loss": 1.669,
      "step": 3928
    },
    {
      "epoch": 0.41445147679324895,
      "grad_norm": 0.648923397064209,
      "learning_rate": 0.0015,
      "loss": 1.6864,
      "step": 3929
    },
    {
      "epoch": 0.41455696202531644,
      "grad_norm": 0.5905894637107849,
      "learning_rate": 0.0015,
      "loss": 1.6805,
      "step": 3930
    },
    {
      "epoch": 0.414662447257384,
      "grad_norm": 0.6553009748458862,
      "learning_rate": 0.0015,
      "loss": 1.6409,
      "step": 3931
    },
    {
      "epoch": 0.4147679324894515,
      "grad_norm": 0.728967010974884,
      "learning_rate": 0.0015,
      "loss": 1.6485,
      "step": 3932
    },
    {
      "epoch": 0.414873417721519,
      "grad_norm": 0.4935154616832733,
      "learning_rate": 0.0015,
      "loss": 1.6606,
      "step": 3933
    },
    {
      "epoch": 0.41497890295358647,
      "grad_norm": 0.5506753921508789,
      "learning_rate": 0.0015,
      "loss": 1.6696,
      "step": 3934
    },
    {
      "epoch": 0.415084388185654,
      "grad_norm": 0.5010538697242737,
      "learning_rate": 0.0015,
      "loss": 1.6827,
      "step": 3935
    },
    {
      "epoch": 0.4151898734177215,
      "grad_norm": 0.5543103814125061,
      "learning_rate": 0.0015,
      "loss": 1.6587,
      "step": 3936
    },
    {
      "epoch": 0.415295358649789,
      "grad_norm": 0.5275358557701111,
      "learning_rate": 0.0015,
      "loss": 1.6995,
      "step": 3937
    },
    {
      "epoch": 0.41540084388185655,
      "grad_norm": 0.5093750357627869,
      "learning_rate": 0.0015,
      "loss": 1.6232,
      "step": 3938
    },
    {
      "epoch": 0.41550632911392404,
      "grad_norm": 0.5419853925704956,
      "learning_rate": 0.0015,
      "loss": 1.6734,
      "step": 3939
    },
    {
      "epoch": 0.41561181434599154,
      "grad_norm": 0.5158277153968811,
      "learning_rate": 0.0015,
      "loss": 1.6718,
      "step": 3940
    },
    {
      "epoch": 0.4157172995780591,
      "grad_norm": 0.5883190035820007,
      "learning_rate": 0.0015,
      "loss": 1.6727,
      "step": 3941
    },
    {
      "epoch": 0.4158227848101266,
      "grad_norm": 0.5016283988952637,
      "learning_rate": 0.0015,
      "loss": 1.7008,
      "step": 3942
    },
    {
      "epoch": 0.41592827004219407,
      "grad_norm": 0.6370705962181091,
      "learning_rate": 0.0015,
      "loss": 1.6594,
      "step": 3943
    },
    {
      "epoch": 0.4160337552742616,
      "grad_norm": 0.5819518566131592,
      "learning_rate": 0.0015,
      "loss": 1.6843,
      "step": 3944
    },
    {
      "epoch": 0.4161392405063291,
      "grad_norm": 0.540444016456604,
      "learning_rate": 0.0015,
      "loss": 1.6469,
      "step": 3945
    },
    {
      "epoch": 0.4162447257383966,
      "grad_norm": 0.5240791440010071,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 3946
    },
    {
      "epoch": 0.41635021097046415,
      "grad_norm": 0.5113990306854248,
      "learning_rate": 0.0015,
      "loss": 1.6945,
      "step": 3947
    },
    {
      "epoch": 0.41645569620253164,
      "grad_norm": 0.599273145198822,
      "learning_rate": 0.0015,
      "loss": 1.6811,
      "step": 3948
    },
    {
      "epoch": 0.41656118143459914,
      "grad_norm": 0.510252833366394,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 3949
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.6712703704833984,
      "learning_rate": 0.0015,
      "loss": 1.6848,
      "step": 3950
    },
    {
      "epoch": 0.4167721518987342,
      "grad_norm": 0.45828548073768616,
      "learning_rate": 0.0015,
      "loss": 1.6814,
      "step": 3951
    },
    {
      "epoch": 0.41687763713080167,
      "grad_norm": 0.5784968137741089,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 3952
    },
    {
      "epoch": 0.4169831223628692,
      "grad_norm": 0.5279408693313599,
      "learning_rate": 0.0015,
      "loss": 1.7094,
      "step": 3953
    },
    {
      "epoch": 0.4170886075949367,
      "grad_norm": 0.4787351191043854,
      "learning_rate": 0.0015,
      "loss": 1.666,
      "step": 3954
    },
    {
      "epoch": 0.4171940928270042,
      "grad_norm": 0.6963359713554382,
      "learning_rate": 0.0015,
      "loss": 1.6586,
      "step": 3955
    },
    {
      "epoch": 0.41729957805907175,
      "grad_norm": 0.623216450214386,
      "learning_rate": 0.0015,
      "loss": 1.6478,
      "step": 3956
    },
    {
      "epoch": 0.41740506329113924,
      "grad_norm": 0.5411447286605835,
      "learning_rate": 0.0015,
      "loss": 1.67,
      "step": 3957
    },
    {
      "epoch": 0.41751054852320674,
      "grad_norm": 0.5261410474777222,
      "learning_rate": 0.0015,
      "loss": 1.6986,
      "step": 3958
    },
    {
      "epoch": 0.4176160337552743,
      "grad_norm": 0.5788530111312866,
      "learning_rate": 0.0015,
      "loss": 1.6235,
      "step": 3959
    },
    {
      "epoch": 0.4177215189873418,
      "grad_norm": 0.5429050922393799,
      "learning_rate": 0.0015,
      "loss": 1.6809,
      "step": 3960
    },
    {
      "epoch": 0.41782700421940927,
      "grad_norm": 0.551701009273529,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 3961
    },
    {
      "epoch": 0.4179324894514768,
      "grad_norm": 0.8163561820983887,
      "learning_rate": 0.0015,
      "loss": 1.6448,
      "step": 3962
    },
    {
      "epoch": 0.4180379746835443,
      "grad_norm": 0.5044242739677429,
      "learning_rate": 0.0015,
      "loss": 1.6468,
      "step": 3963
    },
    {
      "epoch": 0.4181434599156118,
      "grad_norm": 0.6471570730209351,
      "learning_rate": 0.0015,
      "loss": 1.6875,
      "step": 3964
    },
    {
      "epoch": 0.41824894514767935,
      "grad_norm": 0.7842706441879272,
      "learning_rate": 0.0015,
      "loss": 1.6914,
      "step": 3965
    },
    {
      "epoch": 0.41835443037974684,
      "grad_norm": 0.568332850933075,
      "learning_rate": 0.0015,
      "loss": 1.6541,
      "step": 3966
    },
    {
      "epoch": 0.41845991561181434,
      "grad_norm": 0.5272161364555359,
      "learning_rate": 0.0015,
      "loss": 1.6552,
      "step": 3967
    },
    {
      "epoch": 0.41856540084388183,
      "grad_norm": 0.5295596718788147,
      "learning_rate": 0.0015,
      "loss": 1.6694,
      "step": 3968
    },
    {
      "epoch": 0.4186708860759494,
      "grad_norm": 0.5039467811584473,
      "learning_rate": 0.0015,
      "loss": 1.6773,
      "step": 3969
    },
    {
      "epoch": 0.41877637130801687,
      "grad_norm": 0.5752148628234863,
      "learning_rate": 0.0015,
      "loss": 1.6357,
      "step": 3970
    },
    {
      "epoch": 0.41888185654008436,
      "grad_norm": 0.5082814693450928,
      "learning_rate": 0.0015,
      "loss": 1.6872,
      "step": 3971
    },
    {
      "epoch": 0.4189873417721519,
      "grad_norm": 0.501728892326355,
      "learning_rate": 0.0015,
      "loss": 1.6604,
      "step": 3972
    },
    {
      "epoch": 0.4190928270042194,
      "grad_norm": 0.45531946420669556,
      "learning_rate": 0.0015,
      "loss": 1.6869,
      "step": 3973
    },
    {
      "epoch": 0.4191983122362869,
      "grad_norm": 0.5419966578483582,
      "learning_rate": 0.0015,
      "loss": 1.6608,
      "step": 3974
    },
    {
      "epoch": 0.41930379746835444,
      "grad_norm": 0.5244120955467224,
      "learning_rate": 0.0015,
      "loss": 1.6713,
      "step": 3975
    },
    {
      "epoch": 0.41940928270042194,
      "grad_norm": 0.574030876159668,
      "learning_rate": 0.0015,
      "loss": 1.6646,
      "step": 3976
    },
    {
      "epoch": 0.41951476793248943,
      "grad_norm": 0.46261152625083923,
      "learning_rate": 0.0015,
      "loss": 1.6704,
      "step": 3977
    },
    {
      "epoch": 0.419620253164557,
      "grad_norm": 0.5169928073883057,
      "learning_rate": 0.0015,
      "loss": 1.6222,
      "step": 3978
    },
    {
      "epoch": 0.41972573839662447,
      "grad_norm": 0.44602909684181213,
      "learning_rate": 0.0015,
      "loss": 1.677,
      "step": 3979
    },
    {
      "epoch": 0.41983122362869196,
      "grad_norm": 0.4354664087295532,
      "learning_rate": 0.0015,
      "loss": 1.7032,
      "step": 3980
    },
    {
      "epoch": 0.4199367088607595,
      "grad_norm": 0.4882993698120117,
      "learning_rate": 0.0015,
      "loss": 1.6481,
      "step": 3981
    },
    {
      "epoch": 0.420042194092827,
      "grad_norm": 0.4510979652404785,
      "learning_rate": 0.0015,
      "loss": 1.6696,
      "step": 3982
    },
    {
      "epoch": 0.4201476793248945,
      "grad_norm": 0.5278876423835754,
      "learning_rate": 0.0015,
      "loss": 1.6578,
      "step": 3983
    },
    {
      "epoch": 0.42025316455696204,
      "grad_norm": 0.6052321195602417,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 3984
    },
    {
      "epoch": 0.42035864978902954,
      "grad_norm": 0.5470100045204163,
      "learning_rate": 0.0015,
      "loss": 1.6523,
      "step": 3985
    },
    {
      "epoch": 0.42046413502109703,
      "grad_norm": 0.5092055201530457,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 3986
    },
    {
      "epoch": 0.4205696202531646,
      "grad_norm": 0.5102941393852234,
      "learning_rate": 0.0015,
      "loss": 1.643,
      "step": 3987
    },
    {
      "epoch": 0.42067510548523207,
      "grad_norm": 0.48482275009155273,
      "learning_rate": 0.0015,
      "loss": 1.665,
      "step": 3988
    },
    {
      "epoch": 0.42078059071729956,
      "grad_norm": 0.5031700730323792,
      "learning_rate": 0.0015,
      "loss": 1.6675,
      "step": 3989
    },
    {
      "epoch": 0.4208860759493671,
      "grad_norm": 0.4970076382160187,
      "learning_rate": 0.0015,
      "loss": 1.6725,
      "step": 3990
    },
    {
      "epoch": 0.4209915611814346,
      "grad_norm": 0.4832526445388794,
      "learning_rate": 0.0015,
      "loss": 1.6727,
      "step": 3991
    },
    {
      "epoch": 0.4210970464135021,
      "grad_norm": 0.5345883369445801,
      "learning_rate": 0.0015,
      "loss": 1.6683,
      "step": 3992
    },
    {
      "epoch": 0.42120253164556964,
      "grad_norm": 0.5662575364112854,
      "learning_rate": 0.0015,
      "loss": 1.6635,
      "step": 3993
    },
    {
      "epoch": 0.42130801687763714,
      "grad_norm": 0.5999237298965454,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 3994
    },
    {
      "epoch": 0.42141350210970463,
      "grad_norm": 0.5003252029418945,
      "learning_rate": 0.0015,
      "loss": 1.6645,
      "step": 3995
    },
    {
      "epoch": 0.4215189873417722,
      "grad_norm": 0.5532550811767578,
      "learning_rate": 0.0015,
      "loss": 1.6893,
      "step": 3996
    },
    {
      "epoch": 0.42162447257383967,
      "grad_norm": 0.6924083232879639,
      "learning_rate": 0.0015,
      "loss": 1.6921,
      "step": 3997
    },
    {
      "epoch": 0.42172995780590716,
      "grad_norm": 0.582459032535553,
      "learning_rate": 0.0015,
      "loss": 1.6621,
      "step": 3998
    },
    {
      "epoch": 0.4218354430379747,
      "grad_norm": 0.6234803199768066,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 3999
    },
    {
      "epoch": 0.4219409282700422,
      "grad_norm": 0.6062827110290527,
      "learning_rate": 0.0015,
      "loss": 1.6683,
      "step": 4000
    },
    {
      "epoch": 0.4220464135021097,
      "grad_norm": 0.5449821352958679,
      "learning_rate": 0.0015,
      "loss": 1.6439,
      "step": 4001
    },
    {
      "epoch": 0.4221518987341772,
      "grad_norm": 0.6629012227058411,
      "learning_rate": 0.0015,
      "loss": 1.6638,
      "step": 4002
    },
    {
      "epoch": 0.42225738396624474,
      "grad_norm": 0.5163016319274902,
      "learning_rate": 0.0015,
      "loss": 1.6389,
      "step": 4003
    },
    {
      "epoch": 0.42236286919831223,
      "grad_norm": 0.6885299682617188,
      "learning_rate": 0.0015,
      "loss": 1.6826,
      "step": 4004
    },
    {
      "epoch": 0.4224683544303797,
      "grad_norm": 0.7004706859588623,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 4005
    },
    {
      "epoch": 0.42257383966244727,
      "grad_norm": 0.573915958404541,
      "learning_rate": 0.0015,
      "loss": 1.6827,
      "step": 4006
    },
    {
      "epoch": 0.42267932489451476,
      "grad_norm": 0.657016396522522,
      "learning_rate": 0.0015,
      "loss": 1.6699,
      "step": 4007
    },
    {
      "epoch": 0.42278481012658226,
      "grad_norm": 0.6307787299156189,
      "learning_rate": 0.0015,
      "loss": 1.6404,
      "step": 4008
    },
    {
      "epoch": 0.4228902953586498,
      "grad_norm": 0.6289223432540894,
      "learning_rate": 0.0015,
      "loss": 1.6671,
      "step": 4009
    },
    {
      "epoch": 0.4229957805907173,
      "grad_norm": 0.4873087406158447,
      "learning_rate": 0.0015,
      "loss": 1.6915,
      "step": 4010
    },
    {
      "epoch": 0.4231012658227848,
      "grad_norm": 0.5554637312889099,
      "learning_rate": 0.0015,
      "loss": 1.6478,
      "step": 4011
    },
    {
      "epoch": 0.42320675105485234,
      "grad_norm": 0.508516788482666,
      "learning_rate": 0.0015,
      "loss": 1.6355,
      "step": 4012
    },
    {
      "epoch": 0.42331223628691983,
      "grad_norm": 0.5363007187843323,
      "learning_rate": 0.0015,
      "loss": 1.6457,
      "step": 4013
    },
    {
      "epoch": 0.4234177215189873,
      "grad_norm": 0.4818873405456543,
      "learning_rate": 0.0015,
      "loss": 1.6696,
      "step": 4014
    },
    {
      "epoch": 0.42352320675105487,
      "grad_norm": 0.45898446440696716,
      "learning_rate": 0.0015,
      "loss": 1.6888,
      "step": 4015
    },
    {
      "epoch": 0.42362869198312236,
      "grad_norm": 0.5161341428756714,
      "learning_rate": 0.0015,
      "loss": 1.6806,
      "step": 4016
    },
    {
      "epoch": 0.42373417721518986,
      "grad_norm": 0.448438435792923,
      "learning_rate": 0.0015,
      "loss": 1.6496,
      "step": 4017
    },
    {
      "epoch": 0.4238396624472574,
      "grad_norm": 0.4921002686023712,
      "learning_rate": 0.0015,
      "loss": 1.6778,
      "step": 4018
    },
    {
      "epoch": 0.4239451476793249,
      "grad_norm": 0.5122791528701782,
      "learning_rate": 0.0015,
      "loss": 1.6428,
      "step": 4019
    },
    {
      "epoch": 0.4240506329113924,
      "grad_norm": 0.5509474873542786,
      "learning_rate": 0.0015,
      "loss": 1.7036,
      "step": 4020
    },
    {
      "epoch": 0.42415611814345994,
      "grad_norm": 0.5201058387756348,
      "learning_rate": 0.0015,
      "loss": 1.6388,
      "step": 4021
    },
    {
      "epoch": 0.42426160337552743,
      "grad_norm": 0.4961116313934326,
      "learning_rate": 0.0015,
      "loss": 1.6838,
      "step": 4022
    },
    {
      "epoch": 0.4243670886075949,
      "grad_norm": 0.5693700313568115,
      "learning_rate": 0.0015,
      "loss": 1.6423,
      "step": 4023
    },
    {
      "epoch": 0.42447257383966247,
      "grad_norm": 0.550399124622345,
      "learning_rate": 0.0015,
      "loss": 1.6541,
      "step": 4024
    },
    {
      "epoch": 0.42457805907172996,
      "grad_norm": 0.475479394197464,
      "learning_rate": 0.0015,
      "loss": 1.6854,
      "step": 4025
    },
    {
      "epoch": 0.42468354430379746,
      "grad_norm": 0.5067935585975647,
      "learning_rate": 0.0015,
      "loss": 1.6734,
      "step": 4026
    },
    {
      "epoch": 0.424789029535865,
      "grad_norm": 0.5102432370185852,
      "learning_rate": 0.0015,
      "loss": 1.6296,
      "step": 4027
    },
    {
      "epoch": 0.4248945147679325,
      "grad_norm": 0.5276506543159485,
      "learning_rate": 0.0015,
      "loss": 1.6656,
      "step": 4028
    },
    {
      "epoch": 0.425,
      "grad_norm": 0.5145576596260071,
      "learning_rate": 0.0015,
      "loss": 1.6396,
      "step": 4029
    },
    {
      "epoch": 0.42510548523206754,
      "grad_norm": 0.5211308598518372,
      "learning_rate": 0.0015,
      "loss": 1.664,
      "step": 4030
    },
    {
      "epoch": 0.42521097046413503,
      "grad_norm": 0.5144537687301636,
      "learning_rate": 0.0015,
      "loss": 1.6439,
      "step": 4031
    },
    {
      "epoch": 0.4253164556962025,
      "grad_norm": 0.5701487064361572,
      "learning_rate": 0.0015,
      "loss": 1.6957,
      "step": 4032
    },
    {
      "epoch": 0.42542194092827,
      "grad_norm": 0.508618175983429,
      "learning_rate": 0.0015,
      "loss": 1.6771,
      "step": 4033
    },
    {
      "epoch": 0.42552742616033756,
      "grad_norm": 0.5428481101989746,
      "learning_rate": 0.0015,
      "loss": 1.6609,
      "step": 4034
    },
    {
      "epoch": 0.42563291139240506,
      "grad_norm": 0.5794505476951599,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 4035
    },
    {
      "epoch": 0.42573839662447255,
      "grad_norm": 0.5958222150802612,
      "learning_rate": 0.0015,
      "loss": 1.6602,
      "step": 4036
    },
    {
      "epoch": 0.4258438818565401,
      "grad_norm": 0.4970637559890747,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 4037
    },
    {
      "epoch": 0.4259493670886076,
      "grad_norm": 0.6706885099411011,
      "learning_rate": 0.0015,
      "loss": 1.6809,
      "step": 4038
    },
    {
      "epoch": 0.4260548523206751,
      "grad_norm": 0.6228687763214111,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 4039
    },
    {
      "epoch": 0.42616033755274263,
      "grad_norm": 0.5116509199142456,
      "learning_rate": 0.0015,
      "loss": 1.6557,
      "step": 4040
    },
    {
      "epoch": 0.4262658227848101,
      "grad_norm": 0.5696230530738831,
      "learning_rate": 0.0015,
      "loss": 1.6606,
      "step": 4041
    },
    {
      "epoch": 0.4263713080168776,
      "grad_norm": 0.5878680944442749,
      "learning_rate": 0.0015,
      "loss": 1.6875,
      "step": 4042
    },
    {
      "epoch": 0.42647679324894516,
      "grad_norm": 0.5070853233337402,
      "learning_rate": 0.0015,
      "loss": 1.6489,
      "step": 4043
    },
    {
      "epoch": 0.42658227848101266,
      "grad_norm": 0.5454328060150146,
      "learning_rate": 0.0015,
      "loss": 1.6657,
      "step": 4044
    },
    {
      "epoch": 0.42668776371308015,
      "grad_norm": 0.6633443236351013,
      "learning_rate": 0.0015,
      "loss": 1.7154,
      "step": 4045
    },
    {
      "epoch": 0.4267932489451477,
      "grad_norm": 0.4916249215602875,
      "learning_rate": 0.0015,
      "loss": 1.6465,
      "step": 4046
    },
    {
      "epoch": 0.4268987341772152,
      "grad_norm": 0.710806667804718,
      "learning_rate": 0.0015,
      "loss": 1.6683,
      "step": 4047
    },
    {
      "epoch": 0.4270042194092827,
      "grad_norm": 0.6222331523895264,
      "learning_rate": 0.0015,
      "loss": 1.675,
      "step": 4048
    },
    {
      "epoch": 0.42710970464135023,
      "grad_norm": 0.467486172914505,
      "learning_rate": 0.0015,
      "loss": 1.6515,
      "step": 4049
    },
    {
      "epoch": 0.4272151898734177,
      "grad_norm": 0.6178696751594543,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 4050
    },
    {
      "epoch": 0.4273206751054852,
      "grad_norm": 0.516190230846405,
      "learning_rate": 0.0015,
      "loss": 1.6403,
      "step": 4051
    },
    {
      "epoch": 0.42742616033755276,
      "grad_norm": 0.520756721496582,
      "learning_rate": 0.0015,
      "loss": 1.6978,
      "step": 4052
    },
    {
      "epoch": 0.42753164556962026,
      "grad_norm": 0.7068386673927307,
      "learning_rate": 0.0015,
      "loss": 1.6823,
      "step": 4053
    },
    {
      "epoch": 0.42763713080168775,
      "grad_norm": 0.510657787322998,
      "learning_rate": 0.0015,
      "loss": 1.6672,
      "step": 4054
    },
    {
      "epoch": 0.4277426160337553,
      "grad_norm": 0.5260396003723145,
      "learning_rate": 0.0015,
      "loss": 1.6803,
      "step": 4055
    },
    {
      "epoch": 0.4278481012658228,
      "grad_norm": 0.779559850692749,
      "learning_rate": 0.0015,
      "loss": 1.6562,
      "step": 4056
    },
    {
      "epoch": 0.4279535864978903,
      "grad_norm": 0.5969604253768921,
      "learning_rate": 0.0015,
      "loss": 1.6619,
      "step": 4057
    },
    {
      "epoch": 0.42805907172995783,
      "grad_norm": 0.5348222255706787,
      "learning_rate": 0.0015,
      "loss": 1.6853,
      "step": 4058
    },
    {
      "epoch": 0.4281645569620253,
      "grad_norm": 0.6839258074760437,
      "learning_rate": 0.0015,
      "loss": 1.7112,
      "step": 4059
    },
    {
      "epoch": 0.4282700421940928,
      "grad_norm": 0.4636399447917938,
      "learning_rate": 0.0015,
      "loss": 1.6606,
      "step": 4060
    },
    {
      "epoch": 0.42837552742616036,
      "grad_norm": 0.6213461756706238,
      "learning_rate": 0.0015,
      "loss": 1.6851,
      "step": 4061
    },
    {
      "epoch": 0.42848101265822786,
      "grad_norm": 0.5466126799583435,
      "learning_rate": 0.0015,
      "loss": 1.667,
      "step": 4062
    },
    {
      "epoch": 0.42858649789029535,
      "grad_norm": 0.4445098340511322,
      "learning_rate": 0.0015,
      "loss": 1.6253,
      "step": 4063
    },
    {
      "epoch": 0.4286919831223629,
      "grad_norm": 0.5658165216445923,
      "learning_rate": 0.0015,
      "loss": 1.6474,
      "step": 4064
    },
    {
      "epoch": 0.4287974683544304,
      "grad_norm": 0.547552227973938,
      "learning_rate": 0.0015,
      "loss": 1.7097,
      "step": 4065
    },
    {
      "epoch": 0.4289029535864979,
      "grad_norm": 0.6399693489074707,
      "learning_rate": 0.0015,
      "loss": 1.6581,
      "step": 4066
    },
    {
      "epoch": 0.4290084388185654,
      "grad_norm": 0.47368189692497253,
      "learning_rate": 0.0015,
      "loss": 1.6551,
      "step": 4067
    },
    {
      "epoch": 0.4291139240506329,
      "grad_norm": 0.504464864730835,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 4068
    },
    {
      "epoch": 0.4292194092827004,
      "grad_norm": 0.4534807801246643,
      "learning_rate": 0.0015,
      "loss": 1.709,
      "step": 4069
    },
    {
      "epoch": 0.4293248945147679,
      "grad_norm": 0.5343837141990662,
      "learning_rate": 0.0015,
      "loss": 1.6656,
      "step": 4070
    },
    {
      "epoch": 0.42943037974683546,
      "grad_norm": 0.4665624797344208,
      "learning_rate": 0.0015,
      "loss": 1.7028,
      "step": 4071
    },
    {
      "epoch": 0.42953586497890295,
      "grad_norm": 0.5926090478897095,
      "learning_rate": 0.0015,
      "loss": 1.6518,
      "step": 4072
    },
    {
      "epoch": 0.42964135021097044,
      "grad_norm": 0.5662062764167786,
      "learning_rate": 0.0015,
      "loss": 1.6537,
      "step": 4073
    },
    {
      "epoch": 0.429746835443038,
      "grad_norm": 0.5438088774681091,
      "learning_rate": 0.0015,
      "loss": 1.6588,
      "step": 4074
    },
    {
      "epoch": 0.4298523206751055,
      "grad_norm": 0.7898258566856384,
      "learning_rate": 0.0015,
      "loss": 1.7159,
      "step": 4075
    },
    {
      "epoch": 0.429957805907173,
      "grad_norm": 0.6432499885559082,
      "learning_rate": 0.0015,
      "loss": 1.6759,
      "step": 4076
    },
    {
      "epoch": 0.4300632911392405,
      "grad_norm": 0.5525928139686584,
      "learning_rate": 0.0015,
      "loss": 1.6838,
      "step": 4077
    },
    {
      "epoch": 0.430168776371308,
      "grad_norm": 0.6819308400154114,
      "learning_rate": 0.0015,
      "loss": 1.6729,
      "step": 4078
    },
    {
      "epoch": 0.4302742616033755,
      "grad_norm": 0.5028029084205627,
      "learning_rate": 0.0015,
      "loss": 1.6731,
      "step": 4079
    },
    {
      "epoch": 0.43037974683544306,
      "grad_norm": 0.7023353576660156,
      "learning_rate": 0.0015,
      "loss": 1.6505,
      "step": 4080
    },
    {
      "epoch": 0.43048523206751055,
      "grad_norm": 0.5001094937324524,
      "learning_rate": 0.0015,
      "loss": 1.6696,
      "step": 4081
    },
    {
      "epoch": 0.43059071729957804,
      "grad_norm": 0.6945774555206299,
      "learning_rate": 0.0015,
      "loss": 1.696,
      "step": 4082
    },
    {
      "epoch": 0.4306962025316456,
      "grad_norm": 0.7422482967376709,
      "learning_rate": 0.0015,
      "loss": 1.7003,
      "step": 4083
    },
    {
      "epoch": 0.4308016877637131,
      "grad_norm": 0.5532791018486023,
      "learning_rate": 0.0015,
      "loss": 1.6301,
      "step": 4084
    },
    {
      "epoch": 0.4309071729957806,
      "grad_norm": 0.6867392659187317,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 4085
    },
    {
      "epoch": 0.4310126582278481,
      "grad_norm": 0.7010858654975891,
      "learning_rate": 0.0015,
      "loss": 1.6825,
      "step": 4086
    },
    {
      "epoch": 0.4311181434599156,
      "grad_norm": 0.5495920181274414,
      "learning_rate": 0.0015,
      "loss": 1.684,
      "step": 4087
    },
    {
      "epoch": 0.4312236286919831,
      "grad_norm": 0.778850257396698,
      "learning_rate": 0.0015,
      "loss": 1.6569,
      "step": 4088
    },
    {
      "epoch": 0.43132911392405066,
      "grad_norm": 0.7368293404579163,
      "learning_rate": 0.0015,
      "loss": 1.6376,
      "step": 4089
    },
    {
      "epoch": 0.43143459915611815,
      "grad_norm": 0.49126002192497253,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 4090
    },
    {
      "epoch": 0.43154008438818564,
      "grad_norm": 0.7283414006233215,
      "learning_rate": 0.0015,
      "loss": 1.7055,
      "step": 4091
    },
    {
      "epoch": 0.4316455696202532,
      "grad_norm": 0.547957181930542,
      "learning_rate": 0.0015,
      "loss": 1.6671,
      "step": 4092
    },
    {
      "epoch": 0.4317510548523207,
      "grad_norm": 0.5564680099487305,
      "learning_rate": 0.0015,
      "loss": 1.6707,
      "step": 4093
    },
    {
      "epoch": 0.4318565400843882,
      "grad_norm": 0.5678845047950745,
      "learning_rate": 0.0015,
      "loss": 1.6775,
      "step": 4094
    },
    {
      "epoch": 0.4319620253164557,
      "grad_norm": 0.614836573600769,
      "learning_rate": 0.0015,
      "loss": 1.6653,
      "step": 4095
    },
    {
      "epoch": 0.4320675105485232,
      "grad_norm": 0.5518618226051331,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 4096
    },
    {
      "epoch": 0.4321729957805907,
      "grad_norm": 0.49032047390937805,
      "learning_rate": 0.0015,
      "loss": 1.6562,
      "step": 4097
    },
    {
      "epoch": 0.43227848101265826,
      "grad_norm": 0.5626969933509827,
      "learning_rate": 0.0015,
      "loss": 1.677,
      "step": 4098
    },
    {
      "epoch": 0.43238396624472575,
      "grad_norm": 0.4826257824897766,
      "learning_rate": 0.0015,
      "loss": 1.6186,
      "step": 4099
    },
    {
      "epoch": 0.43248945147679324,
      "grad_norm": 0.5447328090667725,
      "learning_rate": 0.0015,
      "loss": 1.6441,
      "step": 4100
    },
    {
      "epoch": 0.43259493670886073,
      "grad_norm": 0.523389458656311,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 4101
    },
    {
      "epoch": 0.4327004219409283,
      "grad_norm": 0.46412360668182373,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 4102
    },
    {
      "epoch": 0.4328059071729958,
      "grad_norm": 0.49491190910339355,
      "learning_rate": 0.0015,
      "loss": 1.6035,
      "step": 4103
    },
    {
      "epoch": 0.43291139240506327,
      "grad_norm": 0.478269100189209,
      "learning_rate": 0.0015,
      "loss": 1.6544,
      "step": 4104
    },
    {
      "epoch": 0.4330168776371308,
      "grad_norm": 0.4965742528438568,
      "learning_rate": 0.0015,
      "loss": 1.6399,
      "step": 4105
    },
    {
      "epoch": 0.4331223628691983,
      "grad_norm": 0.49467846751213074,
      "learning_rate": 0.0015,
      "loss": 1.6821,
      "step": 4106
    },
    {
      "epoch": 0.4332278481012658,
      "grad_norm": 0.5212065577507019,
      "learning_rate": 0.0015,
      "loss": 1.6796,
      "step": 4107
    },
    {
      "epoch": 0.43333333333333335,
      "grad_norm": 0.484543114900589,
      "learning_rate": 0.0015,
      "loss": 1.6438,
      "step": 4108
    },
    {
      "epoch": 0.43343881856540084,
      "grad_norm": 0.47857120633125305,
      "learning_rate": 0.0015,
      "loss": 1.6884,
      "step": 4109
    },
    {
      "epoch": 0.43354430379746833,
      "grad_norm": 0.497016042470932,
      "learning_rate": 0.0015,
      "loss": 1.6906,
      "step": 4110
    },
    {
      "epoch": 0.4336497890295359,
      "grad_norm": 0.5423179864883423,
      "learning_rate": 0.0015,
      "loss": 1.6412,
      "step": 4111
    },
    {
      "epoch": 0.4337552742616034,
      "grad_norm": 0.6989853382110596,
      "learning_rate": 0.0015,
      "loss": 1.6309,
      "step": 4112
    },
    {
      "epoch": 0.43386075949367087,
      "grad_norm": 0.6580510139465332,
      "learning_rate": 0.0015,
      "loss": 1.671,
      "step": 4113
    },
    {
      "epoch": 0.4339662447257384,
      "grad_norm": 0.5103244185447693,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 4114
    },
    {
      "epoch": 0.4340717299578059,
      "grad_norm": 0.6883593797683716,
      "learning_rate": 0.0015,
      "loss": 1.6612,
      "step": 4115
    },
    {
      "epoch": 0.4341772151898734,
      "grad_norm": 0.48387381434440613,
      "learning_rate": 0.0015,
      "loss": 1.6506,
      "step": 4116
    },
    {
      "epoch": 0.43428270042194095,
      "grad_norm": 0.8011125326156616,
      "learning_rate": 0.0015,
      "loss": 1.6675,
      "step": 4117
    },
    {
      "epoch": 0.43438818565400844,
      "grad_norm": 0.7306374907493591,
      "learning_rate": 0.0015,
      "loss": 1.6583,
      "step": 4118
    },
    {
      "epoch": 0.43449367088607593,
      "grad_norm": 0.6169559955596924,
      "learning_rate": 0.0015,
      "loss": 1.6843,
      "step": 4119
    },
    {
      "epoch": 0.4345991561181435,
      "grad_norm": 0.8769050240516663,
      "learning_rate": 0.0015,
      "loss": 1.6846,
      "step": 4120
    },
    {
      "epoch": 0.434704641350211,
      "grad_norm": 0.5737379193305969,
      "learning_rate": 0.0015,
      "loss": 1.6742,
      "step": 4121
    },
    {
      "epoch": 0.43481012658227847,
      "grad_norm": 0.6177200675010681,
      "learning_rate": 0.0015,
      "loss": 1.6509,
      "step": 4122
    },
    {
      "epoch": 0.434915611814346,
      "grad_norm": 0.5053166151046753,
      "learning_rate": 0.0015,
      "loss": 1.6283,
      "step": 4123
    },
    {
      "epoch": 0.4350210970464135,
      "grad_norm": 0.6265692114830017,
      "learning_rate": 0.0015,
      "loss": 1.6743,
      "step": 4124
    },
    {
      "epoch": 0.435126582278481,
      "grad_norm": 0.47219255566596985,
      "learning_rate": 0.0015,
      "loss": 1.6459,
      "step": 4125
    },
    {
      "epoch": 0.43523206751054855,
      "grad_norm": 0.6443513631820679,
      "learning_rate": 0.0015,
      "loss": 1.6619,
      "step": 4126
    },
    {
      "epoch": 0.43533755274261604,
      "grad_norm": 0.5777150392532349,
      "learning_rate": 0.0015,
      "loss": 1.6495,
      "step": 4127
    },
    {
      "epoch": 0.43544303797468353,
      "grad_norm": 0.5461890697479248,
      "learning_rate": 0.0015,
      "loss": 1.6613,
      "step": 4128
    },
    {
      "epoch": 0.4355485232067511,
      "grad_norm": 0.5516852736473083,
      "learning_rate": 0.0015,
      "loss": 1.6762,
      "step": 4129
    },
    {
      "epoch": 0.4356540084388186,
      "grad_norm": 0.5438812375068665,
      "learning_rate": 0.0015,
      "loss": 1.6687,
      "step": 4130
    },
    {
      "epoch": 0.43575949367088607,
      "grad_norm": 0.5853663682937622,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 4131
    },
    {
      "epoch": 0.43586497890295356,
      "grad_norm": 0.6321829557418823,
      "learning_rate": 0.0015,
      "loss": 1.6493,
      "step": 4132
    },
    {
      "epoch": 0.4359704641350211,
      "grad_norm": 0.5404390692710876,
      "learning_rate": 0.0015,
      "loss": 1.6841,
      "step": 4133
    },
    {
      "epoch": 0.4360759493670886,
      "grad_norm": 0.5971294045448303,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 4134
    },
    {
      "epoch": 0.4361814345991561,
      "grad_norm": 0.5202499628067017,
      "learning_rate": 0.0015,
      "loss": 1.6258,
      "step": 4135
    },
    {
      "epoch": 0.43628691983122364,
      "grad_norm": 0.5141414999961853,
      "learning_rate": 0.0015,
      "loss": 1.6493,
      "step": 4136
    },
    {
      "epoch": 0.43639240506329113,
      "grad_norm": 0.6311295628547668,
      "learning_rate": 0.0015,
      "loss": 1.649,
      "step": 4137
    },
    {
      "epoch": 0.4364978902953586,
      "grad_norm": 0.56368488073349,
      "learning_rate": 0.0015,
      "loss": 1.6611,
      "step": 4138
    },
    {
      "epoch": 0.4366033755274262,
      "grad_norm": 0.5762485265731812,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 4139
    },
    {
      "epoch": 0.43670886075949367,
      "grad_norm": 0.6775660514831543,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 4140
    },
    {
      "epoch": 0.43681434599156116,
      "grad_norm": 0.7037628889083862,
      "learning_rate": 0.0015,
      "loss": 1.6521,
      "step": 4141
    },
    {
      "epoch": 0.4369198312236287,
      "grad_norm": 0.6395363211631775,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 4142
    },
    {
      "epoch": 0.4370253164556962,
      "grad_norm": 0.630176842212677,
      "learning_rate": 0.0015,
      "loss": 1.675,
      "step": 4143
    },
    {
      "epoch": 0.4371308016877637,
      "grad_norm": 0.549116849899292,
      "learning_rate": 0.0015,
      "loss": 1.6775,
      "step": 4144
    },
    {
      "epoch": 0.43723628691983124,
      "grad_norm": 0.6416967511177063,
      "learning_rate": 0.0015,
      "loss": 1.6469,
      "step": 4145
    },
    {
      "epoch": 0.43734177215189873,
      "grad_norm": 0.47405681014060974,
      "learning_rate": 0.0015,
      "loss": 1.6487,
      "step": 4146
    },
    {
      "epoch": 0.4374472573839662,
      "grad_norm": 0.5568174123764038,
      "learning_rate": 0.0015,
      "loss": 1.6764,
      "step": 4147
    },
    {
      "epoch": 0.4375527426160338,
      "grad_norm": 0.49878600239753723,
      "learning_rate": 0.0015,
      "loss": 1.7047,
      "step": 4148
    },
    {
      "epoch": 0.43765822784810127,
      "grad_norm": 0.5601268410682678,
      "learning_rate": 0.0015,
      "loss": 1.6497,
      "step": 4149
    },
    {
      "epoch": 0.43776371308016876,
      "grad_norm": 0.6269808411598206,
      "learning_rate": 0.0015,
      "loss": 1.6658,
      "step": 4150
    },
    {
      "epoch": 0.4378691983122363,
      "grad_norm": 0.5254350304603577,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 4151
    },
    {
      "epoch": 0.4379746835443038,
      "grad_norm": 0.5711461901664734,
      "learning_rate": 0.0015,
      "loss": 1.6834,
      "step": 4152
    },
    {
      "epoch": 0.4380801687763713,
      "grad_norm": 0.640809178352356,
      "learning_rate": 0.0015,
      "loss": 1.6707,
      "step": 4153
    },
    {
      "epoch": 0.43818565400843884,
      "grad_norm": 0.665875256061554,
      "learning_rate": 0.0015,
      "loss": 1.6188,
      "step": 4154
    },
    {
      "epoch": 0.43829113924050633,
      "grad_norm": 0.6013915538787842,
      "learning_rate": 0.0015,
      "loss": 1.661,
      "step": 4155
    },
    {
      "epoch": 0.4383966244725738,
      "grad_norm": 0.5874208807945251,
      "learning_rate": 0.0015,
      "loss": 1.6708,
      "step": 4156
    },
    {
      "epoch": 0.4385021097046414,
      "grad_norm": 0.5663081407546997,
      "learning_rate": 0.0015,
      "loss": 1.6761,
      "step": 4157
    },
    {
      "epoch": 0.43860759493670887,
      "grad_norm": 0.7340829968452454,
      "learning_rate": 0.0015,
      "loss": 1.6359,
      "step": 4158
    },
    {
      "epoch": 0.43871308016877636,
      "grad_norm": 0.649359405040741,
      "learning_rate": 0.0015,
      "loss": 1.6562,
      "step": 4159
    },
    {
      "epoch": 0.4388185654008439,
      "grad_norm": 0.5673907995223999,
      "learning_rate": 0.0015,
      "loss": 1.6347,
      "step": 4160
    },
    {
      "epoch": 0.4389240506329114,
      "grad_norm": 0.5553560256958008,
      "learning_rate": 0.0015,
      "loss": 1.6557,
      "step": 4161
    },
    {
      "epoch": 0.4390295358649789,
      "grad_norm": 0.5494565367698669,
      "learning_rate": 0.0015,
      "loss": 1.6755,
      "step": 4162
    },
    {
      "epoch": 0.43913502109704644,
      "grad_norm": 0.653363823890686,
      "learning_rate": 0.0015,
      "loss": 1.6753,
      "step": 4163
    },
    {
      "epoch": 0.43924050632911393,
      "grad_norm": 0.6180105209350586,
      "learning_rate": 0.0015,
      "loss": 1.6262,
      "step": 4164
    },
    {
      "epoch": 0.4393459915611814,
      "grad_norm": 0.5122617483139038,
      "learning_rate": 0.0015,
      "loss": 1.6662,
      "step": 4165
    },
    {
      "epoch": 0.4394514767932489,
      "grad_norm": 0.4645233452320099,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 4166
    },
    {
      "epoch": 0.43955696202531647,
      "grad_norm": 0.5114132165908813,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 4167
    },
    {
      "epoch": 0.43966244725738396,
      "grad_norm": 0.46179649233818054,
      "learning_rate": 0.0015,
      "loss": 1.657,
      "step": 4168
    },
    {
      "epoch": 0.43976793248945145,
      "grad_norm": 0.5351398587226868,
      "learning_rate": 0.0015,
      "loss": 1.67,
      "step": 4169
    },
    {
      "epoch": 0.439873417721519,
      "grad_norm": 0.5336154103279114,
      "learning_rate": 0.0015,
      "loss": 1.6925,
      "step": 4170
    },
    {
      "epoch": 0.4399789029535865,
      "grad_norm": 0.5214482545852661,
      "learning_rate": 0.0015,
      "loss": 1.6554,
      "step": 4171
    },
    {
      "epoch": 0.440084388185654,
      "grad_norm": 0.5894679427146912,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 4172
    },
    {
      "epoch": 0.44018987341772153,
      "grad_norm": 0.6982522010803223,
      "learning_rate": 0.0015,
      "loss": 1.6913,
      "step": 4173
    },
    {
      "epoch": 0.440295358649789,
      "grad_norm": 0.671720564365387,
      "learning_rate": 0.0015,
      "loss": 1.6704,
      "step": 4174
    },
    {
      "epoch": 0.4404008438818565,
      "grad_norm": 0.5447161197662354,
      "learning_rate": 0.0015,
      "loss": 1.6359,
      "step": 4175
    },
    {
      "epoch": 0.44050632911392407,
      "grad_norm": 0.583261251449585,
      "learning_rate": 0.0015,
      "loss": 1.6277,
      "step": 4176
    },
    {
      "epoch": 0.44061181434599156,
      "grad_norm": 0.59373939037323,
      "learning_rate": 0.0015,
      "loss": 1.6726,
      "step": 4177
    },
    {
      "epoch": 0.44071729957805905,
      "grad_norm": 0.6165011525154114,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 4178
    },
    {
      "epoch": 0.4408227848101266,
      "grad_norm": 0.5901514887809753,
      "learning_rate": 0.0015,
      "loss": 1.6789,
      "step": 4179
    },
    {
      "epoch": 0.4409282700421941,
      "grad_norm": 0.7335607409477234,
      "learning_rate": 0.0015,
      "loss": 1.6407,
      "step": 4180
    },
    {
      "epoch": 0.4410337552742616,
      "grad_norm": 0.6217585206031799,
      "learning_rate": 0.0015,
      "loss": 1.6664,
      "step": 4181
    },
    {
      "epoch": 0.44113924050632913,
      "grad_norm": 0.47675660252571106,
      "learning_rate": 0.0015,
      "loss": 1.6681,
      "step": 4182
    },
    {
      "epoch": 0.4412447257383966,
      "grad_norm": 0.6070347428321838,
      "learning_rate": 0.0015,
      "loss": 1.6508,
      "step": 4183
    },
    {
      "epoch": 0.4413502109704641,
      "grad_norm": 0.5597372651100159,
      "learning_rate": 0.0015,
      "loss": 1.6563,
      "step": 4184
    },
    {
      "epoch": 0.44145569620253167,
      "grad_norm": 0.5119388699531555,
      "learning_rate": 0.0015,
      "loss": 1.6509,
      "step": 4185
    },
    {
      "epoch": 0.44156118143459916,
      "grad_norm": 0.5131213665008545,
      "learning_rate": 0.0015,
      "loss": 1.6509,
      "step": 4186
    },
    {
      "epoch": 0.44166666666666665,
      "grad_norm": 0.5310493111610413,
      "learning_rate": 0.0015,
      "loss": 1.6765,
      "step": 4187
    },
    {
      "epoch": 0.4417721518987342,
      "grad_norm": 0.6309808492660522,
      "learning_rate": 0.0015,
      "loss": 1.6822,
      "step": 4188
    },
    {
      "epoch": 0.4418776371308017,
      "grad_norm": 0.5191870331764221,
      "learning_rate": 0.0015,
      "loss": 1.6573,
      "step": 4189
    },
    {
      "epoch": 0.4419831223628692,
      "grad_norm": 0.5047029852867126,
      "learning_rate": 0.0015,
      "loss": 1.6729,
      "step": 4190
    },
    {
      "epoch": 0.44208860759493673,
      "grad_norm": 0.5043741464614868,
      "learning_rate": 0.0015,
      "loss": 1.6488,
      "step": 4191
    },
    {
      "epoch": 0.4421940928270042,
      "grad_norm": 0.5557615756988525,
      "learning_rate": 0.0015,
      "loss": 1.6806,
      "step": 4192
    },
    {
      "epoch": 0.4422995780590717,
      "grad_norm": 0.5199964046478271,
      "learning_rate": 0.0015,
      "loss": 1.6463,
      "step": 4193
    },
    {
      "epoch": 0.44240506329113927,
      "grad_norm": 0.6202685236930847,
      "learning_rate": 0.0015,
      "loss": 1.6954,
      "step": 4194
    },
    {
      "epoch": 0.44251054852320676,
      "grad_norm": 0.49905499815940857,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 4195
    },
    {
      "epoch": 0.44261603375527425,
      "grad_norm": 0.6017361283302307,
      "learning_rate": 0.0015,
      "loss": 1.6308,
      "step": 4196
    },
    {
      "epoch": 0.44272151898734174,
      "grad_norm": 0.6024526357650757,
      "learning_rate": 0.0015,
      "loss": 1.6564,
      "step": 4197
    },
    {
      "epoch": 0.4428270042194093,
      "grad_norm": 0.5642350912094116,
      "learning_rate": 0.0015,
      "loss": 1.6663,
      "step": 4198
    },
    {
      "epoch": 0.4429324894514768,
      "grad_norm": 0.6400055289268494,
      "learning_rate": 0.0015,
      "loss": 1.6394,
      "step": 4199
    },
    {
      "epoch": 0.4430379746835443,
      "grad_norm": 0.48151862621307373,
      "learning_rate": 0.0015,
      "loss": 1.6997,
      "step": 4200
    },
    {
      "epoch": 0.4431434599156118,
      "grad_norm": 0.6156302094459534,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 4201
    },
    {
      "epoch": 0.4432489451476793,
      "grad_norm": 0.5519492626190186,
      "learning_rate": 0.0015,
      "loss": 1.6594,
      "step": 4202
    },
    {
      "epoch": 0.4433544303797468,
      "grad_norm": 0.5420129299163818,
      "learning_rate": 0.0015,
      "loss": 1.6267,
      "step": 4203
    },
    {
      "epoch": 0.44345991561181436,
      "grad_norm": 0.5757969617843628,
      "learning_rate": 0.0015,
      "loss": 1.6709,
      "step": 4204
    },
    {
      "epoch": 0.44356540084388185,
      "grad_norm": 0.5687883496284485,
      "learning_rate": 0.0015,
      "loss": 1.6236,
      "step": 4205
    },
    {
      "epoch": 0.44367088607594934,
      "grad_norm": 0.5870119333267212,
      "learning_rate": 0.0015,
      "loss": 1.6791,
      "step": 4206
    },
    {
      "epoch": 0.4437763713080169,
      "grad_norm": 0.5836820006370544,
      "learning_rate": 0.0015,
      "loss": 1.6354,
      "step": 4207
    },
    {
      "epoch": 0.4438818565400844,
      "grad_norm": 0.5888475775718689,
      "learning_rate": 0.0015,
      "loss": 1.6497,
      "step": 4208
    },
    {
      "epoch": 0.4439873417721519,
      "grad_norm": 0.6259241700172424,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 4209
    },
    {
      "epoch": 0.4440928270042194,
      "grad_norm": 0.5528420209884644,
      "learning_rate": 0.0015,
      "loss": 1.6648,
      "step": 4210
    },
    {
      "epoch": 0.4441983122362869,
      "grad_norm": 0.5624960064888,
      "learning_rate": 0.0015,
      "loss": 1.682,
      "step": 4211
    },
    {
      "epoch": 0.4443037974683544,
      "grad_norm": 0.571373462677002,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 4212
    },
    {
      "epoch": 0.44440928270042196,
      "grad_norm": 0.5219664573669434,
      "learning_rate": 0.0015,
      "loss": 1.6802,
      "step": 4213
    },
    {
      "epoch": 0.44451476793248945,
      "grad_norm": 0.59315425157547,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 4214
    },
    {
      "epoch": 0.44462025316455694,
      "grad_norm": 0.6662893295288086,
      "learning_rate": 0.0015,
      "loss": 1.6371,
      "step": 4215
    },
    {
      "epoch": 0.4447257383966245,
      "grad_norm": 0.6315421462059021,
      "learning_rate": 0.0015,
      "loss": 1.6688,
      "step": 4216
    },
    {
      "epoch": 0.444831223628692,
      "grad_norm": 0.45515403151512146,
      "learning_rate": 0.0015,
      "loss": 1.6564,
      "step": 4217
    },
    {
      "epoch": 0.4449367088607595,
      "grad_norm": 0.7152280211448669,
      "learning_rate": 0.0015,
      "loss": 1.6295,
      "step": 4218
    },
    {
      "epoch": 0.445042194092827,
      "grad_norm": 0.7748788595199585,
      "learning_rate": 0.0015,
      "loss": 1.6777,
      "step": 4219
    },
    {
      "epoch": 0.4451476793248945,
      "grad_norm": 0.6827630996704102,
      "learning_rate": 0.0015,
      "loss": 1.6737,
      "step": 4220
    },
    {
      "epoch": 0.445253164556962,
      "grad_norm": 0.5411749482154846,
      "learning_rate": 0.0015,
      "loss": 1.6607,
      "step": 4221
    },
    {
      "epoch": 0.44535864978902956,
      "grad_norm": 0.6788249015808105,
      "learning_rate": 0.0015,
      "loss": 1.6741,
      "step": 4222
    },
    {
      "epoch": 0.44546413502109705,
      "grad_norm": 0.5095070600509644,
      "learning_rate": 0.0015,
      "loss": 1.6767,
      "step": 4223
    },
    {
      "epoch": 0.44556962025316454,
      "grad_norm": 0.5586391687393188,
      "learning_rate": 0.0015,
      "loss": 1.6403,
      "step": 4224
    },
    {
      "epoch": 0.4456751054852321,
      "grad_norm": 0.5388392806053162,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 4225
    },
    {
      "epoch": 0.4457805907172996,
      "grad_norm": 0.5191367864608765,
      "learning_rate": 0.0015,
      "loss": 1.6517,
      "step": 4226
    },
    {
      "epoch": 0.4458860759493671,
      "grad_norm": 0.5413681864738464,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 4227
    },
    {
      "epoch": 0.4459915611814346,
      "grad_norm": 0.44679126143455505,
      "learning_rate": 0.0015,
      "loss": 1.669,
      "step": 4228
    },
    {
      "epoch": 0.4460970464135021,
      "grad_norm": 0.5609684586524963,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 4229
    },
    {
      "epoch": 0.4462025316455696,
      "grad_norm": 0.4728687107563019,
      "learning_rate": 0.0015,
      "loss": 1.6526,
      "step": 4230
    },
    {
      "epoch": 0.4463080168776371,
      "grad_norm": 0.6355165243148804,
      "learning_rate": 0.0015,
      "loss": 1.7058,
      "step": 4231
    },
    {
      "epoch": 0.44641350210970465,
      "grad_norm": 0.4868917763233185,
      "learning_rate": 0.0015,
      "loss": 1.6585,
      "step": 4232
    },
    {
      "epoch": 0.44651898734177214,
      "grad_norm": 0.5380212664604187,
      "learning_rate": 0.0015,
      "loss": 1.6511,
      "step": 4233
    },
    {
      "epoch": 0.44662447257383964,
      "grad_norm": 0.5393729209899902,
      "learning_rate": 0.0015,
      "loss": 1.6858,
      "step": 4234
    },
    {
      "epoch": 0.4467299578059072,
      "grad_norm": 0.559303879737854,
      "learning_rate": 0.0015,
      "loss": 1.6569,
      "step": 4235
    },
    {
      "epoch": 0.4468354430379747,
      "grad_norm": 0.5218153595924377,
      "learning_rate": 0.0015,
      "loss": 1.6294,
      "step": 4236
    },
    {
      "epoch": 0.44694092827004217,
      "grad_norm": 0.5358302593231201,
      "learning_rate": 0.0015,
      "loss": 1.6735,
      "step": 4237
    },
    {
      "epoch": 0.4470464135021097,
      "grad_norm": 0.6105931401252747,
      "learning_rate": 0.0015,
      "loss": 1.6052,
      "step": 4238
    },
    {
      "epoch": 0.4471518987341772,
      "grad_norm": 0.5396553874015808,
      "learning_rate": 0.0015,
      "loss": 1.6401,
      "step": 4239
    },
    {
      "epoch": 0.4472573839662447,
      "grad_norm": 0.47959256172180176,
      "learning_rate": 0.0015,
      "loss": 1.6405,
      "step": 4240
    },
    {
      "epoch": 0.44736286919831225,
      "grad_norm": 0.6182860732078552,
      "learning_rate": 0.0015,
      "loss": 1.6376,
      "step": 4241
    },
    {
      "epoch": 0.44746835443037974,
      "grad_norm": 0.552215039730072,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 4242
    },
    {
      "epoch": 0.44757383966244724,
      "grad_norm": 0.59195876121521,
      "learning_rate": 0.0015,
      "loss": 1.6673,
      "step": 4243
    },
    {
      "epoch": 0.4476793248945148,
      "grad_norm": 0.5790740847587585,
      "learning_rate": 0.0015,
      "loss": 1.6685,
      "step": 4244
    },
    {
      "epoch": 0.4477848101265823,
      "grad_norm": 0.5079163908958435,
      "learning_rate": 0.0015,
      "loss": 1.6931,
      "step": 4245
    },
    {
      "epoch": 0.44789029535864977,
      "grad_norm": 0.6338910460472107,
      "learning_rate": 0.0015,
      "loss": 1.6364,
      "step": 4246
    },
    {
      "epoch": 0.4479957805907173,
      "grad_norm": 0.7259997725486755,
      "learning_rate": 0.0015,
      "loss": 1.6124,
      "step": 4247
    },
    {
      "epoch": 0.4481012658227848,
      "grad_norm": 0.5896361470222473,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 4248
    },
    {
      "epoch": 0.4482067510548523,
      "grad_norm": 0.6032668948173523,
      "learning_rate": 0.0015,
      "loss": 1.6878,
      "step": 4249
    },
    {
      "epoch": 0.44831223628691985,
      "grad_norm": 0.6996051669120789,
      "learning_rate": 0.0015,
      "loss": 1.6764,
      "step": 4250
    },
    {
      "epoch": 0.44841772151898734,
      "grad_norm": 0.4835391342639923,
      "learning_rate": 0.0015,
      "loss": 1.6569,
      "step": 4251
    },
    {
      "epoch": 0.44852320675105484,
      "grad_norm": 0.6426658630371094,
      "learning_rate": 0.0015,
      "loss": 1.6771,
      "step": 4252
    },
    {
      "epoch": 0.4486286919831224,
      "grad_norm": 0.6918848156929016,
      "learning_rate": 0.0015,
      "loss": 1.6617,
      "step": 4253
    },
    {
      "epoch": 0.4487341772151899,
      "grad_norm": 0.5265471339225769,
      "learning_rate": 0.0015,
      "loss": 1.6584,
      "step": 4254
    },
    {
      "epoch": 0.44883966244725737,
      "grad_norm": 0.7319307923316956,
      "learning_rate": 0.0015,
      "loss": 1.7017,
      "step": 4255
    },
    {
      "epoch": 0.4489451476793249,
      "grad_norm": 0.8076063990592957,
      "learning_rate": 0.0015,
      "loss": 1.6984,
      "step": 4256
    },
    {
      "epoch": 0.4490506329113924,
      "grad_norm": 0.48832979798316956,
      "learning_rate": 0.0015,
      "loss": 1.6249,
      "step": 4257
    },
    {
      "epoch": 0.4491561181434599,
      "grad_norm": 0.8370751142501831,
      "learning_rate": 0.0015,
      "loss": 1.6584,
      "step": 4258
    },
    {
      "epoch": 0.44926160337552745,
      "grad_norm": 0.6099622249603271,
      "learning_rate": 0.0015,
      "loss": 1.6869,
      "step": 4259
    },
    {
      "epoch": 0.44936708860759494,
      "grad_norm": 0.7295295000076294,
      "learning_rate": 0.0015,
      "loss": 1.6452,
      "step": 4260
    },
    {
      "epoch": 0.44947257383966244,
      "grad_norm": 0.8976764678955078,
      "learning_rate": 0.0015,
      "loss": 1.6872,
      "step": 4261
    },
    {
      "epoch": 0.44957805907173,
      "grad_norm": 0.5718755125999451,
      "learning_rate": 0.0015,
      "loss": 1.6774,
      "step": 4262
    },
    {
      "epoch": 0.4496835443037975,
      "grad_norm": 0.6952678561210632,
      "learning_rate": 0.0015,
      "loss": 1.6282,
      "step": 4263
    },
    {
      "epoch": 0.44978902953586497,
      "grad_norm": 0.9510587453842163,
      "learning_rate": 0.0015,
      "loss": 1.6858,
      "step": 4264
    },
    {
      "epoch": 0.44989451476793246,
      "grad_norm": 0.5682377219200134,
      "learning_rate": 0.0015,
      "loss": 1.6521,
      "step": 4265
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8752272129058838,
      "learning_rate": 0.0015,
      "loss": 1.665,
      "step": 4266
    },
    {
      "epoch": 0.4501054852320675,
      "grad_norm": 0.8616081476211548,
      "learning_rate": 0.0015,
      "loss": 1.627,
      "step": 4267
    },
    {
      "epoch": 0.450210970464135,
      "grad_norm": 0.5683330297470093,
      "learning_rate": 0.0015,
      "loss": 1.644,
      "step": 4268
    },
    {
      "epoch": 0.45031645569620254,
      "grad_norm": 0.6447488069534302,
      "learning_rate": 0.0015,
      "loss": 1.6388,
      "step": 4269
    },
    {
      "epoch": 0.45042194092827004,
      "grad_norm": 0.7037097811698914,
      "learning_rate": 0.0015,
      "loss": 1.6874,
      "step": 4270
    },
    {
      "epoch": 0.45052742616033753,
      "grad_norm": 0.6476287841796875,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 4271
    },
    {
      "epoch": 0.4506329113924051,
      "grad_norm": 0.6518765687942505,
      "learning_rate": 0.0015,
      "loss": 1.6615,
      "step": 4272
    },
    {
      "epoch": 0.45073839662447257,
      "grad_norm": 0.5910608768463135,
      "learning_rate": 0.0015,
      "loss": 1.6516,
      "step": 4273
    },
    {
      "epoch": 0.45084388185654006,
      "grad_norm": 0.4365657866001129,
      "learning_rate": 0.0015,
      "loss": 1.6485,
      "step": 4274
    },
    {
      "epoch": 0.4509493670886076,
      "grad_norm": 0.6063331961631775,
      "learning_rate": 0.0015,
      "loss": 1.637,
      "step": 4275
    },
    {
      "epoch": 0.4510548523206751,
      "grad_norm": 0.45042088627815247,
      "learning_rate": 0.0015,
      "loss": 1.6264,
      "step": 4276
    },
    {
      "epoch": 0.4511603375527426,
      "grad_norm": 0.5372198224067688,
      "learning_rate": 0.0015,
      "loss": 1.6842,
      "step": 4277
    },
    {
      "epoch": 0.45126582278481014,
      "grad_norm": 0.5263211727142334,
      "learning_rate": 0.0015,
      "loss": 1.6348,
      "step": 4278
    },
    {
      "epoch": 0.45137130801687764,
      "grad_norm": 0.7315890789031982,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 4279
    },
    {
      "epoch": 0.45147679324894513,
      "grad_norm": 0.6854990720748901,
      "learning_rate": 0.0015,
      "loss": 1.6867,
      "step": 4280
    },
    {
      "epoch": 0.4515822784810127,
      "grad_norm": 0.5408880710601807,
      "learning_rate": 0.0015,
      "loss": 1.6473,
      "step": 4281
    },
    {
      "epoch": 0.45168776371308017,
      "grad_norm": 0.5557799935340881,
      "learning_rate": 0.0015,
      "loss": 1.7033,
      "step": 4282
    },
    {
      "epoch": 0.45179324894514766,
      "grad_norm": 0.5755519866943359,
      "learning_rate": 0.0015,
      "loss": 1.6741,
      "step": 4283
    },
    {
      "epoch": 0.4518987341772152,
      "grad_norm": 0.4988951086997986,
      "learning_rate": 0.0015,
      "loss": 1.6618,
      "step": 4284
    },
    {
      "epoch": 0.4520042194092827,
      "grad_norm": 0.612045407295227,
      "learning_rate": 0.0015,
      "loss": 1.6337,
      "step": 4285
    },
    {
      "epoch": 0.4521097046413502,
      "grad_norm": 0.5338757634162903,
      "learning_rate": 0.0015,
      "loss": 1.6275,
      "step": 4286
    },
    {
      "epoch": 0.45221518987341774,
      "grad_norm": 0.58327716588974,
      "learning_rate": 0.0015,
      "loss": 1.6386,
      "step": 4287
    },
    {
      "epoch": 0.45232067510548524,
      "grad_norm": 0.5607724785804749,
      "learning_rate": 0.0015,
      "loss": 1.6941,
      "step": 4288
    },
    {
      "epoch": 0.45242616033755273,
      "grad_norm": 0.5698756575584412,
      "learning_rate": 0.0015,
      "loss": 1.6878,
      "step": 4289
    },
    {
      "epoch": 0.4525316455696203,
      "grad_norm": 0.5892707705497742,
      "learning_rate": 0.0015,
      "loss": 1.6408,
      "step": 4290
    },
    {
      "epoch": 0.45263713080168777,
      "grad_norm": 0.5143903493881226,
      "learning_rate": 0.0015,
      "loss": 1.6467,
      "step": 4291
    },
    {
      "epoch": 0.45274261603375526,
      "grad_norm": 0.5467579364776611,
      "learning_rate": 0.0015,
      "loss": 1.6578,
      "step": 4292
    },
    {
      "epoch": 0.4528481012658228,
      "grad_norm": 0.49005115032196045,
      "learning_rate": 0.0015,
      "loss": 1.6943,
      "step": 4293
    },
    {
      "epoch": 0.4529535864978903,
      "grad_norm": 0.6001521944999695,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 4294
    },
    {
      "epoch": 0.4530590717299578,
      "grad_norm": 0.5293867588043213,
      "learning_rate": 0.0015,
      "loss": 1.653,
      "step": 4295
    },
    {
      "epoch": 0.4531645569620253,
      "grad_norm": 0.46255889534950256,
      "learning_rate": 0.0015,
      "loss": 1.6647,
      "step": 4296
    },
    {
      "epoch": 0.45327004219409284,
      "grad_norm": 0.6147642135620117,
      "learning_rate": 0.0015,
      "loss": 1.6334,
      "step": 4297
    },
    {
      "epoch": 0.45337552742616033,
      "grad_norm": 0.6130396723747253,
      "learning_rate": 0.0015,
      "loss": 1.6642,
      "step": 4298
    },
    {
      "epoch": 0.4534810126582278,
      "grad_norm": 0.4362105429172516,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 4299
    },
    {
      "epoch": 0.45358649789029537,
      "grad_norm": 0.7801738977432251,
      "learning_rate": 0.0015,
      "loss": 1.6513,
      "step": 4300
    },
    {
      "epoch": 0.45369198312236286,
      "grad_norm": 0.7096121311187744,
      "learning_rate": 0.0015,
      "loss": 1.6991,
      "step": 4301
    },
    {
      "epoch": 0.45379746835443036,
      "grad_norm": 0.5057912468910217,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 4302
    },
    {
      "epoch": 0.4539029535864979,
      "grad_norm": 0.9966181516647339,
      "learning_rate": 0.0015,
      "loss": 1.6613,
      "step": 4303
    },
    {
      "epoch": 0.4540084388185654,
      "grad_norm": 0.7544216513633728,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 4304
    },
    {
      "epoch": 0.4541139240506329,
      "grad_norm": 0.5713809728622437,
      "learning_rate": 0.0015,
      "loss": 1.6053,
      "step": 4305
    },
    {
      "epoch": 0.45421940928270044,
      "grad_norm": 0.9811868071556091,
      "learning_rate": 0.0015,
      "loss": 1.6968,
      "step": 4306
    },
    {
      "epoch": 0.45432489451476793,
      "grad_norm": 0.7886048555374146,
      "learning_rate": 0.0015,
      "loss": 1.6897,
      "step": 4307
    },
    {
      "epoch": 0.4544303797468354,
      "grad_norm": 0.6720010042190552,
      "learning_rate": 0.0015,
      "loss": 1.6582,
      "step": 4308
    },
    {
      "epoch": 0.45453586497890297,
      "grad_norm": 0.9948484897613525,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 4309
    },
    {
      "epoch": 0.45464135021097046,
      "grad_norm": 0.5053322315216064,
      "learning_rate": 0.0015,
      "loss": 1.6139,
      "step": 4310
    },
    {
      "epoch": 0.45474683544303796,
      "grad_norm": 0.6952977776527405,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 4311
    },
    {
      "epoch": 0.4548523206751055,
      "grad_norm": 0.48643603920936584,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 4312
    },
    {
      "epoch": 0.454957805907173,
      "grad_norm": 0.7242816686630249,
      "learning_rate": 0.0015,
      "loss": 1.6639,
      "step": 4313
    },
    {
      "epoch": 0.4550632911392405,
      "grad_norm": 0.5559763312339783,
      "learning_rate": 0.0015,
      "loss": 1.6218,
      "step": 4314
    },
    {
      "epoch": 0.45516877637130804,
      "grad_norm": 0.5705804824829102,
      "learning_rate": 0.0015,
      "loss": 1.6421,
      "step": 4315
    },
    {
      "epoch": 0.45527426160337553,
      "grad_norm": 0.6401084065437317,
      "learning_rate": 0.0015,
      "loss": 1.6325,
      "step": 4316
    },
    {
      "epoch": 0.455379746835443,
      "grad_norm": 0.5702002644538879,
      "learning_rate": 0.0015,
      "loss": 1.6649,
      "step": 4317
    },
    {
      "epoch": 0.45548523206751057,
      "grad_norm": 0.8799147009849548,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 4318
    },
    {
      "epoch": 0.45559071729957806,
      "grad_norm": 0.6108373403549194,
      "learning_rate": 0.0015,
      "loss": 1.6356,
      "step": 4319
    },
    {
      "epoch": 0.45569620253164556,
      "grad_norm": 0.6824080348014832,
      "learning_rate": 0.0015,
      "loss": 1.6971,
      "step": 4320
    },
    {
      "epoch": 0.4558016877637131,
      "grad_norm": 0.6777485609054565,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 4321
    },
    {
      "epoch": 0.4559071729957806,
      "grad_norm": 0.5570958852767944,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 4322
    },
    {
      "epoch": 0.4560126582278481,
      "grad_norm": 0.6561368703842163,
      "learning_rate": 0.0015,
      "loss": 1.7168,
      "step": 4323
    },
    {
      "epoch": 0.45611814345991564,
      "grad_norm": 0.6748980283737183,
      "learning_rate": 0.0015,
      "loss": 1.6374,
      "step": 4324
    },
    {
      "epoch": 0.45622362869198313,
      "grad_norm": 0.5916723012924194,
      "learning_rate": 0.0015,
      "loss": 1.6766,
      "step": 4325
    },
    {
      "epoch": 0.4563291139240506,
      "grad_norm": 0.4420872926712036,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 4326
    },
    {
      "epoch": 0.45643459915611817,
      "grad_norm": 0.6043102145195007,
      "learning_rate": 0.0015,
      "loss": 1.6976,
      "step": 4327
    },
    {
      "epoch": 0.45654008438818566,
      "grad_norm": 0.4901691675186157,
      "learning_rate": 0.0015,
      "loss": 1.6634,
      "step": 4328
    },
    {
      "epoch": 0.45664556962025316,
      "grad_norm": 0.47698915004730225,
      "learning_rate": 0.0015,
      "loss": 1.6684,
      "step": 4329
    },
    {
      "epoch": 0.45675105485232065,
      "grad_norm": 0.4869694411754608,
      "learning_rate": 0.0015,
      "loss": 1.6374,
      "step": 4330
    },
    {
      "epoch": 0.4568565400843882,
      "grad_norm": 0.46680837869644165,
      "learning_rate": 0.0015,
      "loss": 1.6454,
      "step": 4331
    },
    {
      "epoch": 0.4569620253164557,
      "grad_norm": 0.49599412083625793,
      "learning_rate": 0.0015,
      "loss": 1.6664,
      "step": 4332
    },
    {
      "epoch": 0.4570675105485232,
      "grad_norm": 0.572851300239563,
      "learning_rate": 0.0015,
      "loss": 1.6163,
      "step": 4333
    },
    {
      "epoch": 0.45717299578059073,
      "grad_norm": 0.49066606163978577,
      "learning_rate": 0.0015,
      "loss": 1.6635,
      "step": 4334
    },
    {
      "epoch": 0.4572784810126582,
      "grad_norm": 0.5614000558853149,
      "learning_rate": 0.0015,
      "loss": 1.6465,
      "step": 4335
    },
    {
      "epoch": 0.4573839662447257,
      "grad_norm": 0.6668849587440491,
      "learning_rate": 0.0015,
      "loss": 1.6914,
      "step": 4336
    },
    {
      "epoch": 0.45748945147679326,
      "grad_norm": 0.4986950755119324,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 4337
    },
    {
      "epoch": 0.45759493670886076,
      "grad_norm": 0.5193601250648499,
      "learning_rate": 0.0015,
      "loss": 1.6406,
      "step": 4338
    },
    {
      "epoch": 0.45770042194092825,
      "grad_norm": 0.6118957996368408,
      "learning_rate": 0.0015,
      "loss": 1.6771,
      "step": 4339
    },
    {
      "epoch": 0.4578059071729958,
      "grad_norm": 0.4993157684803009,
      "learning_rate": 0.0015,
      "loss": 1.6348,
      "step": 4340
    },
    {
      "epoch": 0.4579113924050633,
      "grad_norm": 0.47754546999931335,
      "learning_rate": 0.0015,
      "loss": 1.6352,
      "step": 4341
    },
    {
      "epoch": 0.4580168776371308,
      "grad_norm": 0.5599583387374878,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 4342
    },
    {
      "epoch": 0.45812236286919833,
      "grad_norm": 0.5248196721076965,
      "learning_rate": 0.0015,
      "loss": 1.6151,
      "step": 4343
    },
    {
      "epoch": 0.4582278481012658,
      "grad_norm": 0.5224418044090271,
      "learning_rate": 0.0015,
      "loss": 1.6795,
      "step": 4344
    },
    {
      "epoch": 0.4583333333333333,
      "grad_norm": 0.5129813551902771,
      "learning_rate": 0.0015,
      "loss": 1.6573,
      "step": 4345
    },
    {
      "epoch": 0.45843881856540086,
      "grad_norm": 0.5238183736801147,
      "learning_rate": 0.0015,
      "loss": 1.6941,
      "step": 4346
    },
    {
      "epoch": 0.45854430379746836,
      "grad_norm": 0.49654272198677063,
      "learning_rate": 0.0015,
      "loss": 1.6387,
      "step": 4347
    },
    {
      "epoch": 0.45864978902953585,
      "grad_norm": 0.5397877097129822,
      "learning_rate": 0.0015,
      "loss": 1.6485,
      "step": 4348
    },
    {
      "epoch": 0.4587552742616034,
      "grad_norm": 0.47956404089927673,
      "learning_rate": 0.0015,
      "loss": 1.6822,
      "step": 4349
    },
    {
      "epoch": 0.4588607594936709,
      "grad_norm": 0.5442126989364624,
      "learning_rate": 0.0015,
      "loss": 1.6715,
      "step": 4350
    },
    {
      "epoch": 0.4589662447257384,
      "grad_norm": 0.6239200830459595,
      "learning_rate": 0.0015,
      "loss": 1.6973,
      "step": 4351
    },
    {
      "epoch": 0.45907172995780593,
      "grad_norm": 0.528907060623169,
      "learning_rate": 0.0015,
      "loss": 1.6857,
      "step": 4352
    },
    {
      "epoch": 0.4591772151898734,
      "grad_norm": 0.5882149934768677,
      "learning_rate": 0.0015,
      "loss": 1.6559,
      "step": 4353
    },
    {
      "epoch": 0.4592827004219409,
      "grad_norm": 0.677436888217926,
      "learning_rate": 0.0015,
      "loss": 1.6864,
      "step": 4354
    },
    {
      "epoch": 0.45938818565400846,
      "grad_norm": 0.5961720943450928,
      "learning_rate": 0.0015,
      "loss": 1.6884,
      "step": 4355
    },
    {
      "epoch": 0.45949367088607596,
      "grad_norm": 0.46816152334213257,
      "learning_rate": 0.0015,
      "loss": 1.6373,
      "step": 4356
    },
    {
      "epoch": 0.45959915611814345,
      "grad_norm": 0.5927190780639648,
      "learning_rate": 0.0015,
      "loss": 1.6636,
      "step": 4357
    },
    {
      "epoch": 0.459704641350211,
      "grad_norm": 0.5788471102714539,
      "learning_rate": 0.0015,
      "loss": 1.6787,
      "step": 4358
    },
    {
      "epoch": 0.4598101265822785,
      "grad_norm": 0.628187894821167,
      "learning_rate": 0.0015,
      "loss": 1.662,
      "step": 4359
    },
    {
      "epoch": 0.459915611814346,
      "grad_norm": 0.8854433298110962,
      "learning_rate": 0.0015,
      "loss": 1.6552,
      "step": 4360
    },
    {
      "epoch": 0.46002109704641353,
      "grad_norm": 0.48503169417381287,
      "learning_rate": 0.0015,
      "loss": 1.6539,
      "step": 4361
    },
    {
      "epoch": 0.460126582278481,
      "grad_norm": 0.7586188912391663,
      "learning_rate": 0.0015,
      "loss": 1.6212,
      "step": 4362
    },
    {
      "epoch": 0.4602320675105485,
      "grad_norm": 0.5986793041229248,
      "learning_rate": 0.0015,
      "loss": 1.6408,
      "step": 4363
    },
    {
      "epoch": 0.460337552742616,
      "grad_norm": 0.5295442342758179,
      "learning_rate": 0.0015,
      "loss": 1.6626,
      "step": 4364
    },
    {
      "epoch": 0.46044303797468356,
      "grad_norm": 0.5004334449768066,
      "learning_rate": 0.0015,
      "loss": 1.6642,
      "step": 4365
    },
    {
      "epoch": 0.46054852320675105,
      "grad_norm": 0.520413875579834,
      "learning_rate": 0.0015,
      "loss": 1.6191,
      "step": 4366
    },
    {
      "epoch": 0.46065400843881854,
      "grad_norm": 0.552975594997406,
      "learning_rate": 0.0015,
      "loss": 1.6584,
      "step": 4367
    },
    {
      "epoch": 0.4607594936708861,
      "grad_norm": 0.5053632259368896,
      "learning_rate": 0.0015,
      "loss": 1.6398,
      "step": 4368
    },
    {
      "epoch": 0.4608649789029536,
      "grad_norm": 0.7521233558654785,
      "learning_rate": 0.0015,
      "loss": 1.6503,
      "step": 4369
    },
    {
      "epoch": 0.4609704641350211,
      "grad_norm": 0.6210848689079285,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 4370
    },
    {
      "epoch": 0.4610759493670886,
      "grad_norm": 0.5326166749000549,
      "learning_rate": 0.0015,
      "loss": 1.6024,
      "step": 4371
    },
    {
      "epoch": 0.4611814345991561,
      "grad_norm": 0.5367142558097839,
      "learning_rate": 0.0015,
      "loss": 1.6619,
      "step": 4372
    },
    {
      "epoch": 0.4612869198312236,
      "grad_norm": 0.4858086109161377,
      "learning_rate": 0.0015,
      "loss": 1.6264,
      "step": 4373
    },
    {
      "epoch": 0.46139240506329116,
      "grad_norm": 0.541311502456665,
      "learning_rate": 0.0015,
      "loss": 1.6591,
      "step": 4374
    },
    {
      "epoch": 0.46149789029535865,
      "grad_norm": 0.5433955192565918,
      "learning_rate": 0.0015,
      "loss": 1.6647,
      "step": 4375
    },
    {
      "epoch": 0.46160337552742614,
      "grad_norm": 0.46293917298316956,
      "learning_rate": 0.0015,
      "loss": 1.6638,
      "step": 4376
    },
    {
      "epoch": 0.4617088607594937,
      "grad_norm": 0.4877987802028656,
      "learning_rate": 0.0015,
      "loss": 1.6225,
      "step": 4377
    },
    {
      "epoch": 0.4618143459915612,
      "grad_norm": 0.443900465965271,
      "learning_rate": 0.0015,
      "loss": 1.6633,
      "step": 4378
    },
    {
      "epoch": 0.4619198312236287,
      "grad_norm": 0.48014596104621887,
      "learning_rate": 0.0015,
      "loss": 1.6762,
      "step": 4379
    },
    {
      "epoch": 0.4620253164556962,
      "grad_norm": 0.4642293453216553,
      "learning_rate": 0.0015,
      "loss": 1.6567,
      "step": 4380
    },
    {
      "epoch": 0.4621308016877637,
      "grad_norm": 0.453438937664032,
      "learning_rate": 0.0015,
      "loss": 1.6814,
      "step": 4381
    },
    {
      "epoch": 0.4622362869198312,
      "grad_norm": 0.516966700553894,
      "learning_rate": 0.0015,
      "loss": 1.6418,
      "step": 4382
    },
    {
      "epoch": 0.46234177215189876,
      "grad_norm": 0.4654674232006073,
      "learning_rate": 0.0015,
      "loss": 1.6411,
      "step": 4383
    },
    {
      "epoch": 0.46244725738396625,
      "grad_norm": 0.5884807705879211,
      "learning_rate": 0.0015,
      "loss": 1.6194,
      "step": 4384
    },
    {
      "epoch": 0.46255274261603374,
      "grad_norm": 0.5772904753684998,
      "learning_rate": 0.0015,
      "loss": 1.6616,
      "step": 4385
    },
    {
      "epoch": 0.4626582278481013,
      "grad_norm": 0.4588421583175659,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 4386
    },
    {
      "epoch": 0.4627637130801688,
      "grad_norm": 0.5177502632141113,
      "learning_rate": 0.0015,
      "loss": 1.6254,
      "step": 4387
    },
    {
      "epoch": 0.4628691983122363,
      "grad_norm": 0.5180033445358276,
      "learning_rate": 0.0015,
      "loss": 1.6537,
      "step": 4388
    },
    {
      "epoch": 0.4629746835443038,
      "grad_norm": 0.5307671427726746,
      "learning_rate": 0.0015,
      "loss": 1.6451,
      "step": 4389
    },
    {
      "epoch": 0.4630801687763713,
      "grad_norm": 0.5913004875183105,
      "learning_rate": 0.0015,
      "loss": 1.6414,
      "step": 4390
    },
    {
      "epoch": 0.4631856540084388,
      "grad_norm": 0.504273533821106,
      "learning_rate": 0.0015,
      "loss": 1.6423,
      "step": 4391
    },
    {
      "epoch": 0.46329113924050636,
      "grad_norm": 0.5816996097564697,
      "learning_rate": 0.0015,
      "loss": 1.6286,
      "step": 4392
    },
    {
      "epoch": 0.46339662447257385,
      "grad_norm": 0.8179287314414978,
      "learning_rate": 0.0015,
      "loss": 1.6497,
      "step": 4393
    },
    {
      "epoch": 0.46350210970464134,
      "grad_norm": 0.7486006021499634,
      "learning_rate": 0.0015,
      "loss": 1.665,
      "step": 4394
    },
    {
      "epoch": 0.46360759493670883,
      "grad_norm": 0.4848342537879944,
      "learning_rate": 0.0015,
      "loss": 1.6656,
      "step": 4395
    },
    {
      "epoch": 0.4637130801687764,
      "grad_norm": 0.951631486415863,
      "learning_rate": 0.0015,
      "loss": 1.6568,
      "step": 4396
    },
    {
      "epoch": 0.4638185654008439,
      "grad_norm": 0.7860904335975647,
      "learning_rate": 0.0015,
      "loss": 1.6536,
      "step": 4397
    },
    {
      "epoch": 0.46392405063291137,
      "grad_norm": 0.5099451541900635,
      "learning_rate": 0.0015,
      "loss": 1.6841,
      "step": 4398
    },
    {
      "epoch": 0.4640295358649789,
      "grad_norm": 0.9489009976387024,
      "learning_rate": 0.0015,
      "loss": 1.6274,
      "step": 4399
    },
    {
      "epoch": 0.4641350210970464,
      "grad_norm": 0.7093892693519592,
      "learning_rate": 0.0015,
      "loss": 1.6884,
      "step": 4400
    },
    {
      "epoch": 0.4642405063291139,
      "grad_norm": 0.6286365985870361,
      "learning_rate": 0.0015,
      "loss": 1.6354,
      "step": 4401
    },
    {
      "epoch": 0.46434599156118145,
      "grad_norm": 0.8489190936088562,
      "learning_rate": 0.0015,
      "loss": 1.6849,
      "step": 4402
    },
    {
      "epoch": 0.46445147679324894,
      "grad_norm": 0.5102296471595764,
      "learning_rate": 0.0015,
      "loss": 1.7035,
      "step": 4403
    },
    {
      "epoch": 0.46455696202531643,
      "grad_norm": 0.8167166709899902,
      "learning_rate": 0.0015,
      "loss": 1.6814,
      "step": 4404
    },
    {
      "epoch": 0.464662447257384,
      "grad_norm": 0.702418863773346,
      "learning_rate": 0.0015,
      "loss": 1.6658,
      "step": 4405
    },
    {
      "epoch": 0.4647679324894515,
      "grad_norm": 0.49861350655555725,
      "learning_rate": 0.0015,
      "loss": 1.6625,
      "step": 4406
    },
    {
      "epoch": 0.46487341772151897,
      "grad_norm": 0.5561988353729248,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 4407
    },
    {
      "epoch": 0.4649789029535865,
      "grad_norm": 0.5472853183746338,
      "learning_rate": 0.0015,
      "loss": 1.6755,
      "step": 4408
    },
    {
      "epoch": 0.465084388185654,
      "grad_norm": 0.5895664095878601,
      "learning_rate": 0.0015,
      "loss": 1.6528,
      "step": 4409
    },
    {
      "epoch": 0.4651898734177215,
      "grad_norm": 0.7193066477775574,
      "learning_rate": 0.0015,
      "loss": 1.6648,
      "step": 4410
    },
    {
      "epoch": 0.46529535864978905,
      "grad_norm": 0.5323827266693115,
      "learning_rate": 0.0015,
      "loss": 1.6458,
      "step": 4411
    },
    {
      "epoch": 0.46540084388185654,
      "grad_norm": 0.5354639887809753,
      "learning_rate": 0.0015,
      "loss": 1.6918,
      "step": 4412
    },
    {
      "epoch": 0.46550632911392403,
      "grad_norm": 0.5667396187782288,
      "learning_rate": 0.0015,
      "loss": 1.6149,
      "step": 4413
    },
    {
      "epoch": 0.4656118143459916,
      "grad_norm": 0.569074809551239,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 4414
    },
    {
      "epoch": 0.4657172995780591,
      "grad_norm": 0.5345011353492737,
      "learning_rate": 0.0015,
      "loss": 1.6651,
      "step": 4415
    },
    {
      "epoch": 0.46582278481012657,
      "grad_norm": 0.6622505784034729,
      "learning_rate": 0.0015,
      "loss": 1.6819,
      "step": 4416
    },
    {
      "epoch": 0.4659282700421941,
      "grad_norm": 0.5924422740936279,
      "learning_rate": 0.0015,
      "loss": 1.6242,
      "step": 4417
    },
    {
      "epoch": 0.4660337552742616,
      "grad_norm": 0.5936039686203003,
      "learning_rate": 0.0015,
      "loss": 1.6564,
      "step": 4418
    },
    {
      "epoch": 0.4661392405063291,
      "grad_norm": 0.5590488910675049,
      "learning_rate": 0.0015,
      "loss": 1.6627,
      "step": 4419
    },
    {
      "epoch": 0.46624472573839665,
      "grad_norm": 0.4982949495315552,
      "learning_rate": 0.0015,
      "loss": 1.6343,
      "step": 4420
    },
    {
      "epoch": 0.46635021097046414,
      "grad_norm": 0.511516809463501,
      "learning_rate": 0.0015,
      "loss": 1.6492,
      "step": 4421
    },
    {
      "epoch": 0.46645569620253163,
      "grad_norm": 0.6481553912162781,
      "learning_rate": 0.0015,
      "loss": 1.6536,
      "step": 4422
    },
    {
      "epoch": 0.4665611814345992,
      "grad_norm": 0.577252209186554,
      "learning_rate": 0.0015,
      "loss": 1.6732,
      "step": 4423
    },
    {
      "epoch": 0.4666666666666667,
      "grad_norm": 0.5631588697433472,
      "learning_rate": 0.0015,
      "loss": 1.6768,
      "step": 4424
    },
    {
      "epoch": 0.46677215189873417,
      "grad_norm": 0.5087147951126099,
      "learning_rate": 0.0015,
      "loss": 1.6449,
      "step": 4425
    },
    {
      "epoch": 0.4668776371308017,
      "grad_norm": 0.6231689453125,
      "learning_rate": 0.0015,
      "loss": 1.6368,
      "step": 4426
    },
    {
      "epoch": 0.4669831223628692,
      "grad_norm": 0.5108701586723328,
      "learning_rate": 0.0015,
      "loss": 1.6494,
      "step": 4427
    },
    {
      "epoch": 0.4670886075949367,
      "grad_norm": 0.5215011239051819,
      "learning_rate": 0.0015,
      "loss": 1.6576,
      "step": 4428
    },
    {
      "epoch": 0.4671940928270042,
      "grad_norm": 0.5362898111343384,
      "learning_rate": 0.0015,
      "loss": 1.6811,
      "step": 4429
    },
    {
      "epoch": 0.46729957805907174,
      "grad_norm": 0.5128081440925598,
      "learning_rate": 0.0015,
      "loss": 1.6234,
      "step": 4430
    },
    {
      "epoch": 0.46740506329113923,
      "grad_norm": 0.645807683467865,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 4431
    },
    {
      "epoch": 0.4675105485232067,
      "grad_norm": 0.6203972101211548,
      "learning_rate": 0.0015,
      "loss": 1.6856,
      "step": 4432
    },
    {
      "epoch": 0.4676160337552743,
      "grad_norm": 0.6224325895309448,
      "learning_rate": 0.0015,
      "loss": 1.6468,
      "step": 4433
    },
    {
      "epoch": 0.46772151898734177,
      "grad_norm": 0.9138966798782349,
      "learning_rate": 0.0015,
      "loss": 1.6826,
      "step": 4434
    },
    {
      "epoch": 0.46782700421940926,
      "grad_norm": 0.7643339037895203,
      "learning_rate": 0.0015,
      "loss": 1.626,
      "step": 4435
    },
    {
      "epoch": 0.4679324894514768,
      "grad_norm": 0.4755401611328125,
      "learning_rate": 0.0015,
      "loss": 1.6263,
      "step": 4436
    },
    {
      "epoch": 0.4680379746835443,
      "grad_norm": 0.6597535014152527,
      "learning_rate": 0.0015,
      "loss": 1.6512,
      "step": 4437
    },
    {
      "epoch": 0.4681434599156118,
      "grad_norm": 0.6575095653533936,
      "learning_rate": 0.0015,
      "loss": 1.6671,
      "step": 4438
    },
    {
      "epoch": 0.46824894514767934,
      "grad_norm": 0.5518959760665894,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 4439
    },
    {
      "epoch": 0.46835443037974683,
      "grad_norm": 0.6916687488555908,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 4440
    },
    {
      "epoch": 0.4684599156118143,
      "grad_norm": 0.5765088796615601,
      "learning_rate": 0.0015,
      "loss": 1.6447,
      "step": 4441
    },
    {
      "epoch": 0.4685654008438819,
      "grad_norm": 0.7362664341926575,
      "learning_rate": 0.0015,
      "loss": 1.6807,
      "step": 4442
    },
    {
      "epoch": 0.46867088607594937,
      "grad_norm": 0.8707485795021057,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 4443
    },
    {
      "epoch": 0.46877637130801686,
      "grad_norm": 0.5041620135307312,
      "learning_rate": 0.0015,
      "loss": 1.6382,
      "step": 4444
    },
    {
      "epoch": 0.4688818565400844,
      "grad_norm": 0.8869928121566772,
      "learning_rate": 0.0015,
      "loss": 1.6819,
      "step": 4445
    },
    {
      "epoch": 0.4689873417721519,
      "grad_norm": 0.9099951386451721,
      "learning_rate": 0.0015,
      "loss": 1.6215,
      "step": 4446
    },
    {
      "epoch": 0.4690928270042194,
      "grad_norm": 0.5322218537330627,
      "learning_rate": 0.0015,
      "loss": 1.6714,
      "step": 4447
    },
    {
      "epoch": 0.46919831223628694,
      "grad_norm": 0.8943033814430237,
      "learning_rate": 0.0015,
      "loss": 1.693,
      "step": 4448
    },
    {
      "epoch": 0.46930379746835443,
      "grad_norm": 0.6124472618103027,
      "learning_rate": 0.0015,
      "loss": 1.6572,
      "step": 4449
    },
    {
      "epoch": 0.4694092827004219,
      "grad_norm": 0.673154890537262,
      "learning_rate": 0.0015,
      "loss": 1.6399,
      "step": 4450
    },
    {
      "epoch": 0.4695147679324895,
      "grad_norm": 0.715014636516571,
      "learning_rate": 0.0015,
      "loss": 1.6534,
      "step": 4451
    },
    {
      "epoch": 0.46962025316455697,
      "grad_norm": 0.5323740243911743,
      "learning_rate": 0.0015,
      "loss": 1.6249,
      "step": 4452
    },
    {
      "epoch": 0.46972573839662446,
      "grad_norm": 0.6634638905525208,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 4453
    },
    {
      "epoch": 0.469831223628692,
      "grad_norm": 0.8106464147567749,
      "learning_rate": 0.0015,
      "loss": 1.6049,
      "step": 4454
    },
    {
      "epoch": 0.4699367088607595,
      "grad_norm": 0.5180960893630981,
      "learning_rate": 0.0015,
      "loss": 1.6292,
      "step": 4455
    },
    {
      "epoch": 0.470042194092827,
      "grad_norm": 0.7639739513397217,
      "learning_rate": 0.0015,
      "loss": 1.6442,
      "step": 4456
    },
    {
      "epoch": 0.47014767932489454,
      "grad_norm": 0.6277756094932556,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 4457
    },
    {
      "epoch": 0.47025316455696203,
      "grad_norm": 0.612694501876831,
      "learning_rate": 0.0015,
      "loss": 1.6344,
      "step": 4458
    },
    {
      "epoch": 0.4703586497890295,
      "grad_norm": 0.6587413549423218,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 4459
    },
    {
      "epoch": 0.4704641350210971,
      "grad_norm": 0.550410807132721,
      "learning_rate": 0.0015,
      "loss": 1.6931,
      "step": 4460
    },
    {
      "epoch": 0.47056962025316457,
      "grad_norm": 0.6902750134468079,
      "learning_rate": 0.0015,
      "loss": 1.6545,
      "step": 4461
    },
    {
      "epoch": 0.47067510548523206,
      "grad_norm": 0.6775808930397034,
      "learning_rate": 0.0015,
      "loss": 1.6586,
      "step": 4462
    },
    {
      "epoch": 0.47078059071729955,
      "grad_norm": 0.5754194259643555,
      "learning_rate": 0.0015,
      "loss": 1.6369,
      "step": 4463
    },
    {
      "epoch": 0.4708860759493671,
      "grad_norm": 0.8526375889778137,
      "learning_rate": 0.0015,
      "loss": 1.654,
      "step": 4464
    },
    {
      "epoch": 0.4709915611814346,
      "grad_norm": 0.667432963848114,
      "learning_rate": 0.0015,
      "loss": 1.6348,
      "step": 4465
    },
    {
      "epoch": 0.4710970464135021,
      "grad_norm": 0.48168909549713135,
      "learning_rate": 0.0015,
      "loss": 1.6636,
      "step": 4466
    },
    {
      "epoch": 0.47120253164556963,
      "grad_norm": 0.5759777426719666,
      "learning_rate": 0.0015,
      "loss": 1.637,
      "step": 4467
    },
    {
      "epoch": 0.4713080168776371,
      "grad_norm": 0.5050199627876282,
      "learning_rate": 0.0015,
      "loss": 1.6402,
      "step": 4468
    },
    {
      "epoch": 0.4714135021097046,
      "grad_norm": 0.5121300220489502,
      "learning_rate": 0.0015,
      "loss": 1.6902,
      "step": 4469
    },
    {
      "epoch": 0.47151898734177217,
      "grad_norm": 0.546608030796051,
      "learning_rate": 0.0015,
      "loss": 1.6852,
      "step": 4470
    },
    {
      "epoch": 0.47162447257383966,
      "grad_norm": 0.5345836281776428,
      "learning_rate": 0.0015,
      "loss": 1.6391,
      "step": 4471
    },
    {
      "epoch": 0.47172995780590715,
      "grad_norm": 0.5008557438850403,
      "learning_rate": 0.0015,
      "loss": 1.6374,
      "step": 4472
    },
    {
      "epoch": 0.4718354430379747,
      "grad_norm": 0.64007169008255,
      "learning_rate": 0.0015,
      "loss": 1.6716,
      "step": 4473
    },
    {
      "epoch": 0.4719409282700422,
      "grad_norm": 0.6776872277259827,
      "learning_rate": 0.0015,
      "loss": 1.7013,
      "step": 4474
    },
    {
      "epoch": 0.4720464135021097,
      "grad_norm": 0.602365255355835,
      "learning_rate": 0.0015,
      "loss": 1.6872,
      "step": 4475
    },
    {
      "epoch": 0.47215189873417723,
      "grad_norm": 0.5588603019714355,
      "learning_rate": 0.0015,
      "loss": 1.6333,
      "step": 4476
    },
    {
      "epoch": 0.4722573839662447,
      "grad_norm": 0.5542494654655457,
      "learning_rate": 0.0015,
      "loss": 1.668,
      "step": 4477
    },
    {
      "epoch": 0.4723628691983122,
      "grad_norm": 0.5087571144104004,
      "learning_rate": 0.0015,
      "loss": 1.6803,
      "step": 4478
    },
    {
      "epoch": 0.47246835443037977,
      "grad_norm": 0.6984165906906128,
      "learning_rate": 0.0015,
      "loss": 1.6555,
      "step": 4479
    },
    {
      "epoch": 0.47257383966244726,
      "grad_norm": 0.5771257281303406,
      "learning_rate": 0.0015,
      "loss": 1.6652,
      "step": 4480
    },
    {
      "epoch": 0.47267932489451475,
      "grad_norm": 0.5852707028388977,
      "learning_rate": 0.0015,
      "loss": 1.6296,
      "step": 4481
    },
    {
      "epoch": 0.4727848101265823,
      "grad_norm": 0.6187868118286133,
      "learning_rate": 0.0015,
      "loss": 1.6424,
      "step": 4482
    },
    {
      "epoch": 0.4728902953586498,
      "grad_norm": 0.5727245807647705,
      "learning_rate": 0.0015,
      "loss": 1.6306,
      "step": 4483
    },
    {
      "epoch": 0.4729957805907173,
      "grad_norm": 0.5415666103363037,
      "learning_rate": 0.0015,
      "loss": 1.6305,
      "step": 4484
    },
    {
      "epoch": 0.47310126582278483,
      "grad_norm": 0.5851130485534668,
      "learning_rate": 0.0015,
      "loss": 1.6555,
      "step": 4485
    },
    {
      "epoch": 0.4732067510548523,
      "grad_norm": 0.5078525543212891,
      "learning_rate": 0.0015,
      "loss": 1.6474,
      "step": 4486
    },
    {
      "epoch": 0.4733122362869198,
      "grad_norm": 0.4962426722049713,
      "learning_rate": 0.0015,
      "loss": 1.6397,
      "step": 4487
    },
    {
      "epoch": 0.47341772151898737,
      "grad_norm": 0.5256374478340149,
      "learning_rate": 0.0015,
      "loss": 1.6701,
      "step": 4488
    },
    {
      "epoch": 0.47352320675105486,
      "grad_norm": 0.6503275632858276,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 4489
    },
    {
      "epoch": 0.47362869198312235,
      "grad_norm": 0.70665043592453,
      "learning_rate": 0.0015,
      "loss": 1.6993,
      "step": 4490
    },
    {
      "epoch": 0.4737341772151899,
      "grad_norm": 0.4727391302585602,
      "learning_rate": 0.0015,
      "loss": 1.6407,
      "step": 4491
    },
    {
      "epoch": 0.4738396624472574,
      "grad_norm": 0.7271082401275635,
      "learning_rate": 0.0015,
      "loss": 1.6665,
      "step": 4492
    },
    {
      "epoch": 0.4739451476793249,
      "grad_norm": 0.6968852877616882,
      "learning_rate": 0.0015,
      "loss": 1.6323,
      "step": 4493
    },
    {
      "epoch": 0.4740506329113924,
      "grad_norm": 0.5151492357254028,
      "learning_rate": 0.0015,
      "loss": 1.6287,
      "step": 4494
    },
    {
      "epoch": 0.4741561181434599,
      "grad_norm": 0.8903449177742004,
      "learning_rate": 0.0015,
      "loss": 1.6769,
      "step": 4495
    },
    {
      "epoch": 0.4742616033755274,
      "grad_norm": 0.7018641233444214,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 4496
    },
    {
      "epoch": 0.4743670886075949,
      "grad_norm": 0.5498849153518677,
      "learning_rate": 0.0015,
      "loss": 1.6276,
      "step": 4497
    },
    {
      "epoch": 0.47447257383966246,
      "grad_norm": 0.6099660992622375,
      "learning_rate": 0.0015,
      "loss": 1.6588,
      "step": 4498
    },
    {
      "epoch": 0.47457805907172995,
      "grad_norm": 0.5692307353019714,
      "learning_rate": 0.0015,
      "loss": 1.6566,
      "step": 4499
    },
    {
      "epoch": 0.47468354430379744,
      "grad_norm": 0.49323442578315735,
      "learning_rate": 0.0015,
      "loss": 1.6528,
      "step": 4500
    },
    {
      "epoch": 0.474789029535865,
      "grad_norm": 0.5160009264945984,
      "learning_rate": 0.0015,
      "loss": 1.6727,
      "step": 4501
    },
    {
      "epoch": 0.4748945147679325,
      "grad_norm": 0.5155899524688721,
      "learning_rate": 0.0015,
      "loss": 1.6691,
      "step": 4502
    },
    {
      "epoch": 0.475,
      "grad_norm": 0.5003455281257629,
      "learning_rate": 0.0015,
      "loss": 1.6442,
      "step": 4503
    },
    {
      "epoch": 0.4751054852320675,
      "grad_norm": 0.47251251339912415,
      "learning_rate": 0.0015,
      "loss": 1.6633,
      "step": 4504
    },
    {
      "epoch": 0.475210970464135,
      "grad_norm": 0.4946487247943878,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 4505
    },
    {
      "epoch": 0.4753164556962025,
      "grad_norm": 0.4878123104572296,
      "learning_rate": 0.0015,
      "loss": 1.6432,
      "step": 4506
    },
    {
      "epoch": 0.47542194092827006,
      "grad_norm": 0.5271539688110352,
      "learning_rate": 0.0015,
      "loss": 1.6725,
      "step": 4507
    },
    {
      "epoch": 0.47552742616033755,
      "grad_norm": 0.4943915903568268,
      "learning_rate": 0.0015,
      "loss": 1.6381,
      "step": 4508
    },
    {
      "epoch": 0.47563291139240504,
      "grad_norm": 0.4758211672306061,
      "learning_rate": 0.0015,
      "loss": 1.6582,
      "step": 4509
    },
    {
      "epoch": 0.4757383966244726,
      "grad_norm": 0.46259981393814087,
      "learning_rate": 0.0015,
      "loss": 1.6105,
      "step": 4510
    },
    {
      "epoch": 0.4758438818565401,
      "grad_norm": 0.44713056087493896,
      "learning_rate": 0.0015,
      "loss": 1.6255,
      "step": 4511
    },
    {
      "epoch": 0.4759493670886076,
      "grad_norm": 0.5059557557106018,
      "learning_rate": 0.0015,
      "loss": 1.6363,
      "step": 4512
    },
    {
      "epoch": 0.4760548523206751,
      "grad_norm": 0.5178486704826355,
      "learning_rate": 0.0015,
      "loss": 1.6535,
      "step": 4513
    },
    {
      "epoch": 0.4761603375527426,
      "grad_norm": 0.4474097192287445,
      "learning_rate": 0.0015,
      "loss": 1.6575,
      "step": 4514
    },
    {
      "epoch": 0.4762658227848101,
      "grad_norm": 0.49275943636894226,
      "learning_rate": 0.0015,
      "loss": 1.6745,
      "step": 4515
    },
    {
      "epoch": 0.47637130801687766,
      "grad_norm": 0.5427425503730774,
      "learning_rate": 0.0015,
      "loss": 1.6252,
      "step": 4516
    },
    {
      "epoch": 0.47647679324894515,
      "grad_norm": 0.5037579536437988,
      "learning_rate": 0.0015,
      "loss": 1.6231,
      "step": 4517
    },
    {
      "epoch": 0.47658227848101264,
      "grad_norm": 0.46757787466049194,
      "learning_rate": 0.0015,
      "loss": 1.607,
      "step": 4518
    },
    {
      "epoch": 0.4766877637130802,
      "grad_norm": 0.518986701965332,
      "learning_rate": 0.0015,
      "loss": 1.6044,
      "step": 4519
    },
    {
      "epoch": 0.4767932489451477,
      "grad_norm": 0.483244389295578,
      "learning_rate": 0.0015,
      "loss": 1.6185,
      "step": 4520
    },
    {
      "epoch": 0.4768987341772152,
      "grad_norm": 0.4733765423297882,
      "learning_rate": 0.0015,
      "loss": 1.672,
      "step": 4521
    },
    {
      "epoch": 0.4770042194092827,
      "grad_norm": 0.469696968793869,
      "learning_rate": 0.0015,
      "loss": 1.6809,
      "step": 4522
    },
    {
      "epoch": 0.4771097046413502,
      "grad_norm": 0.5405987501144409,
      "learning_rate": 0.0015,
      "loss": 1.6566,
      "step": 4523
    },
    {
      "epoch": 0.4772151898734177,
      "grad_norm": 0.5330632925033569,
      "learning_rate": 0.0015,
      "loss": 1.6529,
      "step": 4524
    },
    {
      "epoch": 0.47732067510548526,
      "grad_norm": 0.47293856739997864,
      "learning_rate": 0.0015,
      "loss": 1.6202,
      "step": 4525
    },
    {
      "epoch": 0.47742616033755275,
      "grad_norm": 0.5860047340393066,
      "learning_rate": 0.0015,
      "loss": 1.672,
      "step": 4526
    },
    {
      "epoch": 0.47753164556962024,
      "grad_norm": 0.5031240582466125,
      "learning_rate": 0.0015,
      "loss": 1.6216,
      "step": 4527
    },
    {
      "epoch": 0.47763713080168774,
      "grad_norm": 0.6522221565246582,
      "learning_rate": 0.0015,
      "loss": 1.6586,
      "step": 4528
    },
    {
      "epoch": 0.4777426160337553,
      "grad_norm": 0.6988711953163147,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 4529
    },
    {
      "epoch": 0.4778481012658228,
      "grad_norm": 0.6758490800857544,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 4530
    },
    {
      "epoch": 0.47795358649789027,
      "grad_norm": 0.4575705826282501,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 4531
    },
    {
      "epoch": 0.4780590717299578,
      "grad_norm": 0.5926907062530518,
      "learning_rate": 0.0015,
      "loss": 1.6439,
      "step": 4532
    },
    {
      "epoch": 0.4781645569620253,
      "grad_norm": 0.5273928046226501,
      "learning_rate": 0.0015,
      "loss": 1.6814,
      "step": 4533
    },
    {
      "epoch": 0.4782700421940928,
      "grad_norm": 0.5159898996353149,
      "learning_rate": 0.0015,
      "loss": 1.6732,
      "step": 4534
    },
    {
      "epoch": 0.47837552742616035,
      "grad_norm": 0.6231305003166199,
      "learning_rate": 0.0015,
      "loss": 1.6264,
      "step": 4535
    },
    {
      "epoch": 0.47848101265822784,
      "grad_norm": 0.4684787690639496,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 4536
    },
    {
      "epoch": 0.47858649789029534,
      "grad_norm": 0.5544801950454712,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 4537
    },
    {
      "epoch": 0.4786919831223629,
      "grad_norm": 0.5347890257835388,
      "learning_rate": 0.0015,
      "loss": 1.6577,
      "step": 4538
    },
    {
      "epoch": 0.4787974683544304,
      "grad_norm": 0.551781415939331,
      "learning_rate": 0.0015,
      "loss": 1.6513,
      "step": 4539
    },
    {
      "epoch": 0.47890295358649787,
      "grad_norm": 0.5763802528381348,
      "learning_rate": 0.0015,
      "loss": 1.6893,
      "step": 4540
    },
    {
      "epoch": 0.4790084388185654,
      "grad_norm": 0.540147066116333,
      "learning_rate": 0.0015,
      "loss": 1.6221,
      "step": 4541
    },
    {
      "epoch": 0.4791139240506329,
      "grad_norm": 0.5349014401435852,
      "learning_rate": 0.0015,
      "loss": 1.623,
      "step": 4542
    },
    {
      "epoch": 0.4792194092827004,
      "grad_norm": 0.5280075669288635,
      "learning_rate": 0.0015,
      "loss": 1.6606,
      "step": 4543
    },
    {
      "epoch": 0.47932489451476795,
      "grad_norm": 0.5085506439208984,
      "learning_rate": 0.0015,
      "loss": 1.662,
      "step": 4544
    },
    {
      "epoch": 0.47943037974683544,
      "grad_norm": 0.5173304080963135,
      "learning_rate": 0.0015,
      "loss": 1.6438,
      "step": 4545
    },
    {
      "epoch": 0.47953586497890294,
      "grad_norm": 0.6235553026199341,
      "learning_rate": 0.0015,
      "loss": 1.6657,
      "step": 4546
    },
    {
      "epoch": 0.4796413502109705,
      "grad_norm": 0.5665614008903503,
      "learning_rate": 0.0015,
      "loss": 1.655,
      "step": 4547
    },
    {
      "epoch": 0.479746835443038,
      "grad_norm": 0.44198301434516907,
      "learning_rate": 0.0015,
      "loss": 1.6794,
      "step": 4548
    },
    {
      "epoch": 0.47985232067510547,
      "grad_norm": 0.5924791693687439,
      "learning_rate": 0.0015,
      "loss": 1.6639,
      "step": 4549
    },
    {
      "epoch": 0.479957805907173,
      "grad_norm": 0.4326235353946686,
      "learning_rate": 0.0015,
      "loss": 1.6162,
      "step": 4550
    },
    {
      "epoch": 0.4800632911392405,
      "grad_norm": 0.6195024251937866,
      "learning_rate": 0.0015,
      "loss": 1.6312,
      "step": 4551
    },
    {
      "epoch": 0.480168776371308,
      "grad_norm": 0.631605863571167,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 4552
    },
    {
      "epoch": 0.48027426160337555,
      "grad_norm": 0.4493882954120636,
      "learning_rate": 0.0015,
      "loss": 1.6217,
      "step": 4553
    },
    {
      "epoch": 0.48037974683544304,
      "grad_norm": 0.807175874710083,
      "learning_rate": 0.0015,
      "loss": 1.6678,
      "step": 4554
    },
    {
      "epoch": 0.48048523206751054,
      "grad_norm": 0.8161116242408752,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 4555
    },
    {
      "epoch": 0.4805907172995781,
      "grad_norm": 0.463988333940506,
      "learning_rate": 0.0015,
      "loss": 1.6616,
      "step": 4556
    },
    {
      "epoch": 0.4806962025316456,
      "grad_norm": 0.6695824265480042,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 4557
    },
    {
      "epoch": 0.48080168776371307,
      "grad_norm": 0.5819977521896362,
      "learning_rate": 0.0015,
      "loss": 1.6552,
      "step": 4558
    },
    {
      "epoch": 0.48090717299578056,
      "grad_norm": 0.5306731462478638,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 4559
    },
    {
      "epoch": 0.4810126582278481,
      "grad_norm": 0.6184477806091309,
      "learning_rate": 0.0015,
      "loss": 1.6615,
      "step": 4560
    },
    {
      "epoch": 0.4811181434599156,
      "grad_norm": 0.507031261920929,
      "learning_rate": 0.0015,
      "loss": 1.6048,
      "step": 4561
    },
    {
      "epoch": 0.4812236286919831,
      "grad_norm": 0.6315982937812805,
      "learning_rate": 0.0015,
      "loss": 1.6079,
      "step": 4562
    },
    {
      "epoch": 0.48132911392405064,
      "grad_norm": 0.6824647784233093,
      "learning_rate": 0.0015,
      "loss": 1.6289,
      "step": 4563
    },
    {
      "epoch": 0.48143459915611814,
      "grad_norm": 0.49028703570365906,
      "learning_rate": 0.0015,
      "loss": 1.6405,
      "step": 4564
    },
    {
      "epoch": 0.48154008438818563,
      "grad_norm": 0.5238708257675171,
      "learning_rate": 0.0015,
      "loss": 1.5992,
      "step": 4565
    },
    {
      "epoch": 0.4816455696202532,
      "grad_norm": 0.5775980353355408,
      "learning_rate": 0.0015,
      "loss": 1.6615,
      "step": 4566
    },
    {
      "epoch": 0.48175105485232067,
      "grad_norm": 0.45707377791404724,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 4567
    },
    {
      "epoch": 0.48185654008438816,
      "grad_norm": 0.5543075203895569,
      "learning_rate": 0.0015,
      "loss": 1.6355,
      "step": 4568
    },
    {
      "epoch": 0.4819620253164557,
      "grad_norm": 0.6481205821037292,
      "learning_rate": 0.0015,
      "loss": 1.6028,
      "step": 4569
    },
    {
      "epoch": 0.4820675105485232,
      "grad_norm": 0.5371230840682983,
      "learning_rate": 0.0015,
      "loss": 1.6352,
      "step": 4570
    },
    {
      "epoch": 0.4821729957805907,
      "grad_norm": 0.5408508777618408,
      "learning_rate": 0.0015,
      "loss": 1.6286,
      "step": 4571
    },
    {
      "epoch": 0.48227848101265824,
      "grad_norm": 0.46124279499053955,
      "learning_rate": 0.0015,
      "loss": 1.606,
      "step": 4572
    },
    {
      "epoch": 0.48238396624472574,
      "grad_norm": 0.4997161328792572,
      "learning_rate": 0.0015,
      "loss": 1.6171,
      "step": 4573
    },
    {
      "epoch": 0.48248945147679323,
      "grad_norm": 0.5767737030982971,
      "learning_rate": 0.0015,
      "loss": 1.6953,
      "step": 4574
    },
    {
      "epoch": 0.4825949367088608,
      "grad_norm": 0.4893116354942322,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 4575
    },
    {
      "epoch": 0.48270042194092827,
      "grad_norm": 0.47082701325416565,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 4576
    },
    {
      "epoch": 0.48280590717299576,
      "grad_norm": 0.5216884613037109,
      "learning_rate": 0.0015,
      "loss": 1.6195,
      "step": 4577
    },
    {
      "epoch": 0.4829113924050633,
      "grad_norm": 0.5208689570426941,
      "learning_rate": 0.0015,
      "loss": 1.6425,
      "step": 4578
    },
    {
      "epoch": 0.4830168776371308,
      "grad_norm": 0.4750952422618866,
      "learning_rate": 0.0015,
      "loss": 1.6755,
      "step": 4579
    },
    {
      "epoch": 0.4831223628691983,
      "grad_norm": 0.5047564506530762,
      "learning_rate": 0.0015,
      "loss": 1.6673,
      "step": 4580
    },
    {
      "epoch": 0.48322784810126584,
      "grad_norm": 0.5329234004020691,
      "learning_rate": 0.0015,
      "loss": 1.598,
      "step": 4581
    },
    {
      "epoch": 0.48333333333333334,
      "grad_norm": 0.489246129989624,
      "learning_rate": 0.0015,
      "loss": 1.6191,
      "step": 4582
    },
    {
      "epoch": 0.48343881856540083,
      "grad_norm": 0.5560640692710876,
      "learning_rate": 0.0015,
      "loss": 1.6317,
      "step": 4583
    },
    {
      "epoch": 0.4835443037974684,
      "grad_norm": 0.4825168251991272,
      "learning_rate": 0.0015,
      "loss": 1.6698,
      "step": 4584
    },
    {
      "epoch": 0.48364978902953587,
      "grad_norm": 0.523622453212738,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 4585
    },
    {
      "epoch": 0.48375527426160336,
      "grad_norm": 0.5375120639801025,
      "learning_rate": 0.0015,
      "loss": 1.6471,
      "step": 4586
    },
    {
      "epoch": 0.4838607594936709,
      "grad_norm": 0.4922407269477844,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 4587
    },
    {
      "epoch": 0.4839662447257384,
      "grad_norm": 0.5101624131202698,
      "learning_rate": 0.0015,
      "loss": 1.5948,
      "step": 4588
    },
    {
      "epoch": 0.4840717299578059,
      "grad_norm": 0.5333686470985413,
      "learning_rate": 0.0015,
      "loss": 1.639,
      "step": 4589
    },
    {
      "epoch": 0.48417721518987344,
      "grad_norm": 0.5021880269050598,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 4590
    },
    {
      "epoch": 0.48428270042194094,
      "grad_norm": 0.4982892870903015,
      "learning_rate": 0.0015,
      "loss": 1.6505,
      "step": 4591
    },
    {
      "epoch": 0.48438818565400843,
      "grad_norm": 0.5568966865539551,
      "learning_rate": 0.0015,
      "loss": 1.6312,
      "step": 4592
    },
    {
      "epoch": 0.4844936708860759,
      "grad_norm": 0.5701368451118469,
      "learning_rate": 0.0015,
      "loss": 1.6386,
      "step": 4593
    },
    {
      "epoch": 0.48459915611814347,
      "grad_norm": 0.5515512228012085,
      "learning_rate": 0.0015,
      "loss": 1.6112,
      "step": 4594
    },
    {
      "epoch": 0.48470464135021096,
      "grad_norm": 0.670576810836792,
      "learning_rate": 0.0015,
      "loss": 1.6643,
      "step": 4595
    },
    {
      "epoch": 0.48481012658227846,
      "grad_norm": 0.6857293844223022,
      "learning_rate": 0.0015,
      "loss": 1.6653,
      "step": 4596
    },
    {
      "epoch": 0.484915611814346,
      "grad_norm": 0.6367458701133728,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 4597
    },
    {
      "epoch": 0.4850210970464135,
      "grad_norm": 0.5497254729270935,
      "learning_rate": 0.0015,
      "loss": 1.6578,
      "step": 4598
    },
    {
      "epoch": 0.485126582278481,
      "grad_norm": 0.6092268228530884,
      "learning_rate": 0.0015,
      "loss": 1.6968,
      "step": 4599
    },
    {
      "epoch": 0.48523206751054854,
      "grad_norm": 0.6367563009262085,
      "learning_rate": 0.0015,
      "loss": 1.6674,
      "step": 4600
    },
    {
      "epoch": 0.48533755274261603,
      "grad_norm": 0.5361546277999878,
      "learning_rate": 0.0015,
      "loss": 1.6887,
      "step": 4601
    },
    {
      "epoch": 0.4854430379746835,
      "grad_norm": 0.7194101214408875,
      "learning_rate": 0.0015,
      "loss": 1.6575,
      "step": 4602
    },
    {
      "epoch": 0.48554852320675107,
      "grad_norm": 0.5410552620887756,
      "learning_rate": 0.0015,
      "loss": 1.6047,
      "step": 4603
    },
    {
      "epoch": 0.48565400843881856,
      "grad_norm": 0.689990222454071,
      "learning_rate": 0.0015,
      "loss": 1.631,
      "step": 4604
    },
    {
      "epoch": 0.48575949367088606,
      "grad_norm": 0.7121012806892395,
      "learning_rate": 0.0015,
      "loss": 1.6747,
      "step": 4605
    },
    {
      "epoch": 0.4858649789029536,
      "grad_norm": 0.48668235540390015,
      "learning_rate": 0.0015,
      "loss": 1.6728,
      "step": 4606
    },
    {
      "epoch": 0.4859704641350211,
      "grad_norm": 0.6600933074951172,
      "learning_rate": 0.0015,
      "loss": 1.6645,
      "step": 4607
    },
    {
      "epoch": 0.4860759493670886,
      "grad_norm": 0.6586343050003052,
      "learning_rate": 0.0015,
      "loss": 1.6371,
      "step": 4608
    },
    {
      "epoch": 0.48618143459915614,
      "grad_norm": 0.5716184377670288,
      "learning_rate": 0.0015,
      "loss": 1.659,
      "step": 4609
    },
    {
      "epoch": 0.48628691983122363,
      "grad_norm": 0.9299889802932739,
      "learning_rate": 0.0015,
      "loss": 1.684,
      "step": 4610
    },
    {
      "epoch": 0.4863924050632911,
      "grad_norm": 0.6306243538856506,
      "learning_rate": 0.0015,
      "loss": 1.6159,
      "step": 4611
    },
    {
      "epoch": 0.48649789029535867,
      "grad_norm": 0.5388479828834534,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 4612
    },
    {
      "epoch": 0.48660337552742616,
      "grad_norm": 0.6615060567855835,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 4613
    },
    {
      "epoch": 0.48670886075949366,
      "grad_norm": 0.4851313531398773,
      "learning_rate": 0.0015,
      "loss": 1.6217,
      "step": 4614
    },
    {
      "epoch": 0.4868143459915612,
      "grad_norm": 0.5432100296020508,
      "learning_rate": 0.0015,
      "loss": 1.6692,
      "step": 4615
    },
    {
      "epoch": 0.4869198312236287,
      "grad_norm": 0.625106155872345,
      "learning_rate": 0.0015,
      "loss": 1.6601,
      "step": 4616
    },
    {
      "epoch": 0.4870253164556962,
      "grad_norm": 0.4862165153026581,
      "learning_rate": 0.0015,
      "loss": 1.6729,
      "step": 4617
    },
    {
      "epoch": 0.48713080168776374,
      "grad_norm": 0.5295156240463257,
      "learning_rate": 0.0015,
      "loss": 1.5984,
      "step": 4618
    },
    {
      "epoch": 0.48723628691983123,
      "grad_norm": 0.5893027186393738,
      "learning_rate": 0.0015,
      "loss": 1.5798,
      "step": 4619
    },
    {
      "epoch": 0.4873417721518987,
      "grad_norm": 0.4951131343841553,
      "learning_rate": 0.0015,
      "loss": 1.6432,
      "step": 4620
    },
    {
      "epoch": 0.48744725738396627,
      "grad_norm": 0.6460741758346558,
      "learning_rate": 0.0015,
      "loss": 1.6633,
      "step": 4621
    },
    {
      "epoch": 0.48755274261603376,
      "grad_norm": 0.5386523008346558,
      "learning_rate": 0.0015,
      "loss": 1.6184,
      "step": 4622
    },
    {
      "epoch": 0.48765822784810126,
      "grad_norm": 0.6607567667961121,
      "learning_rate": 0.0015,
      "loss": 1.6348,
      "step": 4623
    },
    {
      "epoch": 0.4877637130801688,
      "grad_norm": 0.7228679060935974,
      "learning_rate": 0.0015,
      "loss": 1.6527,
      "step": 4624
    },
    {
      "epoch": 0.4878691983122363,
      "grad_norm": 0.566042959690094,
      "learning_rate": 0.0015,
      "loss": 1.6481,
      "step": 4625
    },
    {
      "epoch": 0.4879746835443038,
      "grad_norm": 0.6492327451705933,
      "learning_rate": 0.0015,
      "loss": 1.6848,
      "step": 4626
    },
    {
      "epoch": 0.4880801687763713,
      "grad_norm": 0.6564940214157104,
      "learning_rate": 0.0015,
      "loss": 1.681,
      "step": 4627
    },
    {
      "epoch": 0.48818565400843883,
      "grad_norm": 0.5838112235069275,
      "learning_rate": 0.0015,
      "loss": 1.6521,
      "step": 4628
    },
    {
      "epoch": 0.4882911392405063,
      "grad_norm": 0.5898457765579224,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 4629
    },
    {
      "epoch": 0.4883966244725738,
      "grad_norm": 0.6300853490829468,
      "learning_rate": 0.0015,
      "loss": 1.6843,
      "step": 4630
    },
    {
      "epoch": 0.48850210970464136,
      "grad_norm": 0.6254135370254517,
      "learning_rate": 0.0015,
      "loss": 1.6537,
      "step": 4631
    },
    {
      "epoch": 0.48860759493670886,
      "grad_norm": 0.5185118913650513,
      "learning_rate": 0.0015,
      "loss": 1.6439,
      "step": 4632
    },
    {
      "epoch": 0.48871308016877635,
      "grad_norm": 0.4746887981891632,
      "learning_rate": 0.0015,
      "loss": 1.6455,
      "step": 4633
    },
    {
      "epoch": 0.4888185654008439,
      "grad_norm": 0.48079928755760193,
      "learning_rate": 0.0015,
      "loss": 1.6467,
      "step": 4634
    },
    {
      "epoch": 0.4889240506329114,
      "grad_norm": 0.6092884540557861,
      "learning_rate": 0.0015,
      "loss": 1.6342,
      "step": 4635
    },
    {
      "epoch": 0.4890295358649789,
      "grad_norm": 0.6327087879180908,
      "learning_rate": 0.0015,
      "loss": 1.6639,
      "step": 4636
    },
    {
      "epoch": 0.48913502109704643,
      "grad_norm": 0.47679081559181213,
      "learning_rate": 0.0015,
      "loss": 1.6481,
      "step": 4637
    },
    {
      "epoch": 0.4892405063291139,
      "grad_norm": 0.5547592043876648,
      "learning_rate": 0.0015,
      "loss": 1.659,
      "step": 4638
    },
    {
      "epoch": 0.4893459915611814,
      "grad_norm": 0.5467628836631775,
      "learning_rate": 0.0015,
      "loss": 1.6528,
      "step": 4639
    },
    {
      "epoch": 0.48945147679324896,
      "grad_norm": 0.5302047729492188,
      "learning_rate": 0.0015,
      "loss": 1.6743,
      "step": 4640
    },
    {
      "epoch": 0.48955696202531646,
      "grad_norm": 0.6105912923812866,
      "learning_rate": 0.0015,
      "loss": 1.6672,
      "step": 4641
    },
    {
      "epoch": 0.48966244725738395,
      "grad_norm": 0.5176563262939453,
      "learning_rate": 0.0015,
      "loss": 1.6191,
      "step": 4642
    },
    {
      "epoch": 0.4897679324894515,
      "grad_norm": 0.5613191723823547,
      "learning_rate": 0.0015,
      "loss": 1.6513,
      "step": 4643
    },
    {
      "epoch": 0.489873417721519,
      "grad_norm": 0.7750377058982849,
      "learning_rate": 0.0015,
      "loss": 1.6598,
      "step": 4644
    },
    {
      "epoch": 0.4899789029535865,
      "grad_norm": 0.5680108666419983,
      "learning_rate": 0.0015,
      "loss": 1.6847,
      "step": 4645
    },
    {
      "epoch": 0.49008438818565403,
      "grad_norm": 0.6858214139938354,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 4646
    },
    {
      "epoch": 0.4901898734177215,
      "grad_norm": 0.9767922163009644,
      "learning_rate": 0.0015,
      "loss": 1.6738,
      "step": 4647
    },
    {
      "epoch": 0.490295358649789,
      "grad_norm": 0.7666829824447632,
      "learning_rate": 0.0015,
      "loss": 1.6439,
      "step": 4648
    },
    {
      "epoch": 0.49040084388185656,
      "grad_norm": 0.475898414850235,
      "learning_rate": 0.0015,
      "loss": 1.6195,
      "step": 4649
    },
    {
      "epoch": 0.49050632911392406,
      "grad_norm": 0.6132450699806213,
      "learning_rate": 0.0015,
      "loss": 1.6599,
      "step": 4650
    },
    {
      "epoch": 0.49061181434599155,
      "grad_norm": 0.5402424335479736,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 4651
    },
    {
      "epoch": 0.4907172995780591,
      "grad_norm": 0.5422917008399963,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 4652
    },
    {
      "epoch": 0.4908227848101266,
      "grad_norm": 0.5834674835205078,
      "learning_rate": 0.0015,
      "loss": 1.6266,
      "step": 4653
    },
    {
      "epoch": 0.4909282700421941,
      "grad_norm": 0.5989320278167725,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 4654
    },
    {
      "epoch": 0.49103375527426163,
      "grad_norm": 0.494384229183197,
      "learning_rate": 0.0015,
      "loss": 1.6282,
      "step": 4655
    },
    {
      "epoch": 0.4911392405063291,
      "grad_norm": 0.5339090824127197,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 4656
    },
    {
      "epoch": 0.4912447257383966,
      "grad_norm": 0.5074661374092102,
      "learning_rate": 0.0015,
      "loss": 1.6351,
      "step": 4657
    },
    {
      "epoch": 0.4913502109704641,
      "grad_norm": 0.45555511116981506,
      "learning_rate": 0.0015,
      "loss": 1.6501,
      "step": 4658
    },
    {
      "epoch": 0.49145569620253166,
      "grad_norm": 0.5435980558395386,
      "learning_rate": 0.0015,
      "loss": 1.6333,
      "step": 4659
    },
    {
      "epoch": 0.49156118143459915,
      "grad_norm": 0.4912318289279938,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 4660
    },
    {
      "epoch": 0.49166666666666664,
      "grad_norm": 0.5439279079437256,
      "learning_rate": 0.0015,
      "loss": 1.6591,
      "step": 4661
    },
    {
      "epoch": 0.4917721518987342,
      "grad_norm": 0.5071859359741211,
      "learning_rate": 0.0015,
      "loss": 1.6196,
      "step": 4662
    },
    {
      "epoch": 0.4918776371308017,
      "grad_norm": 0.5272972583770752,
      "learning_rate": 0.0015,
      "loss": 1.6231,
      "step": 4663
    },
    {
      "epoch": 0.4919831223628692,
      "grad_norm": 0.6428019404411316,
      "learning_rate": 0.0015,
      "loss": 1.6345,
      "step": 4664
    },
    {
      "epoch": 0.4920886075949367,
      "grad_norm": 0.473804771900177,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 4665
    },
    {
      "epoch": 0.4921940928270042,
      "grad_norm": 0.5919150114059448,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 4666
    },
    {
      "epoch": 0.4922995780590717,
      "grad_norm": 0.5289254784584045,
      "learning_rate": 0.0015,
      "loss": 1.6599,
      "step": 4667
    },
    {
      "epoch": 0.49240506329113926,
      "grad_norm": 0.7686335444450378,
      "learning_rate": 0.0015,
      "loss": 1.6529,
      "step": 4668
    },
    {
      "epoch": 0.49251054852320675,
      "grad_norm": 0.560459554195404,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 4669
    },
    {
      "epoch": 0.49261603375527424,
      "grad_norm": 0.5330448150634766,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 4670
    },
    {
      "epoch": 0.4927215189873418,
      "grad_norm": 0.49647217988967896,
      "learning_rate": 0.0015,
      "loss": 1.6425,
      "step": 4671
    },
    {
      "epoch": 0.4928270042194093,
      "grad_norm": 0.5947394967079163,
      "learning_rate": 0.0015,
      "loss": 1.6126,
      "step": 4672
    },
    {
      "epoch": 0.4929324894514768,
      "grad_norm": 0.5732641816139221,
      "learning_rate": 0.0015,
      "loss": 1.6616,
      "step": 4673
    },
    {
      "epoch": 0.4930379746835443,
      "grad_norm": 0.47771984338760376,
      "learning_rate": 0.0015,
      "loss": 1.6787,
      "step": 4674
    },
    {
      "epoch": 0.4931434599156118,
      "grad_norm": 0.5954535007476807,
      "learning_rate": 0.0015,
      "loss": 1.6261,
      "step": 4675
    },
    {
      "epoch": 0.4932489451476793,
      "grad_norm": 0.6330942511558533,
      "learning_rate": 0.0015,
      "loss": 1.6123,
      "step": 4676
    },
    {
      "epoch": 0.49335443037974686,
      "grad_norm": 0.5292167067527771,
      "learning_rate": 0.0015,
      "loss": 1.6349,
      "step": 4677
    },
    {
      "epoch": 0.49345991561181435,
      "grad_norm": 0.5785264372825623,
      "learning_rate": 0.0015,
      "loss": 1.6203,
      "step": 4678
    },
    {
      "epoch": 0.49356540084388184,
      "grad_norm": 0.6675677299499512,
      "learning_rate": 0.0015,
      "loss": 1.6629,
      "step": 4679
    },
    {
      "epoch": 0.4936708860759494,
      "grad_norm": 0.6064327359199524,
      "learning_rate": 0.0015,
      "loss": 1.6139,
      "step": 4680
    },
    {
      "epoch": 0.4937763713080169,
      "grad_norm": 0.5640125274658203,
      "learning_rate": 0.0015,
      "loss": 1.6328,
      "step": 4681
    },
    {
      "epoch": 0.4938818565400844,
      "grad_norm": 0.5230752229690552,
      "learning_rate": 0.0015,
      "loss": 1.612,
      "step": 4682
    },
    {
      "epoch": 0.4939873417721519,
      "grad_norm": 0.6954069137573242,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 4683
    },
    {
      "epoch": 0.4940928270042194,
      "grad_norm": 0.5776487588882446,
      "learning_rate": 0.0015,
      "loss": 1.6094,
      "step": 4684
    },
    {
      "epoch": 0.4941983122362869,
      "grad_norm": 0.5650107860565186,
      "learning_rate": 0.0015,
      "loss": 1.5985,
      "step": 4685
    },
    {
      "epoch": 0.49430379746835446,
      "grad_norm": 0.5918284058570862,
      "learning_rate": 0.0015,
      "loss": 1.6491,
      "step": 4686
    },
    {
      "epoch": 0.49440928270042195,
      "grad_norm": 0.5596849918365479,
      "learning_rate": 0.0015,
      "loss": 1.6727,
      "step": 4687
    },
    {
      "epoch": 0.49451476793248944,
      "grad_norm": 0.6597175002098083,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 4688
    },
    {
      "epoch": 0.494620253164557,
      "grad_norm": 0.5488419532775879,
      "learning_rate": 0.0015,
      "loss": 1.6222,
      "step": 4689
    },
    {
      "epoch": 0.4947257383966245,
      "grad_norm": 0.5714768767356873,
      "learning_rate": 0.0015,
      "loss": 1.6793,
      "step": 4690
    },
    {
      "epoch": 0.494831223628692,
      "grad_norm": 0.7337178587913513,
      "learning_rate": 0.0015,
      "loss": 1.6811,
      "step": 4691
    },
    {
      "epoch": 0.49493670886075947,
      "grad_norm": 0.5429230332374573,
      "learning_rate": 0.0015,
      "loss": 1.6785,
      "step": 4692
    },
    {
      "epoch": 0.495042194092827,
      "grad_norm": 0.5269771218299866,
      "learning_rate": 0.0015,
      "loss": 1.6384,
      "step": 4693
    },
    {
      "epoch": 0.4951476793248945,
      "grad_norm": 0.5756490230560303,
      "learning_rate": 0.0015,
      "loss": 1.6645,
      "step": 4694
    },
    {
      "epoch": 0.495253164556962,
      "grad_norm": 0.49909448623657227,
      "learning_rate": 0.0015,
      "loss": 1.6276,
      "step": 4695
    },
    {
      "epoch": 0.49535864978902955,
      "grad_norm": 0.7548465728759766,
      "learning_rate": 0.0015,
      "loss": 1.6664,
      "step": 4696
    },
    {
      "epoch": 0.49546413502109704,
      "grad_norm": 0.7117580771446228,
      "learning_rate": 0.0015,
      "loss": 1.6349,
      "step": 4697
    },
    {
      "epoch": 0.49556962025316453,
      "grad_norm": 0.6276371479034424,
      "learning_rate": 0.0015,
      "loss": 1.619,
      "step": 4698
    },
    {
      "epoch": 0.4956751054852321,
      "grad_norm": 0.6697174906730652,
      "learning_rate": 0.0015,
      "loss": 1.6496,
      "step": 4699
    },
    {
      "epoch": 0.4957805907172996,
      "grad_norm": 0.9186984300613403,
      "learning_rate": 0.0015,
      "loss": 1.6335,
      "step": 4700
    },
    {
      "epoch": 0.49588607594936707,
      "grad_norm": 0.7723588943481445,
      "learning_rate": 0.0015,
      "loss": 1.658,
      "step": 4701
    },
    {
      "epoch": 0.4959915611814346,
      "grad_norm": 0.6396031379699707,
      "learning_rate": 0.0015,
      "loss": 1.661,
      "step": 4702
    },
    {
      "epoch": 0.4960970464135021,
      "grad_norm": 0.47685229778289795,
      "learning_rate": 0.0015,
      "loss": 1.6174,
      "step": 4703
    },
    {
      "epoch": 0.4962025316455696,
      "grad_norm": 0.5432143807411194,
      "learning_rate": 0.0015,
      "loss": 1.6531,
      "step": 4704
    },
    {
      "epoch": 0.49630801687763715,
      "grad_norm": 0.48465240001678467,
      "learning_rate": 0.0015,
      "loss": 1.6282,
      "step": 4705
    },
    {
      "epoch": 0.49641350210970464,
      "grad_norm": 0.6186333894729614,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 4706
    },
    {
      "epoch": 0.49651898734177213,
      "grad_norm": 0.5811643004417419,
      "learning_rate": 0.0015,
      "loss": 1.6903,
      "step": 4707
    },
    {
      "epoch": 0.4966244725738397,
      "grad_norm": 0.492428719997406,
      "learning_rate": 0.0015,
      "loss": 1.6558,
      "step": 4708
    },
    {
      "epoch": 0.4967299578059072,
      "grad_norm": 0.5256556272506714,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 4709
    },
    {
      "epoch": 0.49683544303797467,
      "grad_norm": 0.5144762396812439,
      "learning_rate": 0.0015,
      "loss": 1.6486,
      "step": 4710
    },
    {
      "epoch": 0.4969409282700422,
      "grad_norm": 0.49647825956344604,
      "learning_rate": 0.0015,
      "loss": 1.6574,
      "step": 4711
    },
    {
      "epoch": 0.4970464135021097,
      "grad_norm": 0.5308802127838135,
      "learning_rate": 0.0015,
      "loss": 1.6385,
      "step": 4712
    },
    {
      "epoch": 0.4971518987341772,
      "grad_norm": 0.4832683801651001,
      "learning_rate": 0.0015,
      "loss": 1.617,
      "step": 4713
    },
    {
      "epoch": 0.49725738396624475,
      "grad_norm": 0.48039448261260986,
      "learning_rate": 0.0015,
      "loss": 1.653,
      "step": 4714
    },
    {
      "epoch": 0.49736286919831224,
      "grad_norm": 0.5602380037307739,
      "learning_rate": 0.0015,
      "loss": 1.6386,
      "step": 4715
    },
    {
      "epoch": 0.49746835443037973,
      "grad_norm": 0.6170874834060669,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 4716
    },
    {
      "epoch": 0.4975738396624473,
      "grad_norm": 0.6201112270355225,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 4717
    },
    {
      "epoch": 0.4976793248945148,
      "grad_norm": 0.5356724262237549,
      "learning_rate": 0.0015,
      "loss": 1.6842,
      "step": 4718
    },
    {
      "epoch": 0.49778481012658227,
      "grad_norm": 0.7496811151504517,
      "learning_rate": 0.0015,
      "loss": 1.6512,
      "step": 4719
    },
    {
      "epoch": 0.4978902953586498,
      "grad_norm": 0.8356648087501526,
      "learning_rate": 0.0015,
      "loss": 1.6168,
      "step": 4720
    },
    {
      "epoch": 0.4979957805907173,
      "grad_norm": 0.5207118988037109,
      "learning_rate": 0.0015,
      "loss": 1.6345,
      "step": 4721
    },
    {
      "epoch": 0.4981012658227848,
      "grad_norm": 0.5726657509803772,
      "learning_rate": 0.0015,
      "loss": 1.6463,
      "step": 4722
    },
    {
      "epoch": 0.49820675105485235,
      "grad_norm": 0.6330668926239014,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 4723
    },
    {
      "epoch": 0.49831223628691984,
      "grad_norm": 0.4804544448852539,
      "learning_rate": 0.0015,
      "loss": 1.6554,
      "step": 4724
    },
    {
      "epoch": 0.49841772151898733,
      "grad_norm": 0.6218848824501038,
      "learning_rate": 0.0015,
      "loss": 1.6458,
      "step": 4725
    },
    {
      "epoch": 0.4985232067510548,
      "grad_norm": 0.45861953496932983,
      "learning_rate": 0.0015,
      "loss": 1.6344,
      "step": 4726
    },
    {
      "epoch": 0.4986286919831224,
      "grad_norm": 0.744053840637207,
      "learning_rate": 0.0015,
      "loss": 1.6163,
      "step": 4727
    },
    {
      "epoch": 0.49873417721518987,
      "grad_norm": 1.0661535263061523,
      "learning_rate": 0.0015,
      "loss": 1.6813,
      "step": 4728
    },
    {
      "epoch": 0.49883966244725736,
      "grad_norm": 0.6184759140014648,
      "learning_rate": 0.0015,
      "loss": 1.6245,
      "step": 4729
    },
    {
      "epoch": 0.4989451476793249,
      "grad_norm": 0.6284570097923279,
      "learning_rate": 0.0015,
      "loss": 1.6343,
      "step": 4730
    },
    {
      "epoch": 0.4990506329113924,
      "grad_norm": 0.7645490169525146,
      "learning_rate": 0.0015,
      "loss": 1.6667,
      "step": 4731
    },
    {
      "epoch": 0.4991561181434599,
      "grad_norm": 0.5360997319221497,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 4732
    },
    {
      "epoch": 0.49926160337552744,
      "grad_norm": 0.6242209672927856,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 4733
    },
    {
      "epoch": 0.49936708860759493,
      "grad_norm": 0.6186540722846985,
      "learning_rate": 0.0015,
      "loss": 1.6483,
      "step": 4734
    },
    {
      "epoch": 0.4994725738396624,
      "grad_norm": 0.5110530257225037,
      "learning_rate": 0.0015,
      "loss": 1.6626,
      "step": 4735
    },
    {
      "epoch": 0.49957805907173,
      "grad_norm": 0.6061602234840393,
      "learning_rate": 0.0015,
      "loss": 1.6434,
      "step": 4736
    },
    {
      "epoch": 0.49968354430379747,
      "grad_norm": 0.5486086010932922,
      "learning_rate": 0.0015,
      "loss": 1.6149,
      "step": 4737
    },
    {
      "epoch": 0.49978902953586496,
      "grad_norm": 0.4641488492488861,
      "learning_rate": 0.0015,
      "loss": 1.5949,
      "step": 4738
    },
    {
      "epoch": 0.4998945147679325,
      "grad_norm": 0.6686543226242065,
      "learning_rate": 0.0015,
      "loss": 1.6386,
      "step": 4739
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5206875801086426,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 4740
    },
    {
      "epoch": 0.5001054852320675,
      "grad_norm": 0.643432080745697,
      "learning_rate": 0.0015,
      "loss": 1.6758,
      "step": 4741
    },
    {
      "epoch": 0.500210970464135,
      "grad_norm": 0.8481569886207581,
      "learning_rate": 0.0015,
      "loss": 1.6349,
      "step": 4742
    },
    {
      "epoch": 0.5003164556962025,
      "grad_norm": 0.519643247127533,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 4743
    },
    {
      "epoch": 0.5004219409282701,
      "grad_norm": 0.567777693271637,
      "learning_rate": 0.0015,
      "loss": 1.6879,
      "step": 4744
    },
    {
      "epoch": 0.5005274261603375,
      "grad_norm": 0.5984745025634766,
      "learning_rate": 0.0015,
      "loss": 1.6387,
      "step": 4745
    },
    {
      "epoch": 0.5006329113924051,
      "grad_norm": 0.43486952781677246,
      "learning_rate": 0.0015,
      "loss": 1.666,
      "step": 4746
    },
    {
      "epoch": 0.5007383966244726,
      "grad_norm": 0.5758285522460938,
      "learning_rate": 0.0015,
      "loss": 1.6617,
      "step": 4747
    },
    {
      "epoch": 0.50084388185654,
      "grad_norm": 0.6878244280815125,
      "learning_rate": 0.0015,
      "loss": 1.6734,
      "step": 4748
    },
    {
      "epoch": 0.5009493670886076,
      "grad_norm": 0.529334306716919,
      "learning_rate": 0.0015,
      "loss": 1.6676,
      "step": 4749
    },
    {
      "epoch": 0.5010548523206751,
      "grad_norm": 0.7365033030509949,
      "learning_rate": 0.0015,
      "loss": 1.6557,
      "step": 4750
    },
    {
      "epoch": 0.5011603375527426,
      "grad_norm": 0.8762412071228027,
      "learning_rate": 0.0015,
      "loss": 1.6714,
      "step": 4751
    },
    {
      "epoch": 0.5012658227848101,
      "grad_norm": 0.4611147344112396,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 4752
    },
    {
      "epoch": 0.5013713080168777,
      "grad_norm": 0.7059435248374939,
      "learning_rate": 0.0015,
      "loss": 1.6626,
      "step": 4753
    },
    {
      "epoch": 0.5014767932489451,
      "grad_norm": 0.6425533294677734,
      "learning_rate": 0.0015,
      "loss": 1.6433,
      "step": 4754
    },
    {
      "epoch": 0.5015822784810127,
      "grad_norm": 0.50490802526474,
      "learning_rate": 0.0015,
      "loss": 1.629,
      "step": 4755
    },
    {
      "epoch": 0.5016877637130802,
      "grad_norm": 0.5531265139579773,
      "learning_rate": 0.0015,
      "loss": 1.596,
      "step": 4756
    },
    {
      "epoch": 0.5017932489451477,
      "grad_norm": 0.514526903629303,
      "learning_rate": 0.0015,
      "loss": 1.6577,
      "step": 4757
    },
    {
      "epoch": 0.5018987341772152,
      "grad_norm": 0.6004590392112732,
      "learning_rate": 0.0015,
      "loss": 1.6317,
      "step": 4758
    },
    {
      "epoch": 0.5020042194092827,
      "grad_norm": 0.5942376852035522,
      "learning_rate": 0.0015,
      "loss": 1.6236,
      "step": 4759
    },
    {
      "epoch": 0.5021097046413502,
      "grad_norm": 0.5438563823699951,
      "learning_rate": 0.0015,
      "loss": 1.651,
      "step": 4760
    },
    {
      "epoch": 0.5022151898734177,
      "grad_norm": 0.5856278538703918,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 4761
    },
    {
      "epoch": 0.5023206751054853,
      "grad_norm": 0.5899941921234131,
      "learning_rate": 0.0015,
      "loss": 1.6684,
      "step": 4762
    },
    {
      "epoch": 0.5024261603375527,
      "grad_norm": 0.5922358632087708,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 4763
    },
    {
      "epoch": 0.5025316455696203,
      "grad_norm": 0.6511942148208618,
      "learning_rate": 0.0015,
      "loss": 1.6443,
      "step": 4764
    },
    {
      "epoch": 0.5026371308016878,
      "grad_norm": 0.5567985773086548,
      "learning_rate": 0.0015,
      "loss": 1.6243,
      "step": 4765
    },
    {
      "epoch": 0.5027426160337553,
      "grad_norm": 0.55110764503479,
      "learning_rate": 0.0015,
      "loss": 1.6096,
      "step": 4766
    },
    {
      "epoch": 0.5028481012658228,
      "grad_norm": 0.505946934223175,
      "learning_rate": 0.0015,
      "loss": 1.6504,
      "step": 4767
    },
    {
      "epoch": 0.5029535864978903,
      "grad_norm": 0.5708931684494019,
      "learning_rate": 0.0015,
      "loss": 1.6218,
      "step": 4768
    },
    {
      "epoch": 0.5030590717299578,
      "grad_norm": 0.49275654554367065,
      "learning_rate": 0.0015,
      "loss": 1.683,
      "step": 4769
    },
    {
      "epoch": 0.5031645569620253,
      "grad_norm": 0.5598103404045105,
      "learning_rate": 0.0015,
      "loss": 1.6453,
      "step": 4770
    },
    {
      "epoch": 0.5032700421940929,
      "grad_norm": 0.5133016109466553,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 4771
    },
    {
      "epoch": 0.5033755274261603,
      "grad_norm": 0.5452487468719482,
      "learning_rate": 0.0015,
      "loss": 1.7077,
      "step": 4772
    },
    {
      "epoch": 0.5034810126582279,
      "grad_norm": 0.59088534116745,
      "learning_rate": 0.0015,
      "loss": 1.6465,
      "step": 4773
    },
    {
      "epoch": 0.5035864978902953,
      "grad_norm": 0.4592975378036499,
      "learning_rate": 0.0015,
      "loss": 1.6184,
      "step": 4774
    },
    {
      "epoch": 0.5036919831223629,
      "grad_norm": 0.6327562928199768,
      "learning_rate": 0.0015,
      "loss": 1.6795,
      "step": 4775
    },
    {
      "epoch": 0.5037974683544304,
      "grad_norm": 0.511746346950531,
      "learning_rate": 0.0015,
      "loss": 1.656,
      "step": 4776
    },
    {
      "epoch": 0.5039029535864978,
      "grad_norm": 0.6945894360542297,
      "learning_rate": 0.0015,
      "loss": 1.6304,
      "step": 4777
    },
    {
      "epoch": 0.5040084388185654,
      "grad_norm": 0.7663832306861877,
      "learning_rate": 0.0015,
      "loss": 1.6725,
      "step": 4778
    },
    {
      "epoch": 0.5041139240506329,
      "grad_norm": 0.5099506974220276,
      "learning_rate": 0.0015,
      "loss": 1.6322,
      "step": 4779
    },
    {
      "epoch": 0.5042194092827004,
      "grad_norm": 0.666743278503418,
      "learning_rate": 0.0015,
      "loss": 1.6362,
      "step": 4780
    },
    {
      "epoch": 0.5043248945147679,
      "grad_norm": 0.5317825675010681,
      "learning_rate": 0.0015,
      "loss": 1.6101,
      "step": 4781
    },
    {
      "epoch": 0.5044303797468355,
      "grad_norm": 0.5478038191795349,
      "learning_rate": 0.0015,
      "loss": 1.6717,
      "step": 4782
    },
    {
      "epoch": 0.5045358649789029,
      "grad_norm": 0.4518919885158539,
      "learning_rate": 0.0015,
      "loss": 1.6503,
      "step": 4783
    },
    {
      "epoch": 0.5046413502109705,
      "grad_norm": 0.5621270537376404,
      "learning_rate": 0.0015,
      "loss": 1.6205,
      "step": 4784
    },
    {
      "epoch": 0.504746835443038,
      "grad_norm": 0.6832736134529114,
      "learning_rate": 0.0015,
      "loss": 1.6376,
      "step": 4785
    },
    {
      "epoch": 0.5048523206751054,
      "grad_norm": 0.5388296842575073,
      "learning_rate": 0.0015,
      "loss": 1.6243,
      "step": 4786
    },
    {
      "epoch": 0.504957805907173,
      "grad_norm": 0.584375262260437,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 4787
    },
    {
      "epoch": 0.5050632911392405,
      "grad_norm": 0.4646458625793457,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 4788
    },
    {
      "epoch": 0.505168776371308,
      "grad_norm": 0.6249669194221497,
      "learning_rate": 0.0015,
      "loss": 1.5907,
      "step": 4789
    },
    {
      "epoch": 0.5052742616033755,
      "grad_norm": 0.534736156463623,
      "learning_rate": 0.0015,
      "loss": 1.6697,
      "step": 4790
    },
    {
      "epoch": 0.5053797468354431,
      "grad_norm": 0.57759028673172,
      "learning_rate": 0.0015,
      "loss": 1.6691,
      "step": 4791
    },
    {
      "epoch": 0.5054852320675105,
      "grad_norm": 0.557238757610321,
      "learning_rate": 0.0015,
      "loss": 1.6001,
      "step": 4792
    },
    {
      "epoch": 0.505590717299578,
      "grad_norm": 0.5487537980079651,
      "learning_rate": 0.0015,
      "loss": 1.6694,
      "step": 4793
    },
    {
      "epoch": 0.5056962025316456,
      "grad_norm": 0.5272213816642761,
      "learning_rate": 0.0015,
      "loss": 1.7048,
      "step": 4794
    },
    {
      "epoch": 0.505801687763713,
      "grad_norm": 0.5248759984970093,
      "learning_rate": 0.0015,
      "loss": 1.6327,
      "step": 4795
    },
    {
      "epoch": 0.5059071729957806,
      "grad_norm": 0.5483149290084839,
      "learning_rate": 0.0015,
      "loss": 1.6294,
      "step": 4796
    },
    {
      "epoch": 0.5060126582278481,
      "grad_norm": 0.5631021857261658,
      "learning_rate": 0.0015,
      "loss": 1.6312,
      "step": 4797
    },
    {
      "epoch": 0.5061181434599156,
      "grad_norm": 0.497884064912796,
      "learning_rate": 0.0015,
      "loss": 1.594,
      "step": 4798
    },
    {
      "epoch": 0.5062236286919831,
      "grad_norm": 0.6340633034706116,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 4799
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 0.542874276638031,
      "learning_rate": 0.0015,
      "loss": 1.6131,
      "step": 4800
    },
    {
      "epoch": 0.5064345991561181,
      "grad_norm": 0.6450417041778564,
      "learning_rate": 0.0015,
      "loss": 1.6165,
      "step": 4801
    },
    {
      "epoch": 0.5065400843881857,
      "grad_norm": 0.486097127199173,
      "learning_rate": 0.0015,
      "loss": 1.6436,
      "step": 4802
    },
    {
      "epoch": 0.5066455696202532,
      "grad_norm": 0.6658429503440857,
      "learning_rate": 0.0015,
      "loss": 1.6454,
      "step": 4803
    },
    {
      "epoch": 0.5067510548523206,
      "grad_norm": 0.6243430972099304,
      "learning_rate": 0.0015,
      "loss": 1.6151,
      "step": 4804
    },
    {
      "epoch": 0.5068565400843882,
      "grad_norm": 0.4700041115283966,
      "learning_rate": 0.0015,
      "loss": 1.6298,
      "step": 4805
    },
    {
      "epoch": 0.5069620253164557,
      "grad_norm": 0.667640209197998,
      "learning_rate": 0.0015,
      "loss": 1.639,
      "step": 4806
    },
    {
      "epoch": 0.5070675105485232,
      "grad_norm": 0.5673599243164062,
      "learning_rate": 0.0015,
      "loss": 1.6693,
      "step": 4807
    },
    {
      "epoch": 0.5071729957805907,
      "grad_norm": 0.5642109513282776,
      "learning_rate": 0.0015,
      "loss": 1.6456,
      "step": 4808
    },
    {
      "epoch": 0.5072784810126583,
      "grad_norm": 0.5934702754020691,
      "learning_rate": 0.0015,
      "loss": 1.6516,
      "step": 4809
    },
    {
      "epoch": 0.5073839662447257,
      "grad_norm": 0.5233937501907349,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 4810
    },
    {
      "epoch": 0.5074894514767933,
      "grad_norm": 0.5223708748817444,
      "learning_rate": 0.0015,
      "loss": 1.6375,
      "step": 4811
    },
    {
      "epoch": 0.5075949367088608,
      "grad_norm": 0.5699358582496643,
      "learning_rate": 0.0015,
      "loss": 1.6364,
      "step": 4812
    },
    {
      "epoch": 0.5077004219409282,
      "grad_norm": 0.5589818358421326,
      "learning_rate": 0.0015,
      "loss": 1.6891,
      "step": 4813
    },
    {
      "epoch": 0.5078059071729958,
      "grad_norm": 0.4734857380390167,
      "learning_rate": 0.0015,
      "loss": 1.668,
      "step": 4814
    },
    {
      "epoch": 0.5079113924050633,
      "grad_norm": 0.506865918636322,
      "learning_rate": 0.0015,
      "loss": 1.6596,
      "step": 4815
    },
    {
      "epoch": 0.5080168776371308,
      "grad_norm": 0.6197706460952759,
      "learning_rate": 0.0015,
      "loss": 1.6259,
      "step": 4816
    },
    {
      "epoch": 0.5081223628691983,
      "grad_norm": 0.6048730611801147,
      "learning_rate": 0.0015,
      "loss": 1.6418,
      "step": 4817
    },
    {
      "epoch": 0.5082278481012659,
      "grad_norm": 0.5068001747131348,
      "learning_rate": 0.0015,
      "loss": 1.6979,
      "step": 4818
    },
    {
      "epoch": 0.5083333333333333,
      "grad_norm": 0.5281519889831543,
      "learning_rate": 0.0015,
      "loss": 1.6754,
      "step": 4819
    },
    {
      "epoch": 0.5084388185654009,
      "grad_norm": 0.5379570126533508,
      "learning_rate": 0.0015,
      "loss": 1.6316,
      "step": 4820
    },
    {
      "epoch": 0.5085443037974684,
      "grad_norm": 0.5386903882026672,
      "learning_rate": 0.0015,
      "loss": 1.666,
      "step": 4821
    },
    {
      "epoch": 0.5086497890295358,
      "grad_norm": 0.5416865348815918,
      "learning_rate": 0.0015,
      "loss": 1.6314,
      "step": 4822
    },
    {
      "epoch": 0.5087552742616034,
      "grad_norm": 0.5622119307518005,
      "learning_rate": 0.0015,
      "loss": 1.6555,
      "step": 4823
    },
    {
      "epoch": 0.5088607594936709,
      "grad_norm": 0.823489248752594,
      "learning_rate": 0.0015,
      "loss": 1.6658,
      "step": 4824
    },
    {
      "epoch": 0.5089662447257384,
      "grad_norm": 0.5296961665153503,
      "learning_rate": 0.0015,
      "loss": 1.6504,
      "step": 4825
    },
    {
      "epoch": 0.5090717299578059,
      "grad_norm": 0.6514083743095398,
      "learning_rate": 0.0015,
      "loss": 1.6075,
      "step": 4826
    },
    {
      "epoch": 0.5091772151898735,
      "grad_norm": 0.7773762345314026,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 4827
    },
    {
      "epoch": 0.5092827004219409,
      "grad_norm": 0.6508411169052124,
      "learning_rate": 0.0015,
      "loss": 1.6495,
      "step": 4828
    },
    {
      "epoch": 0.5093881856540085,
      "grad_norm": 0.45788708329200745,
      "learning_rate": 0.0015,
      "loss": 1.6281,
      "step": 4829
    },
    {
      "epoch": 0.509493670886076,
      "grad_norm": 0.73536217212677,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 4830
    },
    {
      "epoch": 0.5095991561181434,
      "grad_norm": 0.7188004851341248,
      "learning_rate": 0.0015,
      "loss": 1.6507,
      "step": 4831
    },
    {
      "epoch": 0.509704641350211,
      "grad_norm": 0.6408347487449646,
      "learning_rate": 0.0015,
      "loss": 1.6482,
      "step": 4832
    },
    {
      "epoch": 0.5098101265822785,
      "grad_norm": 0.5778459906578064,
      "learning_rate": 0.0015,
      "loss": 1.5965,
      "step": 4833
    },
    {
      "epoch": 0.509915611814346,
      "grad_norm": 1.011449933052063,
      "learning_rate": 0.0015,
      "loss": 1.6383,
      "step": 4834
    },
    {
      "epoch": 0.5100210970464135,
      "grad_norm": 0.954604983329773,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 4835
    },
    {
      "epoch": 0.5101265822784811,
      "grad_norm": 0.4972829222679138,
      "learning_rate": 0.0015,
      "loss": 1.6097,
      "step": 4836
    },
    {
      "epoch": 0.5102320675105485,
      "grad_norm": 0.7298729419708252,
      "learning_rate": 0.0015,
      "loss": 1.6101,
      "step": 4837
    },
    {
      "epoch": 0.510337552742616,
      "grad_norm": 0.6276075839996338,
      "learning_rate": 0.0015,
      "loss": 1.6452,
      "step": 4838
    },
    {
      "epoch": 0.5104430379746835,
      "grad_norm": 0.582192063331604,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 4839
    },
    {
      "epoch": 0.510548523206751,
      "grad_norm": 0.5772690176963806,
      "learning_rate": 0.0015,
      "loss": 1.6852,
      "step": 4840
    },
    {
      "epoch": 0.5106540084388186,
      "grad_norm": 0.6315447092056274,
      "learning_rate": 0.0015,
      "loss": 1.6337,
      "step": 4841
    },
    {
      "epoch": 0.510759493670886,
      "grad_norm": 0.49861079454421997,
      "learning_rate": 0.0015,
      "loss": 1.6454,
      "step": 4842
    },
    {
      "epoch": 0.5108649789029536,
      "grad_norm": 0.5529313683509827,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 4843
    },
    {
      "epoch": 0.5109704641350211,
      "grad_norm": 0.5122413635253906,
      "learning_rate": 0.0015,
      "loss": 1.6475,
      "step": 4844
    },
    {
      "epoch": 0.5110759493670886,
      "grad_norm": 0.6824837327003479,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 4845
    },
    {
      "epoch": 0.5111814345991561,
      "grad_norm": 0.5071849822998047,
      "learning_rate": 0.0015,
      "loss": 1.6658,
      "step": 4846
    },
    {
      "epoch": 0.5112869198312237,
      "grad_norm": 0.5284269452095032,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 4847
    },
    {
      "epoch": 0.5113924050632911,
      "grad_norm": 0.6052508354187012,
      "learning_rate": 0.0015,
      "loss": 1.6621,
      "step": 4848
    },
    {
      "epoch": 0.5114978902953586,
      "grad_norm": 0.5683450102806091,
      "learning_rate": 0.0015,
      "loss": 1.6405,
      "step": 4849
    },
    {
      "epoch": 0.5116033755274262,
      "grad_norm": 0.5185849070549011,
      "learning_rate": 0.0015,
      "loss": 1.6658,
      "step": 4850
    },
    {
      "epoch": 0.5117088607594936,
      "grad_norm": 0.5939214825630188,
      "learning_rate": 0.0015,
      "loss": 1.6825,
      "step": 4851
    },
    {
      "epoch": 0.5118143459915612,
      "grad_norm": 0.5073562860488892,
      "learning_rate": 0.0015,
      "loss": 1.665,
      "step": 4852
    },
    {
      "epoch": 0.5119198312236287,
      "grad_norm": 0.5370201468467712,
      "learning_rate": 0.0015,
      "loss": 1.6071,
      "step": 4853
    },
    {
      "epoch": 0.5120253164556962,
      "grad_norm": 0.5023263096809387,
      "learning_rate": 0.0015,
      "loss": 1.6767,
      "step": 4854
    },
    {
      "epoch": 0.5121308016877637,
      "grad_norm": 0.5580131411552429,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 4855
    },
    {
      "epoch": 0.5122362869198313,
      "grad_norm": 0.48221713304519653,
      "learning_rate": 0.0015,
      "loss": 1.6257,
      "step": 4856
    },
    {
      "epoch": 0.5123417721518987,
      "grad_norm": 0.5159028172492981,
      "learning_rate": 0.0015,
      "loss": 1.624,
      "step": 4857
    },
    {
      "epoch": 0.5124472573839662,
      "grad_norm": 0.6387952566146851,
      "learning_rate": 0.0015,
      "loss": 1.6069,
      "step": 4858
    },
    {
      "epoch": 0.5125527426160338,
      "grad_norm": 0.5266417860984802,
      "learning_rate": 0.0015,
      "loss": 1.5892,
      "step": 4859
    },
    {
      "epoch": 0.5126582278481012,
      "grad_norm": 0.5424797534942627,
      "learning_rate": 0.0015,
      "loss": 1.642,
      "step": 4860
    },
    {
      "epoch": 0.5127637130801688,
      "grad_norm": 0.6165294051170349,
      "learning_rate": 0.0015,
      "loss": 1.6258,
      "step": 4861
    },
    {
      "epoch": 0.5128691983122363,
      "grad_norm": 0.5575075745582581,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 4862
    },
    {
      "epoch": 0.5129746835443038,
      "grad_norm": 0.5405846238136292,
      "learning_rate": 0.0015,
      "loss": 1.6547,
      "step": 4863
    },
    {
      "epoch": 0.5130801687763713,
      "grad_norm": 0.546652615070343,
      "learning_rate": 0.0015,
      "loss": 1.6493,
      "step": 4864
    },
    {
      "epoch": 0.5131856540084389,
      "grad_norm": 0.5599865317344666,
      "learning_rate": 0.0015,
      "loss": 1.6049,
      "step": 4865
    },
    {
      "epoch": 0.5132911392405063,
      "grad_norm": 0.5863654017448425,
      "learning_rate": 0.0015,
      "loss": 1.6331,
      "step": 4866
    },
    {
      "epoch": 0.5133966244725738,
      "grad_norm": 0.5758659839630127,
      "learning_rate": 0.0015,
      "loss": 1.597,
      "step": 4867
    },
    {
      "epoch": 0.5135021097046414,
      "grad_norm": 0.6346961855888367,
      "learning_rate": 0.0015,
      "loss": 1.6356,
      "step": 4868
    },
    {
      "epoch": 0.5136075949367088,
      "grad_norm": 0.5564453601837158,
      "learning_rate": 0.0015,
      "loss": 1.627,
      "step": 4869
    },
    {
      "epoch": 0.5137130801687764,
      "grad_norm": 0.6270017027854919,
      "learning_rate": 0.0015,
      "loss": 1.633,
      "step": 4870
    },
    {
      "epoch": 0.5138185654008439,
      "grad_norm": 0.5450406670570374,
      "learning_rate": 0.0015,
      "loss": 1.6603,
      "step": 4871
    },
    {
      "epoch": 0.5139240506329114,
      "grad_norm": 0.5726514458656311,
      "learning_rate": 0.0015,
      "loss": 1.6238,
      "step": 4872
    },
    {
      "epoch": 0.5140295358649789,
      "grad_norm": 0.5533527731895447,
      "learning_rate": 0.0015,
      "loss": 1.6888,
      "step": 4873
    },
    {
      "epoch": 0.5141350210970465,
      "grad_norm": 0.6688078045845032,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 4874
    },
    {
      "epoch": 0.5142405063291139,
      "grad_norm": 0.5136394500732422,
      "learning_rate": 0.0015,
      "loss": 1.6096,
      "step": 4875
    },
    {
      "epoch": 0.5143459915611814,
      "grad_norm": 0.5708884000778198,
      "learning_rate": 0.0015,
      "loss": 1.6575,
      "step": 4876
    },
    {
      "epoch": 0.514451476793249,
      "grad_norm": 0.6574445366859436,
      "learning_rate": 0.0015,
      "loss": 1.644,
      "step": 4877
    },
    {
      "epoch": 0.5145569620253164,
      "grad_norm": 0.4927603304386139,
      "learning_rate": 0.0015,
      "loss": 1.6781,
      "step": 4878
    },
    {
      "epoch": 0.514662447257384,
      "grad_norm": 0.6540015339851379,
      "learning_rate": 0.0015,
      "loss": 1.669,
      "step": 4879
    },
    {
      "epoch": 0.5147679324894515,
      "grad_norm": 0.6037447452545166,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 4880
    },
    {
      "epoch": 0.514873417721519,
      "grad_norm": 0.5282309651374817,
      "learning_rate": 0.0015,
      "loss": 1.6535,
      "step": 4881
    },
    {
      "epoch": 0.5149789029535865,
      "grad_norm": 0.6613801717758179,
      "learning_rate": 0.0015,
      "loss": 1.6405,
      "step": 4882
    },
    {
      "epoch": 0.515084388185654,
      "grad_norm": 0.6758782863616943,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 4883
    },
    {
      "epoch": 0.5151898734177215,
      "grad_norm": 0.6558852791786194,
      "learning_rate": 0.0015,
      "loss": 1.6456,
      "step": 4884
    },
    {
      "epoch": 0.515295358649789,
      "grad_norm": 0.5303662419319153,
      "learning_rate": 0.0015,
      "loss": 1.6178,
      "step": 4885
    },
    {
      "epoch": 0.5154008438818566,
      "grad_norm": 0.6675775051116943,
      "learning_rate": 0.0015,
      "loss": 1.6559,
      "step": 4886
    },
    {
      "epoch": 0.515506329113924,
      "grad_norm": 0.6730503439903259,
      "learning_rate": 0.0015,
      "loss": 1.6317,
      "step": 4887
    },
    {
      "epoch": 0.5156118143459916,
      "grad_norm": 0.4933164715766907,
      "learning_rate": 0.0015,
      "loss": 1.6638,
      "step": 4888
    },
    {
      "epoch": 0.5157172995780591,
      "grad_norm": 0.7364172339439392,
      "learning_rate": 0.0015,
      "loss": 1.6215,
      "step": 4889
    },
    {
      "epoch": 0.5158227848101266,
      "grad_norm": 0.5155131816864014,
      "learning_rate": 0.0015,
      "loss": 1.6181,
      "step": 4890
    },
    {
      "epoch": 0.5159282700421941,
      "grad_norm": 0.6437991261482239,
      "learning_rate": 0.0015,
      "loss": 1.6047,
      "step": 4891
    },
    {
      "epoch": 0.5160337552742617,
      "grad_norm": 0.6346082091331482,
      "learning_rate": 0.0015,
      "loss": 1.6714,
      "step": 4892
    },
    {
      "epoch": 0.5161392405063291,
      "grad_norm": 0.5846510529518127,
      "learning_rate": 0.0015,
      "loss": 1.6392,
      "step": 4893
    },
    {
      "epoch": 0.5162447257383966,
      "grad_norm": 0.7635490894317627,
      "learning_rate": 0.0015,
      "loss": 1.6601,
      "step": 4894
    },
    {
      "epoch": 0.5163502109704642,
      "grad_norm": 0.7379271984100342,
      "learning_rate": 0.0015,
      "loss": 1.6253,
      "step": 4895
    },
    {
      "epoch": 0.5164556962025316,
      "grad_norm": 0.5082283616065979,
      "learning_rate": 0.0015,
      "loss": 1.6866,
      "step": 4896
    },
    {
      "epoch": 0.5165611814345992,
      "grad_norm": 0.8635253310203552,
      "learning_rate": 0.0015,
      "loss": 1.648,
      "step": 4897
    },
    {
      "epoch": 0.5166666666666667,
      "grad_norm": 0.7167245149612427,
      "learning_rate": 0.0015,
      "loss": 1.6734,
      "step": 4898
    },
    {
      "epoch": 0.5167721518987342,
      "grad_norm": 0.5967913269996643,
      "learning_rate": 0.0015,
      "loss": 1.6351,
      "step": 4899
    },
    {
      "epoch": 0.5168776371308017,
      "grad_norm": 0.6967950463294983,
      "learning_rate": 0.0015,
      "loss": 1.6612,
      "step": 4900
    },
    {
      "epoch": 0.5169831223628693,
      "grad_norm": 0.5675439238548279,
      "learning_rate": 0.0015,
      "loss": 1.6622,
      "step": 4901
    },
    {
      "epoch": 0.5170886075949367,
      "grad_norm": 0.6265988945960999,
      "learning_rate": 0.0015,
      "loss": 1.6105,
      "step": 4902
    },
    {
      "epoch": 0.5171940928270042,
      "grad_norm": 0.6491437554359436,
      "learning_rate": 0.0015,
      "loss": 1.6309,
      "step": 4903
    },
    {
      "epoch": 0.5172995780590718,
      "grad_norm": 0.5626640319824219,
      "learning_rate": 0.0015,
      "loss": 1.642,
      "step": 4904
    },
    {
      "epoch": 0.5174050632911392,
      "grad_norm": 0.8992338180541992,
      "learning_rate": 0.0015,
      "loss": 1.6709,
      "step": 4905
    },
    {
      "epoch": 0.5175105485232068,
      "grad_norm": 0.7224078178405762,
      "learning_rate": 0.0015,
      "loss": 1.6374,
      "step": 4906
    },
    {
      "epoch": 0.5176160337552742,
      "grad_norm": 0.58208829164505,
      "learning_rate": 0.0015,
      "loss": 1.5838,
      "step": 4907
    },
    {
      "epoch": 0.5177215189873418,
      "grad_norm": 0.9976869821548462,
      "learning_rate": 0.0015,
      "loss": 1.6737,
      "step": 4908
    },
    {
      "epoch": 0.5178270042194093,
      "grad_norm": 0.7647194266319275,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 4909
    },
    {
      "epoch": 0.5179324894514767,
      "grad_norm": 0.6444775462150574,
      "learning_rate": 0.0015,
      "loss": 1.6813,
      "step": 4910
    },
    {
      "epoch": 0.5180379746835443,
      "grad_norm": 0.905405580997467,
      "learning_rate": 0.0015,
      "loss": 1.6476,
      "step": 4911
    },
    {
      "epoch": 0.5181434599156118,
      "grad_norm": 0.582550048828125,
      "learning_rate": 0.0015,
      "loss": 1.6165,
      "step": 4912
    },
    {
      "epoch": 0.5182489451476793,
      "grad_norm": 0.6779419183731079,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 4913
    },
    {
      "epoch": 0.5183544303797468,
      "grad_norm": 0.7489119172096252,
      "learning_rate": 0.0015,
      "loss": 1.6171,
      "step": 4914
    },
    {
      "epoch": 0.5184599156118144,
      "grad_norm": 0.5231460332870483,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 4915
    },
    {
      "epoch": 0.5185654008438818,
      "grad_norm": 0.5889079570770264,
      "learning_rate": 0.0015,
      "loss": 1.6552,
      "step": 4916
    },
    {
      "epoch": 0.5186708860759494,
      "grad_norm": 0.6349860429763794,
      "learning_rate": 0.0015,
      "loss": 1.6536,
      "step": 4917
    },
    {
      "epoch": 0.5187763713080169,
      "grad_norm": 0.5343783497810364,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 4918
    },
    {
      "epoch": 0.5188818565400843,
      "grad_norm": 0.5360980033874512,
      "learning_rate": 0.0015,
      "loss": 1.6209,
      "step": 4919
    },
    {
      "epoch": 0.5189873417721519,
      "grad_norm": 0.5116620659828186,
      "learning_rate": 0.0015,
      "loss": 1.6622,
      "step": 4920
    },
    {
      "epoch": 0.5190928270042194,
      "grad_norm": 0.478810578584671,
      "learning_rate": 0.0015,
      "loss": 1.6475,
      "step": 4921
    },
    {
      "epoch": 0.5191983122362869,
      "grad_norm": 0.523101270198822,
      "learning_rate": 0.0015,
      "loss": 1.6779,
      "step": 4922
    },
    {
      "epoch": 0.5193037974683544,
      "grad_norm": 0.4698803722858429,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 4923
    },
    {
      "epoch": 0.519409282700422,
      "grad_norm": 0.5162774920463562,
      "learning_rate": 0.0015,
      "loss": 1.5954,
      "step": 4924
    },
    {
      "epoch": 0.5195147679324894,
      "grad_norm": 0.45604798197746277,
      "learning_rate": 0.0015,
      "loss": 1.6294,
      "step": 4925
    },
    {
      "epoch": 0.519620253164557,
      "grad_norm": 0.5285675525665283,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 4926
    },
    {
      "epoch": 0.5197257383966245,
      "grad_norm": 0.5319159626960754,
      "learning_rate": 0.0015,
      "loss": 1.619,
      "step": 4927
    },
    {
      "epoch": 0.5198312236286919,
      "grad_norm": 0.5538773536682129,
      "learning_rate": 0.0015,
      "loss": 1.6532,
      "step": 4928
    },
    {
      "epoch": 0.5199367088607595,
      "grad_norm": 0.5458510518074036,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 4929
    },
    {
      "epoch": 0.520042194092827,
      "grad_norm": 0.5238522887229919,
      "learning_rate": 0.0015,
      "loss": 1.6354,
      "step": 4930
    },
    {
      "epoch": 0.5201476793248945,
      "grad_norm": 0.5384401082992554,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 4931
    },
    {
      "epoch": 0.520253164556962,
      "grad_norm": 0.44328439235687256,
      "learning_rate": 0.0015,
      "loss": 1.6557,
      "step": 4932
    },
    {
      "epoch": 0.5203586497890296,
      "grad_norm": 0.5922768115997314,
      "learning_rate": 0.0015,
      "loss": 1.6737,
      "step": 4933
    },
    {
      "epoch": 0.520464135021097,
      "grad_norm": 0.5494799017906189,
      "learning_rate": 0.0015,
      "loss": 1.637,
      "step": 4934
    },
    {
      "epoch": 0.5205696202531646,
      "grad_norm": 0.6524472832679749,
      "learning_rate": 0.0015,
      "loss": 1.5953,
      "step": 4935
    },
    {
      "epoch": 0.5206751054852321,
      "grad_norm": 0.5072101950645447,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 4936
    },
    {
      "epoch": 0.5207805907172995,
      "grad_norm": 0.5628498196601868,
      "learning_rate": 0.0015,
      "loss": 1.6373,
      "step": 4937
    },
    {
      "epoch": 0.5208860759493671,
      "grad_norm": 0.5973775386810303,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 4938
    },
    {
      "epoch": 0.5209915611814346,
      "grad_norm": 0.5659748315811157,
      "learning_rate": 0.0015,
      "loss": 1.6514,
      "step": 4939
    },
    {
      "epoch": 0.5210970464135021,
      "grad_norm": 0.5384728312492371,
      "learning_rate": 0.0015,
      "loss": 1.6249,
      "step": 4940
    },
    {
      "epoch": 0.5212025316455696,
      "grad_norm": 0.572237491607666,
      "learning_rate": 0.0015,
      "loss": 1.6696,
      "step": 4941
    },
    {
      "epoch": 0.5213080168776372,
      "grad_norm": 0.7187435030937195,
      "learning_rate": 0.0015,
      "loss": 1.6518,
      "step": 4942
    },
    {
      "epoch": 0.5214135021097046,
      "grad_norm": 0.5063067078590393,
      "learning_rate": 0.0015,
      "loss": 1.635,
      "step": 4943
    },
    {
      "epoch": 0.5215189873417722,
      "grad_norm": 0.5599090456962585,
      "learning_rate": 0.0015,
      "loss": 1.6486,
      "step": 4944
    },
    {
      "epoch": 0.5216244725738397,
      "grad_norm": 0.5347083210945129,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 4945
    },
    {
      "epoch": 0.5217299578059071,
      "grad_norm": 0.6137127876281738,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 4946
    },
    {
      "epoch": 0.5218354430379747,
      "grad_norm": 0.6572962403297424,
      "learning_rate": 0.0015,
      "loss": 1.6544,
      "step": 4947
    },
    {
      "epoch": 0.5219409282700422,
      "grad_norm": 0.5356124043464661,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 4948
    },
    {
      "epoch": 0.5220464135021097,
      "grad_norm": 0.5502074956893921,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 4949
    },
    {
      "epoch": 0.5221518987341772,
      "grad_norm": 0.647217333316803,
      "learning_rate": 0.0015,
      "loss": 1.6601,
      "step": 4950
    },
    {
      "epoch": 0.5222573839662448,
      "grad_norm": 0.5100189447402954,
      "learning_rate": 0.0015,
      "loss": 1.6567,
      "step": 4951
    },
    {
      "epoch": 0.5223628691983122,
      "grad_norm": 0.49887460470199585,
      "learning_rate": 0.0015,
      "loss": 1.6334,
      "step": 4952
    },
    {
      "epoch": 0.5224683544303798,
      "grad_norm": 0.4950576424598694,
      "learning_rate": 0.0015,
      "loss": 1.5806,
      "step": 4953
    },
    {
      "epoch": 0.5225738396624473,
      "grad_norm": 0.6040221452713013,
      "learning_rate": 0.0015,
      "loss": 1.6344,
      "step": 4954
    },
    {
      "epoch": 0.5226793248945147,
      "grad_norm": 0.6119283437728882,
      "learning_rate": 0.0015,
      "loss": 1.6442,
      "step": 4955
    },
    {
      "epoch": 0.5227848101265823,
      "grad_norm": 0.5912838578224182,
      "learning_rate": 0.0015,
      "loss": 1.6273,
      "step": 4956
    },
    {
      "epoch": 0.5228902953586498,
      "grad_norm": 0.7666657567024231,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 4957
    },
    {
      "epoch": 0.5229957805907173,
      "grad_norm": 0.6835048198699951,
      "learning_rate": 0.0015,
      "loss": 1.6345,
      "step": 4958
    },
    {
      "epoch": 0.5231012658227848,
      "grad_norm": 0.5428978204727173,
      "learning_rate": 0.0015,
      "loss": 1.5995,
      "step": 4959
    },
    {
      "epoch": 0.5232067510548524,
      "grad_norm": 0.6137120127677917,
      "learning_rate": 0.0015,
      "loss": 1.6267,
      "step": 4960
    },
    {
      "epoch": 0.5233122362869198,
      "grad_norm": 0.6189130544662476,
      "learning_rate": 0.0015,
      "loss": 1.6223,
      "step": 4961
    },
    {
      "epoch": 0.5234177215189874,
      "grad_norm": 0.47976619005203247,
      "learning_rate": 0.0015,
      "loss": 1.633,
      "step": 4962
    },
    {
      "epoch": 0.5235232067510549,
      "grad_norm": 0.6020201444625854,
      "learning_rate": 0.0015,
      "loss": 1.6545,
      "step": 4963
    },
    {
      "epoch": 0.5236286919831223,
      "grad_norm": 0.7014598846435547,
      "learning_rate": 0.0015,
      "loss": 1.6266,
      "step": 4964
    },
    {
      "epoch": 0.5237341772151899,
      "grad_norm": 0.5283344984054565,
      "learning_rate": 0.0015,
      "loss": 1.7071,
      "step": 4965
    },
    {
      "epoch": 0.5238396624472574,
      "grad_norm": 0.7310873866081238,
      "learning_rate": 0.0015,
      "loss": 1.628,
      "step": 4966
    },
    {
      "epoch": 0.5239451476793249,
      "grad_norm": 0.6351194381713867,
      "learning_rate": 0.0015,
      "loss": 1.6009,
      "step": 4967
    },
    {
      "epoch": 0.5240506329113924,
      "grad_norm": 0.6132206320762634,
      "learning_rate": 0.0015,
      "loss": 1.5803,
      "step": 4968
    },
    {
      "epoch": 0.52415611814346,
      "grad_norm": 0.6869380474090576,
      "learning_rate": 0.0015,
      "loss": 1.6447,
      "step": 4969
    },
    {
      "epoch": 0.5242616033755274,
      "grad_norm": 0.7326971888542175,
      "learning_rate": 0.0015,
      "loss": 1.677,
      "step": 4970
    },
    {
      "epoch": 0.524367088607595,
      "grad_norm": 0.7327278256416321,
      "learning_rate": 0.0015,
      "loss": 1.696,
      "step": 4971
    },
    {
      "epoch": 0.5244725738396624,
      "grad_norm": 0.4774857461452484,
      "learning_rate": 0.0015,
      "loss": 1.5709,
      "step": 4972
    },
    {
      "epoch": 0.5245780590717299,
      "grad_norm": 0.6482288241386414,
      "learning_rate": 0.0015,
      "loss": 1.6253,
      "step": 4973
    },
    {
      "epoch": 0.5246835443037975,
      "grad_norm": 0.6765118837356567,
      "learning_rate": 0.0015,
      "loss": 1.6312,
      "step": 4974
    },
    {
      "epoch": 0.5247890295358649,
      "grad_norm": 0.601466715335846,
      "learning_rate": 0.0015,
      "loss": 1.6815,
      "step": 4975
    },
    {
      "epoch": 0.5248945147679325,
      "grad_norm": 0.6466884016990662,
      "learning_rate": 0.0015,
      "loss": 1.6359,
      "step": 4976
    },
    {
      "epoch": 0.525,
      "grad_norm": 0.5307595729827881,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 4977
    },
    {
      "epoch": 0.5251054852320675,
      "grad_norm": 0.6720360517501831,
      "learning_rate": 0.0015,
      "loss": 1.6455,
      "step": 4978
    },
    {
      "epoch": 0.525210970464135,
      "grad_norm": 0.5458463430404663,
      "learning_rate": 0.0015,
      "loss": 1.5929,
      "step": 4979
    },
    {
      "epoch": 0.5253164556962026,
      "grad_norm": 0.5916789770126343,
      "learning_rate": 0.0015,
      "loss": 1.6045,
      "step": 4980
    },
    {
      "epoch": 0.52542194092827,
      "grad_norm": 0.580977737903595,
      "learning_rate": 0.0015,
      "loss": 1.6505,
      "step": 4981
    },
    {
      "epoch": 0.5255274261603375,
      "grad_norm": 0.6447780728340149,
      "learning_rate": 0.0015,
      "loss": 1.6168,
      "step": 4982
    },
    {
      "epoch": 0.5256329113924051,
      "grad_norm": 0.5232994556427002,
      "learning_rate": 0.0015,
      "loss": 1.6191,
      "step": 4983
    },
    {
      "epoch": 0.5257383966244725,
      "grad_norm": 0.8021718859672546,
      "learning_rate": 0.0015,
      "loss": 1.6183,
      "step": 4984
    },
    {
      "epoch": 0.5258438818565401,
      "grad_norm": 0.5380504131317139,
      "learning_rate": 0.0015,
      "loss": 1.6314,
      "step": 4985
    },
    {
      "epoch": 0.5259493670886076,
      "grad_norm": 0.7381505370140076,
      "learning_rate": 0.0015,
      "loss": 1.6183,
      "step": 4986
    },
    {
      "epoch": 0.5260548523206751,
      "grad_norm": 0.7041741609573364,
      "learning_rate": 0.0015,
      "loss": 1.6035,
      "step": 4987
    },
    {
      "epoch": 0.5261603375527426,
      "grad_norm": 0.6164463758468628,
      "learning_rate": 0.0015,
      "loss": 1.6462,
      "step": 4988
    },
    {
      "epoch": 0.5262658227848102,
      "grad_norm": 0.5790919661521912,
      "learning_rate": 0.0015,
      "loss": 1.6714,
      "step": 4989
    },
    {
      "epoch": 0.5263713080168776,
      "grad_norm": 0.7210583090782166,
      "learning_rate": 0.0015,
      "loss": 1.641,
      "step": 4990
    },
    {
      "epoch": 0.5264767932489451,
      "grad_norm": 0.4991927444934845,
      "learning_rate": 0.0015,
      "loss": 1.6432,
      "step": 4991
    },
    {
      "epoch": 0.5265822784810127,
      "grad_norm": 0.6949573159217834,
      "learning_rate": 0.0015,
      "loss": 1.6356,
      "step": 4992
    },
    {
      "epoch": 0.5266877637130801,
      "grad_norm": 0.6759629845619202,
      "learning_rate": 0.0015,
      "loss": 1.6448,
      "step": 4993
    },
    {
      "epoch": 0.5267932489451477,
      "grad_norm": 0.6050535440444946,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 4994
    },
    {
      "epoch": 0.5268987341772152,
      "grad_norm": 0.7116928100585938,
      "learning_rate": 0.0015,
      "loss": 1.6533,
      "step": 4995
    },
    {
      "epoch": 0.5270042194092827,
      "grad_norm": 0.48686733841896057,
      "learning_rate": 0.0015,
      "loss": 1.6318,
      "step": 4996
    },
    {
      "epoch": 0.5271097046413502,
      "grad_norm": 0.5821570158004761,
      "learning_rate": 0.0015,
      "loss": 1.6185,
      "step": 4997
    },
    {
      "epoch": 0.5272151898734178,
      "grad_norm": 0.5504971146583557,
      "learning_rate": 0.0015,
      "loss": 1.6608,
      "step": 4998
    },
    {
      "epoch": 0.5273206751054852,
      "grad_norm": 0.5670160055160522,
      "learning_rate": 0.0015,
      "loss": 1.6732,
      "step": 4999
    },
    {
      "epoch": 0.5274261603375527,
      "grad_norm": 0.7006347179412842,
      "learning_rate": 0.0015,
      "loss": 1.6271,
      "step": 5000
    },
    {
      "epoch": 0.5275316455696203,
      "grad_norm": 0.5208625197410583,
      "learning_rate": 0.0015,
      "loss": 1.6649,
      "step": 5001
    },
    {
      "epoch": 0.5276371308016877,
      "grad_norm": 0.6777536869049072,
      "learning_rate": 0.0015,
      "loss": 1.6677,
      "step": 5002
    },
    {
      "epoch": 0.5277426160337553,
      "grad_norm": 0.7044886350631714,
      "learning_rate": 0.0015,
      "loss": 1.6461,
      "step": 5003
    },
    {
      "epoch": 0.5278481012658228,
      "grad_norm": 0.4330019950866699,
      "learning_rate": 0.0015,
      "loss": 1.626,
      "step": 5004
    },
    {
      "epoch": 0.5279535864978903,
      "grad_norm": 0.6689701676368713,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 5005
    },
    {
      "epoch": 0.5280590717299578,
      "grad_norm": 0.4450782835483551,
      "learning_rate": 0.0015,
      "loss": 1.5885,
      "step": 5006
    },
    {
      "epoch": 0.5281645569620254,
      "grad_norm": 0.7085813879966736,
      "learning_rate": 0.0015,
      "loss": 1.6255,
      "step": 5007
    },
    {
      "epoch": 0.5282700421940928,
      "grad_norm": 0.7168097496032715,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 5008
    },
    {
      "epoch": 0.5283755274261603,
      "grad_norm": 0.47027990221977234,
      "learning_rate": 0.0015,
      "loss": 1.6362,
      "step": 5009
    },
    {
      "epoch": 0.5284810126582279,
      "grad_norm": 0.633066713809967,
      "learning_rate": 0.0015,
      "loss": 1.6409,
      "step": 5010
    },
    {
      "epoch": 0.5285864978902953,
      "grad_norm": 0.6865650415420532,
      "learning_rate": 0.0015,
      "loss": 1.6142,
      "step": 5011
    },
    {
      "epoch": 0.5286919831223629,
      "grad_norm": 0.44040533900260925,
      "learning_rate": 0.0015,
      "loss": 1.6588,
      "step": 5012
    },
    {
      "epoch": 0.5287974683544304,
      "grad_norm": 0.6548410058021545,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 5013
    },
    {
      "epoch": 0.5289029535864979,
      "grad_norm": 0.6385186910629272,
      "learning_rate": 0.0015,
      "loss": 1.6215,
      "step": 5014
    },
    {
      "epoch": 0.5290084388185654,
      "grad_norm": 0.5002903938293457,
      "learning_rate": 0.0015,
      "loss": 1.5852,
      "step": 5015
    },
    {
      "epoch": 0.529113924050633,
      "grad_norm": 0.7450860738754272,
      "learning_rate": 0.0015,
      "loss": 1.6672,
      "step": 5016
    },
    {
      "epoch": 0.5292194092827004,
      "grad_norm": 0.5648825764656067,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 5017
    },
    {
      "epoch": 0.5293248945147679,
      "grad_norm": 0.6444371342658997,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5018
    },
    {
      "epoch": 0.5294303797468355,
      "grad_norm": 0.8067348599433899,
      "learning_rate": 0.0015,
      "loss": 1.6153,
      "step": 5019
    },
    {
      "epoch": 0.5295358649789029,
      "grad_norm": 0.5678833723068237,
      "learning_rate": 0.0015,
      "loss": 1.6489,
      "step": 5020
    },
    {
      "epoch": 0.5296413502109705,
      "grad_norm": 0.5519526600837708,
      "learning_rate": 0.0015,
      "loss": 1.6129,
      "step": 5021
    },
    {
      "epoch": 0.529746835443038,
      "grad_norm": 0.5630823373794556,
      "learning_rate": 0.0015,
      "loss": 1.6246,
      "step": 5022
    },
    {
      "epoch": 0.5298523206751055,
      "grad_norm": 0.48882824182510376,
      "learning_rate": 0.0015,
      "loss": 1.6213,
      "step": 5023
    },
    {
      "epoch": 0.529957805907173,
      "grad_norm": 0.5829674601554871,
      "learning_rate": 0.0015,
      "loss": 1.6195,
      "step": 5024
    },
    {
      "epoch": 0.5300632911392406,
      "grad_norm": 0.5589535236358643,
      "learning_rate": 0.0015,
      "loss": 1.6059,
      "step": 5025
    },
    {
      "epoch": 0.530168776371308,
      "grad_norm": 0.5363820195198059,
      "learning_rate": 0.0015,
      "loss": 1.6193,
      "step": 5026
    },
    {
      "epoch": 0.5302742616033755,
      "grad_norm": 0.5025756359100342,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 5027
    },
    {
      "epoch": 0.5303797468354431,
      "grad_norm": 0.5299649238586426,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 5028
    },
    {
      "epoch": 0.5304852320675105,
      "grad_norm": 0.552548348903656,
      "learning_rate": 0.0015,
      "loss": 1.585,
      "step": 5029
    },
    {
      "epoch": 0.5305907172995781,
      "grad_norm": 0.561339795589447,
      "learning_rate": 0.0015,
      "loss": 1.5962,
      "step": 5030
    },
    {
      "epoch": 0.5306962025316456,
      "grad_norm": 0.5364262461662292,
      "learning_rate": 0.0015,
      "loss": 1.6338,
      "step": 5031
    },
    {
      "epoch": 0.5308016877637131,
      "grad_norm": 0.6337389945983887,
      "learning_rate": 0.0015,
      "loss": 1.6588,
      "step": 5032
    },
    {
      "epoch": 0.5309071729957806,
      "grad_norm": 0.5296711325645447,
      "learning_rate": 0.0015,
      "loss": 1.6531,
      "step": 5033
    },
    {
      "epoch": 0.5310126582278482,
      "grad_norm": 0.5987748503684998,
      "learning_rate": 0.0015,
      "loss": 1.6448,
      "step": 5034
    },
    {
      "epoch": 0.5311181434599156,
      "grad_norm": 0.5677121877670288,
      "learning_rate": 0.0015,
      "loss": 1.6581,
      "step": 5035
    },
    {
      "epoch": 0.5312236286919831,
      "grad_norm": 0.5264080762863159,
      "learning_rate": 0.0015,
      "loss": 1.643,
      "step": 5036
    },
    {
      "epoch": 0.5313291139240506,
      "grad_norm": 0.5983706712722778,
      "learning_rate": 0.0015,
      "loss": 1.6172,
      "step": 5037
    },
    {
      "epoch": 0.5314345991561181,
      "grad_norm": 0.746576726436615,
      "learning_rate": 0.0015,
      "loss": 1.6786,
      "step": 5038
    },
    {
      "epoch": 0.5315400843881857,
      "grad_norm": 0.6579143404960632,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 5039
    },
    {
      "epoch": 0.5316455696202531,
      "grad_norm": 0.4936109781265259,
      "learning_rate": 0.0015,
      "loss": 1.6116,
      "step": 5040
    },
    {
      "epoch": 0.5317510548523207,
      "grad_norm": 0.565092146396637,
      "learning_rate": 0.0015,
      "loss": 1.6366,
      "step": 5041
    },
    {
      "epoch": 0.5318565400843882,
      "grad_norm": 0.5107846260070801,
      "learning_rate": 0.0015,
      "loss": 1.6411,
      "step": 5042
    },
    {
      "epoch": 0.5319620253164556,
      "grad_norm": 0.554541289806366,
      "learning_rate": 0.0015,
      "loss": 1.599,
      "step": 5043
    },
    {
      "epoch": 0.5320675105485232,
      "grad_norm": 0.7011889219284058,
      "learning_rate": 0.0015,
      "loss": 1.6418,
      "step": 5044
    },
    {
      "epoch": 0.5321729957805907,
      "grad_norm": 0.4913009703159332,
      "learning_rate": 0.0015,
      "loss": 1.6105,
      "step": 5045
    },
    {
      "epoch": 0.5322784810126582,
      "grad_norm": 0.5300242304801941,
      "learning_rate": 0.0015,
      "loss": 1.673,
      "step": 5046
    },
    {
      "epoch": 0.5323839662447257,
      "grad_norm": 0.5980767607688904,
      "learning_rate": 0.0015,
      "loss": 1.6262,
      "step": 5047
    },
    {
      "epoch": 0.5324894514767933,
      "grad_norm": 0.444725900888443,
      "learning_rate": 0.0015,
      "loss": 1.594,
      "step": 5048
    },
    {
      "epoch": 0.5325949367088607,
      "grad_norm": 0.5098974108695984,
      "learning_rate": 0.0015,
      "loss": 1.649,
      "step": 5049
    },
    {
      "epoch": 0.5327004219409283,
      "grad_norm": 0.49392300844192505,
      "learning_rate": 0.0015,
      "loss": 1.6269,
      "step": 5050
    },
    {
      "epoch": 0.5328059071729958,
      "grad_norm": 0.5042253732681274,
      "learning_rate": 0.0015,
      "loss": 1.6478,
      "step": 5051
    },
    {
      "epoch": 0.5329113924050632,
      "grad_norm": 0.5419065952301025,
      "learning_rate": 0.0015,
      "loss": 1.639,
      "step": 5052
    },
    {
      "epoch": 0.5330168776371308,
      "grad_norm": 0.5093225240707397,
      "learning_rate": 0.0015,
      "loss": 1.6463,
      "step": 5053
    },
    {
      "epoch": 0.5331223628691983,
      "grad_norm": 0.6130210161209106,
      "learning_rate": 0.0015,
      "loss": 1.6475,
      "step": 5054
    },
    {
      "epoch": 0.5332278481012658,
      "grad_norm": 0.6211649179458618,
      "learning_rate": 0.0015,
      "loss": 1.6419,
      "step": 5055
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.5461240410804749,
      "learning_rate": 0.0015,
      "loss": 1.637,
      "step": 5056
    },
    {
      "epoch": 0.5334388185654009,
      "grad_norm": 0.6642283201217651,
      "learning_rate": 0.0015,
      "loss": 1.6697,
      "step": 5057
    },
    {
      "epoch": 0.5335443037974683,
      "grad_norm": 0.6176104545593262,
      "learning_rate": 0.0015,
      "loss": 1.6579,
      "step": 5058
    },
    {
      "epoch": 0.5336497890295359,
      "grad_norm": 0.587188184261322,
      "learning_rate": 0.0015,
      "loss": 1.629,
      "step": 5059
    },
    {
      "epoch": 0.5337552742616034,
      "grad_norm": 0.6248210668563843,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 5060
    },
    {
      "epoch": 0.5338607594936708,
      "grad_norm": 0.49455729126930237,
      "learning_rate": 0.0015,
      "loss": 1.6565,
      "step": 5061
    },
    {
      "epoch": 0.5339662447257384,
      "grad_norm": 0.7127228379249573,
      "learning_rate": 0.0015,
      "loss": 1.6056,
      "step": 5062
    },
    {
      "epoch": 0.5340717299578059,
      "grad_norm": 0.5559295415878296,
      "learning_rate": 0.0015,
      "loss": 1.614,
      "step": 5063
    },
    {
      "epoch": 0.5341772151898734,
      "grad_norm": 0.5910193920135498,
      "learning_rate": 0.0015,
      "loss": 1.6795,
      "step": 5064
    },
    {
      "epoch": 0.5342827004219409,
      "grad_norm": 0.8118014335632324,
      "learning_rate": 0.0015,
      "loss": 1.6185,
      "step": 5065
    },
    {
      "epoch": 0.5343881856540085,
      "grad_norm": 0.6377518177032471,
      "learning_rate": 0.0015,
      "loss": 1.6347,
      "step": 5066
    },
    {
      "epoch": 0.5344936708860759,
      "grad_norm": 0.5281541347503662,
      "learning_rate": 0.0015,
      "loss": 1.623,
      "step": 5067
    },
    {
      "epoch": 0.5345991561181435,
      "grad_norm": 0.7031581997871399,
      "learning_rate": 0.0015,
      "loss": 1.6607,
      "step": 5068
    },
    {
      "epoch": 0.534704641350211,
      "grad_norm": 0.48935553431510925,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 5069
    },
    {
      "epoch": 0.5348101265822784,
      "grad_norm": 0.8359583020210266,
      "learning_rate": 0.0015,
      "loss": 1.6005,
      "step": 5070
    },
    {
      "epoch": 0.534915611814346,
      "grad_norm": 0.897618293762207,
      "learning_rate": 0.0015,
      "loss": 1.6434,
      "step": 5071
    },
    {
      "epoch": 0.5350210970464135,
      "grad_norm": 0.4645049571990967,
      "learning_rate": 0.0015,
      "loss": 1.6397,
      "step": 5072
    },
    {
      "epoch": 0.535126582278481,
      "grad_norm": 0.790713906288147,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5073
    },
    {
      "epoch": 0.5352320675105485,
      "grad_norm": 0.5804443359375,
      "learning_rate": 0.0015,
      "loss": 1.6751,
      "step": 5074
    },
    {
      "epoch": 0.5353375527426161,
      "grad_norm": 0.5091162323951721,
      "learning_rate": 0.0015,
      "loss": 1.6703,
      "step": 5075
    },
    {
      "epoch": 0.5354430379746835,
      "grad_norm": 0.5991224646568298,
      "learning_rate": 0.0015,
      "loss": 1.6468,
      "step": 5076
    },
    {
      "epoch": 0.5355485232067511,
      "grad_norm": 0.5701445937156677,
      "learning_rate": 0.0015,
      "loss": 1.6427,
      "step": 5077
    },
    {
      "epoch": 0.5356540084388186,
      "grad_norm": 0.4461466073989868,
      "learning_rate": 0.0015,
      "loss": 1.653,
      "step": 5078
    },
    {
      "epoch": 0.535759493670886,
      "grad_norm": 0.5141837000846863,
      "learning_rate": 0.0015,
      "loss": 1.6234,
      "step": 5079
    },
    {
      "epoch": 0.5358649789029536,
      "grad_norm": 0.5066314339637756,
      "learning_rate": 0.0015,
      "loss": 1.6366,
      "step": 5080
    },
    {
      "epoch": 0.5359704641350211,
      "grad_norm": 0.5522492527961731,
      "learning_rate": 0.0015,
      "loss": 1.6229,
      "step": 5081
    },
    {
      "epoch": 0.5360759493670886,
      "grad_norm": 0.4731145203113556,
      "learning_rate": 0.0015,
      "loss": 1.6476,
      "step": 5082
    },
    {
      "epoch": 0.5361814345991561,
      "grad_norm": 0.6168615221977234,
      "learning_rate": 0.0015,
      "loss": 1.6154,
      "step": 5083
    },
    {
      "epoch": 0.5362869198312237,
      "grad_norm": 0.6220751404762268,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 5084
    },
    {
      "epoch": 0.5363924050632911,
      "grad_norm": 0.5353846549987793,
      "learning_rate": 0.0015,
      "loss": 1.6454,
      "step": 5085
    },
    {
      "epoch": 0.5364978902953587,
      "grad_norm": 0.5959115624427795,
      "learning_rate": 0.0015,
      "loss": 1.5894,
      "step": 5086
    },
    {
      "epoch": 0.5366033755274262,
      "grad_norm": 0.6136371493339539,
      "learning_rate": 0.0015,
      "loss": 1.594,
      "step": 5087
    },
    {
      "epoch": 0.5367088607594936,
      "grad_norm": 0.5750802159309387,
      "learning_rate": 0.0015,
      "loss": 1.6766,
      "step": 5088
    },
    {
      "epoch": 0.5368143459915612,
      "grad_norm": 0.6107059717178345,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 5089
    },
    {
      "epoch": 0.5369198312236287,
      "grad_norm": 0.5770487189292908,
      "learning_rate": 0.0015,
      "loss": 1.6736,
      "step": 5090
    },
    {
      "epoch": 0.5370253164556962,
      "grad_norm": 0.49546897411346436,
      "learning_rate": 0.0015,
      "loss": 1.6664,
      "step": 5091
    },
    {
      "epoch": 0.5371308016877637,
      "grad_norm": 0.6239323616027832,
      "learning_rate": 0.0015,
      "loss": 1.5621,
      "step": 5092
    },
    {
      "epoch": 0.5372362869198313,
      "grad_norm": 0.5572430491447449,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 5093
    },
    {
      "epoch": 0.5373417721518987,
      "grad_norm": 0.5014874935150146,
      "learning_rate": 0.0015,
      "loss": 1.6049,
      "step": 5094
    },
    {
      "epoch": 0.5374472573839663,
      "grad_norm": 0.5005751252174377,
      "learning_rate": 0.0015,
      "loss": 1.6052,
      "step": 5095
    },
    {
      "epoch": 0.5375527426160338,
      "grad_norm": 0.5073537826538086,
      "learning_rate": 0.0015,
      "loss": 1.6233,
      "step": 5096
    },
    {
      "epoch": 0.5376582278481012,
      "grad_norm": 0.5209341049194336,
      "learning_rate": 0.0015,
      "loss": 1.6447,
      "step": 5097
    },
    {
      "epoch": 0.5377637130801688,
      "grad_norm": 0.5044030547142029,
      "learning_rate": 0.0015,
      "loss": 1.6346,
      "step": 5098
    },
    {
      "epoch": 0.5378691983122363,
      "grad_norm": 0.5082756280899048,
      "learning_rate": 0.0015,
      "loss": 1.6456,
      "step": 5099
    },
    {
      "epoch": 0.5379746835443038,
      "grad_norm": 0.5281136631965637,
      "learning_rate": 0.0015,
      "loss": 1.5906,
      "step": 5100
    },
    {
      "epoch": 0.5380801687763713,
      "grad_norm": 0.4830903112888336,
      "learning_rate": 0.0015,
      "loss": 1.6463,
      "step": 5101
    },
    {
      "epoch": 0.5381856540084389,
      "grad_norm": 0.5429915189743042,
      "learning_rate": 0.0015,
      "loss": 1.6571,
      "step": 5102
    },
    {
      "epoch": 0.5382911392405063,
      "grad_norm": 0.5040405988693237,
      "learning_rate": 0.0015,
      "loss": 1.6247,
      "step": 5103
    },
    {
      "epoch": 0.5383966244725739,
      "grad_norm": 0.7375873327255249,
      "learning_rate": 0.0015,
      "loss": 1.6126,
      "step": 5104
    },
    {
      "epoch": 0.5385021097046413,
      "grad_norm": 0.5359514355659485,
      "learning_rate": 0.0015,
      "loss": 1.5928,
      "step": 5105
    },
    {
      "epoch": 0.5386075949367088,
      "grad_norm": 0.5888318419456482,
      "learning_rate": 0.0015,
      "loss": 1.6362,
      "step": 5106
    },
    {
      "epoch": 0.5387130801687764,
      "grad_norm": 0.5835240483283997,
      "learning_rate": 0.0015,
      "loss": 1.679,
      "step": 5107
    },
    {
      "epoch": 0.5388185654008438,
      "grad_norm": 0.5429015755653381,
      "learning_rate": 0.0015,
      "loss": 1.6443,
      "step": 5108
    },
    {
      "epoch": 0.5389240506329114,
      "grad_norm": 0.6073294878005981,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 5109
    },
    {
      "epoch": 0.5390295358649789,
      "grad_norm": 0.59377521276474,
      "learning_rate": 0.0015,
      "loss": 1.593,
      "step": 5110
    },
    {
      "epoch": 0.5391350210970464,
      "grad_norm": 0.5483939051628113,
      "learning_rate": 0.0015,
      "loss": 1.6507,
      "step": 5111
    },
    {
      "epoch": 0.5392405063291139,
      "grad_norm": 0.5142219662666321,
      "learning_rate": 0.0015,
      "loss": 1.6261,
      "step": 5112
    },
    {
      "epoch": 0.5393459915611815,
      "grad_norm": 0.5235267877578735,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 5113
    },
    {
      "epoch": 0.5394514767932489,
      "grad_norm": 0.5899415016174316,
      "learning_rate": 0.0015,
      "loss": 1.6229,
      "step": 5114
    },
    {
      "epoch": 0.5395569620253164,
      "grad_norm": 0.5644646883010864,
      "learning_rate": 0.0015,
      "loss": 1.608,
      "step": 5115
    },
    {
      "epoch": 0.539662447257384,
      "grad_norm": 0.5044073462486267,
      "learning_rate": 0.0015,
      "loss": 1.6276,
      "step": 5116
    },
    {
      "epoch": 0.5397679324894514,
      "grad_norm": 0.6431909799575806,
      "learning_rate": 0.0015,
      "loss": 1.6187,
      "step": 5117
    },
    {
      "epoch": 0.539873417721519,
      "grad_norm": 0.6481715440750122,
      "learning_rate": 0.0015,
      "loss": 1.6481,
      "step": 5118
    },
    {
      "epoch": 0.5399789029535865,
      "grad_norm": 0.49285146594047546,
      "learning_rate": 0.0015,
      "loss": 1.6267,
      "step": 5119
    },
    {
      "epoch": 0.540084388185654,
      "grad_norm": 0.671585202217102,
      "learning_rate": 0.0015,
      "loss": 1.6086,
      "step": 5120
    },
    {
      "epoch": 0.5401898734177215,
      "grad_norm": 0.679915189743042,
      "learning_rate": 0.0015,
      "loss": 1.6235,
      "step": 5121
    },
    {
      "epoch": 0.5402953586497891,
      "grad_norm": 0.5546087026596069,
      "learning_rate": 0.0015,
      "loss": 1.6149,
      "step": 5122
    },
    {
      "epoch": 0.5404008438818565,
      "grad_norm": 0.8165748119354248,
      "learning_rate": 0.0015,
      "loss": 1.617,
      "step": 5123
    },
    {
      "epoch": 0.540506329113924,
      "grad_norm": 0.693085253238678,
      "learning_rate": 0.0015,
      "loss": 1.6316,
      "step": 5124
    },
    {
      "epoch": 0.5406118143459916,
      "grad_norm": 0.5697019696235657,
      "learning_rate": 0.0015,
      "loss": 1.6325,
      "step": 5125
    },
    {
      "epoch": 0.540717299578059,
      "grad_norm": 0.5882627367973328,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 5126
    },
    {
      "epoch": 0.5408227848101266,
      "grad_norm": 0.616887629032135,
      "learning_rate": 0.0015,
      "loss": 1.6275,
      "step": 5127
    },
    {
      "epoch": 0.5409282700421941,
      "grad_norm": 0.46310409903526306,
      "learning_rate": 0.0015,
      "loss": 1.6289,
      "step": 5128
    },
    {
      "epoch": 0.5410337552742616,
      "grad_norm": 0.5147756338119507,
      "learning_rate": 0.0015,
      "loss": 1.6384,
      "step": 5129
    },
    {
      "epoch": 0.5411392405063291,
      "grad_norm": 0.5554012060165405,
      "learning_rate": 0.0015,
      "loss": 1.6357,
      "step": 5130
    },
    {
      "epoch": 0.5412447257383967,
      "grad_norm": 0.5255956053733826,
      "learning_rate": 0.0015,
      "loss": 1.6538,
      "step": 5131
    },
    {
      "epoch": 0.5413502109704641,
      "grad_norm": 0.5103606581687927,
      "learning_rate": 0.0015,
      "loss": 1.6756,
      "step": 5132
    },
    {
      "epoch": 0.5414556962025316,
      "grad_norm": 0.5117038488388062,
      "learning_rate": 0.0015,
      "loss": 1.6254,
      "step": 5133
    },
    {
      "epoch": 0.5415611814345992,
      "grad_norm": 0.532572865486145,
      "learning_rate": 0.0015,
      "loss": 1.6394,
      "step": 5134
    },
    {
      "epoch": 0.5416666666666666,
      "grad_norm": 0.5013951063156128,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 5135
    },
    {
      "epoch": 0.5417721518987342,
      "grad_norm": 0.5052446126937866,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 5136
    },
    {
      "epoch": 0.5418776371308017,
      "grad_norm": 0.4503644108772278,
      "learning_rate": 0.0015,
      "loss": 1.6458,
      "step": 5137
    },
    {
      "epoch": 0.5419831223628692,
      "grad_norm": 0.4754565954208374,
      "learning_rate": 0.0015,
      "loss": 1.6398,
      "step": 5138
    },
    {
      "epoch": 0.5420886075949367,
      "grad_norm": 0.4887687563896179,
      "learning_rate": 0.0015,
      "loss": 1.642,
      "step": 5139
    },
    {
      "epoch": 0.5421940928270043,
      "grad_norm": 0.5226295590400696,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 5140
    },
    {
      "epoch": 0.5422995780590717,
      "grad_norm": 0.5884148478507996,
      "learning_rate": 0.0015,
      "loss": 1.6196,
      "step": 5141
    },
    {
      "epoch": 0.5424050632911392,
      "grad_norm": 0.6582362055778503,
      "learning_rate": 0.0015,
      "loss": 1.6639,
      "step": 5142
    },
    {
      "epoch": 0.5425105485232068,
      "grad_norm": 0.514098048210144,
      "learning_rate": 0.0015,
      "loss": 1.6572,
      "step": 5143
    },
    {
      "epoch": 0.5426160337552742,
      "grad_norm": 0.5832661986351013,
      "learning_rate": 0.0015,
      "loss": 1.6223,
      "step": 5144
    },
    {
      "epoch": 0.5427215189873418,
      "grad_norm": 0.6021525263786316,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 5145
    },
    {
      "epoch": 0.5428270042194093,
      "grad_norm": 0.5034875273704529,
      "learning_rate": 0.0015,
      "loss": 1.6254,
      "step": 5146
    },
    {
      "epoch": 0.5429324894514768,
      "grad_norm": 0.5350997447967529,
      "learning_rate": 0.0015,
      "loss": 1.6697,
      "step": 5147
    },
    {
      "epoch": 0.5430379746835443,
      "grad_norm": 0.5650411248207092,
      "learning_rate": 0.0015,
      "loss": 1.603,
      "step": 5148
    },
    {
      "epoch": 0.5431434599156119,
      "grad_norm": 0.4724980890750885,
      "learning_rate": 0.0015,
      "loss": 1.6373,
      "step": 5149
    },
    {
      "epoch": 0.5432489451476793,
      "grad_norm": 0.5728057622909546,
      "learning_rate": 0.0015,
      "loss": 1.6602,
      "step": 5150
    },
    {
      "epoch": 0.5433544303797468,
      "grad_norm": 0.573001503944397,
      "learning_rate": 0.0015,
      "loss": 1.6035,
      "step": 5151
    },
    {
      "epoch": 0.5434599156118144,
      "grad_norm": 0.6479495167732239,
      "learning_rate": 0.0015,
      "loss": 1.6581,
      "step": 5152
    },
    {
      "epoch": 0.5435654008438818,
      "grad_norm": 0.6770612001419067,
      "learning_rate": 0.0015,
      "loss": 1.6246,
      "step": 5153
    },
    {
      "epoch": 0.5436708860759494,
      "grad_norm": 0.5020831823348999,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 5154
    },
    {
      "epoch": 0.5437763713080169,
      "grad_norm": 0.8978963494300842,
      "learning_rate": 0.0015,
      "loss": 1.6686,
      "step": 5155
    },
    {
      "epoch": 0.5438818565400844,
      "grad_norm": 0.8261247873306274,
      "learning_rate": 0.0015,
      "loss": 1.604,
      "step": 5156
    },
    {
      "epoch": 0.5439873417721519,
      "grad_norm": 0.5348539352416992,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5157
    },
    {
      "epoch": 0.5440928270042195,
      "grad_norm": 0.952217698097229,
      "learning_rate": 0.0015,
      "loss": 1.6277,
      "step": 5158
    },
    {
      "epoch": 0.5441983122362869,
      "grad_norm": 0.9551669955253601,
      "learning_rate": 0.0015,
      "loss": 1.6583,
      "step": 5159
    },
    {
      "epoch": 0.5443037974683544,
      "grad_norm": 0.5607153177261353,
      "learning_rate": 0.0015,
      "loss": 1.6342,
      "step": 5160
    },
    {
      "epoch": 0.544409282700422,
      "grad_norm": 1.0147234201431274,
      "learning_rate": 0.0015,
      "loss": 1.6858,
      "step": 5161
    },
    {
      "epoch": 0.5445147679324894,
      "grad_norm": 0.7260619401931763,
      "learning_rate": 0.0015,
      "loss": 1.6721,
      "step": 5162
    },
    {
      "epoch": 0.544620253164557,
      "grad_norm": 0.7279376983642578,
      "learning_rate": 0.0015,
      "loss": 1.6074,
      "step": 5163
    },
    {
      "epoch": 0.5447257383966245,
      "grad_norm": 0.9652494788169861,
      "learning_rate": 0.0015,
      "loss": 1.6398,
      "step": 5164
    },
    {
      "epoch": 0.544831223628692,
      "grad_norm": 0.6029044985771179,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 5165
    },
    {
      "epoch": 0.5449367088607595,
      "grad_norm": 0.9093937873840332,
      "learning_rate": 0.0015,
      "loss": 1.607,
      "step": 5166
    },
    {
      "epoch": 0.5450421940928271,
      "grad_norm": 0.808569610118866,
      "learning_rate": 0.0015,
      "loss": 1.644,
      "step": 5167
    },
    {
      "epoch": 0.5451476793248945,
      "grad_norm": 0.5974361300468445,
      "learning_rate": 0.0015,
      "loss": 1.6277,
      "step": 5168
    },
    {
      "epoch": 0.545253164556962,
      "grad_norm": 0.6575527787208557,
      "learning_rate": 0.0015,
      "loss": 1.6639,
      "step": 5169
    },
    {
      "epoch": 0.5453586497890295,
      "grad_norm": 0.5841599106788635,
      "learning_rate": 0.0015,
      "loss": 1.5768,
      "step": 5170
    },
    {
      "epoch": 0.545464135021097,
      "grad_norm": 0.6911558508872986,
      "learning_rate": 0.0015,
      "loss": 1.6134,
      "step": 5171
    },
    {
      "epoch": 0.5455696202531646,
      "grad_norm": 0.6017004251480103,
      "learning_rate": 0.0015,
      "loss": 1.6437,
      "step": 5172
    },
    {
      "epoch": 0.545675105485232,
      "grad_norm": 0.5507343411445618,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 5173
    },
    {
      "epoch": 0.5457805907172996,
      "grad_norm": 0.6274258494377136,
      "learning_rate": 0.0015,
      "loss": 1.6637,
      "step": 5174
    },
    {
      "epoch": 0.5458860759493671,
      "grad_norm": 0.6216654181480408,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 5175
    },
    {
      "epoch": 0.5459915611814345,
      "grad_norm": 0.6118292212486267,
      "learning_rate": 0.0015,
      "loss": 1.6315,
      "step": 5176
    },
    {
      "epoch": 0.5460970464135021,
      "grad_norm": 0.5252754092216492,
      "learning_rate": 0.0015,
      "loss": 1.6619,
      "step": 5177
    },
    {
      "epoch": 0.5462025316455696,
      "grad_norm": 0.5460687875747681,
      "learning_rate": 0.0015,
      "loss": 1.6939,
      "step": 5178
    },
    {
      "epoch": 0.5463080168776371,
      "grad_norm": 0.6485275626182556,
      "learning_rate": 0.0015,
      "loss": 1.6295,
      "step": 5179
    },
    {
      "epoch": 0.5464135021097046,
      "grad_norm": 0.7728783488273621,
      "learning_rate": 0.0015,
      "loss": 1.6491,
      "step": 5180
    },
    {
      "epoch": 0.5465189873417722,
      "grad_norm": 0.6418474316596985,
      "learning_rate": 0.0015,
      "loss": 1.6404,
      "step": 5181
    },
    {
      "epoch": 0.5466244725738396,
      "grad_norm": 0.6431019902229309,
      "learning_rate": 0.0015,
      "loss": 1.633,
      "step": 5182
    },
    {
      "epoch": 0.5467299578059072,
      "grad_norm": 0.5730225443840027,
      "learning_rate": 0.0015,
      "loss": 1.6144,
      "step": 5183
    },
    {
      "epoch": 0.5468354430379747,
      "grad_norm": 0.5020373463630676,
      "learning_rate": 0.0015,
      "loss": 1.6465,
      "step": 5184
    },
    {
      "epoch": 0.5469409282700421,
      "grad_norm": 0.6241316199302673,
      "learning_rate": 0.0015,
      "loss": 1.6155,
      "step": 5185
    },
    {
      "epoch": 0.5470464135021097,
      "grad_norm": 0.6311151385307312,
      "learning_rate": 0.0015,
      "loss": 1.6609,
      "step": 5186
    },
    {
      "epoch": 0.5471518987341772,
      "grad_norm": 0.5612585544586182,
      "learning_rate": 0.0015,
      "loss": 1.6147,
      "step": 5187
    },
    {
      "epoch": 0.5472573839662447,
      "grad_norm": 0.6356328129768372,
      "learning_rate": 0.0015,
      "loss": 1.5925,
      "step": 5188
    },
    {
      "epoch": 0.5473628691983122,
      "grad_norm": 0.7096157670021057,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 5189
    },
    {
      "epoch": 0.5474683544303798,
      "grad_norm": 1.1665655374526978,
      "learning_rate": 0.0015,
      "loss": 1.6176,
      "step": 5190
    },
    {
      "epoch": 0.5475738396624472,
      "grad_norm": 0.582997739315033,
      "learning_rate": 0.0015,
      "loss": 1.645,
      "step": 5191
    },
    {
      "epoch": 0.5476793248945148,
      "grad_norm": 0.7276178002357483,
      "learning_rate": 0.0015,
      "loss": 1.6477,
      "step": 5192
    },
    {
      "epoch": 0.5477848101265823,
      "grad_norm": 0.6223065257072449,
      "learning_rate": 0.0015,
      "loss": 1.6278,
      "step": 5193
    },
    {
      "epoch": 0.5478902953586497,
      "grad_norm": 0.5812250971794128,
      "learning_rate": 0.0015,
      "loss": 1.655,
      "step": 5194
    },
    {
      "epoch": 0.5479957805907173,
      "grad_norm": 0.7441231608390808,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5195
    },
    {
      "epoch": 0.5481012658227848,
      "grad_norm": 0.5721290707588196,
      "learning_rate": 0.0015,
      "loss": 1.667,
      "step": 5196
    },
    {
      "epoch": 0.5482067510548523,
      "grad_norm": 0.5468820333480835,
      "learning_rate": 0.0015,
      "loss": 1.6474,
      "step": 5197
    },
    {
      "epoch": 0.5483122362869198,
      "grad_norm": 0.6002963185310364,
      "learning_rate": 0.0015,
      "loss": 1.6186,
      "step": 5198
    },
    {
      "epoch": 0.5484177215189874,
      "grad_norm": 0.5369439721107483,
      "learning_rate": 0.0015,
      "loss": 1.6254,
      "step": 5199
    },
    {
      "epoch": 0.5485232067510548,
      "grad_norm": 0.7350257039070129,
      "learning_rate": 0.0015,
      "loss": 1.6316,
      "step": 5200
    },
    {
      "epoch": 0.5486286919831224,
      "grad_norm": 0.558731198310852,
      "learning_rate": 0.0015,
      "loss": 1.5774,
      "step": 5201
    },
    {
      "epoch": 0.5487341772151899,
      "grad_norm": 0.5892989635467529,
      "learning_rate": 0.0015,
      "loss": 1.6431,
      "step": 5202
    },
    {
      "epoch": 0.5488396624472573,
      "grad_norm": 0.6152883172035217,
      "learning_rate": 0.0015,
      "loss": 1.6195,
      "step": 5203
    },
    {
      "epoch": 0.5489451476793249,
      "grad_norm": 0.5052489042282104,
      "learning_rate": 0.0015,
      "loss": 1.6132,
      "step": 5204
    },
    {
      "epoch": 0.5490506329113924,
      "grad_norm": 0.6244813203811646,
      "learning_rate": 0.0015,
      "loss": 1.6328,
      "step": 5205
    },
    {
      "epoch": 0.5491561181434599,
      "grad_norm": 0.6771020889282227,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 5206
    },
    {
      "epoch": 0.5492616033755274,
      "grad_norm": 0.5401136875152588,
      "learning_rate": 0.0015,
      "loss": 1.6074,
      "step": 5207
    },
    {
      "epoch": 0.549367088607595,
      "grad_norm": 0.5238175392150879,
      "learning_rate": 0.0015,
      "loss": 1.6482,
      "step": 5208
    },
    {
      "epoch": 0.5494725738396624,
      "grad_norm": 0.6082211136817932,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 5209
    },
    {
      "epoch": 0.54957805907173,
      "grad_norm": 0.5441169738769531,
      "learning_rate": 0.0015,
      "loss": 1.623,
      "step": 5210
    },
    {
      "epoch": 0.5496835443037975,
      "grad_norm": 0.5103268623352051,
      "learning_rate": 0.0015,
      "loss": 1.6481,
      "step": 5211
    },
    {
      "epoch": 0.549789029535865,
      "grad_norm": 0.45983317494392395,
      "learning_rate": 0.0015,
      "loss": 1.5892,
      "step": 5212
    },
    {
      "epoch": 0.5498945147679325,
      "grad_norm": 0.6080833673477173,
      "learning_rate": 0.0015,
      "loss": 1.6702,
      "step": 5213
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5307289958000183,
      "learning_rate": 0.0015,
      "loss": 1.6438,
      "step": 5214
    },
    {
      "epoch": 0.5501054852320675,
      "grad_norm": 0.5215746760368347,
      "learning_rate": 0.0015,
      "loss": 1.6425,
      "step": 5215
    },
    {
      "epoch": 0.550210970464135,
      "grad_norm": 0.5658318996429443,
      "learning_rate": 0.0015,
      "loss": 1.6604,
      "step": 5216
    },
    {
      "epoch": 0.5503164556962026,
      "grad_norm": 0.5127245187759399,
      "learning_rate": 0.0015,
      "loss": 1.6281,
      "step": 5217
    },
    {
      "epoch": 0.55042194092827,
      "grad_norm": 0.46836280822753906,
      "learning_rate": 0.0015,
      "loss": 1.6318,
      "step": 5218
    },
    {
      "epoch": 0.5505274261603376,
      "grad_norm": 0.5214895009994507,
      "learning_rate": 0.0015,
      "loss": 1.6509,
      "step": 5219
    },
    {
      "epoch": 0.5506329113924051,
      "grad_norm": 0.5331636667251587,
      "learning_rate": 0.0015,
      "loss": 1.5931,
      "step": 5220
    },
    {
      "epoch": 0.5507383966244725,
      "grad_norm": 0.5437734127044678,
      "learning_rate": 0.0015,
      "loss": 1.6069,
      "step": 5221
    },
    {
      "epoch": 0.5508438818565401,
      "grad_norm": 0.4964209496974945,
      "learning_rate": 0.0015,
      "loss": 1.6017,
      "step": 5222
    },
    {
      "epoch": 0.5509493670886076,
      "grad_norm": 0.5475440621376038,
      "learning_rate": 0.0015,
      "loss": 1.5937,
      "step": 5223
    },
    {
      "epoch": 0.5510548523206751,
      "grad_norm": 0.45664384961128235,
      "learning_rate": 0.0015,
      "loss": 1.6396,
      "step": 5224
    },
    {
      "epoch": 0.5511603375527426,
      "grad_norm": 0.5754068493843079,
      "learning_rate": 0.0015,
      "loss": 1.6341,
      "step": 5225
    },
    {
      "epoch": 0.5512658227848102,
      "grad_norm": 0.5156245827674866,
      "learning_rate": 0.0015,
      "loss": 1.6283,
      "step": 5226
    },
    {
      "epoch": 0.5513713080168776,
      "grad_norm": 0.47930315136909485,
      "learning_rate": 0.0015,
      "loss": 1.5954,
      "step": 5227
    },
    {
      "epoch": 0.5514767932489452,
      "grad_norm": 0.6200971007347107,
      "learning_rate": 0.0015,
      "loss": 1.6742,
      "step": 5228
    },
    {
      "epoch": 0.5515822784810127,
      "grad_norm": 0.540560781955719,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 5229
    },
    {
      "epoch": 0.5516877637130801,
      "grad_norm": 0.618925929069519,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 5230
    },
    {
      "epoch": 0.5517932489451477,
      "grad_norm": 0.6381429433822632,
      "learning_rate": 0.0015,
      "loss": 1.6592,
      "step": 5231
    },
    {
      "epoch": 0.5518987341772152,
      "grad_norm": 0.6361313462257385,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 5232
    },
    {
      "epoch": 0.5520042194092827,
      "grad_norm": 0.5562092065811157,
      "learning_rate": 0.0015,
      "loss": 1.5821,
      "step": 5233
    },
    {
      "epoch": 0.5521097046413502,
      "grad_norm": 0.6058759093284607,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 5234
    },
    {
      "epoch": 0.5522151898734177,
      "grad_norm": 0.5033120512962341,
      "learning_rate": 0.0015,
      "loss": 1.6419,
      "step": 5235
    },
    {
      "epoch": 0.5523206751054852,
      "grad_norm": 0.5114763975143433,
      "learning_rate": 0.0015,
      "loss": 1.6345,
      "step": 5236
    },
    {
      "epoch": 0.5524261603375528,
      "grad_norm": 0.5686569213867188,
      "learning_rate": 0.0015,
      "loss": 1.6294,
      "step": 5237
    },
    {
      "epoch": 0.5525316455696202,
      "grad_norm": 0.48996034264564514,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5238
    },
    {
      "epoch": 0.5526371308016877,
      "grad_norm": 0.5686507821083069,
      "learning_rate": 0.0015,
      "loss": 1.6487,
      "step": 5239
    },
    {
      "epoch": 0.5527426160337553,
      "grad_norm": 0.5463162660598755,
      "learning_rate": 0.0015,
      "loss": 1.6673,
      "step": 5240
    },
    {
      "epoch": 0.5528481012658227,
      "grad_norm": 0.5198691487312317,
      "learning_rate": 0.0015,
      "loss": 1.6433,
      "step": 5241
    },
    {
      "epoch": 0.5529535864978903,
      "grad_norm": 0.6054553985595703,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 5242
    },
    {
      "epoch": 0.5530590717299578,
      "grad_norm": 0.5022717714309692,
      "learning_rate": 0.0015,
      "loss": 1.6126,
      "step": 5243
    },
    {
      "epoch": 0.5531645569620253,
      "grad_norm": 0.564944326877594,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 5244
    },
    {
      "epoch": 0.5532700421940928,
      "grad_norm": 0.5960189700126648,
      "learning_rate": 0.0015,
      "loss": 1.6333,
      "step": 5245
    },
    {
      "epoch": 0.5533755274261604,
      "grad_norm": 0.5729047060012817,
      "learning_rate": 0.0015,
      "loss": 1.6345,
      "step": 5246
    },
    {
      "epoch": 0.5534810126582278,
      "grad_norm": 0.6358652114868164,
      "learning_rate": 0.0015,
      "loss": 1.6406,
      "step": 5247
    },
    {
      "epoch": 0.5535864978902953,
      "grad_norm": 0.6519100666046143,
      "learning_rate": 0.0015,
      "loss": 1.6376,
      "step": 5248
    },
    {
      "epoch": 0.5536919831223629,
      "grad_norm": 0.7309291362762451,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 5249
    },
    {
      "epoch": 0.5537974683544303,
      "grad_norm": 0.5511775612831116,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 5250
    },
    {
      "epoch": 0.5539029535864979,
      "grad_norm": 0.6198769211769104,
      "learning_rate": 0.0015,
      "loss": 1.659,
      "step": 5251
    },
    {
      "epoch": 0.5540084388185654,
      "grad_norm": 0.5296935439109802,
      "learning_rate": 0.0015,
      "loss": 1.6185,
      "step": 5252
    },
    {
      "epoch": 0.5541139240506329,
      "grad_norm": 0.557013988494873,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 5253
    },
    {
      "epoch": 0.5542194092827004,
      "grad_norm": 0.6444569230079651,
      "learning_rate": 0.0015,
      "loss": 1.6477,
      "step": 5254
    },
    {
      "epoch": 0.554324894514768,
      "grad_norm": 0.48016121983528137,
      "learning_rate": 0.0015,
      "loss": 1.6353,
      "step": 5255
    },
    {
      "epoch": 0.5544303797468354,
      "grad_norm": 0.6487542986869812,
      "learning_rate": 0.0015,
      "loss": 1.6025,
      "step": 5256
    },
    {
      "epoch": 0.554535864978903,
      "grad_norm": 0.7259236574172974,
      "learning_rate": 0.0015,
      "loss": 1.6449,
      "step": 5257
    },
    {
      "epoch": 0.5546413502109705,
      "grad_norm": 0.5721498727798462,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 5258
    },
    {
      "epoch": 0.5547468354430379,
      "grad_norm": 0.6241754293441772,
      "learning_rate": 0.0015,
      "loss": 1.651,
      "step": 5259
    },
    {
      "epoch": 0.5548523206751055,
      "grad_norm": 0.5566892027854919,
      "learning_rate": 0.0015,
      "loss": 1.6166,
      "step": 5260
    },
    {
      "epoch": 0.554957805907173,
      "grad_norm": 0.5055679678916931,
      "learning_rate": 0.0015,
      "loss": 1.6346,
      "step": 5261
    },
    {
      "epoch": 0.5550632911392405,
      "grad_norm": 0.5378907322883606,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 5262
    },
    {
      "epoch": 0.555168776371308,
      "grad_norm": 0.47854986786842346,
      "learning_rate": 0.0015,
      "loss": 1.6407,
      "step": 5263
    },
    {
      "epoch": 0.5552742616033756,
      "grad_norm": 0.5072697401046753,
      "learning_rate": 0.0015,
      "loss": 1.6116,
      "step": 5264
    },
    {
      "epoch": 0.555379746835443,
      "grad_norm": 0.5652428269386292,
      "learning_rate": 0.0015,
      "loss": 1.6292,
      "step": 5265
    },
    {
      "epoch": 0.5554852320675105,
      "grad_norm": 0.6590447425842285,
      "learning_rate": 0.0015,
      "loss": 1.629,
      "step": 5266
    },
    {
      "epoch": 0.5555907172995781,
      "grad_norm": 0.5257437229156494,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 5267
    },
    {
      "epoch": 0.5556962025316455,
      "grad_norm": 0.7118067741394043,
      "learning_rate": 0.0015,
      "loss": 1.6545,
      "step": 5268
    },
    {
      "epoch": 0.5558016877637131,
      "grad_norm": 0.6626205444335938,
      "learning_rate": 0.0015,
      "loss": 1.6533,
      "step": 5269
    },
    {
      "epoch": 0.5559071729957806,
      "grad_norm": 0.5884685516357422,
      "learning_rate": 0.0015,
      "loss": 1.6497,
      "step": 5270
    },
    {
      "epoch": 0.5560126582278481,
      "grad_norm": 0.7747166156768799,
      "learning_rate": 0.0015,
      "loss": 1.6286,
      "step": 5271
    },
    {
      "epoch": 0.5561181434599156,
      "grad_norm": 0.9125948548316956,
      "learning_rate": 0.0015,
      "loss": 1.6546,
      "step": 5272
    },
    {
      "epoch": 0.5562236286919832,
      "grad_norm": 0.7135335803031921,
      "learning_rate": 0.0015,
      "loss": 1.5997,
      "step": 5273
    },
    {
      "epoch": 0.5563291139240506,
      "grad_norm": 0.5398093461990356,
      "learning_rate": 0.0015,
      "loss": 1.5996,
      "step": 5274
    },
    {
      "epoch": 0.5564345991561181,
      "grad_norm": 0.6518300175666809,
      "learning_rate": 0.0015,
      "loss": 1.6135,
      "step": 5275
    },
    {
      "epoch": 0.5565400843881857,
      "grad_norm": 0.5473427772521973,
      "learning_rate": 0.0015,
      "loss": 1.6393,
      "step": 5276
    },
    {
      "epoch": 0.5566455696202531,
      "grad_norm": 0.5354215502738953,
      "learning_rate": 0.0015,
      "loss": 1.654,
      "step": 5277
    },
    {
      "epoch": 0.5567510548523207,
      "grad_norm": 0.5475490093231201,
      "learning_rate": 0.0015,
      "loss": 1.6302,
      "step": 5278
    },
    {
      "epoch": 0.5568565400843882,
      "grad_norm": 0.4854798913002014,
      "learning_rate": 0.0015,
      "loss": 1.642,
      "step": 5279
    },
    {
      "epoch": 0.5569620253164557,
      "grad_norm": 0.4936612844467163,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 5280
    },
    {
      "epoch": 0.5570675105485232,
      "grad_norm": 0.5744133591651917,
      "learning_rate": 0.0015,
      "loss": 1.6412,
      "step": 5281
    },
    {
      "epoch": 0.5571729957805908,
      "grad_norm": 0.4862012267112732,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 5282
    },
    {
      "epoch": 0.5572784810126582,
      "grad_norm": 0.5772653222084045,
      "learning_rate": 0.0015,
      "loss": 1.6024,
      "step": 5283
    },
    {
      "epoch": 0.5573839662447257,
      "grad_norm": 0.620054304599762,
      "learning_rate": 0.0015,
      "loss": 1.6391,
      "step": 5284
    },
    {
      "epoch": 0.5574894514767933,
      "grad_norm": 0.6145972013473511,
      "learning_rate": 0.0015,
      "loss": 1.6216,
      "step": 5285
    },
    {
      "epoch": 0.5575949367088607,
      "grad_norm": 0.47198721766471863,
      "learning_rate": 0.0015,
      "loss": 1.6235,
      "step": 5286
    },
    {
      "epoch": 0.5577004219409283,
      "grad_norm": 0.4644085764884949,
      "learning_rate": 0.0015,
      "loss": 1.6061,
      "step": 5287
    },
    {
      "epoch": 0.5578059071729958,
      "grad_norm": 0.4495280683040619,
      "learning_rate": 0.0015,
      "loss": 1.6333,
      "step": 5288
    },
    {
      "epoch": 0.5579113924050633,
      "grad_norm": 0.5854806900024414,
      "learning_rate": 0.0015,
      "loss": 1.6526,
      "step": 5289
    },
    {
      "epoch": 0.5580168776371308,
      "grad_norm": 0.5856172442436218,
      "learning_rate": 0.0015,
      "loss": 1.6654,
      "step": 5290
    },
    {
      "epoch": 0.5581223628691984,
      "grad_norm": 0.5038925409317017,
      "learning_rate": 0.0015,
      "loss": 1.644,
      "step": 5291
    },
    {
      "epoch": 0.5582278481012658,
      "grad_norm": 0.5857557654380798,
      "learning_rate": 0.0015,
      "loss": 1.6041,
      "step": 5292
    },
    {
      "epoch": 0.5583333333333333,
      "grad_norm": 0.5493942499160767,
      "learning_rate": 0.0015,
      "loss": 1.6316,
      "step": 5293
    },
    {
      "epoch": 0.5584388185654009,
      "grad_norm": 0.7882180213928223,
      "learning_rate": 0.0015,
      "loss": 1.652,
      "step": 5294
    },
    {
      "epoch": 0.5585443037974683,
      "grad_norm": 0.9469513297080994,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 5295
    },
    {
      "epoch": 0.5586497890295359,
      "grad_norm": 0.7776630520820618,
      "learning_rate": 0.0015,
      "loss": 1.6557,
      "step": 5296
    },
    {
      "epoch": 0.5587552742616034,
      "grad_norm": 0.5740671753883362,
      "learning_rate": 0.0015,
      "loss": 1.6239,
      "step": 5297
    },
    {
      "epoch": 0.5588607594936709,
      "grad_norm": 0.8411259055137634,
      "learning_rate": 0.0015,
      "loss": 1.6224,
      "step": 5298
    },
    {
      "epoch": 0.5589662447257384,
      "grad_norm": 0.5124502182006836,
      "learning_rate": 0.0015,
      "loss": 1.657,
      "step": 5299
    },
    {
      "epoch": 0.5590717299578059,
      "grad_norm": 0.6385250091552734,
      "learning_rate": 0.0015,
      "loss": 1.6495,
      "step": 5300
    },
    {
      "epoch": 0.5591772151898734,
      "grad_norm": 0.6861783266067505,
      "learning_rate": 0.0015,
      "loss": 1.6141,
      "step": 5301
    },
    {
      "epoch": 0.559282700421941,
      "grad_norm": 0.5415860414505005,
      "learning_rate": 0.0015,
      "loss": 1.6295,
      "step": 5302
    },
    {
      "epoch": 0.5593881856540084,
      "grad_norm": 0.5780659317970276,
      "learning_rate": 0.0015,
      "loss": 1.6418,
      "step": 5303
    },
    {
      "epoch": 0.5594936708860759,
      "grad_norm": 0.6026530265808105,
      "learning_rate": 0.0015,
      "loss": 1.6151,
      "step": 5304
    },
    {
      "epoch": 0.5595991561181435,
      "grad_norm": 0.46989232301712036,
      "learning_rate": 0.0015,
      "loss": 1.6531,
      "step": 5305
    },
    {
      "epoch": 0.5597046413502109,
      "grad_norm": 0.5512782335281372,
      "learning_rate": 0.0015,
      "loss": 1.6447,
      "step": 5306
    },
    {
      "epoch": 0.5598101265822785,
      "grad_norm": 0.5197358727455139,
      "learning_rate": 0.0015,
      "loss": 1.6308,
      "step": 5307
    },
    {
      "epoch": 0.559915611814346,
      "grad_norm": 0.4554623067378998,
      "learning_rate": 0.0015,
      "loss": 1.587,
      "step": 5308
    },
    {
      "epoch": 0.5600210970464135,
      "grad_norm": 0.47824883460998535,
      "learning_rate": 0.0015,
      "loss": 1.6189,
      "step": 5309
    },
    {
      "epoch": 0.560126582278481,
      "grad_norm": 0.5234121680259705,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 5310
    },
    {
      "epoch": 0.5602320675105485,
      "grad_norm": 0.5879049897193909,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5311
    },
    {
      "epoch": 0.560337552742616,
      "grad_norm": 0.47663185000419617,
      "learning_rate": 0.0015,
      "loss": 1.6143,
      "step": 5312
    },
    {
      "epoch": 0.5604430379746835,
      "grad_norm": 0.610676109790802,
      "learning_rate": 0.0015,
      "loss": 1.6077,
      "step": 5313
    },
    {
      "epoch": 0.5605485232067511,
      "grad_norm": 0.515486478805542,
      "learning_rate": 0.0015,
      "loss": 1.5981,
      "step": 5314
    },
    {
      "epoch": 0.5606540084388185,
      "grad_norm": 0.5926443934440613,
      "learning_rate": 0.0015,
      "loss": 1.6243,
      "step": 5315
    },
    {
      "epoch": 0.5607594936708861,
      "grad_norm": 0.6808208227157593,
      "learning_rate": 0.0015,
      "loss": 1.6473,
      "step": 5316
    },
    {
      "epoch": 0.5608649789029536,
      "grad_norm": 0.62575364112854,
      "learning_rate": 0.0015,
      "loss": 1.5944,
      "step": 5317
    },
    {
      "epoch": 0.560970464135021,
      "grad_norm": 0.5204060673713684,
      "learning_rate": 0.0015,
      "loss": 1.6593,
      "step": 5318
    },
    {
      "epoch": 0.5610759493670886,
      "grad_norm": 0.6099985241889954,
      "learning_rate": 0.0015,
      "loss": 1.6265,
      "step": 5319
    },
    {
      "epoch": 0.5611814345991561,
      "grad_norm": 0.7034229636192322,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5320
    },
    {
      "epoch": 0.5612869198312236,
      "grad_norm": 0.5411562919616699,
      "learning_rate": 0.0015,
      "loss": 1.6601,
      "step": 5321
    },
    {
      "epoch": 0.5613924050632911,
      "grad_norm": 0.6399033665657043,
      "learning_rate": 0.0015,
      "loss": 1.6489,
      "step": 5322
    },
    {
      "epoch": 0.5614978902953587,
      "grad_norm": 0.535225510597229,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 5323
    },
    {
      "epoch": 0.5616033755274261,
      "grad_norm": 0.5888151526451111,
      "learning_rate": 0.0015,
      "loss": 1.6392,
      "step": 5324
    },
    {
      "epoch": 0.5617088607594937,
      "grad_norm": 0.6667788624763489,
      "learning_rate": 0.0015,
      "loss": 1.619,
      "step": 5325
    },
    {
      "epoch": 0.5618143459915612,
      "grad_norm": 0.7977067232131958,
      "learning_rate": 0.0015,
      "loss": 1.6292,
      "step": 5326
    },
    {
      "epoch": 0.5619198312236287,
      "grad_norm": 0.5076849460601807,
      "learning_rate": 0.0015,
      "loss": 1.6028,
      "step": 5327
    },
    {
      "epoch": 0.5620253164556962,
      "grad_norm": 0.5596925020217896,
      "learning_rate": 0.0015,
      "loss": 1.6012,
      "step": 5328
    },
    {
      "epoch": 0.5621308016877637,
      "grad_norm": 0.5293904542922974,
      "learning_rate": 0.0015,
      "loss": 1.6392,
      "step": 5329
    },
    {
      "epoch": 0.5622362869198312,
      "grad_norm": 0.598557710647583,
      "learning_rate": 0.0015,
      "loss": 1.6443,
      "step": 5330
    },
    {
      "epoch": 0.5623417721518987,
      "grad_norm": 0.5245181918144226,
      "learning_rate": 0.0015,
      "loss": 1.6348,
      "step": 5331
    },
    {
      "epoch": 0.5624472573839663,
      "grad_norm": 0.5318989157676697,
      "learning_rate": 0.0015,
      "loss": 1.6329,
      "step": 5332
    },
    {
      "epoch": 0.5625527426160337,
      "grad_norm": 0.49922168254852295,
      "learning_rate": 0.0015,
      "loss": 1.6056,
      "step": 5333
    },
    {
      "epoch": 0.5626582278481013,
      "grad_norm": 0.47323620319366455,
      "learning_rate": 0.0015,
      "loss": 1.6365,
      "step": 5334
    },
    {
      "epoch": 0.5627637130801688,
      "grad_norm": 0.6924019455909729,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 5335
    },
    {
      "epoch": 0.5628691983122363,
      "grad_norm": 0.6119868159294128,
      "learning_rate": 0.0015,
      "loss": 1.6402,
      "step": 5336
    },
    {
      "epoch": 0.5629746835443038,
      "grad_norm": 0.48253345489501953,
      "learning_rate": 0.0015,
      "loss": 1.6144,
      "step": 5337
    },
    {
      "epoch": 0.5630801687763713,
      "grad_norm": 0.5210509896278381,
      "learning_rate": 0.0015,
      "loss": 1.613,
      "step": 5338
    },
    {
      "epoch": 0.5631856540084388,
      "grad_norm": 0.5134233236312866,
      "learning_rate": 0.0015,
      "loss": 1.6311,
      "step": 5339
    },
    {
      "epoch": 0.5632911392405063,
      "grad_norm": 0.4461381435394287,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 5340
    },
    {
      "epoch": 0.5633966244725739,
      "grad_norm": 0.5335596203804016,
      "learning_rate": 0.0015,
      "loss": 1.6686,
      "step": 5341
    },
    {
      "epoch": 0.5635021097046413,
      "grad_norm": 0.45899948477745056,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 5342
    },
    {
      "epoch": 0.5636075949367089,
      "grad_norm": 0.5987926721572876,
      "learning_rate": 0.0015,
      "loss": 1.64,
      "step": 5343
    },
    {
      "epoch": 0.5637130801687764,
      "grad_norm": 0.524309515953064,
      "learning_rate": 0.0015,
      "loss": 1.6046,
      "step": 5344
    },
    {
      "epoch": 0.5638185654008439,
      "grad_norm": 0.4791477620601654,
      "learning_rate": 0.0015,
      "loss": 1.6378,
      "step": 5345
    },
    {
      "epoch": 0.5639240506329114,
      "grad_norm": 0.49584996700286865,
      "learning_rate": 0.0015,
      "loss": 1.6134,
      "step": 5346
    },
    {
      "epoch": 0.564029535864979,
      "grad_norm": 0.5700047612190247,
      "learning_rate": 0.0015,
      "loss": 1.5724,
      "step": 5347
    },
    {
      "epoch": 0.5641350210970464,
      "grad_norm": 0.4859113097190857,
      "learning_rate": 0.0015,
      "loss": 1.6003,
      "step": 5348
    },
    {
      "epoch": 0.5642405063291139,
      "grad_norm": 0.5591371059417725,
      "learning_rate": 0.0015,
      "loss": 1.6279,
      "step": 5349
    },
    {
      "epoch": 0.5643459915611815,
      "grad_norm": 0.5093802809715271,
      "learning_rate": 0.0015,
      "loss": 1.6032,
      "step": 5350
    },
    {
      "epoch": 0.5644514767932489,
      "grad_norm": 0.6617925763130188,
      "learning_rate": 0.0015,
      "loss": 1.6107,
      "step": 5351
    },
    {
      "epoch": 0.5645569620253165,
      "grad_norm": 0.6160645484924316,
      "learning_rate": 0.0015,
      "loss": 1.621,
      "step": 5352
    },
    {
      "epoch": 0.564662447257384,
      "grad_norm": 0.5530300736427307,
      "learning_rate": 0.0015,
      "loss": 1.6396,
      "step": 5353
    },
    {
      "epoch": 0.5647679324894515,
      "grad_norm": 0.5163722038269043,
      "learning_rate": 0.0015,
      "loss": 1.6255,
      "step": 5354
    },
    {
      "epoch": 0.564873417721519,
      "grad_norm": 0.5374283194541931,
      "learning_rate": 0.0015,
      "loss": 1.6338,
      "step": 5355
    },
    {
      "epoch": 0.5649789029535865,
      "grad_norm": 0.49130526185035706,
      "learning_rate": 0.0015,
      "loss": 1.5974,
      "step": 5356
    },
    {
      "epoch": 0.565084388185654,
      "grad_norm": 0.5461239814758301,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5357
    },
    {
      "epoch": 0.5651898734177215,
      "grad_norm": 0.5294837951660156,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 5358
    },
    {
      "epoch": 0.5652953586497891,
      "grad_norm": 0.49759408831596375,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 5359
    },
    {
      "epoch": 0.5654008438818565,
      "grad_norm": 0.5359417200088501,
      "learning_rate": 0.0015,
      "loss": 1.6665,
      "step": 5360
    },
    {
      "epoch": 0.5655063291139241,
      "grad_norm": 0.47370314598083496,
      "learning_rate": 0.0015,
      "loss": 1.6385,
      "step": 5361
    },
    {
      "epoch": 0.5656118143459916,
      "grad_norm": 0.5473756790161133,
      "learning_rate": 0.0015,
      "loss": 1.6366,
      "step": 5362
    },
    {
      "epoch": 0.565717299578059,
      "grad_norm": 0.45411214232444763,
      "learning_rate": 0.0015,
      "loss": 1.6972,
      "step": 5363
    },
    {
      "epoch": 0.5658227848101266,
      "grad_norm": 0.4845397472381592,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 5364
    },
    {
      "epoch": 0.5659282700421941,
      "grad_norm": 0.5022138953208923,
      "learning_rate": 0.0015,
      "loss": 1.6173,
      "step": 5365
    },
    {
      "epoch": 0.5660337552742616,
      "grad_norm": 0.6727507710456848,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 5366
    },
    {
      "epoch": 0.5661392405063291,
      "grad_norm": 0.532525360584259,
      "learning_rate": 0.0015,
      "loss": 1.6213,
      "step": 5367
    },
    {
      "epoch": 0.5662447257383966,
      "grad_norm": 0.4713844656944275,
      "learning_rate": 0.0015,
      "loss": 1.6208,
      "step": 5368
    },
    {
      "epoch": 0.5663502109704641,
      "grad_norm": 0.6724514365196228,
      "learning_rate": 0.0015,
      "loss": 1.58,
      "step": 5369
    },
    {
      "epoch": 0.5664556962025317,
      "grad_norm": 0.6096547245979309,
      "learning_rate": 0.0015,
      "loss": 1.6407,
      "step": 5370
    },
    {
      "epoch": 0.5665611814345991,
      "grad_norm": 0.5276410579681396,
      "learning_rate": 0.0015,
      "loss": 1.64,
      "step": 5371
    },
    {
      "epoch": 0.5666666666666667,
      "grad_norm": 0.5840702056884766,
      "learning_rate": 0.0015,
      "loss": 1.6208,
      "step": 5372
    },
    {
      "epoch": 0.5667721518987342,
      "grad_norm": 0.56119304895401,
      "learning_rate": 0.0015,
      "loss": 1.5967,
      "step": 5373
    },
    {
      "epoch": 0.5668776371308016,
      "grad_norm": 0.4861997365951538,
      "learning_rate": 0.0015,
      "loss": 1.6365,
      "step": 5374
    },
    {
      "epoch": 0.5669831223628692,
      "grad_norm": 0.5228224396705627,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5375
    },
    {
      "epoch": 0.5670886075949367,
      "grad_norm": 0.5005890727043152,
      "learning_rate": 0.0015,
      "loss": 1.5852,
      "step": 5376
    },
    {
      "epoch": 0.5671940928270042,
      "grad_norm": 0.5229915380477905,
      "learning_rate": 0.0015,
      "loss": 1.6726,
      "step": 5377
    },
    {
      "epoch": 0.5672995780590717,
      "grad_norm": 0.5707440972328186,
      "learning_rate": 0.0015,
      "loss": 1.5763,
      "step": 5378
    },
    {
      "epoch": 0.5674050632911393,
      "grad_norm": 0.5765089988708496,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 5379
    },
    {
      "epoch": 0.5675105485232067,
      "grad_norm": 0.5625852346420288,
      "learning_rate": 0.0015,
      "loss": 1.6154,
      "step": 5380
    },
    {
      "epoch": 0.5676160337552743,
      "grad_norm": 0.563363254070282,
      "learning_rate": 0.0015,
      "loss": 1.5958,
      "step": 5381
    },
    {
      "epoch": 0.5677215189873418,
      "grad_norm": 0.6581175923347473,
      "learning_rate": 0.0015,
      "loss": 1.5799,
      "step": 5382
    },
    {
      "epoch": 0.5678270042194092,
      "grad_norm": 0.5173786878585815,
      "learning_rate": 0.0015,
      "loss": 1.6262,
      "step": 5383
    },
    {
      "epoch": 0.5679324894514768,
      "grad_norm": 0.5384842753410339,
      "learning_rate": 0.0015,
      "loss": 1.6297,
      "step": 5384
    },
    {
      "epoch": 0.5680379746835443,
      "grad_norm": 0.7846764922142029,
      "learning_rate": 0.0015,
      "loss": 1.6302,
      "step": 5385
    },
    {
      "epoch": 0.5681434599156118,
      "grad_norm": 0.895563006401062,
      "learning_rate": 0.0015,
      "loss": 1.6506,
      "step": 5386
    },
    {
      "epoch": 0.5682489451476793,
      "grad_norm": 0.5403962135314941,
      "learning_rate": 0.0015,
      "loss": 1.6369,
      "step": 5387
    },
    {
      "epoch": 0.5683544303797469,
      "grad_norm": 0.9266017079353333,
      "learning_rate": 0.0015,
      "loss": 1.6305,
      "step": 5388
    },
    {
      "epoch": 0.5684599156118143,
      "grad_norm": 0.7382924556732178,
      "learning_rate": 0.0015,
      "loss": 1.6365,
      "step": 5389
    },
    {
      "epoch": 0.5685654008438819,
      "grad_norm": 0.6781361699104309,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 5390
    },
    {
      "epoch": 0.5686708860759494,
      "grad_norm": 1.043846607208252,
      "learning_rate": 0.0015,
      "loss": 1.6228,
      "step": 5391
    },
    {
      "epoch": 0.5687763713080168,
      "grad_norm": 0.4971105456352234,
      "learning_rate": 0.0015,
      "loss": 1.6222,
      "step": 5392
    },
    {
      "epoch": 0.5688818565400844,
      "grad_norm": 0.8194665312767029,
      "learning_rate": 0.0015,
      "loss": 1.6542,
      "step": 5393
    },
    {
      "epoch": 0.5689873417721519,
      "grad_norm": 0.6555083990097046,
      "learning_rate": 0.0015,
      "loss": 1.6323,
      "step": 5394
    },
    {
      "epoch": 0.5690928270042194,
      "grad_norm": 0.6417784094810486,
      "learning_rate": 0.0015,
      "loss": 1.6116,
      "step": 5395
    },
    {
      "epoch": 0.5691983122362869,
      "grad_norm": 0.7614520788192749,
      "learning_rate": 0.0015,
      "loss": 1.6269,
      "step": 5396
    },
    {
      "epoch": 0.5693037974683545,
      "grad_norm": 0.6229608654975891,
      "learning_rate": 0.0015,
      "loss": 1.655,
      "step": 5397
    },
    {
      "epoch": 0.5694092827004219,
      "grad_norm": 0.7862270474433899,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 5398
    },
    {
      "epoch": 0.5695147679324895,
      "grad_norm": 0.6895025968551636,
      "learning_rate": 0.0015,
      "loss": 1.604,
      "step": 5399
    },
    {
      "epoch": 0.569620253164557,
      "grad_norm": 0.6094022393226624,
      "learning_rate": 0.0015,
      "loss": 1.5997,
      "step": 5400
    },
    {
      "epoch": 0.5697257383966244,
      "grad_norm": 0.6876747608184814,
      "learning_rate": 0.0015,
      "loss": 1.6291,
      "step": 5401
    },
    {
      "epoch": 0.569831223628692,
      "grad_norm": 0.6110780239105225,
      "learning_rate": 0.0015,
      "loss": 1.6281,
      "step": 5402
    },
    {
      "epoch": 0.5699367088607595,
      "grad_norm": 0.5970811247825623,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 5403
    },
    {
      "epoch": 0.570042194092827,
      "grad_norm": 0.5428264141082764,
      "learning_rate": 0.0015,
      "loss": 1.6744,
      "step": 5404
    },
    {
      "epoch": 0.5701476793248945,
      "grad_norm": 0.6532614827156067,
      "learning_rate": 0.0015,
      "loss": 1.6178,
      "step": 5405
    },
    {
      "epoch": 0.5702531645569621,
      "grad_norm": 0.4627121388912201,
      "learning_rate": 0.0015,
      "loss": 1.584,
      "step": 5406
    },
    {
      "epoch": 0.5703586497890295,
      "grad_norm": 0.5733729004859924,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 5407
    },
    {
      "epoch": 0.570464135021097,
      "grad_norm": 0.5524259209632874,
      "learning_rate": 0.0015,
      "loss": 1.6239,
      "step": 5408
    },
    {
      "epoch": 0.5705696202531646,
      "grad_norm": 0.49608471989631653,
      "learning_rate": 0.0015,
      "loss": 1.6361,
      "step": 5409
    },
    {
      "epoch": 0.570675105485232,
      "grad_norm": 0.521388053894043,
      "learning_rate": 0.0015,
      "loss": 1.5991,
      "step": 5410
    },
    {
      "epoch": 0.5707805907172996,
      "grad_norm": 0.5712475776672363,
      "learning_rate": 0.0015,
      "loss": 1.6184,
      "step": 5411
    },
    {
      "epoch": 0.5708860759493671,
      "grad_norm": 0.47182101011276245,
      "learning_rate": 0.0015,
      "loss": 1.6278,
      "step": 5412
    },
    {
      "epoch": 0.5709915611814346,
      "grad_norm": 0.5659692883491516,
      "learning_rate": 0.0015,
      "loss": 1.6434,
      "step": 5413
    },
    {
      "epoch": 0.5710970464135021,
      "grad_norm": 0.7929262518882751,
      "learning_rate": 0.0015,
      "loss": 1.5769,
      "step": 5414
    },
    {
      "epoch": 0.5712025316455697,
      "grad_norm": 0.5173774361610413,
      "learning_rate": 0.0015,
      "loss": 1.6585,
      "step": 5415
    },
    {
      "epoch": 0.5713080168776371,
      "grad_norm": 0.5311462879180908,
      "learning_rate": 0.0015,
      "loss": 1.611,
      "step": 5416
    },
    {
      "epoch": 0.5714135021097047,
      "grad_norm": 0.5997411012649536,
      "learning_rate": 0.0015,
      "loss": 1.6346,
      "step": 5417
    },
    {
      "epoch": 0.5715189873417722,
      "grad_norm": 0.5332018733024597,
      "learning_rate": 0.0015,
      "loss": 1.6247,
      "step": 5418
    },
    {
      "epoch": 0.5716244725738396,
      "grad_norm": 0.5726498365402222,
      "learning_rate": 0.0015,
      "loss": 1.603,
      "step": 5419
    },
    {
      "epoch": 0.5717299578059072,
      "grad_norm": 0.550566554069519,
      "learning_rate": 0.0015,
      "loss": 1.6147,
      "step": 5420
    },
    {
      "epoch": 0.5718354430379747,
      "grad_norm": 0.4947754144668579,
      "learning_rate": 0.0015,
      "loss": 1.6544,
      "step": 5421
    },
    {
      "epoch": 0.5719409282700422,
      "grad_norm": 0.4676307737827301,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5422
    },
    {
      "epoch": 0.5720464135021097,
      "grad_norm": 0.4953192472457886,
      "learning_rate": 0.0015,
      "loss": 1.5834,
      "step": 5423
    },
    {
      "epoch": 0.5721518987341773,
      "grad_norm": 0.49735572934150696,
      "learning_rate": 0.0015,
      "loss": 1.601,
      "step": 5424
    },
    {
      "epoch": 0.5722573839662447,
      "grad_norm": 0.6120027303695679,
      "learning_rate": 0.0015,
      "loss": 1.6274,
      "step": 5425
    },
    {
      "epoch": 0.5723628691983123,
      "grad_norm": 0.5690551400184631,
      "learning_rate": 0.0015,
      "loss": 1.64,
      "step": 5426
    },
    {
      "epoch": 0.5724683544303798,
      "grad_norm": 0.6098531484603882,
      "learning_rate": 0.0015,
      "loss": 1.6364,
      "step": 5427
    },
    {
      "epoch": 0.5725738396624472,
      "grad_norm": 0.5622851252555847,
      "learning_rate": 0.0015,
      "loss": 1.616,
      "step": 5428
    },
    {
      "epoch": 0.5726793248945148,
      "grad_norm": 0.4972100555896759,
      "learning_rate": 0.0015,
      "loss": 1.6117,
      "step": 5429
    },
    {
      "epoch": 0.5727848101265823,
      "grad_norm": 0.6018389463424683,
      "learning_rate": 0.0015,
      "loss": 1.5915,
      "step": 5430
    },
    {
      "epoch": 0.5728902953586498,
      "grad_norm": 0.6689639687538147,
      "learning_rate": 0.0015,
      "loss": 1.6499,
      "step": 5431
    },
    {
      "epoch": 0.5729957805907173,
      "grad_norm": 0.5190061926841736,
      "learning_rate": 0.0015,
      "loss": 1.6036,
      "step": 5432
    },
    {
      "epoch": 0.5731012658227848,
      "grad_norm": 0.6291723847389221,
      "learning_rate": 0.0015,
      "loss": 1.5978,
      "step": 5433
    },
    {
      "epoch": 0.5732067510548523,
      "grad_norm": 0.6594454050064087,
      "learning_rate": 0.0015,
      "loss": 1.6667,
      "step": 5434
    },
    {
      "epoch": 0.5733122362869199,
      "grad_norm": 0.46946433186531067,
      "learning_rate": 0.0015,
      "loss": 1.5823,
      "step": 5435
    },
    {
      "epoch": 0.5734177215189873,
      "grad_norm": 0.8045597076416016,
      "learning_rate": 0.0015,
      "loss": 1.6093,
      "step": 5436
    },
    {
      "epoch": 0.5735232067510548,
      "grad_norm": 1.0402706861495972,
      "learning_rate": 0.0015,
      "loss": 1.5974,
      "step": 5437
    },
    {
      "epoch": 0.5736286919831224,
      "grad_norm": 0.4941970407962799,
      "learning_rate": 0.0015,
      "loss": 1.6368,
      "step": 5438
    },
    {
      "epoch": 0.5737341772151898,
      "grad_norm": 0.8785650730133057,
      "learning_rate": 0.0015,
      "loss": 1.6295,
      "step": 5439
    },
    {
      "epoch": 0.5738396624472574,
      "grad_norm": 0.8148274421691895,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5440
    },
    {
      "epoch": 0.5739451476793249,
      "grad_norm": 0.5330122709274292,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 5441
    },
    {
      "epoch": 0.5740506329113924,
      "grad_norm": 0.8619452714920044,
      "learning_rate": 0.0015,
      "loss": 1.6285,
      "step": 5442
    },
    {
      "epoch": 0.5741561181434599,
      "grad_norm": 0.6316558122634888,
      "learning_rate": 0.0015,
      "loss": 1.635,
      "step": 5443
    },
    {
      "epoch": 0.5742616033755275,
      "grad_norm": 0.7529088854789734,
      "learning_rate": 0.0015,
      "loss": 1.6314,
      "step": 5444
    },
    {
      "epoch": 0.5743670886075949,
      "grad_norm": 0.8501078486442566,
      "learning_rate": 0.0015,
      "loss": 1.606,
      "step": 5445
    },
    {
      "epoch": 0.5744725738396624,
      "grad_norm": 0.6048441529273987,
      "learning_rate": 0.0015,
      "loss": 1.6464,
      "step": 5446
    },
    {
      "epoch": 0.57457805907173,
      "grad_norm": 0.7871187925338745,
      "learning_rate": 0.0015,
      "loss": 1.6494,
      "step": 5447
    },
    {
      "epoch": 0.5746835443037974,
      "grad_norm": 0.7705832719802856,
      "learning_rate": 0.0015,
      "loss": 1.636,
      "step": 5448
    },
    {
      "epoch": 0.574789029535865,
      "grad_norm": 0.629921555519104,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 5449
    },
    {
      "epoch": 0.5748945147679325,
      "grad_norm": 0.606794536113739,
      "learning_rate": 0.0015,
      "loss": 1.6081,
      "step": 5450
    },
    {
      "epoch": 0.575,
      "grad_norm": 0.73259037733078,
      "learning_rate": 0.0015,
      "loss": 1.6302,
      "step": 5451
    },
    {
      "epoch": 0.5751054852320675,
      "grad_norm": 0.5201510787010193,
      "learning_rate": 0.0015,
      "loss": 1.5911,
      "step": 5452
    },
    {
      "epoch": 0.575210970464135,
      "grad_norm": 0.6375489234924316,
      "learning_rate": 0.0015,
      "loss": 1.6514,
      "step": 5453
    },
    {
      "epoch": 0.5753164556962025,
      "grad_norm": 0.5524571537971497,
      "learning_rate": 0.0015,
      "loss": 1.6288,
      "step": 5454
    },
    {
      "epoch": 0.57542194092827,
      "grad_norm": 0.6841790080070496,
      "learning_rate": 0.0015,
      "loss": 1.6723,
      "step": 5455
    },
    {
      "epoch": 0.5755274261603376,
      "grad_norm": 0.5184100270271301,
      "learning_rate": 0.0015,
      "loss": 1.6012,
      "step": 5456
    },
    {
      "epoch": 0.575632911392405,
      "grad_norm": 0.5849893689155579,
      "learning_rate": 0.0015,
      "loss": 1.6336,
      "step": 5457
    },
    {
      "epoch": 0.5757383966244726,
      "grad_norm": 0.5877789855003357,
      "learning_rate": 0.0015,
      "loss": 1.6124,
      "step": 5458
    },
    {
      "epoch": 0.5758438818565401,
      "grad_norm": 0.5175401568412781,
      "learning_rate": 0.0015,
      "loss": 1.6113,
      "step": 5459
    },
    {
      "epoch": 0.5759493670886076,
      "grad_norm": 0.553992509841919,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5460
    },
    {
      "epoch": 0.5760548523206751,
      "grad_norm": 0.4861304759979248,
      "learning_rate": 0.0015,
      "loss": 1.5881,
      "step": 5461
    },
    {
      "epoch": 0.5761603375527427,
      "grad_norm": 0.623470664024353,
      "learning_rate": 0.0015,
      "loss": 1.5811,
      "step": 5462
    },
    {
      "epoch": 0.5762658227848101,
      "grad_norm": 0.6538658738136292,
      "learning_rate": 0.0015,
      "loss": 1.6206,
      "step": 5463
    },
    {
      "epoch": 0.5763713080168776,
      "grad_norm": 0.4597054421901703,
      "learning_rate": 0.0015,
      "loss": 1.6034,
      "step": 5464
    },
    {
      "epoch": 0.5764767932489452,
      "grad_norm": 0.6882071495056152,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 5465
    },
    {
      "epoch": 0.5765822784810126,
      "grad_norm": 0.5916186571121216,
      "learning_rate": 0.0015,
      "loss": 1.6543,
      "step": 5466
    },
    {
      "epoch": 0.5766877637130802,
      "grad_norm": 0.5639544725418091,
      "learning_rate": 0.0015,
      "loss": 1.626,
      "step": 5467
    },
    {
      "epoch": 0.5767932489451477,
      "grad_norm": 0.7871806025505066,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 5468
    },
    {
      "epoch": 0.5768987341772152,
      "grad_norm": 0.5329530835151672,
      "learning_rate": 0.0015,
      "loss": 1.6369,
      "step": 5469
    },
    {
      "epoch": 0.5770042194092827,
      "grad_norm": 0.5972363948822021,
      "learning_rate": 0.0015,
      "loss": 1.6298,
      "step": 5470
    },
    {
      "epoch": 0.5771097046413503,
      "grad_norm": 0.5829533934593201,
      "learning_rate": 0.0015,
      "loss": 1.6347,
      "step": 5471
    },
    {
      "epoch": 0.5772151898734177,
      "grad_norm": 0.5822646617889404,
      "learning_rate": 0.0015,
      "loss": 1.6127,
      "step": 5472
    },
    {
      "epoch": 0.5773206751054852,
      "grad_norm": 0.5193725824356079,
      "learning_rate": 0.0015,
      "loss": 1.5926,
      "step": 5473
    },
    {
      "epoch": 0.5774261603375528,
      "grad_norm": 0.5308032035827637,
      "learning_rate": 0.0015,
      "loss": 1.5972,
      "step": 5474
    },
    {
      "epoch": 0.5775316455696202,
      "grad_norm": 0.548362135887146,
      "learning_rate": 0.0015,
      "loss": 1.6194,
      "step": 5475
    },
    {
      "epoch": 0.5776371308016878,
      "grad_norm": 0.5116716027259827,
      "learning_rate": 0.0015,
      "loss": 1.605,
      "step": 5476
    },
    {
      "epoch": 0.5777426160337553,
      "grad_norm": 0.5850107073783875,
      "learning_rate": 0.0015,
      "loss": 1.6641,
      "step": 5477
    },
    {
      "epoch": 0.5778481012658228,
      "grad_norm": 0.5661741495132446,
      "learning_rate": 0.0015,
      "loss": 1.6372,
      "step": 5478
    },
    {
      "epoch": 0.5779535864978903,
      "grad_norm": 0.5660033226013184,
      "learning_rate": 0.0015,
      "loss": 1.6333,
      "step": 5479
    },
    {
      "epoch": 0.5780590717299579,
      "grad_norm": 0.7832927107810974,
      "learning_rate": 0.0015,
      "loss": 1.6677,
      "step": 5480
    },
    {
      "epoch": 0.5781645569620253,
      "grad_norm": 0.6293277740478516,
      "learning_rate": 0.0015,
      "loss": 1.5881,
      "step": 5481
    },
    {
      "epoch": 0.5782700421940928,
      "grad_norm": 0.6669265627861023,
      "learning_rate": 0.0015,
      "loss": 1.6425,
      "step": 5482
    },
    {
      "epoch": 0.5783755274261604,
      "grad_norm": 0.8398557901382446,
      "learning_rate": 0.0015,
      "loss": 1.6499,
      "step": 5483
    },
    {
      "epoch": 0.5784810126582278,
      "grad_norm": 0.5337482690811157,
      "learning_rate": 0.0015,
      "loss": 1.6316,
      "step": 5484
    },
    {
      "epoch": 0.5785864978902954,
      "grad_norm": 1.0220376253128052,
      "learning_rate": 0.0015,
      "loss": 1.6422,
      "step": 5485
    },
    {
      "epoch": 0.5786919831223629,
      "grad_norm": 0.6867162585258484,
      "learning_rate": 0.0015,
      "loss": 1.6111,
      "step": 5486
    },
    {
      "epoch": 0.5787974683544304,
      "grad_norm": 0.783750057220459,
      "learning_rate": 0.0015,
      "loss": 1.6436,
      "step": 5487
    },
    {
      "epoch": 0.5789029535864979,
      "grad_norm": 0.9936336278915405,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5488
    },
    {
      "epoch": 0.5790084388185655,
      "grad_norm": 0.6602028608322144,
      "learning_rate": 0.0015,
      "loss": 1.6845,
      "step": 5489
    },
    {
      "epoch": 0.5791139240506329,
      "grad_norm": 0.8614243268966675,
      "learning_rate": 0.0015,
      "loss": 1.6211,
      "step": 5490
    },
    {
      "epoch": 0.5792194092827004,
      "grad_norm": 0.7624712586402893,
      "learning_rate": 0.0015,
      "loss": 1.6184,
      "step": 5491
    },
    {
      "epoch": 0.579324894514768,
      "grad_norm": 0.5247186422348022,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 5492
    },
    {
      "epoch": 0.5794303797468354,
      "grad_norm": 0.6216135025024414,
      "learning_rate": 0.0015,
      "loss": 1.638,
      "step": 5493
    },
    {
      "epoch": 0.579535864978903,
      "grad_norm": 0.5212397575378418,
      "learning_rate": 0.0015,
      "loss": 1.606,
      "step": 5494
    },
    {
      "epoch": 0.5796413502109705,
      "grad_norm": 0.5704751014709473,
      "learning_rate": 0.0015,
      "loss": 1.6519,
      "step": 5495
    },
    {
      "epoch": 0.579746835443038,
      "grad_norm": 0.5986801385879517,
      "learning_rate": 0.0015,
      "loss": 1.5896,
      "step": 5496
    },
    {
      "epoch": 0.5798523206751055,
      "grad_norm": 0.493880033493042,
      "learning_rate": 0.0015,
      "loss": 1.6191,
      "step": 5497
    },
    {
      "epoch": 0.5799578059071729,
      "grad_norm": 0.4813474714756012,
      "learning_rate": 0.0015,
      "loss": 1.6106,
      "step": 5498
    },
    {
      "epoch": 0.5800632911392405,
      "grad_norm": 0.5063060522079468,
      "learning_rate": 0.0015,
      "loss": 1.6582,
      "step": 5499
    },
    {
      "epoch": 0.580168776371308,
      "grad_norm": 0.45211276412010193,
      "learning_rate": 0.0015,
      "loss": 1.5961,
      "step": 5500
    },
    {
      "epoch": 0.5802742616033755,
      "grad_norm": 0.6631359457969666,
      "learning_rate": 0.0015,
      "loss": 1.6114,
      "step": 5501
    },
    {
      "epoch": 0.580379746835443,
      "grad_norm": 0.5923169255256653,
      "learning_rate": 0.0015,
      "loss": 1.6518,
      "step": 5502
    },
    {
      "epoch": 0.5804852320675106,
      "grad_norm": 0.531610906124115,
      "learning_rate": 0.0015,
      "loss": 1.6384,
      "step": 5503
    },
    {
      "epoch": 0.580590717299578,
      "grad_norm": 0.5216389894485474,
      "learning_rate": 0.0015,
      "loss": 1.6108,
      "step": 5504
    },
    {
      "epoch": 0.5806962025316456,
      "grad_norm": 0.4604436755180359,
      "learning_rate": 0.0015,
      "loss": 1.606,
      "step": 5505
    },
    {
      "epoch": 0.5808016877637131,
      "grad_norm": 0.5579452514648438,
      "learning_rate": 0.0015,
      "loss": 1.6573,
      "step": 5506
    },
    {
      "epoch": 0.5809071729957805,
      "grad_norm": 0.5045171976089478,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 5507
    },
    {
      "epoch": 0.5810126582278481,
      "grad_norm": 0.49698296189308167,
      "learning_rate": 0.0015,
      "loss": 1.5988,
      "step": 5508
    },
    {
      "epoch": 0.5811181434599156,
      "grad_norm": 0.49496710300445557,
      "learning_rate": 0.0015,
      "loss": 1.6378,
      "step": 5509
    },
    {
      "epoch": 0.5812236286919831,
      "grad_norm": 0.6719024777412415,
      "learning_rate": 0.0015,
      "loss": 1.612,
      "step": 5510
    },
    {
      "epoch": 0.5813291139240506,
      "grad_norm": 0.6405683159828186,
      "learning_rate": 0.0015,
      "loss": 1.5809,
      "step": 5511
    },
    {
      "epoch": 0.5814345991561182,
      "grad_norm": 0.5058594346046448,
      "learning_rate": 0.0015,
      "loss": 1.615,
      "step": 5512
    },
    {
      "epoch": 0.5815400843881856,
      "grad_norm": 0.6330063343048096,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 5513
    },
    {
      "epoch": 0.5816455696202532,
      "grad_norm": 0.49307113885879517,
      "learning_rate": 0.0015,
      "loss": 1.6381,
      "step": 5514
    },
    {
      "epoch": 0.5817510548523207,
      "grad_norm": 0.4965069591999054,
      "learning_rate": 0.0015,
      "loss": 1.6238,
      "step": 5515
    },
    {
      "epoch": 0.5818565400843881,
      "grad_norm": 0.5093645453453064,
      "learning_rate": 0.0015,
      "loss": 1.5887,
      "step": 5516
    },
    {
      "epoch": 0.5819620253164557,
      "grad_norm": 0.45555707812309265,
      "learning_rate": 0.0015,
      "loss": 1.6051,
      "step": 5517
    },
    {
      "epoch": 0.5820675105485232,
      "grad_norm": 0.4636574983596802,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5518
    },
    {
      "epoch": 0.5821729957805907,
      "grad_norm": 0.5562624931335449,
      "learning_rate": 0.0015,
      "loss": 1.6118,
      "step": 5519
    },
    {
      "epoch": 0.5822784810126582,
      "grad_norm": 0.5193811655044556,
      "learning_rate": 0.0015,
      "loss": 1.6555,
      "step": 5520
    },
    {
      "epoch": 0.5823839662447258,
      "grad_norm": 0.6242823004722595,
      "learning_rate": 0.0015,
      "loss": 1.6047,
      "step": 5521
    },
    {
      "epoch": 0.5824894514767932,
      "grad_norm": 0.6714182496070862,
      "learning_rate": 0.0015,
      "loss": 1.6045,
      "step": 5522
    },
    {
      "epoch": 0.5825949367088608,
      "grad_norm": 0.4450235962867737,
      "learning_rate": 0.0015,
      "loss": 1.5997,
      "step": 5523
    },
    {
      "epoch": 0.5827004219409283,
      "grad_norm": 0.581916093826294,
      "learning_rate": 0.0015,
      "loss": 1.626,
      "step": 5524
    },
    {
      "epoch": 0.5828059071729957,
      "grad_norm": 0.4704454243183136,
      "learning_rate": 0.0015,
      "loss": 1.608,
      "step": 5525
    },
    {
      "epoch": 0.5829113924050633,
      "grad_norm": 0.5212851762771606,
      "learning_rate": 0.0015,
      "loss": 1.6189,
      "step": 5526
    },
    {
      "epoch": 0.5830168776371308,
      "grad_norm": 0.5930542349815369,
      "learning_rate": 0.0015,
      "loss": 1.6078,
      "step": 5527
    },
    {
      "epoch": 0.5831223628691983,
      "grad_norm": 0.5354674458503723,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5528
    },
    {
      "epoch": 0.5832278481012658,
      "grad_norm": 0.6332305669784546,
      "learning_rate": 0.0015,
      "loss": 1.6052,
      "step": 5529
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.5576563477516174,
      "learning_rate": 0.0015,
      "loss": 1.6536,
      "step": 5530
    },
    {
      "epoch": 0.5834388185654008,
      "grad_norm": 0.6107875108718872,
      "learning_rate": 0.0015,
      "loss": 1.6279,
      "step": 5531
    },
    {
      "epoch": 0.5835443037974684,
      "grad_norm": 0.5109343528747559,
      "learning_rate": 0.0015,
      "loss": 1.5876,
      "step": 5532
    },
    {
      "epoch": 0.5836497890295359,
      "grad_norm": 0.6298612356185913,
      "learning_rate": 0.0015,
      "loss": 1.5928,
      "step": 5533
    },
    {
      "epoch": 0.5837552742616033,
      "grad_norm": 0.6461126208305359,
      "learning_rate": 0.0015,
      "loss": 1.5934,
      "step": 5534
    },
    {
      "epoch": 0.5838607594936709,
      "grad_norm": 0.5158522725105286,
      "learning_rate": 0.0015,
      "loss": 1.6065,
      "step": 5535
    },
    {
      "epoch": 0.5839662447257384,
      "grad_norm": 0.5445433855056763,
      "learning_rate": 0.0015,
      "loss": 1.6041,
      "step": 5536
    },
    {
      "epoch": 0.5840717299578059,
      "grad_norm": 0.6405567526817322,
      "learning_rate": 0.0015,
      "loss": 1.6052,
      "step": 5537
    },
    {
      "epoch": 0.5841772151898734,
      "grad_norm": 0.5018154978752136,
      "learning_rate": 0.0015,
      "loss": 1.6743,
      "step": 5538
    },
    {
      "epoch": 0.584282700421941,
      "grad_norm": 0.5532844066619873,
      "learning_rate": 0.0015,
      "loss": 1.6058,
      "step": 5539
    },
    {
      "epoch": 0.5843881856540084,
      "grad_norm": 0.6547574400901794,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 5540
    },
    {
      "epoch": 0.584493670886076,
      "grad_norm": 0.4973040521144867,
      "learning_rate": 0.0015,
      "loss": 1.627,
      "step": 5541
    },
    {
      "epoch": 0.5845991561181435,
      "grad_norm": 0.5852428674697876,
      "learning_rate": 0.0015,
      "loss": 1.6282,
      "step": 5542
    },
    {
      "epoch": 0.5847046413502109,
      "grad_norm": 0.6409162878990173,
      "learning_rate": 0.0015,
      "loss": 1.6331,
      "step": 5543
    },
    {
      "epoch": 0.5848101265822785,
      "grad_norm": 0.5296412706375122,
      "learning_rate": 0.0015,
      "loss": 1.6087,
      "step": 5544
    },
    {
      "epoch": 0.584915611814346,
      "grad_norm": 0.5370193123817444,
      "learning_rate": 0.0015,
      "loss": 1.6305,
      "step": 5545
    },
    {
      "epoch": 0.5850210970464135,
      "grad_norm": 0.6180422902107239,
      "learning_rate": 0.0015,
      "loss": 1.614,
      "step": 5546
    },
    {
      "epoch": 0.585126582278481,
      "grad_norm": 0.5282862186431885,
      "learning_rate": 0.0015,
      "loss": 1.6094,
      "step": 5547
    },
    {
      "epoch": 0.5852320675105486,
      "grad_norm": 0.5119280815124512,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 5548
    },
    {
      "epoch": 0.585337552742616,
      "grad_norm": 0.664736807346344,
      "learning_rate": 0.0015,
      "loss": 1.6351,
      "step": 5549
    },
    {
      "epoch": 0.5854430379746836,
      "grad_norm": 0.6273255944252014,
      "learning_rate": 0.0015,
      "loss": 1.6662,
      "step": 5550
    },
    {
      "epoch": 0.5855485232067511,
      "grad_norm": 0.4826644957065582,
      "learning_rate": 0.0015,
      "loss": 1.6528,
      "step": 5551
    },
    {
      "epoch": 0.5856540084388185,
      "grad_norm": 0.5993300676345825,
      "learning_rate": 0.0015,
      "loss": 1.6886,
      "step": 5552
    },
    {
      "epoch": 0.5857594936708861,
      "grad_norm": 0.4966449737548828,
      "learning_rate": 0.0015,
      "loss": 1.6166,
      "step": 5553
    },
    {
      "epoch": 0.5858649789029536,
      "grad_norm": 0.4663368761539459,
      "learning_rate": 0.0015,
      "loss": 1.6331,
      "step": 5554
    },
    {
      "epoch": 0.5859704641350211,
      "grad_norm": 0.46478691697120667,
      "learning_rate": 0.0015,
      "loss": 1.6354,
      "step": 5555
    },
    {
      "epoch": 0.5860759493670886,
      "grad_norm": 0.4704135060310364,
      "learning_rate": 0.0015,
      "loss": 1.6025,
      "step": 5556
    },
    {
      "epoch": 0.5861814345991562,
      "grad_norm": 0.484254390001297,
      "learning_rate": 0.0015,
      "loss": 1.6087,
      "step": 5557
    },
    {
      "epoch": 0.5862869198312236,
      "grad_norm": 0.4876977503299713,
      "learning_rate": 0.0015,
      "loss": 1.647,
      "step": 5558
    },
    {
      "epoch": 0.5863924050632912,
      "grad_norm": 0.5591520071029663,
      "learning_rate": 0.0015,
      "loss": 1.6153,
      "step": 5559
    },
    {
      "epoch": 0.5864978902953587,
      "grad_norm": 0.5457459688186646,
      "learning_rate": 0.0015,
      "loss": 1.6292,
      "step": 5560
    },
    {
      "epoch": 0.5866033755274261,
      "grad_norm": 0.5583513975143433,
      "learning_rate": 0.0015,
      "loss": 1.6203,
      "step": 5561
    },
    {
      "epoch": 0.5867088607594937,
      "grad_norm": 0.4454638957977295,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5562
    },
    {
      "epoch": 0.5868143459915611,
      "grad_norm": 0.6796378493309021,
      "learning_rate": 0.0015,
      "loss": 1.6119,
      "step": 5563
    },
    {
      "epoch": 0.5869198312236287,
      "grad_norm": 0.47859227657318115,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 5564
    },
    {
      "epoch": 0.5870253164556962,
      "grad_norm": 0.5425044894218445,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 5565
    },
    {
      "epoch": 0.5871308016877637,
      "grad_norm": 0.5822433233261108,
      "learning_rate": 0.0015,
      "loss": 1.6353,
      "step": 5566
    },
    {
      "epoch": 0.5872362869198312,
      "grad_norm": 0.4745939373970032,
      "learning_rate": 0.0015,
      "loss": 1.6471,
      "step": 5567
    },
    {
      "epoch": 0.5873417721518988,
      "grad_norm": 0.48696377873420715,
      "learning_rate": 0.0015,
      "loss": 1.5975,
      "step": 5568
    },
    {
      "epoch": 0.5874472573839662,
      "grad_norm": 0.5826190710067749,
      "learning_rate": 0.0015,
      "loss": 1.602,
      "step": 5569
    },
    {
      "epoch": 0.5875527426160337,
      "grad_norm": 0.47986218333244324,
      "learning_rate": 0.0015,
      "loss": 1.5995,
      "step": 5570
    },
    {
      "epoch": 0.5876582278481013,
      "grad_norm": 0.5235339403152466,
      "learning_rate": 0.0015,
      "loss": 1.6131,
      "step": 5571
    },
    {
      "epoch": 0.5877637130801687,
      "grad_norm": 0.5208390951156616,
      "learning_rate": 0.0015,
      "loss": 1.6309,
      "step": 5572
    },
    {
      "epoch": 0.5878691983122363,
      "grad_norm": 0.6189125776290894,
      "learning_rate": 0.0015,
      "loss": 1.6259,
      "step": 5573
    },
    {
      "epoch": 0.5879746835443038,
      "grad_norm": 0.6651872396469116,
      "learning_rate": 0.0015,
      "loss": 1.6211,
      "step": 5574
    },
    {
      "epoch": 0.5880801687763713,
      "grad_norm": 0.5567631721496582,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 5575
    },
    {
      "epoch": 0.5881856540084388,
      "grad_norm": 0.5870737433433533,
      "learning_rate": 0.0015,
      "loss": 1.6106,
      "step": 5576
    },
    {
      "epoch": 0.5882911392405064,
      "grad_norm": 0.6546647548675537,
      "learning_rate": 0.0015,
      "loss": 1.6366,
      "step": 5577
    },
    {
      "epoch": 0.5883966244725738,
      "grad_norm": 0.6215531826019287,
      "learning_rate": 0.0015,
      "loss": 1.6081,
      "step": 5578
    },
    {
      "epoch": 0.5885021097046413,
      "grad_norm": 0.7473545074462891,
      "learning_rate": 0.0015,
      "loss": 1.6366,
      "step": 5579
    },
    {
      "epoch": 0.5886075949367089,
      "grad_norm": 0.6567901968955994,
      "learning_rate": 0.0015,
      "loss": 1.5861,
      "step": 5580
    },
    {
      "epoch": 0.5887130801687763,
      "grad_norm": 0.6100008487701416,
      "learning_rate": 0.0015,
      "loss": 1.6193,
      "step": 5581
    },
    {
      "epoch": 0.5888185654008439,
      "grad_norm": 0.5344525575637817,
      "learning_rate": 0.0015,
      "loss": 1.6265,
      "step": 5582
    },
    {
      "epoch": 0.5889240506329114,
      "grad_norm": 0.5835578441619873,
      "learning_rate": 0.0015,
      "loss": 1.5907,
      "step": 5583
    },
    {
      "epoch": 0.5890295358649789,
      "grad_norm": 0.4616762101650238,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 5584
    },
    {
      "epoch": 0.5891350210970464,
      "grad_norm": 0.7263842821121216,
      "learning_rate": 0.0015,
      "loss": 1.651,
      "step": 5585
    },
    {
      "epoch": 0.589240506329114,
      "grad_norm": 0.6085631847381592,
      "learning_rate": 0.0015,
      "loss": 1.6172,
      "step": 5586
    },
    {
      "epoch": 0.5893459915611814,
      "grad_norm": 0.5674893260002136,
      "learning_rate": 0.0015,
      "loss": 1.6151,
      "step": 5587
    },
    {
      "epoch": 0.5894514767932489,
      "grad_norm": 0.5729935765266418,
      "learning_rate": 0.0015,
      "loss": 1.6611,
      "step": 5588
    },
    {
      "epoch": 0.5895569620253165,
      "grad_norm": 0.7947391271591187,
      "learning_rate": 0.0015,
      "loss": 1.6111,
      "step": 5589
    },
    {
      "epoch": 0.5896624472573839,
      "grad_norm": 0.48430517315864563,
      "learning_rate": 0.0015,
      "loss": 1.6017,
      "step": 5590
    },
    {
      "epoch": 0.5897679324894515,
      "grad_norm": 0.7709295749664307,
      "learning_rate": 0.0015,
      "loss": 1.6046,
      "step": 5591
    },
    {
      "epoch": 0.589873417721519,
      "grad_norm": 0.6639266610145569,
      "learning_rate": 0.0015,
      "loss": 1.655,
      "step": 5592
    },
    {
      "epoch": 0.5899789029535865,
      "grad_norm": 0.6433471441268921,
      "learning_rate": 0.0015,
      "loss": 1.6102,
      "step": 5593
    },
    {
      "epoch": 0.590084388185654,
      "grad_norm": 0.6052055954933167,
      "learning_rate": 0.0015,
      "loss": 1.6102,
      "step": 5594
    },
    {
      "epoch": 0.5901898734177216,
      "grad_norm": 0.5959548354148865,
      "learning_rate": 0.0015,
      "loss": 1.6164,
      "step": 5595
    },
    {
      "epoch": 0.590295358649789,
      "grad_norm": 0.6258935332298279,
      "learning_rate": 0.0015,
      "loss": 1.5975,
      "step": 5596
    },
    {
      "epoch": 0.5904008438818565,
      "grad_norm": 0.6642781496047974,
      "learning_rate": 0.0015,
      "loss": 1.6069,
      "step": 5597
    },
    {
      "epoch": 0.5905063291139241,
      "grad_norm": 0.7683665752410889,
      "learning_rate": 0.0015,
      "loss": 1.6317,
      "step": 5598
    },
    {
      "epoch": 0.5906118143459915,
      "grad_norm": 0.5938720107078552,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 5599
    },
    {
      "epoch": 0.5907172995780591,
      "grad_norm": 0.5939382910728455,
      "learning_rate": 0.0015,
      "loss": 1.641,
      "step": 5600
    },
    {
      "epoch": 0.5908227848101266,
      "grad_norm": 0.6148390769958496,
      "learning_rate": 0.0015,
      "loss": 1.6301,
      "step": 5601
    },
    {
      "epoch": 0.5909282700421941,
      "grad_norm": 0.5103157758712769,
      "learning_rate": 0.0015,
      "loss": 1.6503,
      "step": 5602
    },
    {
      "epoch": 0.5910337552742616,
      "grad_norm": 0.6613872051239014,
      "learning_rate": 0.0015,
      "loss": 1.6367,
      "step": 5603
    },
    {
      "epoch": 0.5911392405063292,
      "grad_norm": 0.7048614621162415,
      "learning_rate": 0.0015,
      "loss": 1.6078,
      "step": 5604
    },
    {
      "epoch": 0.5912447257383966,
      "grad_norm": 0.5022804737091064,
      "learning_rate": 0.0015,
      "loss": 1.5921,
      "step": 5605
    },
    {
      "epoch": 0.5913502109704641,
      "grad_norm": 0.7162870764732361,
      "learning_rate": 0.0015,
      "loss": 1.6307,
      "step": 5606
    },
    {
      "epoch": 0.5914556962025317,
      "grad_norm": 0.5350881218910217,
      "learning_rate": 0.0015,
      "loss": 1.6294,
      "step": 5607
    },
    {
      "epoch": 0.5915611814345991,
      "grad_norm": 0.5418235063552856,
      "learning_rate": 0.0015,
      "loss": 1.6208,
      "step": 5608
    },
    {
      "epoch": 0.5916666666666667,
      "grad_norm": 0.5851300358772278,
      "learning_rate": 0.0015,
      "loss": 1.6147,
      "step": 5609
    },
    {
      "epoch": 0.5917721518987342,
      "grad_norm": 0.577905535697937,
      "learning_rate": 0.0015,
      "loss": 1.6272,
      "step": 5610
    },
    {
      "epoch": 0.5918776371308017,
      "grad_norm": 0.5166160464286804,
      "learning_rate": 0.0015,
      "loss": 1.6197,
      "step": 5611
    },
    {
      "epoch": 0.5919831223628692,
      "grad_norm": 0.5385041832923889,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 5612
    },
    {
      "epoch": 0.5920886075949368,
      "grad_norm": 0.6943890452384949,
      "learning_rate": 0.0015,
      "loss": 1.6297,
      "step": 5613
    },
    {
      "epoch": 0.5921940928270042,
      "grad_norm": 0.5998681783676147,
      "learning_rate": 0.0015,
      "loss": 1.6314,
      "step": 5614
    },
    {
      "epoch": 0.5922995780590717,
      "grad_norm": 0.5115747451782227,
      "learning_rate": 0.0015,
      "loss": 1.6429,
      "step": 5615
    },
    {
      "epoch": 0.5924050632911393,
      "grad_norm": 0.5083685517311096,
      "learning_rate": 0.0015,
      "loss": 1.6219,
      "step": 5616
    },
    {
      "epoch": 0.5925105485232067,
      "grad_norm": 0.46947669982910156,
      "learning_rate": 0.0015,
      "loss": 1.6026,
      "step": 5617
    },
    {
      "epoch": 0.5926160337552743,
      "grad_norm": 0.5864566564559937,
      "learning_rate": 0.0015,
      "loss": 1.6297,
      "step": 5618
    },
    {
      "epoch": 0.5927215189873418,
      "grad_norm": 0.6747725605964661,
      "learning_rate": 0.0015,
      "loss": 1.6308,
      "step": 5619
    },
    {
      "epoch": 0.5928270042194093,
      "grad_norm": 0.51981121301651,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5620
    },
    {
      "epoch": 0.5929324894514768,
      "grad_norm": 0.47632312774658203,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 5621
    },
    {
      "epoch": 0.5930379746835444,
      "grad_norm": 0.5157601833343506,
      "learning_rate": 0.0015,
      "loss": 1.6444,
      "step": 5622
    },
    {
      "epoch": 0.5931434599156118,
      "grad_norm": 0.5353105664253235,
      "learning_rate": 0.0015,
      "loss": 1.5813,
      "step": 5623
    },
    {
      "epoch": 0.5932489451476793,
      "grad_norm": 0.5490435361862183,
      "learning_rate": 0.0015,
      "loss": 1.6462,
      "step": 5624
    },
    {
      "epoch": 0.5933544303797469,
      "grad_norm": 0.4852970838546753,
      "learning_rate": 0.0015,
      "loss": 1.645,
      "step": 5625
    },
    {
      "epoch": 0.5934599156118143,
      "grad_norm": 0.503825306892395,
      "learning_rate": 0.0015,
      "loss": 1.6216,
      "step": 5626
    },
    {
      "epoch": 0.5935654008438819,
      "grad_norm": 0.5791821479797363,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 5627
    },
    {
      "epoch": 0.5936708860759494,
      "grad_norm": 0.5218534469604492,
      "learning_rate": 0.0015,
      "loss": 1.5899,
      "step": 5628
    },
    {
      "epoch": 0.5937763713080169,
      "grad_norm": 0.6484091281890869,
      "learning_rate": 0.0015,
      "loss": 1.6264,
      "step": 5629
    },
    {
      "epoch": 0.5938818565400844,
      "grad_norm": 0.6759604215621948,
      "learning_rate": 0.0015,
      "loss": 1.6077,
      "step": 5630
    },
    {
      "epoch": 0.5939873417721518,
      "grad_norm": 0.4621976315975189,
      "learning_rate": 0.0015,
      "loss": 1.638,
      "step": 5631
    },
    {
      "epoch": 0.5940928270042194,
      "grad_norm": 0.7069104313850403,
      "learning_rate": 0.0015,
      "loss": 1.6385,
      "step": 5632
    },
    {
      "epoch": 0.5941983122362869,
      "grad_norm": 0.6323444247245789,
      "learning_rate": 0.0015,
      "loss": 1.6159,
      "step": 5633
    },
    {
      "epoch": 0.5943037974683544,
      "grad_norm": 0.5433825850486755,
      "learning_rate": 0.0015,
      "loss": 1.6355,
      "step": 5634
    },
    {
      "epoch": 0.5944092827004219,
      "grad_norm": 0.7824275493621826,
      "learning_rate": 0.0015,
      "loss": 1.6282,
      "step": 5635
    },
    {
      "epoch": 0.5945147679324895,
      "grad_norm": 0.6480106711387634,
      "learning_rate": 0.0015,
      "loss": 1.6246,
      "step": 5636
    },
    {
      "epoch": 0.5946202531645569,
      "grad_norm": 0.5131890177726746,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 5637
    },
    {
      "epoch": 0.5947257383966245,
      "grad_norm": 0.6877164840698242,
      "learning_rate": 0.0015,
      "loss": 1.6798,
      "step": 5638
    },
    {
      "epoch": 0.594831223628692,
      "grad_norm": 0.5200788378715515,
      "learning_rate": 0.0015,
      "loss": 1.6289,
      "step": 5639
    },
    {
      "epoch": 0.5949367088607594,
      "grad_norm": 0.5588069558143616,
      "learning_rate": 0.0015,
      "loss": 1.6494,
      "step": 5640
    },
    {
      "epoch": 0.595042194092827,
      "grad_norm": 0.7360785007476807,
      "learning_rate": 0.0015,
      "loss": 1.6531,
      "step": 5641
    },
    {
      "epoch": 0.5951476793248945,
      "grad_norm": 0.4863540828227997,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 5642
    },
    {
      "epoch": 0.595253164556962,
      "grad_norm": 0.6577567458152771,
      "learning_rate": 0.0015,
      "loss": 1.6265,
      "step": 5643
    },
    {
      "epoch": 0.5953586497890295,
      "grad_norm": 0.5978899598121643,
      "learning_rate": 0.0015,
      "loss": 1.6343,
      "step": 5644
    },
    {
      "epoch": 0.5954641350210971,
      "grad_norm": 0.4906193017959595,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5645
    },
    {
      "epoch": 0.5955696202531645,
      "grad_norm": 0.5887708067893982,
      "learning_rate": 0.0015,
      "loss": 1.6155,
      "step": 5646
    },
    {
      "epoch": 0.5956751054852321,
      "grad_norm": 0.5471548438072205,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 5647
    },
    {
      "epoch": 0.5957805907172996,
      "grad_norm": 0.5576574206352234,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5648
    },
    {
      "epoch": 0.595886075949367,
      "grad_norm": 0.4525201618671417,
      "learning_rate": 0.0015,
      "loss": 1.64,
      "step": 5649
    },
    {
      "epoch": 0.5959915611814346,
      "grad_norm": 0.579256534576416,
      "learning_rate": 0.0015,
      "loss": 1.616,
      "step": 5650
    },
    {
      "epoch": 0.5960970464135021,
      "grad_norm": 0.4890976846218109,
      "learning_rate": 0.0015,
      "loss": 1.6213,
      "step": 5651
    },
    {
      "epoch": 0.5962025316455696,
      "grad_norm": 0.4881046414375305,
      "learning_rate": 0.0015,
      "loss": 1.6122,
      "step": 5652
    },
    {
      "epoch": 0.5963080168776371,
      "grad_norm": 0.5339732766151428,
      "learning_rate": 0.0015,
      "loss": 1.5761,
      "step": 5653
    },
    {
      "epoch": 0.5964135021097047,
      "grad_norm": 0.48167338967323303,
      "learning_rate": 0.0015,
      "loss": 1.6168,
      "step": 5654
    },
    {
      "epoch": 0.5965189873417721,
      "grad_norm": 0.6510812044143677,
      "learning_rate": 0.0015,
      "loss": 1.6351,
      "step": 5655
    },
    {
      "epoch": 0.5966244725738397,
      "grad_norm": 0.6494855284690857,
      "learning_rate": 0.0015,
      "loss": 1.6396,
      "step": 5656
    },
    {
      "epoch": 0.5967299578059072,
      "grad_norm": 0.5147927403450012,
      "learning_rate": 0.0015,
      "loss": 1.6353,
      "step": 5657
    },
    {
      "epoch": 0.5968354430379746,
      "grad_norm": 0.6979304552078247,
      "learning_rate": 0.0015,
      "loss": 1.6597,
      "step": 5658
    },
    {
      "epoch": 0.5969409282700422,
      "grad_norm": 0.5883269906044006,
      "learning_rate": 0.0015,
      "loss": 1.6104,
      "step": 5659
    },
    {
      "epoch": 0.5970464135021097,
      "grad_norm": 0.4970814883708954,
      "learning_rate": 0.0015,
      "loss": 1.608,
      "step": 5660
    },
    {
      "epoch": 0.5971518987341772,
      "grad_norm": 0.5940549969673157,
      "learning_rate": 0.0015,
      "loss": 1.6499,
      "step": 5661
    },
    {
      "epoch": 0.5972573839662447,
      "grad_norm": 0.5467049479484558,
      "learning_rate": 0.0015,
      "loss": 1.592,
      "step": 5662
    },
    {
      "epoch": 0.5973628691983123,
      "grad_norm": 0.4867047071456909,
      "learning_rate": 0.0015,
      "loss": 1.649,
      "step": 5663
    },
    {
      "epoch": 0.5974683544303797,
      "grad_norm": 0.4736812710762024,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 5664
    },
    {
      "epoch": 0.5975738396624473,
      "grad_norm": 0.5190638899803162,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 5665
    },
    {
      "epoch": 0.5976793248945148,
      "grad_norm": 0.4780181050300598,
      "learning_rate": 0.0015,
      "loss": 1.5853,
      "step": 5666
    },
    {
      "epoch": 0.5977848101265822,
      "grad_norm": 0.503572404384613,
      "learning_rate": 0.0015,
      "loss": 1.6551,
      "step": 5667
    },
    {
      "epoch": 0.5978902953586498,
      "grad_norm": 0.5043089389801025,
      "learning_rate": 0.0015,
      "loss": 1.6613,
      "step": 5668
    },
    {
      "epoch": 0.5979957805907173,
      "grad_norm": 0.5234181880950928,
      "learning_rate": 0.0015,
      "loss": 1.6303,
      "step": 5669
    },
    {
      "epoch": 0.5981012658227848,
      "grad_norm": 0.519684910774231,
      "learning_rate": 0.0015,
      "loss": 1.5951,
      "step": 5670
    },
    {
      "epoch": 0.5982067510548523,
      "grad_norm": 0.4918913245201111,
      "learning_rate": 0.0015,
      "loss": 1.6307,
      "step": 5671
    },
    {
      "epoch": 0.5983122362869199,
      "grad_norm": 0.5128886103630066,
      "learning_rate": 0.0015,
      "loss": 1.6155,
      "step": 5672
    },
    {
      "epoch": 0.5984177215189873,
      "grad_norm": 0.5375071167945862,
      "learning_rate": 0.0015,
      "loss": 1.6597,
      "step": 5673
    },
    {
      "epoch": 0.5985232067510549,
      "grad_norm": 0.47754842042922974,
      "learning_rate": 0.0015,
      "loss": 1.646,
      "step": 5674
    },
    {
      "epoch": 0.5986286919831224,
      "grad_norm": 0.6603662371635437,
      "learning_rate": 0.0015,
      "loss": 1.6129,
      "step": 5675
    },
    {
      "epoch": 0.5987341772151898,
      "grad_norm": 0.7207350730895996,
      "learning_rate": 0.0015,
      "loss": 1.6491,
      "step": 5676
    },
    {
      "epoch": 0.5988396624472574,
      "grad_norm": 0.5981911420822144,
      "learning_rate": 0.0015,
      "loss": 1.5973,
      "step": 5677
    },
    {
      "epoch": 0.5989451476793249,
      "grad_norm": 0.7006922364234924,
      "learning_rate": 0.0015,
      "loss": 1.6631,
      "step": 5678
    },
    {
      "epoch": 0.5990506329113924,
      "grad_norm": 0.7043522000312805,
      "learning_rate": 0.0015,
      "loss": 1.593,
      "step": 5679
    },
    {
      "epoch": 0.5991561181434599,
      "grad_norm": 0.47904908657073975,
      "learning_rate": 0.0015,
      "loss": 1.6443,
      "step": 5680
    },
    {
      "epoch": 0.5992616033755275,
      "grad_norm": 0.6863402128219604,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5681
    },
    {
      "epoch": 0.5993670886075949,
      "grad_norm": 0.8411417603492737,
      "learning_rate": 0.0015,
      "loss": 1.6169,
      "step": 5682
    },
    {
      "epoch": 0.5994725738396625,
      "grad_norm": 0.4999617040157318,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 5683
    },
    {
      "epoch": 0.59957805907173,
      "grad_norm": 0.61651611328125,
      "learning_rate": 0.0015,
      "loss": 1.5913,
      "step": 5684
    },
    {
      "epoch": 0.5996835443037974,
      "grad_norm": 0.6154442429542542,
      "learning_rate": 0.0015,
      "loss": 1.609,
      "step": 5685
    },
    {
      "epoch": 0.599789029535865,
      "grad_norm": 0.5308151841163635,
      "learning_rate": 0.0015,
      "loss": 1.6353,
      "step": 5686
    },
    {
      "epoch": 0.5998945147679325,
      "grad_norm": 0.5686870217323303,
      "learning_rate": 0.0015,
      "loss": 1.6175,
      "step": 5687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5483143329620361,
      "learning_rate": 0.0015,
      "loss": 1.6121,
      "step": 5688
    },
    {
      "epoch": 0.6001054852320675,
      "grad_norm": 0.73167884349823,
      "learning_rate": 0.0015,
      "loss": 1.5947,
      "step": 5689
    },
    {
      "epoch": 0.6002109704641351,
      "grad_norm": 0.46591293811798096,
      "learning_rate": 0.0015,
      "loss": 1.6132,
      "step": 5690
    },
    {
      "epoch": 0.6003164556962025,
      "grad_norm": 0.8627923727035522,
      "learning_rate": 0.0015,
      "loss": 1.6571,
      "step": 5691
    },
    {
      "epoch": 0.6004219409282701,
      "grad_norm": 0.5825620889663696,
      "learning_rate": 0.0015,
      "loss": 1.5811,
      "step": 5692
    },
    {
      "epoch": 0.6005274261603376,
      "grad_norm": 0.6405885815620422,
      "learning_rate": 0.0015,
      "loss": 1.624,
      "step": 5693
    },
    {
      "epoch": 0.600632911392405,
      "grad_norm": 0.7344543933868408,
      "learning_rate": 0.0015,
      "loss": 1.6012,
      "step": 5694
    },
    {
      "epoch": 0.6007383966244726,
      "grad_norm": 0.5269743800163269,
      "learning_rate": 0.0015,
      "loss": 1.5742,
      "step": 5695
    },
    {
      "epoch": 0.60084388185654,
      "grad_norm": 0.6673415899276733,
      "learning_rate": 0.0015,
      "loss": 1.6384,
      "step": 5696
    },
    {
      "epoch": 0.6009493670886076,
      "grad_norm": 0.6590900421142578,
      "learning_rate": 0.0015,
      "loss": 1.627,
      "step": 5697
    },
    {
      "epoch": 0.6010548523206751,
      "grad_norm": 0.5728099942207336,
      "learning_rate": 0.0015,
      "loss": 1.6321,
      "step": 5698
    },
    {
      "epoch": 0.6011603375527426,
      "grad_norm": 0.7162034511566162,
      "learning_rate": 0.0015,
      "loss": 1.6005,
      "step": 5699
    },
    {
      "epoch": 0.6012658227848101,
      "grad_norm": 0.8475410342216492,
      "learning_rate": 0.0015,
      "loss": 1.6492,
      "step": 5700
    },
    {
      "epoch": 0.6013713080168777,
      "grad_norm": 0.5202561616897583,
      "learning_rate": 0.0015,
      "loss": 1.5997,
      "step": 5701
    },
    {
      "epoch": 0.6014767932489451,
      "grad_norm": 0.7218925356864929,
      "learning_rate": 0.0015,
      "loss": 1.6295,
      "step": 5702
    },
    {
      "epoch": 0.6015822784810126,
      "grad_norm": 0.8395099639892578,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 5703
    },
    {
      "epoch": 0.6016877637130802,
      "grad_norm": 0.5557817220687866,
      "learning_rate": 0.0015,
      "loss": 1.6082,
      "step": 5704
    },
    {
      "epoch": 0.6017932489451476,
      "grad_norm": 0.6226076483726501,
      "learning_rate": 0.0015,
      "loss": 1.6395,
      "step": 5705
    },
    {
      "epoch": 0.6018987341772152,
      "grad_norm": 0.7292788028717041,
      "learning_rate": 0.0015,
      "loss": 1.5953,
      "step": 5706
    },
    {
      "epoch": 0.6020042194092827,
      "grad_norm": 0.5449414253234863,
      "learning_rate": 0.0015,
      "loss": 1.5922,
      "step": 5707
    },
    {
      "epoch": 0.6021097046413502,
      "grad_norm": 0.6932792663574219,
      "learning_rate": 0.0015,
      "loss": 1.6076,
      "step": 5708
    },
    {
      "epoch": 0.6022151898734177,
      "grad_norm": 0.5870957374572754,
      "learning_rate": 0.0015,
      "loss": 1.6195,
      "step": 5709
    },
    {
      "epoch": 0.6023206751054853,
      "grad_norm": 0.7367919087409973,
      "learning_rate": 0.0015,
      "loss": 1.6621,
      "step": 5710
    },
    {
      "epoch": 0.6024261603375527,
      "grad_norm": 0.5735125541687012,
      "learning_rate": 0.0015,
      "loss": 1.6422,
      "step": 5711
    },
    {
      "epoch": 0.6025316455696202,
      "grad_norm": 0.590947687625885,
      "learning_rate": 0.0015,
      "loss": 1.6146,
      "step": 5712
    },
    {
      "epoch": 0.6026371308016878,
      "grad_norm": 0.515593945980072,
      "learning_rate": 0.0015,
      "loss": 1.6127,
      "step": 5713
    },
    {
      "epoch": 0.6027426160337552,
      "grad_norm": 0.7016730308532715,
      "learning_rate": 0.0015,
      "loss": 1.6089,
      "step": 5714
    },
    {
      "epoch": 0.6028481012658228,
      "grad_norm": 0.47112807631492615,
      "learning_rate": 0.0015,
      "loss": 1.5973,
      "step": 5715
    },
    {
      "epoch": 0.6029535864978903,
      "grad_norm": 0.5997288227081299,
      "learning_rate": 0.0015,
      "loss": 1.653,
      "step": 5716
    },
    {
      "epoch": 0.6030590717299578,
      "grad_norm": 0.5407869815826416,
      "learning_rate": 0.0015,
      "loss": 1.6433,
      "step": 5717
    },
    {
      "epoch": 0.6031645569620253,
      "grad_norm": 0.527229905128479,
      "learning_rate": 0.0015,
      "loss": 1.6204,
      "step": 5718
    },
    {
      "epoch": 0.6032700421940929,
      "grad_norm": 0.5542311072349548,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 5719
    },
    {
      "epoch": 0.6033755274261603,
      "grad_norm": 0.5649443864822388,
      "learning_rate": 0.0015,
      "loss": 1.6146,
      "step": 5720
    },
    {
      "epoch": 0.6034810126582278,
      "grad_norm": 0.4718017280101776,
      "learning_rate": 0.0015,
      "loss": 1.6105,
      "step": 5721
    },
    {
      "epoch": 0.6035864978902954,
      "grad_norm": 0.5023795366287231,
      "learning_rate": 0.0015,
      "loss": 1.6215,
      "step": 5722
    },
    {
      "epoch": 0.6036919831223628,
      "grad_norm": 0.5101316571235657,
      "learning_rate": 0.0015,
      "loss": 1.6153,
      "step": 5723
    },
    {
      "epoch": 0.6037974683544304,
      "grad_norm": 0.4487985372543335,
      "learning_rate": 0.0015,
      "loss": 1.6342,
      "step": 5724
    },
    {
      "epoch": 0.6039029535864979,
      "grad_norm": 0.5671184062957764,
      "learning_rate": 0.0015,
      "loss": 1.6086,
      "step": 5725
    },
    {
      "epoch": 0.6040084388185654,
      "grad_norm": 0.538108229637146,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 5726
    },
    {
      "epoch": 0.6041139240506329,
      "grad_norm": 0.5328302979469299,
      "learning_rate": 0.0015,
      "loss": 1.6043,
      "step": 5727
    },
    {
      "epoch": 0.6042194092827005,
      "grad_norm": 0.6336845755577087,
      "learning_rate": 0.0015,
      "loss": 1.6115,
      "step": 5728
    },
    {
      "epoch": 0.6043248945147679,
      "grad_norm": 0.6453084349632263,
      "learning_rate": 0.0015,
      "loss": 1.6251,
      "step": 5729
    },
    {
      "epoch": 0.6044303797468354,
      "grad_norm": 0.5204219818115234,
      "learning_rate": 0.0015,
      "loss": 1.6296,
      "step": 5730
    },
    {
      "epoch": 0.604535864978903,
      "grad_norm": 0.4975261390209198,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5731
    },
    {
      "epoch": 0.6046413502109704,
      "grad_norm": 0.6070615649223328,
      "learning_rate": 0.0015,
      "loss": 1.6001,
      "step": 5732
    },
    {
      "epoch": 0.604746835443038,
      "grad_norm": 0.47148597240448,
      "learning_rate": 0.0015,
      "loss": 1.6211,
      "step": 5733
    },
    {
      "epoch": 0.6048523206751055,
      "grad_norm": 0.7071391940116882,
      "learning_rate": 0.0015,
      "loss": 1.603,
      "step": 5734
    },
    {
      "epoch": 0.604957805907173,
      "grad_norm": 0.6131590604782104,
      "learning_rate": 0.0015,
      "loss": 1.6079,
      "step": 5735
    },
    {
      "epoch": 0.6050632911392405,
      "grad_norm": 0.5092592835426331,
      "learning_rate": 0.0015,
      "loss": 1.6459,
      "step": 5736
    },
    {
      "epoch": 0.6051687763713081,
      "grad_norm": 0.667983889579773,
      "learning_rate": 0.0015,
      "loss": 1.6173,
      "step": 5737
    },
    {
      "epoch": 0.6052742616033755,
      "grad_norm": 0.5349023342132568,
      "learning_rate": 0.0015,
      "loss": 1.6159,
      "step": 5738
    },
    {
      "epoch": 0.605379746835443,
      "grad_norm": 0.5172088742256165,
      "learning_rate": 0.0015,
      "loss": 1.6661,
      "step": 5739
    },
    {
      "epoch": 0.6054852320675106,
      "grad_norm": 0.5283232927322388,
      "learning_rate": 0.0015,
      "loss": 1.6037,
      "step": 5740
    },
    {
      "epoch": 0.605590717299578,
      "grad_norm": 0.47503188252449036,
      "learning_rate": 0.0015,
      "loss": 1.5955,
      "step": 5741
    },
    {
      "epoch": 0.6056962025316456,
      "grad_norm": 0.5385279655456543,
      "learning_rate": 0.0015,
      "loss": 1.6232,
      "step": 5742
    },
    {
      "epoch": 0.6058016877637131,
      "grad_norm": 0.5128896832466125,
      "learning_rate": 0.0015,
      "loss": 1.5765,
      "step": 5743
    },
    {
      "epoch": 0.6059071729957806,
      "grad_norm": 0.5275681614875793,
      "learning_rate": 0.0015,
      "loss": 1.592,
      "step": 5744
    },
    {
      "epoch": 0.6060126582278481,
      "grad_norm": 0.5630440711975098,
      "learning_rate": 0.0015,
      "loss": 1.5831,
      "step": 5745
    },
    {
      "epoch": 0.6061181434599157,
      "grad_norm": 0.5439804792404175,
      "learning_rate": 0.0015,
      "loss": 1.6014,
      "step": 5746
    },
    {
      "epoch": 0.6062236286919831,
      "grad_norm": 0.641860842704773,
      "learning_rate": 0.0015,
      "loss": 1.5944,
      "step": 5747
    },
    {
      "epoch": 0.6063291139240506,
      "grad_norm": 0.724949836730957,
      "learning_rate": 0.0015,
      "loss": 1.6154,
      "step": 5748
    },
    {
      "epoch": 0.6064345991561182,
      "grad_norm": 0.5219384431838989,
      "learning_rate": 0.0015,
      "loss": 1.6187,
      "step": 5749
    },
    {
      "epoch": 0.6065400843881856,
      "grad_norm": 0.9701561331748962,
      "learning_rate": 0.0015,
      "loss": 1.5626,
      "step": 5750
    },
    {
      "epoch": 0.6066455696202532,
      "grad_norm": 0.7944713830947876,
      "learning_rate": 0.0015,
      "loss": 1.5801,
      "step": 5751
    },
    {
      "epoch": 0.6067510548523207,
      "grad_norm": 0.5765522718429565,
      "learning_rate": 0.0015,
      "loss": 1.5814,
      "step": 5752
    },
    {
      "epoch": 0.6068565400843882,
      "grad_norm": 0.655381441116333,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 5753
    },
    {
      "epoch": 0.6069620253164557,
      "grad_norm": 0.6935734748840332,
      "learning_rate": 0.0015,
      "loss": 1.6107,
      "step": 5754
    },
    {
      "epoch": 0.6070675105485233,
      "grad_norm": 0.5127397775650024,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 5755
    },
    {
      "epoch": 0.6071729957805907,
      "grad_norm": 0.6720975637435913,
      "learning_rate": 0.0015,
      "loss": 1.5996,
      "step": 5756
    },
    {
      "epoch": 0.6072784810126582,
      "grad_norm": 0.7431419491767883,
      "learning_rate": 0.0015,
      "loss": 1.5775,
      "step": 5757
    },
    {
      "epoch": 0.6073839662447258,
      "grad_norm": 0.47397467494010925,
      "learning_rate": 0.0015,
      "loss": 1.6319,
      "step": 5758
    },
    {
      "epoch": 0.6074894514767932,
      "grad_norm": 0.5727673172950745,
      "learning_rate": 0.0015,
      "loss": 1.6192,
      "step": 5759
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 0.5239375829696655,
      "learning_rate": 0.0015,
      "loss": 1.6203,
      "step": 5760
    },
    {
      "epoch": 0.6077004219409282,
      "grad_norm": 0.5244763493537903,
      "learning_rate": 0.0015,
      "loss": 1.5693,
      "step": 5761
    },
    {
      "epoch": 0.6078059071729958,
      "grad_norm": 0.5762230157852173,
      "learning_rate": 0.0015,
      "loss": 1.6229,
      "step": 5762
    },
    {
      "epoch": 0.6079113924050633,
      "grad_norm": 0.5437536239624023,
      "learning_rate": 0.0015,
      "loss": 1.5972,
      "step": 5763
    },
    {
      "epoch": 0.6080168776371307,
      "grad_norm": 0.5512168407440186,
      "learning_rate": 0.0015,
      "loss": 1.6059,
      "step": 5764
    },
    {
      "epoch": 0.6081223628691983,
      "grad_norm": 0.7002081871032715,
      "learning_rate": 0.0015,
      "loss": 1.604,
      "step": 5765
    },
    {
      "epoch": 0.6082278481012658,
      "grad_norm": 0.45743414759635925,
      "learning_rate": 0.0015,
      "loss": 1.6318,
      "step": 5766
    },
    {
      "epoch": 0.6083333333333333,
      "grad_norm": 0.6875436902046204,
      "learning_rate": 0.0015,
      "loss": 1.6227,
      "step": 5767
    },
    {
      "epoch": 0.6084388185654008,
      "grad_norm": 0.6525773406028748,
      "learning_rate": 0.0015,
      "loss": 1.5984,
      "step": 5768
    },
    {
      "epoch": 0.6085443037974684,
      "grad_norm": 0.7001569867134094,
      "learning_rate": 0.0015,
      "loss": 1.6079,
      "step": 5769
    },
    {
      "epoch": 0.6086497890295358,
      "grad_norm": 0.6135374903678894,
      "learning_rate": 0.0015,
      "loss": 1.6293,
      "step": 5770
    },
    {
      "epoch": 0.6087552742616034,
      "grad_norm": 0.9716602563858032,
      "learning_rate": 0.0015,
      "loss": 1.5995,
      "step": 5771
    },
    {
      "epoch": 0.6088607594936709,
      "grad_norm": 0.8283466696739197,
      "learning_rate": 0.0015,
      "loss": 1.6205,
      "step": 5772
    },
    {
      "epoch": 0.6089662447257383,
      "grad_norm": 0.5932331085205078,
      "learning_rate": 0.0015,
      "loss": 1.6275,
      "step": 5773
    },
    {
      "epoch": 0.6090717299578059,
      "grad_norm": 0.8104361295700073,
      "learning_rate": 0.0015,
      "loss": 1.6086,
      "step": 5774
    },
    {
      "epoch": 0.6091772151898734,
      "grad_norm": 0.5436210632324219,
      "learning_rate": 0.0015,
      "loss": 1.6024,
      "step": 5775
    },
    {
      "epoch": 0.6092827004219409,
      "grad_norm": 0.7614966034889221,
      "learning_rate": 0.0015,
      "loss": 1.6373,
      "step": 5776
    },
    {
      "epoch": 0.6093881856540084,
      "grad_norm": 0.6954220533370972,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 5777
    },
    {
      "epoch": 0.609493670886076,
      "grad_norm": 0.557464063167572,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5778
    },
    {
      "epoch": 0.6095991561181434,
      "grad_norm": 0.5181831121444702,
      "learning_rate": 0.0015,
      "loss": 1.614,
      "step": 5779
    },
    {
      "epoch": 0.609704641350211,
      "grad_norm": 0.544055700302124,
      "learning_rate": 0.0015,
      "loss": 1.6268,
      "step": 5780
    },
    {
      "epoch": 0.6098101265822785,
      "grad_norm": 0.4917396008968353,
      "learning_rate": 0.0015,
      "loss": 1.6202,
      "step": 5781
    },
    {
      "epoch": 0.609915611814346,
      "grad_norm": 0.5461342930793762,
      "learning_rate": 0.0015,
      "loss": 1.6204,
      "step": 5782
    },
    {
      "epoch": 0.6100210970464135,
      "grad_norm": 0.5721676349639893,
      "learning_rate": 0.0015,
      "loss": 1.6446,
      "step": 5783
    },
    {
      "epoch": 0.610126582278481,
      "grad_norm": 0.5346076488494873,
      "learning_rate": 0.0015,
      "loss": 1.6545,
      "step": 5784
    },
    {
      "epoch": 0.6102320675105485,
      "grad_norm": 0.5412622690200806,
      "learning_rate": 0.0015,
      "loss": 1.6004,
      "step": 5785
    },
    {
      "epoch": 0.610337552742616,
      "grad_norm": 0.5639292001724243,
      "learning_rate": 0.0015,
      "loss": 1.588,
      "step": 5786
    },
    {
      "epoch": 0.6104430379746836,
      "grad_norm": 0.4506779611110687,
      "learning_rate": 0.0015,
      "loss": 1.5823,
      "step": 5787
    },
    {
      "epoch": 0.610548523206751,
      "grad_norm": 0.6533520221710205,
      "learning_rate": 0.0015,
      "loss": 1.6031,
      "step": 5788
    },
    {
      "epoch": 0.6106540084388186,
      "grad_norm": 0.6267601251602173,
      "learning_rate": 0.0015,
      "loss": 1.6281,
      "step": 5789
    },
    {
      "epoch": 0.6107594936708861,
      "grad_norm": 0.446891725063324,
      "learning_rate": 0.0015,
      "loss": 1.5948,
      "step": 5790
    },
    {
      "epoch": 0.6108649789029535,
      "grad_norm": 0.6829448938369751,
      "learning_rate": 0.0015,
      "loss": 1.6464,
      "step": 5791
    },
    {
      "epoch": 0.6109704641350211,
      "grad_norm": 0.6592504382133484,
      "learning_rate": 0.0015,
      "loss": 1.6136,
      "step": 5792
    },
    {
      "epoch": 0.6110759493670886,
      "grad_norm": 0.5578280687332153,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 5793
    },
    {
      "epoch": 0.6111814345991561,
      "grad_norm": 0.5601876378059387,
      "learning_rate": 0.0015,
      "loss": 1.6245,
      "step": 5794
    },
    {
      "epoch": 0.6112869198312236,
      "grad_norm": 0.5430405735969543,
      "learning_rate": 0.0015,
      "loss": 1.6,
      "step": 5795
    },
    {
      "epoch": 0.6113924050632912,
      "grad_norm": 0.5691314935684204,
      "learning_rate": 0.0015,
      "loss": 1.6085,
      "step": 5796
    },
    {
      "epoch": 0.6114978902953586,
      "grad_norm": 0.7052751183509827,
      "learning_rate": 0.0015,
      "loss": 1.6144,
      "step": 5797
    },
    {
      "epoch": 0.6116033755274262,
      "grad_norm": 0.6696596145629883,
      "learning_rate": 0.0015,
      "loss": 1.5813,
      "step": 5798
    },
    {
      "epoch": 0.6117088607594937,
      "grad_norm": 0.49856799840927124,
      "learning_rate": 0.0015,
      "loss": 1.6123,
      "step": 5799
    },
    {
      "epoch": 0.6118143459915611,
      "grad_norm": 0.602507472038269,
      "learning_rate": 0.0015,
      "loss": 1.625,
      "step": 5800
    },
    {
      "epoch": 0.6119198312236287,
      "grad_norm": 0.5198401808738708,
      "learning_rate": 0.0015,
      "loss": 1.623,
      "step": 5801
    },
    {
      "epoch": 0.6120253164556962,
      "grad_norm": 0.49069100618362427,
      "learning_rate": 0.0015,
      "loss": 1.6083,
      "step": 5802
    },
    {
      "epoch": 0.6121308016877637,
      "grad_norm": 0.645224392414093,
      "learning_rate": 0.0015,
      "loss": 1.5971,
      "step": 5803
    },
    {
      "epoch": 0.6122362869198312,
      "grad_norm": 0.4484640657901764,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 5804
    },
    {
      "epoch": 0.6123417721518988,
      "grad_norm": 0.47885531187057495,
      "learning_rate": 0.0015,
      "loss": 1.5912,
      "step": 5805
    },
    {
      "epoch": 0.6124472573839662,
      "grad_norm": 0.49233120679855347,
      "learning_rate": 0.0015,
      "loss": 1.6376,
      "step": 5806
    },
    {
      "epoch": 0.6125527426160338,
      "grad_norm": 0.47025009989738464,
      "learning_rate": 0.0015,
      "loss": 1.632,
      "step": 5807
    },
    {
      "epoch": 0.6126582278481013,
      "grad_norm": 0.5202906131744385,
      "learning_rate": 0.0015,
      "loss": 1.6325,
      "step": 5808
    },
    {
      "epoch": 0.6127637130801687,
      "grad_norm": 0.55734783411026,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5809
    },
    {
      "epoch": 0.6128691983122363,
      "grad_norm": 0.5348408818244934,
      "learning_rate": 0.0015,
      "loss": 1.6151,
      "step": 5810
    },
    {
      "epoch": 0.6129746835443038,
      "grad_norm": 0.5444215536117554,
      "learning_rate": 0.0015,
      "loss": 1.6495,
      "step": 5811
    },
    {
      "epoch": 0.6130801687763713,
      "grad_norm": 0.5886389017105103,
      "learning_rate": 0.0015,
      "loss": 1.6202,
      "step": 5812
    },
    {
      "epoch": 0.6131856540084388,
      "grad_norm": 0.49230071902275085,
      "learning_rate": 0.0015,
      "loss": 1.6004,
      "step": 5813
    },
    {
      "epoch": 0.6132911392405064,
      "grad_norm": 0.5946803092956543,
      "learning_rate": 0.0015,
      "loss": 1.6136,
      "step": 5814
    },
    {
      "epoch": 0.6133966244725738,
      "grad_norm": 0.5703949928283691,
      "learning_rate": 0.0015,
      "loss": 1.6559,
      "step": 5815
    },
    {
      "epoch": 0.6135021097046414,
      "grad_norm": 0.5334026217460632,
      "learning_rate": 0.0015,
      "loss": 1.6435,
      "step": 5816
    },
    {
      "epoch": 0.6136075949367089,
      "grad_norm": 0.5981934666633606,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5817
    },
    {
      "epoch": 0.6137130801687763,
      "grad_norm": 0.5421572327613831,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 5818
    },
    {
      "epoch": 0.6138185654008439,
      "grad_norm": 0.4682169258594513,
      "learning_rate": 0.0015,
      "loss": 1.6214,
      "step": 5819
    },
    {
      "epoch": 0.6139240506329114,
      "grad_norm": 0.5654427409172058,
      "learning_rate": 0.0015,
      "loss": 1.634,
      "step": 5820
    },
    {
      "epoch": 0.6140295358649789,
      "grad_norm": 0.5109453201293945,
      "learning_rate": 0.0015,
      "loss": 1.6458,
      "step": 5821
    },
    {
      "epoch": 0.6141350210970464,
      "grad_norm": 0.5720937848091125,
      "learning_rate": 0.0015,
      "loss": 1.6071,
      "step": 5822
    },
    {
      "epoch": 0.614240506329114,
      "grad_norm": 0.5675110220909119,
      "learning_rate": 0.0015,
      "loss": 1.6038,
      "step": 5823
    },
    {
      "epoch": 0.6143459915611814,
      "grad_norm": 0.4660292863845825,
      "learning_rate": 0.0015,
      "loss": 1.6226,
      "step": 5824
    },
    {
      "epoch": 0.614451476793249,
      "grad_norm": 0.5449029207229614,
      "learning_rate": 0.0015,
      "loss": 1.6177,
      "step": 5825
    },
    {
      "epoch": 0.6145569620253165,
      "grad_norm": 0.462939590215683,
      "learning_rate": 0.0015,
      "loss": 1.6128,
      "step": 5826
    },
    {
      "epoch": 0.614662447257384,
      "grad_norm": 0.575222909450531,
      "learning_rate": 0.0015,
      "loss": 1.6242,
      "step": 5827
    },
    {
      "epoch": 0.6147679324894515,
      "grad_norm": 0.5540435910224915,
      "learning_rate": 0.0015,
      "loss": 1.5909,
      "step": 5828
    },
    {
      "epoch": 0.6148734177215189,
      "grad_norm": 0.45478394627571106,
      "learning_rate": 0.0015,
      "loss": 1.6245,
      "step": 5829
    },
    {
      "epoch": 0.6149789029535865,
      "grad_norm": 0.5058201551437378,
      "learning_rate": 0.0015,
      "loss": 1.5887,
      "step": 5830
    },
    {
      "epoch": 0.615084388185654,
      "grad_norm": 0.6100703477859497,
      "learning_rate": 0.0015,
      "loss": 1.6205,
      "step": 5831
    },
    {
      "epoch": 0.6151898734177215,
      "grad_norm": 0.4519583284854889,
      "learning_rate": 0.0015,
      "loss": 1.6471,
      "step": 5832
    },
    {
      "epoch": 0.615295358649789,
      "grad_norm": 0.5411374568939209,
      "learning_rate": 0.0015,
      "loss": 1.5915,
      "step": 5833
    },
    {
      "epoch": 0.6154008438818566,
      "grad_norm": 0.44370853900909424,
      "learning_rate": 0.0015,
      "loss": 1.6349,
      "step": 5834
    },
    {
      "epoch": 0.615506329113924,
      "grad_norm": 0.5247477889060974,
      "learning_rate": 0.0015,
      "loss": 1.6121,
      "step": 5835
    },
    {
      "epoch": 0.6156118143459915,
      "grad_norm": 0.5240331292152405,
      "learning_rate": 0.0015,
      "loss": 1.6613,
      "step": 5836
    },
    {
      "epoch": 0.6157172995780591,
      "grad_norm": 0.4902103841304779,
      "learning_rate": 0.0015,
      "loss": 1.5931,
      "step": 5837
    },
    {
      "epoch": 0.6158227848101265,
      "grad_norm": 0.6526527404785156,
      "learning_rate": 0.0015,
      "loss": 1.6518,
      "step": 5838
    },
    {
      "epoch": 0.6159282700421941,
      "grad_norm": 0.4508664011955261,
      "learning_rate": 0.0015,
      "loss": 1.6285,
      "step": 5839
    },
    {
      "epoch": 0.6160337552742616,
      "grad_norm": 0.6527950167655945,
      "learning_rate": 0.0015,
      "loss": 1.6554,
      "step": 5840
    },
    {
      "epoch": 0.6161392405063291,
      "grad_norm": 0.6767658591270447,
      "learning_rate": 0.0015,
      "loss": 1.6492,
      "step": 5841
    },
    {
      "epoch": 0.6162447257383966,
      "grad_norm": 0.49983566999435425,
      "learning_rate": 0.0015,
      "loss": 1.6543,
      "step": 5842
    },
    {
      "epoch": 0.6163502109704642,
      "grad_norm": 0.6607310175895691,
      "learning_rate": 0.0015,
      "loss": 1.5831,
      "step": 5843
    },
    {
      "epoch": 0.6164556962025316,
      "grad_norm": 0.679019570350647,
      "learning_rate": 0.0015,
      "loss": 1.6287,
      "step": 5844
    },
    {
      "epoch": 0.6165611814345991,
      "grad_norm": 0.5855091214179993,
      "learning_rate": 0.0015,
      "loss": 1.6463,
      "step": 5845
    },
    {
      "epoch": 0.6166666666666667,
      "grad_norm": 0.4927886426448822,
      "learning_rate": 0.0015,
      "loss": 1.6072,
      "step": 5846
    },
    {
      "epoch": 0.6167721518987341,
      "grad_norm": 0.5097840428352356,
      "learning_rate": 0.0015,
      "loss": 1.6259,
      "step": 5847
    },
    {
      "epoch": 0.6168776371308017,
      "grad_norm": 0.4735028147697449,
      "learning_rate": 0.0015,
      "loss": 1.6517,
      "step": 5848
    },
    {
      "epoch": 0.6169831223628692,
      "grad_norm": 0.5716160535812378,
      "learning_rate": 0.0015,
      "loss": 1.6735,
      "step": 5849
    },
    {
      "epoch": 0.6170886075949367,
      "grad_norm": 0.5068883299827576,
      "learning_rate": 0.0015,
      "loss": 1.6179,
      "step": 5850
    },
    {
      "epoch": 0.6171940928270042,
      "grad_norm": 0.6885242462158203,
      "learning_rate": 0.0015,
      "loss": 1.6302,
      "step": 5851
    },
    {
      "epoch": 0.6172995780590718,
      "grad_norm": 0.5677760243415833,
      "learning_rate": 0.0015,
      "loss": 1.6092,
      "step": 5852
    },
    {
      "epoch": 0.6174050632911392,
      "grad_norm": 0.5608734488487244,
      "learning_rate": 0.0015,
      "loss": 1.6143,
      "step": 5853
    },
    {
      "epoch": 0.6175105485232067,
      "grad_norm": 0.5561981797218323,
      "learning_rate": 0.0015,
      "loss": 1.5826,
      "step": 5854
    },
    {
      "epoch": 0.6176160337552743,
      "grad_norm": 0.6651297211647034,
      "learning_rate": 0.0015,
      "loss": 1.6783,
      "step": 5855
    },
    {
      "epoch": 0.6177215189873417,
      "grad_norm": 0.5847839117050171,
      "learning_rate": 0.0015,
      "loss": 1.6284,
      "step": 5856
    },
    {
      "epoch": 0.6178270042194093,
      "grad_norm": 0.7147279381752014,
      "learning_rate": 0.0015,
      "loss": 1.597,
      "step": 5857
    },
    {
      "epoch": 0.6179324894514768,
      "grad_norm": 0.6902298927307129,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 5858
    },
    {
      "epoch": 0.6180379746835443,
      "grad_norm": 0.6074957847595215,
      "learning_rate": 0.0015,
      "loss": 1.6391,
      "step": 5859
    },
    {
      "epoch": 0.6181434599156118,
      "grad_norm": 0.6798166632652283,
      "learning_rate": 0.0015,
      "loss": 1.6323,
      "step": 5860
    },
    {
      "epoch": 0.6182489451476794,
      "grad_norm": 0.6154612302780151,
      "learning_rate": 0.0015,
      "loss": 1.6487,
      "step": 5861
    },
    {
      "epoch": 0.6183544303797468,
      "grad_norm": 0.6916834115982056,
      "learning_rate": 0.0015,
      "loss": 1.6737,
      "step": 5862
    },
    {
      "epoch": 0.6184599156118143,
      "grad_norm": 0.9251354336738586,
      "learning_rate": 0.0015,
      "loss": 1.5949,
      "step": 5863
    },
    {
      "epoch": 0.6185654008438819,
      "grad_norm": 0.7731589674949646,
      "learning_rate": 0.0015,
      "loss": 1.5898,
      "step": 5864
    },
    {
      "epoch": 0.6186708860759493,
      "grad_norm": 0.5317090749740601,
      "learning_rate": 0.0015,
      "loss": 1.6159,
      "step": 5865
    },
    {
      "epoch": 0.6187763713080169,
      "grad_norm": 0.7302789092063904,
      "learning_rate": 0.0015,
      "loss": 1.6303,
      "step": 5866
    },
    {
      "epoch": 0.6188818565400844,
      "grad_norm": 0.6556691527366638,
      "learning_rate": 0.0015,
      "loss": 1.5668,
      "step": 5867
    },
    {
      "epoch": 0.6189873417721519,
      "grad_norm": 0.5228979587554932,
      "learning_rate": 0.0015,
      "loss": 1.5929,
      "step": 5868
    },
    {
      "epoch": 0.6190928270042194,
      "grad_norm": 0.6936362981796265,
      "learning_rate": 0.0015,
      "loss": 1.6343,
      "step": 5869
    },
    {
      "epoch": 0.619198312236287,
      "grad_norm": 0.5519838929176331,
      "learning_rate": 0.0015,
      "loss": 1.6058,
      "step": 5870
    },
    {
      "epoch": 0.6193037974683544,
      "grad_norm": 0.5528889298439026,
      "learning_rate": 0.0015,
      "loss": 1.6553,
      "step": 5871
    },
    {
      "epoch": 0.619409282700422,
      "grad_norm": 0.5852952003479004,
      "learning_rate": 0.0015,
      "loss": 1.6176,
      "step": 5872
    },
    {
      "epoch": 0.6195147679324895,
      "grad_norm": 0.5672625303268433,
      "learning_rate": 0.0015,
      "loss": 1.6019,
      "step": 5873
    },
    {
      "epoch": 0.6196202531645569,
      "grad_norm": 0.5782454013824463,
      "learning_rate": 0.0015,
      "loss": 1.6073,
      "step": 5874
    },
    {
      "epoch": 0.6197257383966245,
      "grad_norm": 0.6658385992050171,
      "learning_rate": 0.0015,
      "loss": 1.6785,
      "step": 5875
    },
    {
      "epoch": 0.619831223628692,
      "grad_norm": 0.5504497289657593,
      "learning_rate": 0.0015,
      "loss": 1.5839,
      "step": 5876
    },
    {
      "epoch": 0.6199367088607595,
      "grad_norm": 0.6169859170913696,
      "learning_rate": 0.0015,
      "loss": 1.6356,
      "step": 5877
    },
    {
      "epoch": 0.620042194092827,
      "grad_norm": 0.7992807030677795,
      "learning_rate": 0.0015,
      "loss": 1.6373,
      "step": 5878
    },
    {
      "epoch": 0.6201476793248946,
      "grad_norm": 0.5499757528305054,
      "learning_rate": 0.0015,
      "loss": 1.6084,
      "step": 5879
    },
    {
      "epoch": 0.620253164556962,
      "grad_norm": 0.6638899445533752,
      "learning_rate": 0.0015,
      "loss": 1.618,
      "step": 5880
    },
    {
      "epoch": 0.6203586497890295,
      "grad_norm": 0.7057284116744995,
      "learning_rate": 0.0015,
      "loss": 1.61,
      "step": 5881
    },
    {
      "epoch": 0.6204641350210971,
      "grad_norm": 0.5079201459884644,
      "learning_rate": 0.0015,
      "loss": 1.6263,
      "step": 5882
    },
    {
      "epoch": 0.6205696202531645,
      "grad_norm": 0.6595799922943115,
      "learning_rate": 0.0015,
      "loss": 1.5922,
      "step": 5883
    },
    {
      "epoch": 0.6206751054852321,
      "grad_norm": 0.5134573578834534,
      "learning_rate": 0.0015,
      "loss": 1.588,
      "step": 5884
    },
    {
      "epoch": 0.6207805907172996,
      "grad_norm": 0.6489550471305847,
      "learning_rate": 0.0015,
      "loss": 1.5965,
      "step": 5885
    },
    {
      "epoch": 0.6208860759493671,
      "grad_norm": 0.5748170614242554,
      "learning_rate": 0.0015,
      "loss": 1.6496,
      "step": 5886
    },
    {
      "epoch": 0.6209915611814346,
      "grad_norm": 0.5478618144989014,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5887
    },
    {
      "epoch": 0.6210970464135022,
      "grad_norm": 0.5693939328193665,
      "learning_rate": 0.0015,
      "loss": 1.5937,
      "step": 5888
    },
    {
      "epoch": 0.6212025316455696,
      "grad_norm": 0.5580261945724487,
      "learning_rate": 0.0015,
      "loss": 1.6273,
      "step": 5889
    },
    {
      "epoch": 0.6213080168776371,
      "grad_norm": 0.6235843300819397,
      "learning_rate": 0.0015,
      "loss": 1.6454,
      "step": 5890
    },
    {
      "epoch": 0.6214135021097047,
      "grad_norm": 0.6731160879135132,
      "learning_rate": 0.0015,
      "loss": 1.5879,
      "step": 5891
    },
    {
      "epoch": 0.6215189873417721,
      "grad_norm": 0.47634485363960266,
      "learning_rate": 0.0015,
      "loss": 1.6291,
      "step": 5892
    },
    {
      "epoch": 0.6216244725738397,
      "grad_norm": 0.6869895458221436,
      "learning_rate": 0.0015,
      "loss": 1.5821,
      "step": 5893
    },
    {
      "epoch": 0.6217299578059071,
      "grad_norm": 0.5553609132766724,
      "learning_rate": 0.0015,
      "loss": 1.5941,
      "step": 5894
    },
    {
      "epoch": 0.6218354430379747,
      "grad_norm": 0.6233806610107422,
      "learning_rate": 0.0015,
      "loss": 1.6253,
      "step": 5895
    },
    {
      "epoch": 0.6219409282700422,
      "grad_norm": 0.7386407852172852,
      "learning_rate": 0.0015,
      "loss": 1.6253,
      "step": 5896
    },
    {
      "epoch": 0.6220464135021097,
      "grad_norm": 0.6310761570930481,
      "learning_rate": 0.0015,
      "loss": 1.6501,
      "step": 5897
    },
    {
      "epoch": 0.6221518987341772,
      "grad_norm": 0.5247330665588379,
      "learning_rate": 0.0015,
      "loss": 1.6417,
      "step": 5898
    },
    {
      "epoch": 0.6222573839662447,
      "grad_norm": 0.644690215587616,
      "learning_rate": 0.0015,
      "loss": 1.5999,
      "step": 5899
    },
    {
      "epoch": 0.6223628691983122,
      "grad_norm": 0.4776199758052826,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5900
    },
    {
      "epoch": 0.6224683544303797,
      "grad_norm": 0.5957555770874023,
      "learning_rate": 0.0015,
      "loss": 1.6078,
      "step": 5901
    },
    {
      "epoch": 0.6225738396624473,
      "grad_norm": 0.6283895373344421,
      "learning_rate": 0.0015,
      "loss": 1.6617,
      "step": 5902
    },
    {
      "epoch": 0.6226793248945147,
      "grad_norm": 0.6419785618782043,
      "learning_rate": 0.0015,
      "loss": 1.5818,
      "step": 5903
    },
    {
      "epoch": 0.6227848101265823,
      "grad_norm": 0.5279337763786316,
      "learning_rate": 0.0015,
      "loss": 1.6173,
      "step": 5904
    },
    {
      "epoch": 0.6228902953586498,
      "grad_norm": 0.7001970410346985,
      "learning_rate": 0.0015,
      "loss": 1.5934,
      "step": 5905
    },
    {
      "epoch": 0.6229957805907173,
      "grad_norm": 0.5482293367385864,
      "learning_rate": 0.0015,
      "loss": 1.6168,
      "step": 5906
    },
    {
      "epoch": 0.6231012658227848,
      "grad_norm": 0.5112176537513733,
      "learning_rate": 0.0015,
      "loss": 1.5851,
      "step": 5907
    },
    {
      "epoch": 0.6232067510548523,
      "grad_norm": 0.5309563279151917,
      "learning_rate": 0.0015,
      "loss": 1.6221,
      "step": 5908
    },
    {
      "epoch": 0.6233122362869198,
      "grad_norm": 0.4815877377986908,
      "learning_rate": 0.0015,
      "loss": 1.621,
      "step": 5909
    },
    {
      "epoch": 0.6234177215189873,
      "grad_norm": 0.5217291712760925,
      "learning_rate": 0.0015,
      "loss": 1.6123,
      "step": 5910
    },
    {
      "epoch": 0.6235232067510549,
      "grad_norm": 0.503728985786438,
      "learning_rate": 0.0015,
      "loss": 1.5732,
      "step": 5911
    },
    {
      "epoch": 0.6236286919831223,
      "grad_norm": 0.5668125152587891,
      "learning_rate": 0.0015,
      "loss": 1.6498,
      "step": 5912
    },
    {
      "epoch": 0.6237341772151899,
      "grad_norm": 0.5331942439079285,
      "learning_rate": 0.0015,
      "loss": 1.5858,
      "step": 5913
    },
    {
      "epoch": 0.6238396624472574,
      "grad_norm": 0.6289189457893372,
      "learning_rate": 0.0015,
      "loss": 1.5957,
      "step": 5914
    },
    {
      "epoch": 0.6239451476793249,
      "grad_norm": 0.5665965676307678,
      "learning_rate": 0.0015,
      "loss": 1.5923,
      "step": 5915
    },
    {
      "epoch": 0.6240506329113924,
      "grad_norm": 0.5158137083053589,
      "learning_rate": 0.0015,
      "loss": 1.6104,
      "step": 5916
    },
    {
      "epoch": 0.62415611814346,
      "grad_norm": 0.5021527409553528,
      "learning_rate": 0.0015,
      "loss": 1.6349,
      "step": 5917
    },
    {
      "epoch": 0.6242616033755274,
      "grad_norm": 0.5725732445716858,
      "learning_rate": 0.0015,
      "loss": 1.6131,
      "step": 5918
    },
    {
      "epoch": 0.6243670886075949,
      "grad_norm": 0.7081011533737183,
      "learning_rate": 0.0015,
      "loss": 1.6149,
      "step": 5919
    },
    {
      "epoch": 0.6244725738396625,
      "grad_norm": 0.5116013884544373,
      "learning_rate": 0.0015,
      "loss": 1.6051,
      "step": 5920
    },
    {
      "epoch": 0.6245780590717299,
      "grad_norm": 0.6848795413970947,
      "learning_rate": 0.0015,
      "loss": 1.6293,
      "step": 5921
    },
    {
      "epoch": 0.6246835443037975,
      "grad_norm": 0.7860782742500305,
      "learning_rate": 0.0015,
      "loss": 1.5985,
      "step": 5922
    },
    {
      "epoch": 0.624789029535865,
      "grad_norm": 0.44655081629753113,
      "learning_rate": 0.0015,
      "loss": 1.625,
      "step": 5923
    },
    {
      "epoch": 0.6248945147679325,
      "grad_norm": 0.7083460688591003,
      "learning_rate": 0.0015,
      "loss": 1.6344,
      "step": 5924
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.7011964917182922,
      "learning_rate": 0.0015,
      "loss": 1.633,
      "step": 5925
    },
    {
      "epoch": 0.6251054852320675,
      "grad_norm": 0.5411670804023743,
      "learning_rate": 0.0015,
      "loss": 1.5601,
      "step": 5926
    },
    {
      "epoch": 0.625210970464135,
      "grad_norm": 0.8851147294044495,
      "learning_rate": 0.0015,
      "loss": 1.6247,
      "step": 5927
    },
    {
      "epoch": 0.6253164556962025,
      "grad_norm": 0.8171315789222717,
      "learning_rate": 0.0015,
      "loss": 1.6091,
      "step": 5928
    },
    {
      "epoch": 0.6254219409282701,
      "grad_norm": 0.6932423710823059,
      "learning_rate": 0.0015,
      "loss": 1.603,
      "step": 5929
    },
    {
      "epoch": 0.6255274261603375,
      "grad_norm": 1.0889321565628052,
      "learning_rate": 0.0015,
      "loss": 1.6293,
      "step": 5930
    },
    {
      "epoch": 0.6256329113924051,
      "grad_norm": 0.7974080443382263,
      "learning_rate": 0.0015,
      "loss": 1.6181,
      "step": 5931
    },
    {
      "epoch": 0.6257383966244726,
      "grad_norm": 0.5604573488235474,
      "learning_rate": 0.0015,
      "loss": 1.6313,
      "step": 5932
    },
    {
      "epoch": 0.62584388185654,
      "grad_norm": 0.8448460698127747,
      "learning_rate": 0.0015,
      "loss": 1.6522,
      "step": 5933
    },
    {
      "epoch": 0.6259493670886076,
      "grad_norm": 0.5236601829528809,
      "learning_rate": 0.0015,
      "loss": 1.6332,
      "step": 5934
    },
    {
      "epoch": 0.6260548523206751,
      "grad_norm": 0.7095980048179626,
      "learning_rate": 0.0015,
      "loss": 1.6279,
      "step": 5935
    },
    {
      "epoch": 0.6261603375527426,
      "grad_norm": 0.628194272518158,
      "learning_rate": 0.0015,
      "loss": 1.6135,
      "step": 5936
    },
    {
      "epoch": 0.6262658227848101,
      "grad_norm": 0.5797224640846252,
      "learning_rate": 0.0015,
      "loss": 1.6225,
      "step": 5937
    },
    {
      "epoch": 0.6263713080168777,
      "grad_norm": 0.5181826949119568,
      "learning_rate": 0.0015,
      "loss": 1.5771,
      "step": 5938
    },
    {
      "epoch": 0.6264767932489451,
      "grad_norm": 0.5678123235702515,
      "learning_rate": 0.0015,
      "loss": 1.5644,
      "step": 5939
    },
    {
      "epoch": 0.6265822784810127,
      "grad_norm": 0.5376322269439697,
      "learning_rate": 0.0015,
      "loss": 1.6502,
      "step": 5940
    },
    {
      "epoch": 0.6266877637130802,
      "grad_norm": 0.6702601909637451,
      "learning_rate": 0.0015,
      "loss": 1.6377,
      "step": 5941
    },
    {
      "epoch": 0.6267932489451477,
      "grad_norm": 0.5268521904945374,
      "learning_rate": 0.0015,
      "loss": 1.6228,
      "step": 5942
    },
    {
      "epoch": 0.6268987341772152,
      "grad_norm": 0.5045256614685059,
      "learning_rate": 0.0015,
      "loss": 1.5645,
      "step": 5943
    },
    {
      "epoch": 0.6270042194092827,
      "grad_norm": 0.5835480690002441,
      "learning_rate": 0.0015,
      "loss": 1.6261,
      "step": 5944
    },
    {
      "epoch": 0.6271097046413502,
      "grad_norm": 0.512340247631073,
      "learning_rate": 0.0015,
      "loss": 1.5797,
      "step": 5945
    },
    {
      "epoch": 0.6272151898734177,
      "grad_norm": 0.4840671122074127,
      "learning_rate": 0.0015,
      "loss": 1.6288,
      "step": 5946
    },
    {
      "epoch": 0.6273206751054853,
      "grad_norm": 0.5077359080314636,
      "learning_rate": 0.0015,
      "loss": 1.6037,
      "step": 5947
    },
    {
      "epoch": 0.6274261603375527,
      "grad_norm": 0.5138733983039856,
      "learning_rate": 0.0015,
      "loss": 1.5926,
      "step": 5948
    },
    {
      "epoch": 0.6275316455696203,
      "grad_norm": 0.5429332852363586,
      "learning_rate": 0.0015,
      "loss": 1.6139,
      "step": 5949
    },
    {
      "epoch": 0.6276371308016878,
      "grad_norm": 0.5711822509765625,
      "learning_rate": 0.0015,
      "loss": 1.6051,
      "step": 5950
    },
    {
      "epoch": 0.6277426160337553,
      "grad_norm": 0.5061172842979431,
      "learning_rate": 0.0015,
      "loss": 1.6585,
      "step": 5951
    },
    {
      "epoch": 0.6278481012658228,
      "grad_norm": 0.572616696357727,
      "learning_rate": 0.0015,
      "loss": 1.6299,
      "step": 5952
    },
    {
      "epoch": 0.6279535864978903,
      "grad_norm": 0.49323344230651855,
      "learning_rate": 0.0015,
      "loss": 1.6045,
      "step": 5953
    },
    {
      "epoch": 0.6280590717299578,
      "grad_norm": 0.5708640217781067,
      "learning_rate": 0.0015,
      "loss": 1.5992,
      "step": 5954
    },
    {
      "epoch": 0.6281645569620253,
      "grad_norm": 0.623921275138855,
      "learning_rate": 0.0015,
      "loss": 1.6263,
      "step": 5955
    },
    {
      "epoch": 0.6282700421940929,
      "grad_norm": 0.6181391477584839,
      "learning_rate": 0.0015,
      "loss": 1.6326,
      "step": 5956
    },
    {
      "epoch": 0.6283755274261603,
      "grad_norm": 0.6751028895378113,
      "learning_rate": 0.0015,
      "loss": 1.5871,
      "step": 5957
    },
    {
      "epoch": 0.6284810126582279,
      "grad_norm": 0.5583983063697815,
      "learning_rate": 0.0015,
      "loss": 1.5975,
      "step": 5958
    },
    {
      "epoch": 0.6285864978902953,
      "grad_norm": 0.6823524832725525,
      "learning_rate": 0.0015,
      "loss": 1.5976,
      "step": 5959
    },
    {
      "epoch": 0.6286919831223629,
      "grad_norm": 0.7429535984992981,
      "learning_rate": 0.0015,
      "loss": 1.6315,
      "step": 5960
    },
    {
      "epoch": 0.6287974683544304,
      "grad_norm": 0.5538409352302551,
      "learning_rate": 0.0015,
      "loss": 1.6202,
      "step": 5961
    },
    {
      "epoch": 0.6289029535864978,
      "grad_norm": 0.9482542276382446,
      "learning_rate": 0.0015,
      "loss": 1.6132,
      "step": 5962
    },
    {
      "epoch": 0.6290084388185654,
      "grad_norm": 0.7959316372871399,
      "learning_rate": 0.0015,
      "loss": 1.6312,
      "step": 5963
    },
    {
      "epoch": 0.6291139240506329,
      "grad_norm": 0.6570045948028564,
      "learning_rate": 0.0015,
      "loss": 1.6305,
      "step": 5964
    },
    {
      "epoch": 0.6292194092827004,
      "grad_norm": 0.6505491733551025,
      "learning_rate": 0.0015,
      "loss": 1.6524,
      "step": 5965
    },
    {
      "epoch": 0.6293248945147679,
      "grad_norm": 0.5877590179443359,
      "learning_rate": 0.0015,
      "loss": 1.6133,
      "step": 5966
    },
    {
      "epoch": 0.6294303797468355,
      "grad_norm": 0.5810253024101257,
      "learning_rate": 0.0015,
      "loss": 1.6161,
      "step": 5967
    },
    {
      "epoch": 0.6295358649789029,
      "grad_norm": 0.571105420589447,
      "learning_rate": 0.0015,
      "loss": 1.5749,
      "step": 5968
    },
    {
      "epoch": 0.6296413502109705,
      "grad_norm": 0.5433521270751953,
      "learning_rate": 0.0015,
      "loss": 1.588,
      "step": 5969
    },
    {
      "epoch": 0.629746835443038,
      "grad_norm": 0.5073991417884827,
      "learning_rate": 0.0015,
      "loss": 1.6402,
      "step": 5970
    },
    {
      "epoch": 0.6298523206751054,
      "grad_norm": 0.6024935245513916,
      "learning_rate": 0.0015,
      "loss": 1.6367,
      "step": 5971
    },
    {
      "epoch": 0.629957805907173,
      "grad_norm": 0.4808545708656311,
      "learning_rate": 0.0015,
      "loss": 1.5751,
      "step": 5972
    },
    {
      "epoch": 0.6300632911392405,
      "grad_norm": 0.5532571077346802,
      "learning_rate": 0.0015,
      "loss": 1.6197,
      "step": 5973
    },
    {
      "epoch": 0.630168776371308,
      "grad_norm": 0.566626250743866,
      "learning_rate": 0.0015,
      "loss": 1.6278,
      "step": 5974
    },
    {
      "epoch": 0.6302742616033755,
      "grad_norm": 0.6555110812187195,
      "learning_rate": 0.0015,
      "loss": 1.6046,
      "step": 5975
    },
    {
      "epoch": 0.6303797468354431,
      "grad_norm": 0.6266216039657593,
      "learning_rate": 0.0015,
      "loss": 1.6078,
      "step": 5976
    },
    {
      "epoch": 0.6304852320675105,
      "grad_norm": 0.5250189304351807,
      "learning_rate": 0.0015,
      "loss": 1.5797,
      "step": 5977
    },
    {
      "epoch": 0.630590717299578,
      "grad_norm": 0.654657244682312,
      "learning_rate": 0.0015,
      "loss": 1.613,
      "step": 5978
    },
    {
      "epoch": 0.6306962025316456,
      "grad_norm": 0.5831957459449768,
      "learning_rate": 0.0015,
      "loss": 1.5601,
      "step": 5979
    },
    {
      "epoch": 0.630801687763713,
      "grad_norm": 0.5356829166412354,
      "learning_rate": 0.0015,
      "loss": 1.5873,
      "step": 5980
    },
    {
      "epoch": 0.6309071729957806,
      "grad_norm": 0.6585840582847595,
      "learning_rate": 0.0015,
      "loss": 1.6178,
      "step": 5981
    },
    {
      "epoch": 0.6310126582278481,
      "grad_norm": 0.5871908068656921,
      "learning_rate": 0.0015,
      "loss": 1.6202,
      "step": 5982
    },
    {
      "epoch": 0.6311181434599156,
      "grad_norm": 0.6139295697212219,
      "learning_rate": 0.0015,
      "loss": 1.5875,
      "step": 5983
    },
    {
      "epoch": 0.6312236286919831,
      "grad_norm": 0.5918750762939453,
      "learning_rate": 0.0015,
      "loss": 1.6258,
      "step": 5984
    },
    {
      "epoch": 0.6313291139240507,
      "grad_norm": 0.600638210773468,
      "learning_rate": 0.0015,
      "loss": 1.6141,
      "step": 5985
    },
    {
      "epoch": 0.6314345991561181,
      "grad_norm": 0.5739317536354065,
      "learning_rate": 0.0015,
      "loss": 1.6324,
      "step": 5986
    },
    {
      "epoch": 0.6315400843881857,
      "grad_norm": 0.7033628225326538,
      "learning_rate": 0.0015,
      "loss": 1.6094,
      "step": 5987
    },
    {
      "epoch": 0.6316455696202532,
      "grad_norm": 0.6276975274085999,
      "learning_rate": 0.0015,
      "loss": 1.6339,
      "step": 5988
    },
    {
      "epoch": 0.6317510548523206,
      "grad_norm": 0.8046818375587463,
      "learning_rate": 0.0015,
      "loss": 1.654,
      "step": 5989
    },
    {
      "epoch": 0.6318565400843882,
      "grad_norm": 0.5364254117012024,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 5990
    },
    {
      "epoch": 0.6319620253164557,
      "grad_norm": 0.6985353231430054,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 5991
    },
    {
      "epoch": 0.6320675105485232,
      "grad_norm": 0.6330892443656921,
      "learning_rate": 0.0015,
      "loss": 1.6067,
      "step": 5992
    },
    {
      "epoch": 0.6321729957805907,
      "grad_norm": 0.5060620307922363,
      "learning_rate": 0.0015,
      "loss": 1.6879,
      "step": 5993
    },
    {
      "epoch": 0.6322784810126583,
      "grad_norm": 0.5791623592376709,
      "learning_rate": 0.0015,
      "loss": 1.6367,
      "step": 5994
    },
    {
      "epoch": 0.6323839662447257,
      "grad_norm": 0.5025582313537598,
      "learning_rate": 0.0015,
      "loss": 1.5957,
      "step": 5995
    },
    {
      "epoch": 0.6324894514767933,
      "grad_norm": 0.662771463394165,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 5996
    },
    {
      "epoch": 0.6325949367088608,
      "grad_norm": 0.6416019201278687,
      "learning_rate": 0.0015,
      "loss": 1.6356,
      "step": 5997
    },
    {
      "epoch": 0.6327004219409282,
      "grad_norm": 0.5294812321662903,
      "learning_rate": 0.0015,
      "loss": 1.6018,
      "step": 5998
    },
    {
      "epoch": 0.6328059071729958,
      "grad_norm": 0.6193150281906128,
      "learning_rate": 0.0015,
      "loss": 1.6267,
      "step": 5999
    },
    {
      "epoch": 0.6329113924050633,
      "grad_norm": 0.6678804755210876,
      "learning_rate": 0.0015,
      "loss": 1.6405,
      "step": 6000
    },
    {
      "epoch": 0.6330168776371308,
      "grad_norm": 0.46960487961769104,
      "learning_rate": 0.0015,
      "loss": 1.6028,
      "step": 6001
    },
    {
      "epoch": 0.6331223628691983,
      "grad_norm": 0.6118060946464539,
      "learning_rate": 0.0015,
      "loss": 1.5838,
      "step": 6002
    },
    {
      "epoch": 0.6332278481012659,
      "grad_norm": 0.6287263035774231,
      "learning_rate": 0.0015,
      "loss": 1.6326,
      "step": 6003
    },
    {
      "epoch": 0.6333333333333333,
      "grad_norm": 0.5200289487838745,
      "learning_rate": 0.0015,
      "loss": 1.6523,
      "step": 6004
    },
    {
      "epoch": 0.6334388185654009,
      "grad_norm": 0.49725160002708435,
      "learning_rate": 0.0015,
      "loss": 1.6041,
      "step": 6005
    },
    {
      "epoch": 0.6335443037974684,
      "grad_norm": 0.49891433119773865,
      "learning_rate": 0.0015,
      "loss": 1.6142,
      "step": 6006
    },
    {
      "epoch": 0.6336497890295358,
      "grad_norm": 0.6071246266365051,
      "learning_rate": 0.0015,
      "loss": 1.6106,
      "step": 6007
    },
    {
      "epoch": 0.6337552742616034,
      "grad_norm": 0.4743974208831787,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 6008
    },
    {
      "epoch": 0.6338607594936709,
      "grad_norm": 0.6092520356178284,
      "learning_rate": 0.0015,
      "loss": 1.643,
      "step": 6009
    },
    {
      "epoch": 0.6339662447257384,
      "grad_norm": 0.6248684525489807,
      "learning_rate": 0.0015,
      "loss": 1.5923,
      "step": 6010
    },
    {
      "epoch": 0.6340717299578059,
      "grad_norm": 0.4810630679130554,
      "learning_rate": 0.0015,
      "loss": 1.5926,
      "step": 6011
    },
    {
      "epoch": 0.6341772151898735,
      "grad_norm": 0.7272651195526123,
      "learning_rate": 0.0015,
      "loss": 1.6445,
      "step": 6012
    },
    {
      "epoch": 0.6342827004219409,
      "grad_norm": 0.8365352749824524,
      "learning_rate": 0.0015,
      "loss": 1.6409,
      "step": 6013
    },
    {
      "epoch": 0.6343881856540085,
      "grad_norm": 0.6300891637802124,
      "learning_rate": 0.0015,
      "loss": 1.6287,
      "step": 6014
    },
    {
      "epoch": 0.634493670886076,
      "grad_norm": 0.5885586142539978,
      "learning_rate": 0.0015,
      "loss": 1.6237,
      "step": 6015
    },
    {
      "epoch": 0.6345991561181434,
      "grad_norm": 0.6359388828277588,
      "learning_rate": 0.0015,
      "loss": 1.6061,
      "step": 6016
    },
    {
      "epoch": 0.634704641350211,
      "grad_norm": 0.5030336380004883,
      "learning_rate": 0.0015,
      "loss": 1.6466,
      "step": 6017
    },
    {
      "epoch": 0.6348101265822785,
      "grad_norm": 0.7077120542526245,
      "learning_rate": 0.0015,
      "loss": 1.5889,
      "step": 6018
    },
    {
      "epoch": 0.634915611814346,
      "grad_norm": 0.5615753531455994,
      "learning_rate": 0.0015,
      "loss": 1.5846,
      "step": 6019
    },
    {
      "epoch": 0.6350210970464135,
      "grad_norm": 0.5832058191299438,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 6020
    },
    {
      "epoch": 0.6351265822784811,
      "grad_norm": 0.7304758429527283,
      "learning_rate": 0.0015,
      "loss": 1.6031,
      "step": 6021
    },
    {
      "epoch": 0.6352320675105485,
      "grad_norm": 0.4968035817146301,
      "learning_rate": 0.0015,
      "loss": 1.5802,
      "step": 6022
    },
    {
      "epoch": 0.635337552742616,
      "grad_norm": 0.5690223574638367,
      "learning_rate": 0.0015,
      "loss": 1.6213,
      "step": 6023
    },
    {
      "epoch": 0.6354430379746835,
      "grad_norm": 0.5893293023109436,
      "learning_rate": 0.0015,
      "loss": 1.5834,
      "step": 6024
    },
    {
      "epoch": 0.635548523206751,
      "grad_norm": 0.45773980021476746,
      "learning_rate": 0.0015,
      "loss": 1.6117,
      "step": 6025
    },
    {
      "epoch": 0.6356540084388186,
      "grad_norm": 0.7095865607261658,
      "learning_rate": 0.0015,
      "loss": 1.5999,
      "step": 6026
    },
    {
      "epoch": 0.635759493670886,
      "grad_norm": 0.6526431441307068,
      "learning_rate": 0.0015,
      "loss": 1.6153,
      "step": 6027
    },
    {
      "epoch": 0.6358649789029536,
      "grad_norm": 0.4970572590827942,
      "learning_rate": 0.0015,
      "loss": 1.6235,
      "step": 6028
    },
    {
      "epoch": 0.6359704641350211,
      "grad_norm": 0.5066674947738647,
      "learning_rate": 0.0015,
      "loss": 1.5961,
      "step": 6029
    },
    {
      "epoch": 0.6360759493670886,
      "grad_norm": 0.5693353414535522,
      "learning_rate": 0.0015,
      "loss": 1.6099,
      "step": 6030
    },
    {
      "epoch": 0.6361814345991561,
      "grad_norm": 0.48609164357185364,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 6031
    },
    {
      "epoch": 0.6362869198312237,
      "grad_norm": 0.5931279063224792,
      "learning_rate": 0.0015,
      "loss": 1.6172,
      "step": 6032
    },
    {
      "epoch": 0.6363924050632911,
      "grad_norm": 0.5124304890632629,
      "learning_rate": 0.0015,
      "loss": 1.61,
      "step": 6033
    },
    {
      "epoch": 0.6364978902953586,
      "grad_norm": 0.5319472551345825,
      "learning_rate": 0.0015,
      "loss": 1.6412,
      "step": 6034
    },
    {
      "epoch": 0.6366033755274262,
      "grad_norm": 0.590932309627533,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 6035
    },
    {
      "epoch": 0.6367088607594936,
      "grad_norm": 0.5260143876075745,
      "learning_rate": 0.0015,
      "loss": 1.6013,
      "step": 6036
    },
    {
      "epoch": 0.6368143459915612,
      "grad_norm": 0.5621371269226074,
      "learning_rate": 0.0015,
      "loss": 1.6234,
      "step": 6037
    },
    {
      "epoch": 0.6369198312236287,
      "grad_norm": 0.6405736207962036,
      "learning_rate": 0.0015,
      "loss": 1.616,
      "step": 6038
    },
    {
      "epoch": 0.6370253164556962,
      "grad_norm": 0.5318799614906311,
      "learning_rate": 0.0015,
      "loss": 1.5909,
      "step": 6039
    },
    {
      "epoch": 0.6371308016877637,
      "grad_norm": 0.79112708568573,
      "learning_rate": 0.0015,
      "loss": 1.6369,
      "step": 6040
    },
    {
      "epoch": 0.6372362869198313,
      "grad_norm": 0.7743318676948547,
      "learning_rate": 0.0015,
      "loss": 1.5944,
      "step": 6041
    },
    {
      "epoch": 0.6373417721518987,
      "grad_norm": 0.5720817446708679,
      "learning_rate": 0.0015,
      "loss": 1.6102,
      "step": 6042
    },
    {
      "epoch": 0.6374472573839662,
      "grad_norm": 0.5739864706993103,
      "learning_rate": 0.0015,
      "loss": 1.6027,
      "step": 6043
    },
    {
      "epoch": 0.6375527426160338,
      "grad_norm": 0.5801692605018616,
      "learning_rate": 0.0015,
      "loss": 1.5922,
      "step": 6044
    },
    {
      "epoch": 0.6376582278481012,
      "grad_norm": 0.5075503587722778,
      "learning_rate": 0.0015,
      "loss": 1.5981,
      "step": 6045
    },
    {
      "epoch": 0.6377637130801688,
      "grad_norm": 0.5207285284996033,
      "learning_rate": 0.0015,
      "loss": 1.5909,
      "step": 6046
    },
    {
      "epoch": 0.6378691983122363,
      "grad_norm": 0.6117684245109558,
      "learning_rate": 0.0015,
      "loss": 1.6199,
      "step": 6047
    },
    {
      "epoch": 0.6379746835443038,
      "grad_norm": 0.513938844203949,
      "learning_rate": 0.0015,
      "loss": 1.6304,
      "step": 6048
    },
    {
      "epoch": 0.6380801687763713,
      "grad_norm": 0.7167081236839294,
      "learning_rate": 0.0015,
      "loss": 1.6011,
      "step": 6049
    },
    {
      "epoch": 0.6381856540084389,
      "grad_norm": 0.5096848607063293,
      "learning_rate": 0.0015,
      "loss": 1.6114,
      "step": 6050
    },
    {
      "epoch": 0.6382911392405063,
      "grad_norm": 0.7479782104492188,
      "learning_rate": 0.0015,
      "loss": 1.6132,
      "step": 6051
    },
    {
      "epoch": 0.6383966244725738,
      "grad_norm": 0.6062121987342834,
      "learning_rate": 0.0015,
      "loss": 1.6583,
      "step": 6052
    },
    {
      "epoch": 0.6385021097046414,
      "grad_norm": 0.5051358938217163,
      "learning_rate": 0.0015,
      "loss": 1.5881,
      "step": 6053
    },
    {
      "epoch": 0.6386075949367088,
      "grad_norm": 0.5258079171180725,
      "learning_rate": 0.0015,
      "loss": 1.6374,
      "step": 6054
    },
    {
      "epoch": 0.6387130801687764,
      "grad_norm": 0.48984280228614807,
      "learning_rate": 0.0015,
      "loss": 1.5783,
      "step": 6055
    },
    {
      "epoch": 0.6388185654008439,
      "grad_norm": 0.49099859595298767,
      "learning_rate": 0.0015,
      "loss": 1.5901,
      "step": 6056
    },
    {
      "epoch": 0.6389240506329114,
      "grad_norm": 0.47011294960975647,
      "learning_rate": 0.0015,
      "loss": 1.622,
      "step": 6057
    },
    {
      "epoch": 0.6390295358649789,
      "grad_norm": 0.5314732789993286,
      "learning_rate": 0.0015,
      "loss": 1.6279,
      "step": 6058
    },
    {
      "epoch": 0.6391350210970465,
      "grad_norm": 0.4974585771560669,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 6059
    },
    {
      "epoch": 0.6392405063291139,
      "grad_norm": 0.5173642635345459,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 6060
    },
    {
      "epoch": 0.6393459915611814,
      "grad_norm": 0.5478307604789734,
      "learning_rate": 0.0015,
      "loss": 1.6224,
      "step": 6061
    },
    {
      "epoch": 0.639451476793249,
      "grad_norm": 0.5353797078132629,
      "learning_rate": 0.0015,
      "loss": 1.5966,
      "step": 6062
    },
    {
      "epoch": 0.6395569620253164,
      "grad_norm": 0.5645928978919983,
      "learning_rate": 0.0015,
      "loss": 1.6005,
      "step": 6063
    },
    {
      "epoch": 0.639662447257384,
      "grad_norm": 0.5212071537971497,
      "learning_rate": 0.0015,
      "loss": 1.6645,
      "step": 6064
    },
    {
      "epoch": 0.6397679324894515,
      "grad_norm": 0.5072381496429443,
      "learning_rate": 0.0015,
      "loss": 1.593,
      "step": 6065
    },
    {
      "epoch": 0.639873417721519,
      "grad_norm": 0.5284490585327148,
      "learning_rate": 0.0015,
      "loss": 1.5814,
      "step": 6066
    },
    {
      "epoch": 0.6399789029535865,
      "grad_norm": 0.5199257135391235,
      "learning_rate": 0.0015,
      "loss": 1.5931,
      "step": 6067
    },
    {
      "epoch": 0.640084388185654,
      "grad_norm": 0.4942372143268585,
      "learning_rate": 0.0015,
      "loss": 1.5915,
      "step": 6068
    },
    {
      "epoch": 0.6401898734177215,
      "grad_norm": 0.5032808780670166,
      "learning_rate": 0.0015,
      "loss": 1.6549,
      "step": 6069
    },
    {
      "epoch": 0.640295358649789,
      "grad_norm": 0.610774040222168,
      "learning_rate": 0.0015,
      "loss": 1.6112,
      "step": 6070
    },
    {
      "epoch": 0.6404008438818566,
      "grad_norm": 0.5463709235191345,
      "learning_rate": 0.0015,
      "loss": 1.6137,
      "step": 6071
    },
    {
      "epoch": 0.640506329113924,
      "grad_norm": 0.48926249146461487,
      "learning_rate": 0.0015,
      "loss": 1.6076,
      "step": 6072
    },
    {
      "epoch": 0.6406118143459916,
      "grad_norm": 0.491427481174469,
      "learning_rate": 0.0015,
      "loss": 1.6413,
      "step": 6073
    },
    {
      "epoch": 0.6407172995780591,
      "grad_norm": 0.4824755787849426,
      "learning_rate": 0.0015,
      "loss": 1.6407,
      "step": 6074
    },
    {
      "epoch": 0.6408227848101266,
      "grad_norm": 0.5169383883476257,
      "learning_rate": 0.0015,
      "loss": 1.5855,
      "step": 6075
    },
    {
      "epoch": 0.6409282700421941,
      "grad_norm": 0.6159296631813049,
      "learning_rate": 0.0015,
      "loss": 1.5848,
      "step": 6076
    },
    {
      "epoch": 0.6410337552742617,
      "grad_norm": 0.4957151710987091,
      "learning_rate": 0.0015,
      "loss": 1.6154,
      "step": 6077
    },
    {
      "epoch": 0.6411392405063291,
      "grad_norm": 0.5977268815040588,
      "learning_rate": 0.0015,
      "loss": 1.6075,
      "step": 6078
    },
    {
      "epoch": 0.6412447257383966,
      "grad_norm": 0.7418320178985596,
      "learning_rate": 0.0015,
      "loss": 1.6379,
      "step": 6079
    },
    {
      "epoch": 0.6413502109704642,
      "grad_norm": 0.5622373223304749,
      "learning_rate": 0.0015,
      "loss": 1.5845,
      "step": 6080
    },
    {
      "epoch": 0.6414556962025316,
      "grad_norm": 0.5169886946678162,
      "learning_rate": 0.0015,
      "loss": 1.5967,
      "step": 6081
    },
    {
      "epoch": 0.6415611814345992,
      "grad_norm": 0.5313584208488464,
      "learning_rate": 0.0015,
      "loss": 1.5757,
      "step": 6082
    },
    {
      "epoch": 0.6416666666666667,
      "grad_norm": 0.6163620352745056,
      "learning_rate": 0.0015,
      "loss": 1.6558,
      "step": 6083
    },
    {
      "epoch": 0.6417721518987342,
      "grad_norm": 0.4772838056087494,
      "learning_rate": 0.0015,
      "loss": 1.59,
      "step": 6084
    },
    {
      "epoch": 0.6418776371308017,
      "grad_norm": 0.5533785223960876,
      "learning_rate": 0.0015,
      "loss": 1.5964,
      "step": 6085
    },
    {
      "epoch": 0.6419831223628693,
      "grad_norm": 0.5664098262786865,
      "learning_rate": 0.0015,
      "loss": 1.6369,
      "step": 6086
    },
    {
      "epoch": 0.6420886075949367,
      "grad_norm": 0.6539817452430725,
      "learning_rate": 0.0015,
      "loss": 1.6367,
      "step": 6087
    },
    {
      "epoch": 0.6421940928270042,
      "grad_norm": 0.5820465087890625,
      "learning_rate": 0.0015,
      "loss": 1.5957,
      "step": 6088
    },
    {
      "epoch": 0.6422995780590718,
      "grad_norm": 0.548473596572876,
      "learning_rate": 0.0015,
      "loss": 1.602,
      "step": 6089
    },
    {
      "epoch": 0.6424050632911392,
      "grad_norm": 0.513701856136322,
      "learning_rate": 0.0015,
      "loss": 1.6172,
      "step": 6090
    },
    {
      "epoch": 0.6425105485232068,
      "grad_norm": 0.5873857736587524,
      "learning_rate": 0.0015,
      "loss": 1.6358,
      "step": 6091
    },
    {
      "epoch": 0.6426160337552742,
      "grad_norm": 0.5551254153251648,
      "learning_rate": 0.0015,
      "loss": 1.6041,
      "step": 6092
    },
    {
      "epoch": 0.6427215189873418,
      "grad_norm": 0.5355796813964844,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 6093
    },
    {
      "epoch": 0.6428270042194093,
      "grad_norm": 0.6306756734848022,
      "learning_rate": 0.0015,
      "loss": 1.6073,
      "step": 6094
    },
    {
      "epoch": 0.6429324894514767,
      "grad_norm": 0.7098020315170288,
      "learning_rate": 0.0015,
      "loss": 1.5638,
      "step": 6095
    },
    {
      "epoch": 0.6430379746835443,
      "grad_norm": 0.6527588963508606,
      "learning_rate": 0.0015,
      "loss": 1.6019,
      "step": 6096
    },
    {
      "epoch": 0.6431434599156118,
      "grad_norm": 0.5478804111480713,
      "learning_rate": 0.0015,
      "loss": 1.5903,
      "step": 6097
    },
    {
      "epoch": 0.6432489451476793,
      "grad_norm": 0.5948776006698608,
      "learning_rate": 0.0015,
      "loss": 1.6301,
      "step": 6098
    },
    {
      "epoch": 0.6433544303797468,
      "grad_norm": 0.8403067588806152,
      "learning_rate": 0.0015,
      "loss": 1.5863,
      "step": 6099
    },
    {
      "epoch": 0.6434599156118144,
      "grad_norm": 0.6364478468894958,
      "learning_rate": 0.0015,
      "loss": 1.6109,
      "step": 6100
    },
    {
      "epoch": 0.6435654008438818,
      "grad_norm": 0.6228508949279785,
      "learning_rate": 0.0015,
      "loss": 1.6484,
      "step": 6101
    },
    {
      "epoch": 0.6436708860759494,
      "grad_norm": 0.7421320676803589,
      "learning_rate": 0.0015,
      "loss": 1.5813,
      "step": 6102
    },
    {
      "epoch": 0.6437763713080169,
      "grad_norm": 0.7950869202613831,
      "learning_rate": 0.0015,
      "loss": 1.5982,
      "step": 6103
    },
    {
      "epoch": 0.6438818565400843,
      "grad_norm": 0.5360313653945923,
      "learning_rate": 0.0015,
      "loss": 1.6181,
      "step": 6104
    },
    {
      "epoch": 0.6439873417721519,
      "grad_norm": 0.6788882613182068,
      "learning_rate": 0.0015,
      "loss": 1.6443,
      "step": 6105
    },
    {
      "epoch": 0.6440928270042194,
      "grad_norm": 0.7845962047576904,
      "learning_rate": 0.0015,
      "loss": 1.6162,
      "step": 6106
    },
    {
      "epoch": 0.6441983122362869,
      "grad_norm": 0.46688973903656006,
      "learning_rate": 0.0015,
      "loss": 1.6479,
      "step": 6107
    },
    {
      "epoch": 0.6443037974683544,
      "grad_norm": 0.9376731514930725,
      "learning_rate": 0.0015,
      "loss": 1.6114,
      "step": 6108
    },
    {
      "epoch": 0.644409282700422,
      "grad_norm": 0.9342179894447327,
      "learning_rate": 0.0015,
      "loss": 1.641,
      "step": 6109
    },
    {
      "epoch": 0.6445147679324894,
      "grad_norm": 0.5573187470436096,
      "learning_rate": 0.0015,
      "loss": 1.6083,
      "step": 6110
    },
    {
      "epoch": 0.644620253164557,
      "grad_norm": 1.011339545249939,
      "learning_rate": 0.0015,
      "loss": 1.621,
      "step": 6111
    },
    {
      "epoch": 0.6447257383966245,
      "grad_norm": 0.795906126499176,
      "learning_rate": 0.0015,
      "loss": 1.5891,
      "step": 6112
    },
    {
      "epoch": 0.6448312236286919,
      "grad_norm": 0.6206145882606506,
      "learning_rate": 0.0015,
      "loss": 1.6364,
      "step": 6113
    },
    {
      "epoch": 0.6449367088607595,
      "grad_norm": 0.7847033143043518,
      "learning_rate": 0.0015,
      "loss": 1.6097,
      "step": 6114
    },
    {
      "epoch": 0.645042194092827,
      "grad_norm": 0.5453957915306091,
      "learning_rate": 0.0015,
      "loss": 1.6232,
      "step": 6115
    },
    {
      "epoch": 0.6451476793248945,
      "grad_norm": 0.6635275483131409,
      "learning_rate": 0.0015,
      "loss": 1.645,
      "step": 6116
    },
    {
      "epoch": 0.645253164556962,
      "grad_norm": 0.6184625625610352,
      "learning_rate": 0.0015,
      "loss": 1.5987,
      "step": 6117
    },
    {
      "epoch": 0.6453586497890296,
      "grad_norm": 0.6471688151359558,
      "learning_rate": 0.0015,
      "loss": 1.6335,
      "step": 6118
    },
    {
      "epoch": 0.645464135021097,
      "grad_norm": 0.49944692850112915,
      "learning_rate": 0.0015,
      "loss": 1.597,
      "step": 6119
    },
    {
      "epoch": 0.6455696202531646,
      "grad_norm": 0.5704482197761536,
      "learning_rate": 0.0015,
      "loss": 1.5783,
      "step": 6120
    },
    {
      "epoch": 0.6456751054852321,
      "grad_norm": 0.6205321550369263,
      "learning_rate": 0.0015,
      "loss": 1.6034,
      "step": 6121
    },
    {
      "epoch": 0.6457805907172995,
      "grad_norm": 0.5599362850189209,
      "learning_rate": 0.0015,
      "loss": 1.6301,
      "step": 6122
    },
    {
      "epoch": 0.6458860759493671,
      "grad_norm": 0.6305397152900696,
      "learning_rate": 0.0015,
      "loss": 1.6074,
      "step": 6123
    },
    {
      "epoch": 0.6459915611814346,
      "grad_norm": 0.442531555891037,
      "learning_rate": 0.0015,
      "loss": 1.6088,
      "step": 6124
    },
    {
      "epoch": 0.6460970464135021,
      "grad_norm": 0.5729601383209229,
      "learning_rate": 0.0015,
      "loss": 1.6051,
      "step": 6125
    },
    {
      "epoch": 0.6462025316455696,
      "grad_norm": 0.5429379343986511,
      "learning_rate": 0.0015,
      "loss": 1.6025,
      "step": 6126
    },
    {
      "epoch": 0.6463080168776372,
      "grad_norm": 0.5254256725311279,
      "learning_rate": 0.0015,
      "loss": 1.5951,
      "step": 6127
    },
    {
      "epoch": 0.6464135021097046,
      "grad_norm": 0.5388456583023071,
      "learning_rate": 0.0015,
      "loss": 1.5848,
      "step": 6128
    },
    {
      "epoch": 0.6465189873417722,
      "grad_norm": 0.5493997931480408,
      "learning_rate": 0.0015,
      "loss": 1.6044,
      "step": 6129
    },
    {
      "epoch": 0.6466244725738397,
      "grad_norm": 0.45452308654785156,
      "learning_rate": 0.0015,
      "loss": 1.6456,
      "step": 6130
    },
    {
      "epoch": 0.6467299578059071,
      "grad_norm": 0.5767956972122192,
      "learning_rate": 0.0015,
      "loss": 1.6493,
      "step": 6131
    },
    {
      "epoch": 0.6468354430379747,
      "grad_norm": 0.5556018948554993,
      "learning_rate": 0.0015,
      "loss": 1.6296,
      "step": 6132
    },
    {
      "epoch": 0.6469409282700422,
      "grad_norm": 0.49602171778678894,
      "learning_rate": 0.0015,
      "loss": 1.6,
      "step": 6133
    },
    {
      "epoch": 0.6470464135021097,
      "grad_norm": 0.6708524823188782,
      "learning_rate": 0.0015,
      "loss": 1.6204,
      "step": 6134
    },
    {
      "epoch": 0.6471518987341772,
      "grad_norm": 0.5771464705467224,
      "learning_rate": 0.0015,
      "loss": 1.5842,
      "step": 6135
    },
    {
      "epoch": 0.6472573839662448,
      "grad_norm": 0.5093113780021667,
      "learning_rate": 0.0015,
      "loss": 1.6388,
      "step": 6136
    },
    {
      "epoch": 0.6473628691983122,
      "grad_norm": 0.6123894453048706,
      "learning_rate": 0.0015,
      "loss": 1.6135,
      "step": 6137
    },
    {
      "epoch": 0.6474683544303798,
      "grad_norm": 0.5309610962867737,
      "learning_rate": 0.0015,
      "loss": 1.62,
      "step": 6138
    },
    {
      "epoch": 0.6475738396624473,
      "grad_norm": 0.5773170590400696,
      "learning_rate": 0.0015,
      "loss": 1.6176,
      "step": 6139
    },
    {
      "epoch": 0.6476793248945147,
      "grad_norm": 0.48332715034484863,
      "learning_rate": 0.0015,
      "loss": 1.5468,
      "step": 6140
    },
    {
      "epoch": 0.6477848101265823,
      "grad_norm": 0.6897808909416199,
      "learning_rate": 0.0015,
      "loss": 1.6216,
      "step": 6141
    },
    {
      "epoch": 0.6478902953586498,
      "grad_norm": 0.7130773663520813,
      "learning_rate": 0.0015,
      "loss": 1.5861,
      "step": 6142
    },
    {
      "epoch": 0.6479957805907173,
      "grad_norm": 0.5731837749481201,
      "learning_rate": 0.0015,
      "loss": 1.5846,
      "step": 6143
    },
    {
      "epoch": 0.6481012658227848,
      "grad_norm": 0.6874064207077026,
      "learning_rate": 0.0015,
      "loss": 1.6256,
      "step": 6144
    },
    {
      "epoch": 0.6482067510548524,
      "grad_norm": 0.7943158745765686,
      "learning_rate": 0.0015,
      "loss": 1.6472,
      "step": 6145
    },
    {
      "epoch": 0.6483122362869198,
      "grad_norm": 0.5174962878227234,
      "learning_rate": 0.0015,
      "loss": 1.6209,
      "step": 6146
    },
    {
      "epoch": 0.6484177215189874,
      "grad_norm": 0.781793475151062,
      "learning_rate": 0.0015,
      "loss": 1.6306,
      "step": 6147
    },
    {
      "epoch": 0.6485232067510549,
      "grad_norm": 0.6787018179893494,
      "learning_rate": 0.0015,
      "loss": 1.6049,
      "step": 6148
    },
    {
      "epoch": 0.6486286919831223,
      "grad_norm": 0.6252360343933105,
      "learning_rate": 0.0015,
      "loss": 1.6248,
      "step": 6149
    },
    {
      "epoch": 0.6487341772151899,
      "grad_norm": 0.7669170498847961,
      "learning_rate": 0.0015,
      "loss": 1.6585,
      "step": 6150
    },
    {
      "epoch": 0.6488396624472574,
      "grad_norm": 0.6429520845413208,
      "learning_rate": 0.0015,
      "loss": 1.6207,
      "step": 6151
    },
    {
      "epoch": 0.6489451476793249,
      "grad_norm": 0.5717629790306091,
      "learning_rate": 0.0015,
      "loss": 1.641,
      "step": 6152
    },
    {
      "epoch": 0.6490506329113924,
      "grad_norm": 0.6288654208183289,
      "learning_rate": 0.0015,
      "loss": 1.5998,
      "step": 6153
    },
    {
      "epoch": 0.64915611814346,
      "grad_norm": 0.5095597505569458,
      "learning_rate": 0.0015,
      "loss": 1.595,
      "step": 6154
    },
    {
      "epoch": 0.6492616033755274,
      "grad_norm": 0.6188340783119202,
      "learning_rate": 0.0015,
      "loss": 1.6095,
      "step": 6155
    },
    {
      "epoch": 0.649367088607595,
      "grad_norm": 0.6024876832962036,
      "learning_rate": 0.0015,
      "loss": 1.5919,
      "step": 6156
    },
    {
      "epoch": 0.6494725738396624,
      "grad_norm": 0.5583590865135193,
      "learning_rate": 0.0015,
      "loss": 1.6147,
      "step": 6157
    },
    {
      "epoch": 0.6495780590717299,
      "grad_norm": 0.4911785125732422,
      "learning_rate": 0.0015,
      "loss": 1.643,
      "step": 6158
    },
    {
      "epoch": 0.6496835443037975,
      "grad_norm": 0.520928144454956,
      "learning_rate": 0.0015,
      "loss": 1.576,
      "step": 6159
    },
    {
      "epoch": 0.6497890295358649,
      "grad_norm": 0.5773823261260986,
      "learning_rate": 0.0015,
      "loss": 1.6427,
      "step": 6160
    },
    {
      "epoch": 0.6498945147679325,
      "grad_norm": 0.5181955695152283,
      "learning_rate": 0.0015,
      "loss": 1.6087,
      "step": 6161
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5335739850997925,
      "learning_rate": 0.0015,
      "loss": 1.602,
      "step": 6162
    },
    {
      "epoch": 0.6501054852320675,
      "grad_norm": 0.5832805037498474,
      "learning_rate": 0.0014979195407665976,
      "loss": 1.5409,
      "step": 6163
    },
    {
      "epoch": 0.650210970464135,
      "grad_norm": 0.5168347358703613,
      "learning_rate": 0.00149584196707361,
      "loss": 1.575,
      "step": 6164
    },
    {
      "epoch": 0.6503164556962026,
      "grad_norm": 0.5147069692611694,
      "learning_rate": 0.0014937672749188704,
      "loss": 1.5896,
      "step": 6165
    },
    {
      "epoch": 0.65042194092827,
      "grad_norm": 0.5709660649299622,
      "learning_rate": 0.0014916954603057643,
      "loss": 1.599,
      "step": 6166
    },
    {
      "epoch": 0.6505274261603375,
      "grad_norm": 0.49591460824012756,
      "learning_rate": 0.0014896265192432194,
      "loss": 1.5824,
      "step": 6167
    },
    {
      "epoch": 0.6506329113924051,
      "grad_norm": 0.48474690318107605,
      "learning_rate": 0.001487560447745699,
      "loss": 1.5815,
      "step": 6168
    },
    {
      "epoch": 0.6507383966244725,
      "grad_norm": 0.5037993788719177,
      "learning_rate": 0.0014854972418331944,
      "loss": 1.6377,
      "step": 6169
    },
    {
      "epoch": 0.6508438818565401,
      "grad_norm": 0.4749622046947479,
      "learning_rate": 0.0014834368975312174,
      "loss": 1.5855,
      "step": 6170
    },
    {
      "epoch": 0.6509493670886076,
      "grad_norm": 0.5159726738929749,
      "learning_rate": 0.0014813794108707917,
      "loss": 1.609,
      "step": 6171
    },
    {
      "epoch": 0.6510548523206751,
      "grad_norm": 0.469255656003952,
      "learning_rate": 0.0014793247778884461,
      "loss": 1.5809,
      "step": 6172
    },
    {
      "epoch": 0.6511603375527426,
      "grad_norm": 0.5568392276763916,
      "learning_rate": 0.0014772729946262069,
      "loss": 1.6286,
      "step": 6173
    },
    {
      "epoch": 0.6512658227848102,
      "grad_norm": 0.6769633293151855,
      "learning_rate": 0.0014752240571315894,
      "loss": 1.6391,
      "step": 6174
    },
    {
      "epoch": 0.6513713080168776,
      "grad_norm": 0.5625841021537781,
      "learning_rate": 0.0014731779614575917,
      "loss": 1.5954,
      "step": 6175
    },
    {
      "epoch": 0.6514767932489451,
      "grad_norm": 0.6080664396286011,
      "learning_rate": 0.0014711347036626854,
      "loss": 1.6225,
      "step": 6176
    },
    {
      "epoch": 0.6515822784810127,
      "grad_norm": 0.7148519158363342,
      "learning_rate": 0.0014690942798108097,
      "loss": 1.6261,
      "step": 6177
    },
    {
      "epoch": 0.6516877637130801,
      "grad_norm": 0.5286668539047241,
      "learning_rate": 0.0014670566859713624,
      "loss": 1.6047,
      "step": 6178
    },
    {
      "epoch": 0.6517932489451477,
      "grad_norm": 0.570422887802124,
      "learning_rate": 0.0014650219182191931,
      "loss": 1.6317,
      "step": 6179
    },
    {
      "epoch": 0.6518987341772152,
      "grad_norm": 0.533623218536377,
      "learning_rate": 0.0014629899726345957,
      "loss": 1.6166,
      "step": 6180
    },
    {
      "epoch": 0.6520042194092827,
      "grad_norm": 0.49662795662879944,
      "learning_rate": 0.0014609608453033007,
      "loss": 1.6382,
      "step": 6181
    },
    {
      "epoch": 0.6521097046413502,
      "grad_norm": 0.5145726799964905,
      "learning_rate": 0.001458934532316467,
      "loss": 1.626,
      "step": 6182
    },
    {
      "epoch": 0.6522151898734178,
      "grad_norm": 0.5761494636535645,
      "learning_rate": 0.0014569110297706755,
      "loss": 1.6098,
      "step": 6183
    },
    {
      "epoch": 0.6523206751054852,
      "grad_norm": 0.4725034832954407,
      "learning_rate": 0.0014548903337679206,
      "loss": 1.6107,
      "step": 6184
    },
    {
      "epoch": 0.6524261603375527,
      "grad_norm": 0.5854225754737854,
      "learning_rate": 0.0014528724404156037,
      "loss": 1.6193,
      "step": 6185
    },
    {
      "epoch": 0.6525316455696203,
      "grad_norm": 0.5390808582305908,
      "learning_rate": 0.0014508573458265248,
      "loss": 1.6286,
      "step": 6186
    },
    {
      "epoch": 0.6526371308016877,
      "grad_norm": 0.6974532008171082,
      "learning_rate": 0.0014488450461188752,
      "loss": 1.5973,
      "step": 6187
    },
    {
      "epoch": 0.6527426160337553,
      "grad_norm": 0.5741993188858032,
      "learning_rate": 0.0014468355374162303,
      "loss": 1.6076,
      "step": 6188
    },
    {
      "epoch": 0.6528481012658228,
      "grad_norm": 0.5944492816925049,
      "learning_rate": 0.001444828815847542,
      "loss": 1.5961,
      "step": 6189
    },
    {
      "epoch": 0.6529535864978903,
      "grad_norm": 0.6624055504798889,
      "learning_rate": 0.0014428248775471316,
      "loss": 1.6427,
      "step": 6190
    },
    {
      "epoch": 0.6530590717299578,
      "grad_norm": 0.5096984505653381,
      "learning_rate": 0.0014408237186546813,
      "loss": 1.6364,
      "step": 6191
    },
    {
      "epoch": 0.6531645569620254,
      "grad_norm": 0.6230143904685974,
      "learning_rate": 0.0014388253353152278,
      "loss": 1.5722,
      "step": 6192
    },
    {
      "epoch": 0.6532700421940928,
      "grad_norm": 0.4900723695755005,
      "learning_rate": 0.0014368297236791545,
      "loss": 1.6226,
      "step": 6193
    },
    {
      "epoch": 0.6533755274261603,
      "grad_norm": 0.5623832941055298,
      "learning_rate": 0.0014348368799021844,
      "loss": 1.5723,
      "step": 6194
    },
    {
      "epoch": 0.6534810126582279,
      "grad_norm": 0.6694051623344421,
      "learning_rate": 0.0014328468001453718,
      "loss": 1.6187,
      "step": 6195
    },
    {
      "epoch": 0.6535864978902953,
      "grad_norm": 0.4565059244632721,
      "learning_rate": 0.001430859480575096,
      "loss": 1.5848,
      "step": 6196
    },
    {
      "epoch": 0.6536919831223629,
      "grad_norm": 0.5141550898551941,
      "learning_rate": 0.0014288749173630535,
      "loss": 1.6061,
      "step": 6197
    },
    {
      "epoch": 0.6537974683544304,
      "grad_norm": 0.46880844235420227,
      "learning_rate": 0.0014268931066862504,
      "loss": 1.6278,
      "step": 6198
    },
    {
      "epoch": 0.6539029535864979,
      "grad_norm": 0.4959510564804077,
      "learning_rate": 0.0014249140447269945,
      "loss": 1.6452,
      "step": 6199
    },
    {
      "epoch": 0.6540084388185654,
      "grad_norm": 0.4055125117301941,
      "learning_rate": 0.00142293772767289,
      "loss": 1.6074,
      "step": 6200
    },
    {
      "epoch": 0.654113924050633,
      "grad_norm": 0.46506351232528687,
      "learning_rate": 0.0014209641517168275,
      "loss": 1.5739,
      "step": 6201
    },
    {
      "epoch": 0.6542194092827004,
      "grad_norm": 0.4965575337409973,
      "learning_rate": 0.001418993313056979,
      "loss": 1.6063,
      "step": 6202
    },
    {
      "epoch": 0.6543248945147679,
      "grad_norm": 0.4766387641429901,
      "learning_rate": 0.0014170252078967885,
      "loss": 1.5947,
      "step": 6203
    },
    {
      "epoch": 0.6544303797468355,
      "grad_norm": 0.5006406307220459,
      "learning_rate": 0.0014150598324449667,
      "loss": 1.6039,
      "step": 6204
    },
    {
      "epoch": 0.6545358649789029,
      "grad_norm": 0.5292978882789612,
      "learning_rate": 0.001413097182915482,
      "loss": 1.6045,
      "step": 6205
    },
    {
      "epoch": 0.6546413502109705,
      "grad_norm": 0.5393363833427429,
      "learning_rate": 0.0014111372555275542,
      "loss": 1.6359,
      "step": 6206
    },
    {
      "epoch": 0.654746835443038,
      "grad_norm": 0.5250663757324219,
      "learning_rate": 0.0014091800465056473,
      "loss": 1.631,
      "step": 6207
    },
    {
      "epoch": 0.6548523206751055,
      "grad_norm": 0.6228719353675842,
      "learning_rate": 0.0014072255520794614,
      "loss": 1.6064,
      "step": 6208
    },
    {
      "epoch": 0.654957805907173,
      "grad_norm": 0.6128799319267273,
      "learning_rate": 0.0014052737684839257,
      "loss": 1.5979,
      "step": 6209
    },
    {
      "epoch": 0.6550632911392406,
      "grad_norm": 0.5618577599525452,
      "learning_rate": 0.001403324691959192,
      "loss": 1.598,
      "step": 6210
    },
    {
      "epoch": 0.655168776371308,
      "grad_norm": 0.5514854788780212,
      "learning_rate": 0.0014013783187506268,
      "loss": 1.618,
      "step": 6211
    },
    {
      "epoch": 0.6552742616033755,
      "grad_norm": 0.5488616824150085,
      "learning_rate": 0.0013994346451088036,
      "loss": 1.6058,
      "step": 6212
    },
    {
      "epoch": 0.6553797468354431,
      "grad_norm": 0.5754488706588745,
      "learning_rate": 0.0013974936672894972,
      "loss": 1.6188,
      "step": 6213
    },
    {
      "epoch": 0.6554852320675105,
      "grad_norm": 0.5307894945144653,
      "learning_rate": 0.0013955553815536747,
      "loss": 1.5847,
      "step": 6214
    },
    {
      "epoch": 0.6555907172995781,
      "grad_norm": 0.5454530119895935,
      "learning_rate": 0.0013936197841674894,
      "loss": 1.6157,
      "step": 6215
    },
    {
      "epoch": 0.6556962025316456,
      "grad_norm": 0.505645215511322,
      "learning_rate": 0.0013916868714022737,
      "loss": 1.6416,
      "step": 6216
    },
    {
      "epoch": 0.6558016877637131,
      "grad_norm": 0.5031726956367493,
      "learning_rate": 0.0013897566395345313,
      "loss": 1.6161,
      "step": 6217
    },
    {
      "epoch": 0.6559071729957806,
      "grad_norm": 0.5135219097137451,
      "learning_rate": 0.0013878290848459301,
      "loss": 1.6263,
      "step": 6218
    },
    {
      "epoch": 0.6560126582278482,
      "grad_norm": 0.5340127944946289,
      "learning_rate": 0.0013859042036232954,
      "loss": 1.5921,
      "step": 6219
    },
    {
      "epoch": 0.6561181434599156,
      "grad_norm": 0.5894213914871216,
      "learning_rate": 0.0013839819921586025,
      "loss": 1.5843,
      "step": 6220
    },
    {
      "epoch": 0.6562236286919831,
      "grad_norm": 0.48287490010261536,
      "learning_rate": 0.00138206244674897,
      "loss": 1.6115,
      "step": 6221
    },
    {
      "epoch": 0.6563291139240506,
      "grad_norm": 0.6512756943702698,
      "learning_rate": 0.0013801455636966516,
      "loss": 1.568,
      "step": 6222
    },
    {
      "epoch": 0.6564345991561181,
      "grad_norm": 0.5339322686195374,
      "learning_rate": 0.0013782313393090303,
      "loss": 1.5928,
      "step": 6223
    },
    {
      "epoch": 0.6565400843881857,
      "grad_norm": 0.5671067237854004,
      "learning_rate": 0.0013763197698986101,
      "loss": 1.5917,
      "step": 6224
    },
    {
      "epoch": 0.6566455696202531,
      "grad_norm": 0.5248642563819885,
      "learning_rate": 0.0013744108517830104,
      "loss": 1.5972,
      "step": 6225
    },
    {
      "epoch": 0.6567510548523207,
      "grad_norm": 0.4802647829055786,
      "learning_rate": 0.0013725045812849569,
      "loss": 1.6163,
      "step": 6226
    },
    {
      "epoch": 0.6568565400843882,
      "grad_norm": 0.5078423619270325,
      "learning_rate": 0.001370600954732276,
      "loss": 1.5962,
      "step": 6227
    },
    {
      "epoch": 0.6569620253164556,
      "grad_norm": 0.49496930837631226,
      "learning_rate": 0.0013686999684578874,
      "loss": 1.5901,
      "step": 6228
    },
    {
      "epoch": 0.6570675105485232,
      "grad_norm": 0.6016704440116882,
      "learning_rate": 0.001366801618799797,
      "loss": 1.5926,
      "step": 6229
    },
    {
      "epoch": 0.6571729957805907,
      "grad_norm": 0.5402101874351501,
      "learning_rate": 0.0013649059021010894,
      "loss": 1.6073,
      "step": 6230
    },
    {
      "epoch": 0.6572784810126582,
      "grad_norm": 0.6456952691078186,
      "learning_rate": 0.0013630128147099215,
      "loss": 1.597,
      "step": 6231
    },
    {
      "epoch": 0.6573839662447257,
      "grad_norm": 0.5368161797523499,
      "learning_rate": 0.0013611223529795156,
      "loss": 1.6236,
      "step": 6232
    },
    {
      "epoch": 0.6574894514767933,
      "grad_norm": 0.5132485032081604,
      "learning_rate": 0.001359234513268151,
      "loss": 1.6261,
      "step": 6233
    },
    {
      "epoch": 0.6575949367088607,
      "grad_norm": 0.5268896818161011,
      "learning_rate": 0.0013573492919391594,
      "loss": 1.6197,
      "step": 6234
    },
    {
      "epoch": 0.6577004219409283,
      "grad_norm": 0.5224751234054565,
      "learning_rate": 0.0013554666853609146,
      "loss": 1.6148,
      "step": 6235
    },
    {
      "epoch": 0.6578059071729958,
      "grad_norm": 0.5029194355010986,
      "learning_rate": 0.001353586689906829,
      "loss": 1.5926,
      "step": 6236
    },
    {
      "epoch": 0.6579113924050632,
      "grad_norm": 0.5476905703544617,
      "learning_rate": 0.0013517093019553442,
      "loss": 1.5828,
      "step": 6237
    },
    {
      "epoch": 0.6580168776371308,
      "grad_norm": 0.47993701696395874,
      "learning_rate": 0.001349834517889925,
      "loss": 1.5973,
      "step": 6238
    },
    {
      "epoch": 0.6581223628691983,
      "grad_norm": 0.4334630072116852,
      "learning_rate": 0.001347962334099052,
      "loss": 1.5764,
      "step": 6239
    },
    {
      "epoch": 0.6582278481012658,
      "grad_norm": 0.5023558735847473,
      "learning_rate": 0.0013460927469762154,
      "loss": 1.5915,
      "step": 6240
    },
    {
      "epoch": 0.6583333333333333,
      "grad_norm": 0.5679836869239807,
      "learning_rate": 0.0013442257529199069,
      "loss": 1.5917,
      "step": 6241
    },
    {
      "epoch": 0.6584388185654009,
      "grad_norm": 0.4634576141834259,
      "learning_rate": 0.0013423613483336142,
      "loss": 1.596,
      "step": 6242
    },
    {
      "epoch": 0.6585443037974683,
      "grad_norm": 0.5537039637565613,
      "learning_rate": 0.001340499529625812,
      "loss": 1.6381,
      "step": 6243
    },
    {
      "epoch": 0.6586497890295359,
      "grad_norm": 0.48177632689476013,
      "learning_rate": 0.0013386402932099575,
      "loss": 1.6078,
      "step": 6244
    },
    {
      "epoch": 0.6587552742616034,
      "grad_norm": 0.5461094379425049,
      "learning_rate": 0.0013367836355044822,
      "loss": 1.6046,
      "step": 6245
    },
    {
      "epoch": 0.6588607594936708,
      "grad_norm": 0.5001852512359619,
      "learning_rate": 0.0013349295529327845,
      "loss": 1.57,
      "step": 6246
    },
    {
      "epoch": 0.6589662447257384,
      "grad_norm": 0.5089578628540039,
      "learning_rate": 0.0013330780419232241,
      "loss": 1.6072,
      "step": 6247
    },
    {
      "epoch": 0.6590717299578059,
      "grad_norm": 0.513290524482727,
      "learning_rate": 0.001331229098909114,
      "loss": 1.5841,
      "step": 6248
    },
    {
      "epoch": 0.6591772151898734,
      "grad_norm": 0.6138255000114441,
      "learning_rate": 0.0013293827203287143,
      "loss": 1.6423,
      "step": 6249
    },
    {
      "epoch": 0.6592827004219409,
      "grad_norm": 0.518519401550293,
      "learning_rate": 0.0013275389026252255,
      "loss": 1.6392,
      "step": 6250
    },
    {
      "epoch": 0.6593881856540085,
      "grad_norm": 0.5507919192314148,
      "learning_rate": 0.0013256976422467803,
      "loss": 1.5877,
      "step": 6251
    },
    {
      "epoch": 0.6594936708860759,
      "grad_norm": 0.5733009576797485,
      "learning_rate": 0.001323858935646439,
      "loss": 1.6122,
      "step": 6252
    },
    {
      "epoch": 0.6595991561181435,
      "grad_norm": 0.5359091758728027,
      "learning_rate": 0.0013220227792821804,
      "loss": 1.5781,
      "step": 6253
    },
    {
      "epoch": 0.659704641350211,
      "grad_norm": 0.6037209630012512,
      "learning_rate": 0.0013201891696168965,
      "loss": 1.6048,
      "step": 6254
    },
    {
      "epoch": 0.6598101265822784,
      "grad_norm": 0.5205094814300537,
      "learning_rate": 0.001318358103118385,
      "loss": 1.6156,
      "step": 6255
    },
    {
      "epoch": 0.659915611814346,
      "grad_norm": 0.5952802300453186,
      "learning_rate": 0.0013165295762593426,
      "loss": 1.6016,
      "step": 6256
    },
    {
      "epoch": 0.6600210970464135,
      "grad_norm": 0.4881475865840912,
      "learning_rate": 0.0013147035855173587,
      "loss": 1.5574,
      "step": 6257
    },
    {
      "epoch": 0.660126582278481,
      "grad_norm": 0.5686355233192444,
      "learning_rate": 0.0013128801273749075,
      "loss": 1.6119,
      "step": 6258
    },
    {
      "epoch": 0.6602320675105485,
      "grad_norm": 0.5342782735824585,
      "learning_rate": 0.0013110591983193423,
      "loss": 1.6093,
      "step": 6259
    },
    {
      "epoch": 0.6603375527426161,
      "grad_norm": 0.5252060294151306,
      "learning_rate": 0.0013092407948428887,
      "loss": 1.6088,
      "step": 6260
    },
    {
      "epoch": 0.6604430379746835,
      "grad_norm": 0.5563986897468567,
      "learning_rate": 0.0013074249134426368,
      "loss": 1.6147,
      "step": 6261
    },
    {
      "epoch": 0.6605485232067511,
      "grad_norm": 0.4771369397640228,
      "learning_rate": 0.0013056115506205354,
      "loss": 1.6167,
      "step": 6262
    },
    {
      "epoch": 0.6606540084388186,
      "grad_norm": 0.5097554326057434,
      "learning_rate": 0.0013038007028833853,
      "loss": 1.5617,
      "step": 6263
    },
    {
      "epoch": 0.660759493670886,
      "grad_norm": 0.560704231262207,
      "learning_rate": 0.001301992366742832,
      "loss": 1.6432,
      "step": 6264
    },
    {
      "epoch": 0.6608649789029536,
      "grad_norm": 0.5001439452171326,
      "learning_rate": 0.0013001865387153588,
      "loss": 1.6021,
      "step": 6265
    },
    {
      "epoch": 0.6609704641350211,
      "grad_norm": 0.49058055877685547,
      "learning_rate": 0.0012983832153222814,
      "loss": 1.6046,
      "step": 6266
    },
    {
      "epoch": 0.6610759493670886,
      "grad_norm": 0.48805370926856995,
      "learning_rate": 0.0012965823930897401,
      "loss": 1.6331,
      "step": 6267
    },
    {
      "epoch": 0.6611814345991561,
      "grad_norm": 0.579522430896759,
      "learning_rate": 0.0012947840685486932,
      "loss": 1.5785,
      "step": 6268
    },
    {
      "epoch": 0.6612869198312237,
      "grad_norm": 0.5598164200782776,
      "learning_rate": 0.0012929882382349102,
      "loss": 1.6291,
      "step": 6269
    },
    {
      "epoch": 0.6613924050632911,
      "grad_norm": 0.5173453688621521,
      "learning_rate": 0.001291194898688966,
      "loss": 1.5746,
      "step": 6270
    },
    {
      "epoch": 0.6614978902953587,
      "grad_norm": 0.5323518514633179,
      "learning_rate": 0.001289404046456233,
      "loss": 1.5848,
      "step": 6271
    },
    {
      "epoch": 0.6616033755274262,
      "grad_norm": 0.5157877802848816,
      "learning_rate": 0.0012876156780868755,
      "loss": 1.6074,
      "step": 6272
    },
    {
      "epoch": 0.6617088607594936,
      "grad_norm": 0.5611578822135925,
      "learning_rate": 0.0012858297901358424,
      "loss": 1.6519,
      "step": 6273
    },
    {
      "epoch": 0.6618143459915612,
      "grad_norm": 0.5870733261108398,
      "learning_rate": 0.001284046379162861,
      "loss": 1.5992,
      "step": 6274
    },
    {
      "epoch": 0.6619198312236287,
      "grad_norm": 0.5836647152900696,
      "learning_rate": 0.0012822654417324305,
      "loss": 1.6038,
      "step": 6275
    },
    {
      "epoch": 0.6620253164556962,
      "grad_norm": 0.5240967869758606,
      "learning_rate": 0.0012804869744138137,
      "loss": 1.6002,
      "step": 6276
    },
    {
      "epoch": 0.6621308016877637,
      "grad_norm": 0.6198570728302002,
      "learning_rate": 0.0012787109737810332,
      "loss": 1.5796,
      "step": 6277
    },
    {
      "epoch": 0.6622362869198313,
      "grad_norm": 0.49905139207839966,
      "learning_rate": 0.0012769374364128628,
      "loss": 1.6228,
      "step": 6278
    },
    {
      "epoch": 0.6623417721518987,
      "grad_norm": 0.6339825391769409,
      "learning_rate": 0.0012751663588928214,
      "loss": 1.5957,
      "step": 6279
    },
    {
      "epoch": 0.6624472573839663,
      "grad_norm": 0.49158141016960144,
      "learning_rate": 0.001273397737809166,
      "loss": 1.6048,
      "step": 6280
    },
    {
      "epoch": 0.6625527426160338,
      "grad_norm": 0.5476797223091125,
      "learning_rate": 0.001271631569754887,
      "loss": 1.5466,
      "step": 6281
    },
    {
      "epoch": 0.6626582278481012,
      "grad_norm": 0.6192284226417542,
      "learning_rate": 0.0012698678513276987,
      "loss": 1.5831,
      "step": 6282
    },
    {
      "epoch": 0.6627637130801688,
      "grad_norm": 0.4934277832508087,
      "learning_rate": 0.0012681065791300351,
      "loss": 1.5746,
      "step": 6283
    },
    {
      "epoch": 0.6628691983122363,
      "grad_norm": 0.6882602572441101,
      "learning_rate": 0.0012663477497690421,
      "loss": 1.6384,
      "step": 6284
    },
    {
      "epoch": 0.6629746835443038,
      "grad_norm": 0.5981414914131165,
      "learning_rate": 0.0012645913598565719,
      "loss": 1.6155,
      "step": 6285
    },
    {
      "epoch": 0.6630801687763713,
      "grad_norm": 0.6074311137199402,
      "learning_rate": 0.0012628374060091757,
      "loss": 1.5985,
      "step": 6286
    },
    {
      "epoch": 0.6631856540084389,
      "grad_norm": 0.6158648133277893,
      "learning_rate": 0.0012610858848480973,
      "loss": 1.6191,
      "step": 6287
    },
    {
      "epoch": 0.6632911392405063,
      "grad_norm": 0.5439046025276184,
      "learning_rate": 0.0012593367929992667,
      "loss": 1.6113,
      "step": 6288
    },
    {
      "epoch": 0.6633966244725739,
      "grad_norm": 0.576534628868103,
      "learning_rate": 0.0012575901270932943,
      "loss": 1.6132,
      "step": 6289
    },
    {
      "epoch": 0.6635021097046413,
      "grad_norm": 0.6843683123588562,
      "learning_rate": 0.001255845883765463,
      "loss": 1.5776,
      "step": 6290
    },
    {
      "epoch": 0.6636075949367088,
      "grad_norm": 0.4963736832141876,
      "learning_rate": 0.0012541040596557229,
      "loss": 1.5631,
      "step": 6291
    },
    {
      "epoch": 0.6637130801687764,
      "grad_norm": 0.5085331201553345,
      "learning_rate": 0.001252364651408684,
      "loss": 1.6155,
      "step": 6292
    },
    {
      "epoch": 0.6638185654008438,
      "grad_norm": 0.6039493680000305,
      "learning_rate": 0.001250627655673611,
      "loss": 1.5934,
      "step": 6293
    },
    {
      "epoch": 0.6639240506329114,
      "grad_norm": 0.530308723449707,
      "learning_rate": 0.0012488930691044145,
      "loss": 1.6114,
      "step": 6294
    },
    {
      "epoch": 0.6640295358649789,
      "grad_norm": 0.4819502532482147,
      "learning_rate": 0.0012471608883596475,
      "loss": 1.5915,
      "step": 6295
    },
    {
      "epoch": 0.6641350210970464,
      "grad_norm": 0.5258892774581909,
      "learning_rate": 0.0012454311101024967,
      "loss": 1.5958,
      "step": 6296
    },
    {
      "epoch": 0.6642405063291139,
      "grad_norm": 0.5121935606002808,
      "learning_rate": 0.0012437037310007774,
      "loss": 1.5899,
      "step": 6297
    },
    {
      "epoch": 0.6643459915611815,
      "grad_norm": 0.4811895489692688,
      "learning_rate": 0.0012419787477269257,
      "loss": 1.6169,
      "step": 6298
    },
    {
      "epoch": 0.6644514767932489,
      "grad_norm": 0.5116422176361084,
      "learning_rate": 0.0012402561569579936,
      "loss": 1.5668,
      "step": 6299
    },
    {
      "epoch": 0.6645569620253164,
      "grad_norm": 0.4468708634376526,
      "learning_rate": 0.001238535955375642,
      "loss": 1.5695,
      "step": 6300
    },
    {
      "epoch": 0.664662447257384,
      "grad_norm": 0.5267335772514343,
      "learning_rate": 0.001236818139666134,
      "loss": 1.5943,
      "step": 6301
    },
    {
      "epoch": 0.6647679324894514,
      "grad_norm": 0.5234312415122986,
      "learning_rate": 0.0012351027065203286,
      "loss": 1.5761,
      "step": 6302
    },
    {
      "epoch": 0.664873417721519,
      "grad_norm": 0.4723226726055145,
      "learning_rate": 0.001233389652633675,
      "loss": 1.5605,
      "step": 6303
    },
    {
      "epoch": 0.6649789029535865,
      "grad_norm": 0.4188882112503052,
      "learning_rate": 0.001231678974706205,
      "loss": 1.5742,
      "step": 6304
    },
    {
      "epoch": 0.665084388185654,
      "grad_norm": 0.5146017074584961,
      "learning_rate": 0.0012299706694425285,
      "loss": 1.5992,
      "step": 6305
    },
    {
      "epoch": 0.6651898734177215,
      "grad_norm": 0.45604467391967773,
      "learning_rate": 0.0012282647335518245,
      "loss": 1.5687,
      "step": 6306
    },
    {
      "epoch": 0.6652953586497891,
      "grad_norm": 0.4664759933948517,
      "learning_rate": 0.0012265611637478377,
      "loss": 1.6354,
      "step": 6307
    },
    {
      "epoch": 0.6654008438818565,
      "grad_norm": 0.4594112038612366,
      "learning_rate": 0.0012248599567488698,
      "loss": 1.601,
      "step": 6308
    },
    {
      "epoch": 0.665506329113924,
      "grad_norm": 0.41812050342559814,
      "learning_rate": 0.0012231611092777745,
      "loss": 1.5699,
      "step": 6309
    },
    {
      "epoch": 0.6656118143459916,
      "grad_norm": 0.5122634768486023,
      "learning_rate": 0.0012214646180619506,
      "loss": 1.6065,
      "step": 6310
    },
    {
      "epoch": 0.665717299578059,
      "grad_norm": 0.4882543683052063,
      "learning_rate": 0.0012197704798333365,
      "loss": 1.6161,
      "step": 6311
    },
    {
      "epoch": 0.6658227848101266,
      "grad_norm": 0.48394256830215454,
      "learning_rate": 0.0012180786913284026,
      "loss": 1.5558,
      "step": 6312
    },
    {
      "epoch": 0.6659282700421941,
      "grad_norm": 0.545910656452179,
      "learning_rate": 0.001216389249288146,
      "loss": 1.5944,
      "step": 6313
    },
    {
      "epoch": 0.6660337552742616,
      "grad_norm": 0.46415403485298157,
      "learning_rate": 0.0012147021504580842,
      "loss": 1.5633,
      "step": 6314
    },
    {
      "epoch": 0.6661392405063291,
      "grad_norm": 0.5876246094703674,
      "learning_rate": 0.0012130173915882478,
      "loss": 1.57,
      "step": 6315
    },
    {
      "epoch": 0.6662447257383967,
      "grad_norm": 0.5909762978553772,
      "learning_rate": 0.0012113349694331762,
      "loss": 1.5887,
      "step": 6316
    },
    {
      "epoch": 0.6663502109704641,
      "grad_norm": 0.5569766163825989,
      "learning_rate": 0.0012096548807519092,
      "loss": 1.6191,
      "step": 6317
    },
    {
      "epoch": 0.6664556962025316,
      "grad_norm": 0.6135898232460022,
      "learning_rate": 0.0012079771223079822,
      "loss": 1.5971,
      "step": 6318
    },
    {
      "epoch": 0.6665611814345992,
      "grad_norm": 0.5652821660041809,
      "learning_rate": 0.0012063016908694193,
      "loss": 1.6093,
      "step": 6319
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.5761544704437256,
      "learning_rate": 0.001204628583208727,
      "loss": 1.5671,
      "step": 6320
    },
    {
      "epoch": 0.6667721518987342,
      "grad_norm": 0.6448853015899658,
      "learning_rate": 0.0012029577961028893,
      "loss": 1.603,
      "step": 6321
    },
    {
      "epoch": 0.6668776371308017,
      "grad_norm": 0.6403090357780457,
      "learning_rate": 0.0012012893263333587,
      "loss": 1.5955,
      "step": 6322
    },
    {
      "epoch": 0.6669831223628692,
      "grad_norm": 0.5241579413414001,
      "learning_rate": 0.0011996231706860535,
      "loss": 1.5721,
      "step": 6323
    },
    {
      "epoch": 0.6670886075949367,
      "grad_norm": 0.6040076017379761,
      "learning_rate": 0.0011979593259513486,
      "loss": 1.6376,
      "step": 6324
    },
    {
      "epoch": 0.6671940928270043,
      "grad_norm": 0.5078284740447998,
      "learning_rate": 0.0011962977889240713,
      "loss": 1.6035,
      "step": 6325
    },
    {
      "epoch": 0.6672995780590717,
      "grad_norm": 0.5907483696937561,
      "learning_rate": 0.001194638556403494,
      "loss": 1.5757,
      "step": 6326
    },
    {
      "epoch": 0.6674050632911392,
      "grad_norm": 0.5162091851234436,
      "learning_rate": 0.0011929816251933286,
      "loss": 1.5983,
      "step": 6327
    },
    {
      "epoch": 0.6675105485232068,
      "grad_norm": 0.5676397085189819,
      "learning_rate": 0.0011913269921017202,
      "loss": 1.5361,
      "step": 6328
    },
    {
      "epoch": 0.6676160337552742,
      "grad_norm": 0.4341263175010681,
      "learning_rate": 0.0011896746539412405,
      "loss": 1.5804,
      "step": 6329
    },
    {
      "epoch": 0.6677215189873418,
      "grad_norm": 0.4882994592189789,
      "learning_rate": 0.0011880246075288827,
      "loss": 1.6195,
      "step": 6330
    },
    {
      "epoch": 0.6678270042194093,
      "grad_norm": 0.4783797562122345,
      "learning_rate": 0.001186376849686054,
      "loss": 1.5526,
      "step": 6331
    },
    {
      "epoch": 0.6679324894514768,
      "grad_norm": 0.5487940907478333,
      "learning_rate": 0.0011847313772385714,
      "loss": 1.6263,
      "step": 6332
    },
    {
      "epoch": 0.6680379746835443,
      "grad_norm": 0.5111870169639587,
      "learning_rate": 0.0011830881870166531,
      "loss": 1.6047,
      "step": 6333
    },
    {
      "epoch": 0.6681434599156119,
      "grad_norm": 0.5094480514526367,
      "learning_rate": 0.0011814472758549144,
      "loss": 1.5745,
      "step": 6334
    },
    {
      "epoch": 0.6682489451476793,
      "grad_norm": 0.48153698444366455,
      "learning_rate": 0.0011798086405923607,
      "loss": 1.561,
      "step": 6335
    },
    {
      "epoch": 0.6683544303797468,
      "grad_norm": 0.5202271342277527,
      "learning_rate": 0.0011781722780723819,
      "loss": 1.6094,
      "step": 6336
    },
    {
      "epoch": 0.6684599156118144,
      "grad_norm": 0.6091830730438232,
      "learning_rate": 0.0011765381851427457,
      "loss": 1.5831,
      "step": 6337
    },
    {
      "epoch": 0.6685654008438818,
      "grad_norm": 0.49341899156570435,
      "learning_rate": 0.0011749063586555919,
      "loss": 1.5811,
      "step": 6338
    },
    {
      "epoch": 0.6686708860759494,
      "grad_norm": 0.617054283618927,
      "learning_rate": 0.0011732767954674265,
      "loss": 1.6074,
      "step": 6339
    },
    {
      "epoch": 0.6687763713080169,
      "grad_norm": 0.48009762167930603,
      "learning_rate": 0.001171649492439115,
      "loss": 1.5744,
      "step": 6340
    },
    {
      "epoch": 0.6688818565400844,
      "grad_norm": 0.5383430123329163,
      "learning_rate": 0.0011700244464358776,
      "loss": 1.5833,
      "step": 6341
    },
    {
      "epoch": 0.6689873417721519,
      "grad_norm": 0.457782119512558,
      "learning_rate": 0.0011684016543272815,
      "loss": 1.6314,
      "step": 6342
    },
    {
      "epoch": 0.6690928270042195,
      "grad_norm": 0.4396111071109772,
      "learning_rate": 0.0011667811129872365,
      "loss": 1.5998,
      "step": 6343
    },
    {
      "epoch": 0.6691983122362869,
      "grad_norm": 0.4802881181240082,
      "learning_rate": 0.0011651628192939872,
      "loss": 1.5782,
      "step": 6344
    },
    {
      "epoch": 0.6693037974683544,
      "grad_norm": 0.4975625276565552,
      "learning_rate": 0.001163546770130109,
      "loss": 1.5992,
      "step": 6345
    },
    {
      "epoch": 0.669409282700422,
      "grad_norm": 0.49034079909324646,
      "learning_rate": 0.0011619329623825006,
      "loss": 1.6007,
      "step": 6346
    },
    {
      "epoch": 0.6695147679324894,
      "grad_norm": 0.5298082232475281,
      "learning_rate": 0.0011603213929423785,
      "loss": 1.5753,
      "step": 6347
    },
    {
      "epoch": 0.669620253164557,
      "grad_norm": 0.519305944442749,
      "learning_rate": 0.001158712058705271,
      "loss": 1.5663,
      "step": 6348
    },
    {
      "epoch": 0.6697257383966245,
      "grad_norm": 0.5319370627403259,
      "learning_rate": 0.0011571049565710122,
      "loss": 1.5674,
      "step": 6349
    },
    {
      "epoch": 0.669831223628692,
      "grad_norm": 0.5216324925422668,
      "learning_rate": 0.0011555000834437364,
      "loss": 1.5919,
      "step": 6350
    },
    {
      "epoch": 0.6699367088607595,
      "grad_norm": 0.5723080039024353,
      "learning_rate": 0.0011538974362318712,
      "loss": 1.5839,
      "step": 6351
    },
    {
      "epoch": 0.6700421940928271,
      "grad_norm": 0.4261742830276489,
      "learning_rate": 0.0011522970118481325,
      "loss": 1.5743,
      "step": 6352
    },
    {
      "epoch": 0.6701476793248945,
      "grad_norm": 0.5841290354728699,
      "learning_rate": 0.0011506988072095183,
      "loss": 1.5658,
      "step": 6353
    },
    {
      "epoch": 0.670253164556962,
      "grad_norm": 0.5021260976791382,
      "learning_rate": 0.0011491028192373023,
      "loss": 1.628,
      "step": 6354
    },
    {
      "epoch": 0.6703586497890295,
      "grad_norm": 0.639585018157959,
      "learning_rate": 0.0011475090448570281,
      "loss": 1.5459,
      "step": 6355
    },
    {
      "epoch": 0.670464135021097,
      "grad_norm": 0.6755632162094116,
      "learning_rate": 0.0011459174809985047,
      "loss": 1.581,
      "step": 6356
    },
    {
      "epoch": 0.6705696202531646,
      "grad_norm": 0.4771636426448822,
      "learning_rate": 0.0011443281245957977,
      "loss": 1.5929,
      "step": 6357
    },
    {
      "epoch": 0.670675105485232,
      "grad_norm": 0.6079390645027161,
      "learning_rate": 0.0011427409725872262,
      "loss": 1.621,
      "step": 6358
    },
    {
      "epoch": 0.6707805907172996,
      "grad_norm": 0.587040901184082,
      "learning_rate": 0.001141156021915355,
      "loss": 1.5891,
      "step": 6359
    },
    {
      "epoch": 0.6708860759493671,
      "grad_norm": 0.6071764826774597,
      "learning_rate": 0.0011395732695269908,
      "loss": 1.6022,
      "step": 6360
    },
    {
      "epoch": 0.6709915611814345,
      "grad_norm": 0.5442952513694763,
      "learning_rate": 0.0011379927123731737,
      "loss": 1.5671,
      "step": 6361
    },
    {
      "epoch": 0.6710970464135021,
      "grad_norm": 0.5130666494369507,
      "learning_rate": 0.0011364143474091727,
      "loss": 1.5919,
      "step": 6362
    },
    {
      "epoch": 0.6712025316455696,
      "grad_norm": 0.6559984087944031,
      "learning_rate": 0.0011348381715944804,
      "loss": 1.5961,
      "step": 6363
    },
    {
      "epoch": 0.6713080168776371,
      "grad_norm": 0.5327663421630859,
      "learning_rate": 0.0011332641818928063,
      "loss": 1.5638,
      "step": 6364
    },
    {
      "epoch": 0.6714135021097046,
      "grad_norm": 0.6795653104782104,
      "learning_rate": 0.001131692375272071,
      "loss": 1.5515,
      "step": 6365
    },
    {
      "epoch": 0.6715189873417722,
      "grad_norm": 0.49332743883132935,
      "learning_rate": 0.0011301227487044005,
      "loss": 1.5454,
      "step": 6366
    },
    {
      "epoch": 0.6716244725738396,
      "grad_norm": 0.6678374409675598,
      "learning_rate": 0.0011285552991661203,
      "loss": 1.5252,
      "step": 6367
    },
    {
      "epoch": 0.6717299578059072,
      "grad_norm": 0.5235859155654907,
      "learning_rate": 0.00112699002363775,
      "loss": 1.5745,
      "step": 6368
    },
    {
      "epoch": 0.6718354430379747,
      "grad_norm": 0.6558351516723633,
      "learning_rate": 0.001125426919103997,
      "loss": 1.5968,
      "step": 6369
    },
    {
      "epoch": 0.6719409282700421,
      "grad_norm": 0.5537667870521545,
      "learning_rate": 0.0011238659825537505,
      "loss": 1.5846,
      "step": 6370
    },
    {
      "epoch": 0.6720464135021097,
      "grad_norm": 0.5901890397071838,
      "learning_rate": 0.0011223072109800768,
      "loss": 1.5461,
      "step": 6371
    },
    {
      "epoch": 0.6721518987341772,
      "grad_norm": 0.6574958562850952,
      "learning_rate": 0.0011207506013802117,
      "loss": 1.5991,
      "step": 6372
    },
    {
      "epoch": 0.6722573839662447,
      "grad_norm": 0.5097714066505432,
      "learning_rate": 0.0011191961507555567,
      "loss": 1.5982,
      "step": 6373
    },
    {
      "epoch": 0.6723628691983122,
      "grad_norm": 0.6988516449928284,
      "learning_rate": 0.0011176438561116713,
      "loss": 1.5623,
      "step": 6374
    },
    {
      "epoch": 0.6724683544303798,
      "grad_norm": 0.4968533217906952,
      "learning_rate": 0.0011160937144582695,
      "loss": 1.546,
      "step": 6375
    },
    {
      "epoch": 0.6725738396624472,
      "grad_norm": 0.7256531715393066,
      "learning_rate": 0.0011145457228092116,
      "loss": 1.5632,
      "step": 6376
    },
    {
      "epoch": 0.6726793248945148,
      "grad_norm": 0.5300629138946533,
      "learning_rate": 0.0011129998781824997,
      "loss": 1.5966,
      "step": 6377
    },
    {
      "epoch": 0.6727848101265823,
      "grad_norm": 0.6823804974555969,
      "learning_rate": 0.0011114561776002726,
      "loss": 1.5763,
      "step": 6378
    },
    {
      "epoch": 0.6728902953586497,
      "grad_norm": 0.7479099631309509,
      "learning_rate": 0.001109914618088799,
      "loss": 1.5844,
      "step": 6379
    },
    {
      "epoch": 0.6729957805907173,
      "grad_norm": 0.4734697937965393,
      "learning_rate": 0.0011083751966784717,
      "loss": 1.5715,
      "step": 6380
    },
    {
      "epoch": 0.6731012658227848,
      "grad_norm": 0.8295102715492249,
      "learning_rate": 0.0011068379104038023,
      "loss": 1.5896,
      "step": 6381
    },
    {
      "epoch": 0.6732067510548523,
      "grad_norm": 0.4697086811065674,
      "learning_rate": 0.0011053027563034162,
      "loss": 1.5633,
      "step": 6382
    },
    {
      "epoch": 0.6733122362869198,
      "grad_norm": 0.622559666633606,
      "learning_rate": 0.001103769731420045,
      "loss": 1.6436,
      "step": 6383
    },
    {
      "epoch": 0.6734177215189874,
      "grad_norm": 0.45600736141204834,
      "learning_rate": 0.0011022388328005232,
      "loss": 1.576,
      "step": 6384
    },
    {
      "epoch": 0.6735232067510548,
      "grad_norm": 0.573523998260498,
      "learning_rate": 0.0011007100574957802,
      "loss": 1.6194,
      "step": 6385
    },
    {
      "epoch": 0.6736286919831224,
      "grad_norm": 0.44580280780792236,
      "learning_rate": 0.0010991834025608363,
      "loss": 1.6139,
      "step": 6386
    },
    {
      "epoch": 0.6737341772151899,
      "grad_norm": 0.6450295448303223,
      "learning_rate": 0.001097658865054796,
      "loss": 1.592,
      "step": 6387
    },
    {
      "epoch": 0.6738396624472573,
      "grad_norm": 0.5917106866836548,
      "learning_rate": 0.001096136442040843,
      "loss": 1.5868,
      "step": 6388
    },
    {
      "epoch": 0.6739451476793249,
      "grad_norm": 0.6550825238227844,
      "learning_rate": 0.0010946161305862348,
      "loss": 1.5981,
      "step": 6389
    },
    {
      "epoch": 0.6740506329113924,
      "grad_norm": 0.8302057385444641,
      "learning_rate": 0.0010930979277622953,
      "loss": 1.5301,
      "step": 6390
    },
    {
      "epoch": 0.6741561181434599,
      "grad_norm": 0.43817853927612305,
      "learning_rate": 0.0010915818306444112,
      "loss": 1.5929,
      "step": 6391
    },
    {
      "epoch": 0.6742616033755274,
      "grad_norm": 0.5853847861289978,
      "learning_rate": 0.0010900678363120256,
      "loss": 1.6225,
      "step": 6392
    },
    {
      "epoch": 0.674367088607595,
      "grad_norm": 0.4759158492088318,
      "learning_rate": 0.001088555941848632,
      "loss": 1.5829,
      "step": 6393
    },
    {
      "epoch": 0.6744725738396624,
      "grad_norm": 0.5427954196929932,
      "learning_rate": 0.0010870461443417694,
      "loss": 1.5835,
      "step": 6394
    },
    {
      "epoch": 0.67457805907173,
      "grad_norm": 0.45315021276474,
      "learning_rate": 0.001085538440883016,
      "loss": 1.574,
      "step": 6395
    },
    {
      "epoch": 0.6746835443037975,
      "grad_norm": 0.5170419812202454,
      "learning_rate": 0.0010840328285679837,
      "loss": 1.5532,
      "step": 6396
    },
    {
      "epoch": 0.674789029535865,
      "grad_norm": 0.5563789010047913,
      "learning_rate": 0.0010825293044963132,
      "loss": 1.6181,
      "step": 6397
    },
    {
      "epoch": 0.6748945147679325,
      "grad_norm": 0.5063940286636353,
      "learning_rate": 0.001081027865771668,
      "loss": 1.5592,
      "step": 6398
    },
    {
      "epoch": 0.675,
      "grad_norm": 0.47441422939300537,
      "learning_rate": 0.001079528509501728,
      "loss": 1.5673,
      "step": 6399
    },
    {
      "epoch": 0.6751054852320675,
      "grad_norm": 0.559227705001831,
      "learning_rate": 0.0010780312327981854,
      "loss": 1.6173,
      "step": 6400
    },
    {
      "epoch": 0.675210970464135,
      "grad_norm": 0.5507892370223999,
      "learning_rate": 0.001076536032776738,
      "loss": 1.5578,
      "step": 6401
    },
    {
      "epoch": 0.6753164556962026,
      "grad_norm": 0.4988034665584564,
      "learning_rate": 0.0010750429065570842,
      "loss": 1.5814,
      "step": 6402
    },
    {
      "epoch": 0.67542194092827,
      "grad_norm": 0.5726801156997681,
      "learning_rate": 0.0010735518512629172,
      "loss": 1.6166,
      "step": 6403
    },
    {
      "epoch": 0.6755274261603376,
      "grad_norm": 0.5148823261260986,
      "learning_rate": 0.00107206286402192,
      "loss": 1.6174,
      "step": 6404
    },
    {
      "epoch": 0.6756329113924051,
      "grad_norm": 0.5496354699134827,
      "learning_rate": 0.0010705759419657585,
      "loss": 1.5994,
      "step": 6405
    },
    {
      "epoch": 0.6757383966244725,
      "grad_norm": 0.5202333331108093,
      "learning_rate": 0.0010690910822300777,
      "loss": 1.5915,
      "step": 6406
    },
    {
      "epoch": 0.6758438818565401,
      "grad_norm": 0.5268917679786682,
      "learning_rate": 0.0010676082819544952,
      "loss": 1.57,
      "step": 6407
    },
    {
      "epoch": 0.6759493670886076,
      "grad_norm": 0.48678407073020935,
      "learning_rate": 0.0010661275382825958,
      "loss": 1.5684,
      "step": 6408
    },
    {
      "epoch": 0.6760548523206751,
      "grad_norm": 0.5924727916717529,
      "learning_rate": 0.0010646488483619261,
      "loss": 1.5661,
      "step": 6409
    },
    {
      "epoch": 0.6761603375527426,
      "grad_norm": 0.6849046349525452,
      "learning_rate": 0.0010631722093439888,
      "loss": 1.5561,
      "step": 6410
    },
    {
      "epoch": 0.6762658227848102,
      "grad_norm": 0.5055809617042542,
      "learning_rate": 0.0010616976183842378,
      "loss": 1.5934,
      "step": 6411
    },
    {
      "epoch": 0.6763713080168776,
      "grad_norm": 0.594955325126648,
      "learning_rate": 0.001060225072642072,
      "loss": 1.5516,
      "step": 6412
    },
    {
      "epoch": 0.6764767932489452,
      "grad_norm": 0.4963608384132385,
      "learning_rate": 0.0010587545692808302,
      "loss": 1.5613,
      "step": 6413
    },
    {
      "epoch": 0.6765822784810127,
      "grad_norm": 0.5594482421875,
      "learning_rate": 0.0010572861054677853,
      "loss": 1.5496,
      "step": 6414
    },
    {
      "epoch": 0.6766877637130801,
      "grad_norm": 0.42274507880210876,
      "learning_rate": 0.0010558196783741396,
      "loss": 1.5883,
      "step": 6415
    },
    {
      "epoch": 0.6767932489451477,
      "grad_norm": 0.572669267654419,
      "learning_rate": 0.0010543552851750187,
      "loss": 1.5471,
      "step": 6416
    },
    {
      "epoch": 0.6768987341772152,
      "grad_norm": 0.4971157908439636,
      "learning_rate": 0.001052892923049466,
      "loss": 1.5903,
      "step": 6417
    },
    {
      "epoch": 0.6770042194092827,
      "grad_norm": 0.7142729163169861,
      "learning_rate": 0.0010514325891804379,
      "loss": 1.5983,
      "step": 6418
    },
    {
      "epoch": 0.6771097046413502,
      "grad_norm": 0.5082519054412842,
      "learning_rate": 0.0010499742807547976,
      "loss": 1.5506,
      "step": 6419
    },
    {
      "epoch": 0.6772151898734177,
      "grad_norm": 0.645077645778656,
      "learning_rate": 0.00104851799496331,
      "loss": 1.5915,
      "step": 6420
    },
    {
      "epoch": 0.6773206751054852,
      "grad_norm": 0.5200485587120056,
      "learning_rate": 0.0010470637290006365,
      "loss": 1.5817,
      "step": 6421
    },
    {
      "epoch": 0.6774261603375528,
      "grad_norm": 0.6938650608062744,
      "learning_rate": 0.00104561148006533,
      "loss": 1.5685,
      "step": 6422
    },
    {
      "epoch": 0.6775316455696202,
      "grad_norm": 0.5358496308326721,
      "learning_rate": 0.0010441612453598276,
      "loss": 1.5528,
      "step": 6423
    },
    {
      "epoch": 0.6776371308016877,
      "grad_norm": 0.5471860766410828,
      "learning_rate": 0.001042713022090448,
      "loss": 1.6125,
      "step": 6424
    },
    {
      "epoch": 0.6777426160337553,
      "grad_norm": 0.5146363973617554,
      "learning_rate": 0.0010412668074673832,
      "loss": 1.6071,
      "step": 6425
    },
    {
      "epoch": 0.6778481012658227,
      "grad_norm": 0.5841184258460999,
      "learning_rate": 0.0010398225987046958,
      "loss": 1.5716,
      "step": 6426
    },
    {
      "epoch": 0.6779535864978903,
      "grad_norm": 0.5139831900596619,
      "learning_rate": 0.001038380393020312,
      "loss": 1.5923,
      "step": 6427
    },
    {
      "epoch": 0.6780590717299578,
      "grad_norm": 0.5693835020065308,
      "learning_rate": 0.0010369401876360166,
      "loss": 1.5781,
      "step": 6428
    },
    {
      "epoch": 0.6781645569620253,
      "grad_norm": 0.4554748833179474,
      "learning_rate": 0.0010355019797774478,
      "loss": 1.5312,
      "step": 6429
    },
    {
      "epoch": 0.6782700421940928,
      "grad_norm": 0.4807449281215668,
      "learning_rate": 0.0010340657666740914,
      "loss": 1.5943,
      "step": 6430
    },
    {
      "epoch": 0.6783755274261604,
      "grad_norm": 0.4526333510875702,
      "learning_rate": 0.0010326315455592766,
      "loss": 1.5456,
      "step": 6431
    },
    {
      "epoch": 0.6784810126582278,
      "grad_norm": 0.5008558034896851,
      "learning_rate": 0.001031199313670169,
      "loss": 1.5741,
      "step": 6432
    },
    {
      "epoch": 0.6785864978902953,
      "grad_norm": 0.5095291137695312,
      "learning_rate": 0.0010297690682477669,
      "loss": 1.5431,
      "step": 6433
    },
    {
      "epoch": 0.6786919831223629,
      "grad_norm": 0.4715951681137085,
      "learning_rate": 0.0010283408065368948,
      "loss": 1.5626,
      "step": 6434
    },
    {
      "epoch": 0.6787974683544303,
      "grad_norm": 0.5276069045066833,
      "learning_rate": 0.0010269145257861987,
      "loss": 1.5411,
      "step": 6435
    },
    {
      "epoch": 0.6789029535864979,
      "grad_norm": 0.4964304268360138,
      "learning_rate": 0.0010254902232481407,
      "loss": 1.5656,
      "step": 6436
    },
    {
      "epoch": 0.6790084388185654,
      "grad_norm": 0.574787437915802,
      "learning_rate": 0.0010240678961789937,
      "loss": 1.5609,
      "step": 6437
    },
    {
      "epoch": 0.6791139240506329,
      "grad_norm": 0.6316975355148315,
      "learning_rate": 0.001022647541838836,
      "loss": 1.5743,
      "step": 6438
    },
    {
      "epoch": 0.6792194092827004,
      "grad_norm": 0.5257666110992432,
      "learning_rate": 0.001021229157491546,
      "loss": 1.5655,
      "step": 6439
    },
    {
      "epoch": 0.679324894514768,
      "grad_norm": 0.49049559235572815,
      "learning_rate": 0.0010198127404047975,
      "loss": 1.5776,
      "step": 6440
    },
    {
      "epoch": 0.6794303797468354,
      "grad_norm": 0.46640104055404663,
      "learning_rate": 0.001018398287850053,
      "loss": 1.5996,
      "step": 6441
    },
    {
      "epoch": 0.679535864978903,
      "grad_norm": 0.46320921182632446,
      "learning_rate": 0.0010169857971025606,
      "loss": 1.5284,
      "step": 6442
    },
    {
      "epoch": 0.6796413502109705,
      "grad_norm": 0.5223776698112488,
      "learning_rate": 0.0010155752654413468,
      "loss": 1.5752,
      "step": 6443
    },
    {
      "epoch": 0.6797468354430379,
      "grad_norm": 0.43690183758735657,
      "learning_rate": 0.0010141666901492116,
      "loss": 1.5857,
      "step": 6444
    },
    {
      "epoch": 0.6798523206751055,
      "grad_norm": 0.5709619522094727,
      "learning_rate": 0.0010127600685127247,
      "loss": 1.5694,
      "step": 6445
    },
    {
      "epoch": 0.679957805907173,
      "grad_norm": 0.42933282256126404,
      "learning_rate": 0.0010113553978222192,
      "loss": 1.5595,
      "step": 6446
    },
    {
      "epoch": 0.6800632911392405,
      "grad_norm": 0.5185548663139343,
      "learning_rate": 0.0010099526753717856,
      "loss": 1.5642,
      "step": 6447
    },
    {
      "epoch": 0.680168776371308,
      "grad_norm": 0.5778734087944031,
      "learning_rate": 0.0010085518984592678,
      "loss": 1.5457,
      "step": 6448
    },
    {
      "epoch": 0.6802742616033756,
      "grad_norm": 0.5577317476272583,
      "learning_rate": 0.0010071530643862578,
      "loss": 1.5403,
      "step": 6449
    },
    {
      "epoch": 0.680379746835443,
      "grad_norm": 0.6061927676200867,
      "learning_rate": 0.0010057561704580897,
      "loss": 1.5824,
      "step": 6450
    },
    {
      "epoch": 0.6804852320675105,
      "grad_norm": 0.5703639984130859,
      "learning_rate": 0.001004361213983836,
      "loss": 1.5662,
      "step": 6451
    },
    {
      "epoch": 0.6805907172995781,
      "grad_norm": 0.49401646852493286,
      "learning_rate": 0.0010029681922762998,
      "loss": 1.6089,
      "step": 6452
    },
    {
      "epoch": 0.6806962025316455,
      "grad_norm": 0.5242007374763489,
      "learning_rate": 0.0010015771026520132,
      "loss": 1.5994,
      "step": 6453
    },
    {
      "epoch": 0.6808016877637131,
      "grad_norm": 0.47871851921081543,
      "learning_rate": 0.0010001879424312286,
      "loss": 1.5509,
      "step": 6454
    },
    {
      "epoch": 0.6809071729957806,
      "grad_norm": 0.5109929442405701,
      "learning_rate": 0.0009988007089379162,
      "loss": 1.5669,
      "step": 6455
    },
    {
      "epoch": 0.6810126582278481,
      "grad_norm": 0.4957762658596039,
      "learning_rate": 0.000997415399499757,
      "loss": 1.5608,
      "step": 6456
    },
    {
      "epoch": 0.6811181434599156,
      "grad_norm": 0.4794304072856903,
      "learning_rate": 0.000996032011448139,
      "loss": 1.592,
      "step": 6457
    },
    {
      "epoch": 0.6812236286919832,
      "grad_norm": 0.4826592803001404,
      "learning_rate": 0.0009946505421181513,
      "loss": 1.5718,
      "step": 6458
    },
    {
      "epoch": 0.6813291139240506,
      "grad_norm": 0.46919921040534973,
      "learning_rate": 0.000993270988848579,
      "loss": 1.5934,
      "step": 6459
    },
    {
      "epoch": 0.6814345991561181,
      "grad_norm": 0.5313397645950317,
      "learning_rate": 0.0009918933489818985,
      "loss": 1.5905,
      "step": 6460
    },
    {
      "epoch": 0.6815400843881857,
      "grad_norm": 0.48957186937332153,
      "learning_rate": 0.000990517619864272,
      "loss": 1.5732,
      "step": 6461
    },
    {
      "epoch": 0.6816455696202531,
      "grad_norm": 0.5460265278816223,
      "learning_rate": 0.0009891437988455425,
      "loss": 1.579,
      "step": 6462
    },
    {
      "epoch": 0.6817510548523207,
      "grad_norm": 0.45310288667678833,
      "learning_rate": 0.0009877718832792285,
      "loss": 1.5617,
      "step": 6463
    },
    {
      "epoch": 0.6818565400843882,
      "grad_norm": 0.43776223063468933,
      "learning_rate": 0.0009864018705225196,
      "loss": 1.5731,
      "step": 6464
    },
    {
      "epoch": 0.6819620253164557,
      "grad_norm": 0.504557728767395,
      "learning_rate": 0.0009850337579362701,
      "loss": 1.5676,
      "step": 6465
    },
    {
      "epoch": 0.6820675105485232,
      "grad_norm": 0.48897743225097656,
      "learning_rate": 0.000983667542884996,
      "loss": 1.5462,
      "step": 6466
    },
    {
      "epoch": 0.6821729957805908,
      "grad_norm": 0.5264372825622559,
      "learning_rate": 0.000982303222736867,
      "loss": 1.5713,
      "step": 6467
    },
    {
      "epoch": 0.6822784810126582,
      "grad_norm": 0.45343002676963806,
      "learning_rate": 0.0009809407948637044,
      "loss": 1.589,
      "step": 6468
    },
    {
      "epoch": 0.6823839662447257,
      "grad_norm": 0.5581714510917664,
      "learning_rate": 0.0009795802566409742,
      "loss": 1.5558,
      "step": 6469
    },
    {
      "epoch": 0.6824894514767933,
      "grad_norm": 0.5224825739860535,
      "learning_rate": 0.0009782216054477827,
      "loss": 1.6053,
      "step": 6470
    },
    {
      "epoch": 0.6825949367088607,
      "grad_norm": 0.4501763880252838,
      "learning_rate": 0.000976864838666871,
      "loss": 1.5557,
      "step": 6471
    },
    {
      "epoch": 0.6827004219409283,
      "grad_norm": 0.4748144745826721,
      "learning_rate": 0.0009755099536846105,
      "loss": 1.6209,
      "step": 6472
    },
    {
      "epoch": 0.6828059071729958,
      "grad_norm": 0.5101902484893799,
      "learning_rate": 0.0009741569478909979,
      "loss": 1.5864,
      "step": 6473
    },
    {
      "epoch": 0.6829113924050633,
      "grad_norm": 0.4839603006839752,
      "learning_rate": 0.0009728058186796492,
      "loss": 1.556,
      "step": 6474
    },
    {
      "epoch": 0.6830168776371308,
      "grad_norm": 0.47969716787338257,
      "learning_rate": 0.0009714565634477962,
      "loss": 1.5887,
      "step": 6475
    },
    {
      "epoch": 0.6831223628691984,
      "grad_norm": 0.43456801772117615,
      "learning_rate": 0.00097010917959628,
      "loss": 1.6041,
      "step": 6476
    },
    {
      "epoch": 0.6832278481012658,
      "grad_norm": 0.4625198543071747,
      "learning_rate": 0.0009687636645295469,
      "loss": 1.5682,
      "step": 6477
    },
    {
      "epoch": 0.6833333333333333,
      "grad_norm": 0.5103845596313477,
      "learning_rate": 0.0009674200156556436,
      "loss": 1.5815,
      "step": 6478
    },
    {
      "epoch": 0.6834388185654009,
      "grad_norm": 0.6014232635498047,
      "learning_rate": 0.0009660782303862109,
      "loss": 1.5554,
      "step": 6479
    },
    {
      "epoch": 0.6835443037974683,
      "grad_norm": 0.539760172367096,
      "learning_rate": 0.0009647383061364801,
      "loss": 1.5845,
      "step": 6480
    },
    {
      "epoch": 0.6836497890295359,
      "grad_norm": 0.6342958807945251,
      "learning_rate": 0.0009634002403252678,
      "loss": 1.5832,
      "step": 6481
    },
    {
      "epoch": 0.6837552742616034,
      "grad_norm": 0.5717540979385376,
      "learning_rate": 0.00096206403037497,
      "loss": 1.5361,
      "step": 6482
    },
    {
      "epoch": 0.6838607594936709,
      "grad_norm": 0.5399227142333984,
      "learning_rate": 0.000960729673711558,
      "loss": 1.5326,
      "step": 6483
    },
    {
      "epoch": 0.6839662447257384,
      "grad_norm": 0.5392372012138367,
      "learning_rate": 0.0009593971677645735,
      "loss": 1.604,
      "step": 6484
    },
    {
      "epoch": 0.6840717299578059,
      "grad_norm": 0.5298331379890442,
      "learning_rate": 0.0009580665099671228,
      "loss": 1.5759,
      "step": 6485
    },
    {
      "epoch": 0.6841772151898734,
      "grad_norm": 0.4869494140148163,
      "learning_rate": 0.000956737697755873,
      "loss": 1.5695,
      "step": 6486
    },
    {
      "epoch": 0.684282700421941,
      "grad_norm": 0.5326877236366272,
      "learning_rate": 0.0009554107285710461,
      "loss": 1.5787,
      "step": 6487
    },
    {
      "epoch": 0.6843881856540084,
      "grad_norm": 0.5505571365356445,
      "learning_rate": 0.0009540855998564147,
      "loss": 1.607,
      "step": 6488
    },
    {
      "epoch": 0.6844936708860759,
      "grad_norm": 0.5273861289024353,
      "learning_rate": 0.0009527623090592963,
      "loss": 1.5825,
      "step": 6489
    },
    {
      "epoch": 0.6845991561181435,
      "grad_norm": 0.5127564072608948,
      "learning_rate": 0.0009514408536305495,
      "loss": 1.5864,
      "step": 6490
    },
    {
      "epoch": 0.6847046413502109,
      "grad_norm": 0.5200434327125549,
      "learning_rate": 0.0009501212310245682,
      "loss": 1.5992,
      "step": 6491
    },
    {
      "epoch": 0.6848101265822785,
      "grad_norm": 0.5136992931365967,
      "learning_rate": 0.0009488034386992771,
      "loss": 1.6061,
      "step": 6492
    },
    {
      "epoch": 0.684915611814346,
      "grad_norm": 0.4220972955226898,
      "learning_rate": 0.0009474874741161266,
      "loss": 1.5787,
      "step": 6493
    },
    {
      "epoch": 0.6850210970464135,
      "grad_norm": 0.46680107712745667,
      "learning_rate": 0.0009461733347400879,
      "loss": 1.5614,
      "step": 6494
    },
    {
      "epoch": 0.685126582278481,
      "grad_norm": 0.4657742381095886,
      "learning_rate": 0.0009448610180396485,
      "loss": 1.6051,
      "step": 6495
    },
    {
      "epoch": 0.6852320675105485,
      "grad_norm": 0.4843524694442749,
      "learning_rate": 0.0009435505214868068,
      "loss": 1.582,
      "step": 6496
    },
    {
      "epoch": 0.685337552742616,
      "grad_norm": 0.45350903272628784,
      "learning_rate": 0.0009422418425570675,
      "loss": 1.6047,
      "step": 6497
    },
    {
      "epoch": 0.6854430379746835,
      "grad_norm": 0.47634002566337585,
      "learning_rate": 0.000940934978729437,
      "loss": 1.565,
      "step": 6498
    },
    {
      "epoch": 0.6855485232067511,
      "grad_norm": 0.46629229187965393,
      "learning_rate": 0.0009396299274864177,
      "loss": 1.6145,
      "step": 6499
    },
    {
      "epoch": 0.6856540084388185,
      "grad_norm": 0.5664366483688354,
      "learning_rate": 0.0009383266863140042,
      "loss": 1.5605,
      "step": 6500
    },
    {
      "epoch": 0.6857594936708861,
      "grad_norm": 0.4939168691635132,
      "learning_rate": 0.0009370252527016777,
      "loss": 1.5816,
      "step": 6501
    },
    {
      "epoch": 0.6858649789029536,
      "grad_norm": 0.5229809880256653,
      "learning_rate": 0.0009357256241424013,
      "loss": 1.5639,
      "step": 6502
    },
    {
      "epoch": 0.685970464135021,
      "grad_norm": 0.49933311343193054,
      "learning_rate": 0.0009344277981326158,
      "loss": 1.6144,
      "step": 6503
    },
    {
      "epoch": 0.6860759493670886,
      "grad_norm": 0.5490395426750183,
      "learning_rate": 0.0009331317721722339,
      "loss": 1.6002,
      "step": 6504
    },
    {
      "epoch": 0.6861814345991561,
      "grad_norm": 0.4353964626789093,
      "learning_rate": 0.0009318375437646361,
      "loss": 1.5893,
      "step": 6505
    },
    {
      "epoch": 0.6862869198312236,
      "grad_norm": 0.5495443940162659,
      "learning_rate": 0.0009305451104166652,
      "loss": 1.6009,
      "step": 6506
    },
    {
      "epoch": 0.6863924050632911,
      "grad_norm": 0.4636639952659607,
      "learning_rate": 0.0009292544696386228,
      "loss": 1.5774,
      "step": 6507
    },
    {
      "epoch": 0.6864978902953587,
      "grad_norm": 0.6736634373664856,
      "learning_rate": 0.0009279656189442628,
      "loss": 1.5518,
      "step": 6508
    },
    {
      "epoch": 0.6866033755274261,
      "grad_norm": 0.5159350633621216,
      "learning_rate": 0.0009266785558507877,
      "loss": 1.6017,
      "step": 6509
    },
    {
      "epoch": 0.6867088607594937,
      "grad_norm": 0.501348614692688,
      "learning_rate": 0.000925393277878844,
      "loss": 1.5657,
      "step": 6510
    },
    {
      "epoch": 0.6868143459915612,
      "grad_norm": 0.5065723061561584,
      "learning_rate": 0.0009241097825525162,
      "loss": 1.569,
      "step": 6511
    },
    {
      "epoch": 0.6869198312236287,
      "grad_norm": 0.4902362525463104,
      "learning_rate": 0.0009228280673993236,
      "loss": 1.5852,
      "step": 6512
    },
    {
      "epoch": 0.6870253164556962,
      "grad_norm": 0.46798020601272583,
      "learning_rate": 0.0009215481299502144,
      "loss": 1.5695,
      "step": 6513
    },
    {
      "epoch": 0.6871308016877637,
      "grad_norm": 0.5281558036804199,
      "learning_rate": 0.0009202699677395614,
      "loss": 1.5829,
      "step": 6514
    },
    {
      "epoch": 0.6872362869198312,
      "grad_norm": 0.5488610863685608,
      "learning_rate": 0.000918993578305157,
      "loss": 1.5587,
      "step": 6515
    },
    {
      "epoch": 0.6873417721518987,
      "grad_norm": 0.5047544240951538,
      "learning_rate": 0.0009177189591882088,
      "loss": 1.5715,
      "step": 6516
    },
    {
      "epoch": 0.6874472573839663,
      "grad_norm": 0.5690421462059021,
      "learning_rate": 0.0009164461079333344,
      "loss": 1.6138,
      "step": 6517
    },
    {
      "epoch": 0.6875527426160337,
      "grad_norm": 0.5395167469978333,
      "learning_rate": 0.0009151750220885573,
      "loss": 1.5522,
      "step": 6518
    },
    {
      "epoch": 0.6876582278481013,
      "grad_norm": 0.4937494695186615,
      "learning_rate": 0.0009139056992053016,
      "loss": 1.5303,
      "step": 6519
    },
    {
      "epoch": 0.6877637130801688,
      "grad_norm": 0.6328351497650146,
      "learning_rate": 0.0009126381368383879,
      "loss": 1.5712,
      "step": 6520
    },
    {
      "epoch": 0.6878691983122363,
      "grad_norm": 0.5024483799934387,
      "learning_rate": 0.0009113723325460275,
      "loss": 1.5727,
      "step": 6521
    },
    {
      "epoch": 0.6879746835443038,
      "grad_norm": 0.44090771675109863,
      "learning_rate": 0.000910108283889819,
      "loss": 1.5399,
      "step": 6522
    },
    {
      "epoch": 0.6880801687763713,
      "grad_norm": 0.5520035028457642,
      "learning_rate": 0.0009088459884347425,
      "loss": 1.5385,
      "step": 6523
    },
    {
      "epoch": 0.6881856540084388,
      "grad_norm": 0.4542565643787384,
      "learning_rate": 0.0009075854437491562,
      "loss": 1.5719,
      "step": 6524
    },
    {
      "epoch": 0.6882911392405063,
      "grad_norm": 0.51065993309021,
      "learning_rate": 0.0009063266474047897,
      "loss": 1.58,
      "step": 6525
    },
    {
      "epoch": 0.6883966244725739,
      "grad_norm": 0.4812704920768738,
      "learning_rate": 0.0009050695969767418,
      "loss": 1.575,
      "step": 6526
    },
    {
      "epoch": 0.6885021097046413,
      "grad_norm": 0.4445286989212036,
      "learning_rate": 0.0009038142900434738,
      "loss": 1.5785,
      "step": 6527
    },
    {
      "epoch": 0.6886075949367089,
      "grad_norm": 0.4607788324356079,
      "learning_rate": 0.0009025607241868057,
      "loss": 1.5669,
      "step": 6528
    },
    {
      "epoch": 0.6887130801687764,
      "grad_norm": 0.44700443744659424,
      "learning_rate": 0.000901308896991912,
      "loss": 1.5556,
      "step": 6529
    },
    {
      "epoch": 0.6888185654008439,
      "grad_norm": 0.4597567021846771,
      "learning_rate": 0.0009000588060473156,
      "loss": 1.5587,
      "step": 6530
    },
    {
      "epoch": 0.6889240506329114,
      "grad_norm": 0.480972021818161,
      "learning_rate": 0.0008988104489448847,
      "loss": 1.5182,
      "step": 6531
    },
    {
      "epoch": 0.689029535864979,
      "grad_norm": 0.44952064752578735,
      "learning_rate": 0.0008975638232798275,
      "loss": 1.5398,
      "step": 6532
    },
    {
      "epoch": 0.6891350210970464,
      "grad_norm": 0.5170565247535706,
      "learning_rate": 0.0008963189266506873,
      "loss": 1.5435,
      "step": 6533
    },
    {
      "epoch": 0.6892405063291139,
      "grad_norm": 0.4291788935661316,
      "learning_rate": 0.000895075756659338,
      "loss": 1.5614,
      "step": 6534
    },
    {
      "epoch": 0.6893459915611815,
      "grad_norm": 0.571100115776062,
      "learning_rate": 0.0008938343109109804,
      "loss": 1.606,
      "step": 6535
    },
    {
      "epoch": 0.6894514767932489,
      "grad_norm": 0.4427125155925751,
      "learning_rate": 0.0008925945870141361,
      "loss": 1.5812,
      "step": 6536
    },
    {
      "epoch": 0.6895569620253165,
      "grad_norm": 0.4918278455734253,
      "learning_rate": 0.0008913565825806437,
      "loss": 1.593,
      "step": 6537
    },
    {
      "epoch": 0.689662447257384,
      "grad_norm": 0.4837626516819,
      "learning_rate": 0.0008901202952256545,
      "loss": 1.5812,
      "step": 6538
    },
    {
      "epoch": 0.6897679324894515,
      "grad_norm": 0.4777262210845947,
      "learning_rate": 0.000888885722567627,
      "loss": 1.6016,
      "step": 6539
    },
    {
      "epoch": 0.689873417721519,
      "grad_norm": 0.4948436915874481,
      "learning_rate": 0.0008876528622283235,
      "loss": 1.5775,
      "step": 6540
    },
    {
      "epoch": 0.6899789029535865,
      "grad_norm": 0.49163320660591125,
      "learning_rate": 0.0008864217118328042,
      "loss": 1.5581,
      "step": 6541
    },
    {
      "epoch": 0.690084388185654,
      "grad_norm": 0.5227521657943726,
      "learning_rate": 0.0008851922690094236,
      "loss": 1.5612,
      "step": 6542
    },
    {
      "epoch": 0.6901898734177215,
      "grad_norm": 0.5023651123046875,
      "learning_rate": 0.0008839645313898255,
      "loss": 1.5396,
      "step": 6543
    },
    {
      "epoch": 0.6902953586497891,
      "grad_norm": 0.5832700133323669,
      "learning_rate": 0.0008827384966089386,
      "loss": 1.5889,
      "step": 6544
    },
    {
      "epoch": 0.6904008438818565,
      "grad_norm": 0.5308181047439575,
      "learning_rate": 0.0008815141623049723,
      "loss": 1.5654,
      "step": 6545
    },
    {
      "epoch": 0.6905063291139241,
      "grad_norm": 0.5272881984710693,
      "learning_rate": 0.0008802915261194108,
      "loss": 1.5759,
      "step": 6546
    },
    {
      "epoch": 0.6906118143459916,
      "grad_norm": 0.5909549593925476,
      "learning_rate": 0.00087907058569701,
      "loss": 1.5925,
      "step": 6547
    },
    {
      "epoch": 0.690717299578059,
      "grad_norm": 0.4635099768638611,
      "learning_rate": 0.0008778513386857928,
      "loss": 1.5697,
      "step": 6548
    },
    {
      "epoch": 0.6908227848101266,
      "grad_norm": 0.5166774988174438,
      "learning_rate": 0.0008766337827370438,
      "loss": 1.5476,
      "step": 6549
    },
    {
      "epoch": 0.6909282700421941,
      "grad_norm": 0.4826642870903015,
      "learning_rate": 0.0008754179155053053,
      "loss": 1.5229,
      "step": 6550
    },
    {
      "epoch": 0.6910337552742616,
      "grad_norm": 0.4578114449977875,
      "learning_rate": 0.0008742037346483729,
      "loss": 1.5979,
      "step": 6551
    },
    {
      "epoch": 0.6911392405063291,
      "grad_norm": 0.5516717433929443,
      "learning_rate": 0.00087299123782729,
      "loss": 1.5412,
      "step": 6552
    },
    {
      "epoch": 0.6912447257383966,
      "grad_norm": 0.44047558307647705,
      "learning_rate": 0.0008717804227063454,
      "loss": 1.5782,
      "step": 6553
    },
    {
      "epoch": 0.6913502109704641,
      "grad_norm": 0.46936869621276855,
      "learning_rate": 0.0008705712869530661,
      "loss": 1.5913,
      "step": 6554
    },
    {
      "epoch": 0.6914556962025317,
      "grad_norm": 0.4681449234485626,
      "learning_rate": 0.0008693638282382152,
      "loss": 1.5446,
      "step": 6555
    },
    {
      "epoch": 0.6915611814345991,
      "grad_norm": 0.5338939428329468,
      "learning_rate": 0.0008681580442357857,
      "loss": 1.5814,
      "step": 6556
    },
    {
      "epoch": 0.6916666666666667,
      "grad_norm": 0.5230638384819031,
      "learning_rate": 0.000866953932622997,
      "loss": 1.5655,
      "step": 6557
    },
    {
      "epoch": 0.6917721518987342,
      "grad_norm": 0.48675185441970825,
      "learning_rate": 0.0008657514910802905,
      "loss": 1.5576,
      "step": 6558
    },
    {
      "epoch": 0.6918776371308016,
      "grad_norm": 0.5382813811302185,
      "learning_rate": 0.000864550717291324,
      "loss": 1.5421,
      "step": 6559
    },
    {
      "epoch": 0.6919831223628692,
      "grad_norm": 0.5913761258125305,
      "learning_rate": 0.0008633516089429683,
      "loss": 1.5629,
      "step": 6560
    },
    {
      "epoch": 0.6920886075949367,
      "grad_norm": 0.6232990622520447,
      "learning_rate": 0.0008621541637253029,
      "loss": 1.5848,
      "step": 6561
    },
    {
      "epoch": 0.6921940928270042,
      "grad_norm": 0.4897494614124298,
      "learning_rate": 0.0008609583793316104,
      "loss": 1.5525,
      "step": 6562
    },
    {
      "epoch": 0.6922995780590717,
      "grad_norm": 0.5203216075897217,
      "learning_rate": 0.0008597642534583734,
      "loss": 1.522,
      "step": 6563
    },
    {
      "epoch": 0.6924050632911393,
      "grad_norm": 0.5382830500602722,
      "learning_rate": 0.0008585717838052689,
      "loss": 1.5407,
      "step": 6564
    },
    {
      "epoch": 0.6925105485232067,
      "grad_norm": 0.5280312299728394,
      "learning_rate": 0.0008573809680751646,
      "loss": 1.5775,
      "step": 6565
    },
    {
      "epoch": 0.6926160337552743,
      "grad_norm": 0.5594722628593445,
      "learning_rate": 0.0008561918039741143,
      "loss": 1.5636,
      "step": 6566
    },
    {
      "epoch": 0.6927215189873418,
      "grad_norm": 0.4756142497062683,
      "learning_rate": 0.0008550042892113534,
      "loss": 1.5952,
      "step": 6567
    },
    {
      "epoch": 0.6928270042194092,
      "grad_norm": 0.6093976497650146,
      "learning_rate": 0.0008538184214992943,
      "loss": 1.5593,
      "step": 6568
    },
    {
      "epoch": 0.6929324894514768,
      "grad_norm": 0.46203407645225525,
      "learning_rate": 0.0008526341985535229,
      "loss": 1.5786,
      "step": 6569
    },
    {
      "epoch": 0.6930379746835443,
      "grad_norm": 0.5525863766670227,
      "learning_rate": 0.0008514516180927928,
      "loss": 1.5517,
      "step": 6570
    },
    {
      "epoch": 0.6931434599156118,
      "grad_norm": 0.49556243419647217,
      "learning_rate": 0.0008502706778390216,
      "loss": 1.5781,
      "step": 6571
    },
    {
      "epoch": 0.6932489451476793,
      "grad_norm": 0.5265051126480103,
      "learning_rate": 0.0008490913755172875,
      "loss": 1.5707,
      "step": 6572
    },
    {
      "epoch": 0.6933544303797469,
      "grad_norm": 0.5411252379417419,
      "learning_rate": 0.0008479137088558226,
      "loss": 1.5391,
      "step": 6573
    },
    {
      "epoch": 0.6934599156118143,
      "grad_norm": 0.47814619541168213,
      "learning_rate": 0.0008467376755860108,
      "loss": 1.5752,
      "step": 6574
    },
    {
      "epoch": 0.6935654008438819,
      "grad_norm": 0.47830256819725037,
      "learning_rate": 0.0008455632734423824,
      "loss": 1.5643,
      "step": 6575
    },
    {
      "epoch": 0.6936708860759494,
      "grad_norm": 0.607549250125885,
      "learning_rate": 0.0008443905001626097,
      "loss": 1.5168,
      "step": 6576
    },
    {
      "epoch": 0.6937763713080168,
      "grad_norm": 0.45923298597335815,
      "learning_rate": 0.0008432193534875027,
      "loss": 1.5424,
      "step": 6577
    },
    {
      "epoch": 0.6938818565400844,
      "grad_norm": 0.7608394026756287,
      "learning_rate": 0.0008420498311610049,
      "loss": 1.5573,
      "step": 6578
    },
    {
      "epoch": 0.6939873417721519,
      "grad_norm": 0.6764566898345947,
      "learning_rate": 0.0008408819309301891,
      "loss": 1.5478,
      "step": 6579
    },
    {
      "epoch": 0.6940928270042194,
      "grad_norm": 0.5462896227836609,
      "learning_rate": 0.0008397156505452524,
      "loss": 1.5959,
      "step": 6580
    },
    {
      "epoch": 0.6941983122362869,
      "grad_norm": 0.6780955195426941,
      "learning_rate": 0.0008385509877595129,
      "loss": 1.5372,
      "step": 6581
    },
    {
      "epoch": 0.6943037974683545,
      "grad_norm": 0.5146687626838684,
      "learning_rate": 0.0008373879403294043,
      "loss": 1.5595,
      "step": 6582
    },
    {
      "epoch": 0.6944092827004219,
      "grad_norm": 0.6291818022727966,
      "learning_rate": 0.0008362265060144721,
      "loss": 1.5619,
      "step": 6583
    },
    {
      "epoch": 0.6945147679324895,
      "grad_norm": 0.5552919507026672,
      "learning_rate": 0.0008350666825773697,
      "loss": 1.5476,
      "step": 6584
    },
    {
      "epoch": 0.694620253164557,
      "grad_norm": 0.6432161927223206,
      "learning_rate": 0.0008339084677838532,
      "loss": 1.5743,
      "step": 6585
    },
    {
      "epoch": 0.6947257383966244,
      "grad_norm": 0.49259400367736816,
      "learning_rate": 0.0008327518594027778,
      "loss": 1.5471,
      "step": 6586
    },
    {
      "epoch": 0.694831223628692,
      "grad_norm": 0.6837813258171082,
      "learning_rate": 0.0008315968552060928,
      "loss": 1.5693,
      "step": 6587
    },
    {
      "epoch": 0.6949367088607595,
      "grad_norm": 0.4736441671848297,
      "learning_rate": 0.0008304434529688382,
      "loss": 1.5642,
      "step": 6588
    },
    {
      "epoch": 0.695042194092827,
      "grad_norm": 0.5561773180961609,
      "learning_rate": 0.0008292916504691397,
      "loss": 1.5588,
      "step": 6589
    },
    {
      "epoch": 0.6951476793248945,
      "grad_norm": 0.4884398579597473,
      "learning_rate": 0.0008281414454882051,
      "loss": 1.6241,
      "step": 6590
    },
    {
      "epoch": 0.6952531645569621,
      "grad_norm": 0.5623233318328857,
      "learning_rate": 0.000826992835810319,
      "loss": 1.5559,
      "step": 6591
    },
    {
      "epoch": 0.6953586497890295,
      "grad_norm": 0.4664805829524994,
      "learning_rate": 0.0008258458192228395,
      "loss": 1.5807,
      "step": 6592
    },
    {
      "epoch": 0.695464135021097,
      "grad_norm": 0.5534698367118835,
      "learning_rate": 0.0008247003935161936,
      "loss": 1.5873,
      "step": 6593
    },
    {
      "epoch": 0.6955696202531646,
      "grad_norm": 0.4882756173610687,
      "learning_rate": 0.0008235565564838727,
      "loss": 1.5443,
      "step": 6594
    },
    {
      "epoch": 0.695675105485232,
      "grad_norm": 0.5256763696670532,
      "learning_rate": 0.0008224143059224287,
      "loss": 1.5387,
      "step": 6595
    },
    {
      "epoch": 0.6957805907172996,
      "grad_norm": 0.45350080728530884,
      "learning_rate": 0.0008212736396314697,
      "loss": 1.5834,
      "step": 6596
    },
    {
      "epoch": 0.6958860759493671,
      "grad_norm": 0.55854731798172,
      "learning_rate": 0.0008201345554136556,
      "loss": 1.5803,
      "step": 6597
    },
    {
      "epoch": 0.6959915611814346,
      "grad_norm": 0.44322511553764343,
      "learning_rate": 0.0008189970510746938,
      "loss": 1.5674,
      "step": 6598
    },
    {
      "epoch": 0.6960970464135021,
      "grad_norm": 0.47145506739616394,
      "learning_rate": 0.0008178611244233354,
      "loss": 1.5709,
      "step": 6599
    },
    {
      "epoch": 0.6962025316455697,
      "grad_norm": 0.4970199763774872,
      "learning_rate": 0.0008167267732713704,
      "loss": 1.5835,
      "step": 6600
    },
    {
      "epoch": 0.6963080168776371,
      "grad_norm": 0.5133510231971741,
      "learning_rate": 0.0008155939954336242,
      "loss": 1.5401,
      "step": 6601
    },
    {
      "epoch": 0.6964135021097047,
      "grad_norm": 0.4554741680622101,
      "learning_rate": 0.0008144627887279526,
      "loss": 1.549,
      "step": 6602
    },
    {
      "epoch": 0.6965189873417722,
      "grad_norm": 0.5472342371940613,
      "learning_rate": 0.0008133331509752381,
      "loss": 1.5807,
      "step": 6603
    },
    {
      "epoch": 0.6966244725738396,
      "grad_norm": 0.5181341171264648,
      "learning_rate": 0.0008122050799993857,
      "loss": 1.5762,
      "step": 6604
    },
    {
      "epoch": 0.6967299578059072,
      "grad_norm": 0.5634217858314514,
      "learning_rate": 0.0008110785736273183,
      "loss": 1.5564,
      "step": 6605
    },
    {
      "epoch": 0.6968354430379747,
      "grad_norm": 0.5411979556083679,
      "learning_rate": 0.0008099536296889731,
      "loss": 1.5235,
      "step": 6606
    },
    {
      "epoch": 0.6969409282700422,
      "grad_norm": 0.4497706890106201,
      "learning_rate": 0.0008088302460172971,
      "loss": 1.553,
      "step": 6607
    },
    {
      "epoch": 0.6970464135021097,
      "grad_norm": 0.5087087154388428,
      "learning_rate": 0.0008077084204482425,
      "loss": 1.5056,
      "step": 6608
    },
    {
      "epoch": 0.6971518987341773,
      "grad_norm": 0.5459806323051453,
      "learning_rate": 0.0008065881508207637,
      "loss": 1.5539,
      "step": 6609
    },
    {
      "epoch": 0.6972573839662447,
      "grad_norm": 0.5198091864585876,
      "learning_rate": 0.0008054694349768117,
      "loss": 1.6082,
      "step": 6610
    },
    {
      "epoch": 0.6973628691983123,
      "grad_norm": 0.5953304171562195,
      "learning_rate": 0.000804352270761331,
      "loss": 1.5139,
      "step": 6611
    },
    {
      "epoch": 0.6974683544303798,
      "grad_norm": 0.577004611492157,
      "learning_rate": 0.0008032366560222553,
      "loss": 1.569,
      "step": 6612
    },
    {
      "epoch": 0.6975738396624472,
      "grad_norm": 0.5869729518890381,
      "learning_rate": 0.0008021225886105027,
      "loss": 1.608,
      "step": 6613
    },
    {
      "epoch": 0.6976793248945148,
      "grad_norm": 0.48819971084594727,
      "learning_rate": 0.0008010100663799726,
      "loss": 1.5605,
      "step": 6614
    },
    {
      "epoch": 0.6977848101265823,
      "grad_norm": 0.5203465223312378,
      "learning_rate": 0.0007998990871875402,
      "loss": 1.5404,
      "step": 6615
    },
    {
      "epoch": 0.6978902953586498,
      "grad_norm": 0.5282949209213257,
      "learning_rate": 0.0007987896488930539,
      "loss": 1.5809,
      "step": 6616
    },
    {
      "epoch": 0.6979957805907173,
      "grad_norm": 0.48796385526657104,
      "learning_rate": 0.0007976817493593302,
      "loss": 1.5812,
      "step": 6617
    },
    {
      "epoch": 0.6981012658227848,
      "grad_norm": 0.5409809350967407,
      "learning_rate": 0.0007965753864521494,
      "loss": 1.5467,
      "step": 6618
    },
    {
      "epoch": 0.6982067510548523,
      "grad_norm": 0.4720846116542816,
      "learning_rate": 0.0007954705580402523,
      "loss": 1.5555,
      "step": 6619
    },
    {
      "epoch": 0.6983122362869199,
      "grad_norm": 0.5541364550590515,
      "learning_rate": 0.0007943672619953359,
      "loss": 1.5407,
      "step": 6620
    },
    {
      "epoch": 0.6984177215189873,
      "grad_norm": 0.4393485188484192,
      "learning_rate": 0.0007932654961920488,
      "loss": 1.5584,
      "step": 6621
    },
    {
      "epoch": 0.6985232067510548,
      "grad_norm": 0.4939269721508026,
      "learning_rate": 0.0007921652585079873,
      "loss": 1.5851,
      "step": 6622
    },
    {
      "epoch": 0.6986286919831224,
      "grad_norm": 0.5119544267654419,
      "learning_rate": 0.0007910665468236916,
      "loss": 1.5439,
      "step": 6623
    },
    {
      "epoch": 0.6987341772151898,
      "grad_norm": 0.47787803411483765,
      "learning_rate": 0.0007899693590226415,
      "loss": 1.5496,
      "step": 6624
    },
    {
      "epoch": 0.6988396624472574,
      "grad_norm": 0.48977434635162354,
      "learning_rate": 0.0007888736929912525,
      "loss": 1.5652,
      "step": 6625
    },
    {
      "epoch": 0.6989451476793249,
      "grad_norm": 0.505007803440094,
      "learning_rate": 0.0007877795466188712,
      "loss": 1.5323,
      "step": 6626
    },
    {
      "epoch": 0.6990506329113924,
      "grad_norm": 0.48375818133354187,
      "learning_rate": 0.0007866869177977721,
      "loss": 1.5483,
      "step": 6627
    },
    {
      "epoch": 0.6991561181434599,
      "grad_norm": 0.5514267683029175,
      "learning_rate": 0.0007855958044231527,
      "loss": 1.5648,
      "step": 6628
    },
    {
      "epoch": 0.6992616033755275,
      "grad_norm": 0.5183826088905334,
      "learning_rate": 0.0007845062043931298,
      "loss": 1.5337,
      "step": 6629
    },
    {
      "epoch": 0.6993670886075949,
      "grad_norm": 0.4456199109554291,
      "learning_rate": 0.0007834181156087356,
      "loss": 1.5538,
      "step": 6630
    },
    {
      "epoch": 0.6994725738396624,
      "grad_norm": 0.7880083918571472,
      "learning_rate": 0.0007823315359739137,
      "loss": 1.5667,
      "step": 6631
    },
    {
      "epoch": 0.69957805907173,
      "grad_norm": 0.48204830288887024,
      "learning_rate": 0.0007812464633955144,
      "loss": 1.5504,
      "step": 6632
    },
    {
      "epoch": 0.6996835443037974,
      "grad_norm": 0.6027332544326782,
      "learning_rate": 0.0007801628957832918,
      "loss": 1.5507,
      "step": 6633
    },
    {
      "epoch": 0.699789029535865,
      "grad_norm": 0.4713844060897827,
      "learning_rate": 0.0007790808310498984,
      "loss": 1.5784,
      "step": 6634
    },
    {
      "epoch": 0.6998945147679325,
      "grad_norm": 0.6253005862236023,
      "learning_rate": 0.0007780002671108819,
      "loss": 1.554,
      "step": 6635
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46921494603157043,
      "learning_rate": 0.0007769212018846818,
      "loss": 1.533,
      "step": 6636
    },
    {
      "epoch": 0.7001054852320675,
      "grad_norm": 0.4672645926475525,
      "learning_rate": 0.0007758436332926237,
      "loss": 1.5921,
      "step": 6637
    },
    {
      "epoch": 0.700210970464135,
      "grad_norm": 0.492880642414093,
      "learning_rate": 0.000774767559258917,
      "loss": 1.5715,
      "step": 6638
    },
    {
      "epoch": 0.7003164556962025,
      "grad_norm": 0.5257558226585388,
      "learning_rate": 0.0007736929777106497,
      "loss": 1.544,
      "step": 6639
    },
    {
      "epoch": 0.70042194092827,
      "grad_norm": 0.4666813611984253,
      "learning_rate": 0.0007726198865777852,
      "loss": 1.5434,
      "step": 6640
    },
    {
      "epoch": 0.7005274261603376,
      "grad_norm": 0.4529757499694824,
      "learning_rate": 0.000771548283793158,
      "loss": 1.5697,
      "step": 6641
    },
    {
      "epoch": 0.700632911392405,
      "grad_norm": 0.4499722123146057,
      "learning_rate": 0.000770478167292469,
      "loss": 1.5468,
      "step": 6642
    },
    {
      "epoch": 0.7007383966244726,
      "grad_norm": 0.4351818859577179,
      "learning_rate": 0.0007694095350142834,
      "loss": 1.5656,
      "step": 6643
    },
    {
      "epoch": 0.7008438818565401,
      "grad_norm": 0.4683161973953247,
      "learning_rate": 0.0007683423849000246,
      "loss": 1.5193,
      "step": 6644
    },
    {
      "epoch": 0.7009493670886076,
      "grad_norm": 0.4370659589767456,
      "learning_rate": 0.0007672767148939714,
      "loss": 1.5203,
      "step": 6645
    },
    {
      "epoch": 0.7010548523206751,
      "grad_norm": 0.4852113723754883,
      "learning_rate": 0.0007662125229432543,
      "loss": 1.5383,
      "step": 6646
    },
    {
      "epoch": 0.7011603375527427,
      "grad_norm": 0.4382689297199249,
      "learning_rate": 0.0007651498069978504,
      "loss": 1.5637,
      "step": 6647
    },
    {
      "epoch": 0.7012658227848101,
      "grad_norm": 0.47351962327957153,
      "learning_rate": 0.0007640885650105806,
      "loss": 1.5911,
      "step": 6648
    },
    {
      "epoch": 0.7013713080168776,
      "grad_norm": 0.46821945905685425,
      "learning_rate": 0.000763028794937105,
      "loss": 1.5085,
      "step": 6649
    },
    {
      "epoch": 0.7014767932489452,
      "grad_norm": 0.5545336604118347,
      "learning_rate": 0.0007619704947359191,
      "loss": 1.5797,
      "step": 6650
    },
    {
      "epoch": 0.7015822784810126,
      "grad_norm": 0.4496048092842102,
      "learning_rate": 0.0007609136623683499,
      "loss": 1.5488,
      "step": 6651
    },
    {
      "epoch": 0.7016877637130802,
      "grad_norm": 0.5856061577796936,
      "learning_rate": 0.0007598582957985525,
      "loss": 1.5727,
      "step": 6652
    },
    {
      "epoch": 0.7017932489451477,
      "grad_norm": 0.5694226026535034,
      "learning_rate": 0.000758804392993505,
      "loss": 1.5304,
      "step": 6653
    },
    {
      "epoch": 0.7018987341772152,
      "grad_norm": 0.4599759578704834,
      "learning_rate": 0.0007577519519230052,
      "loss": 1.5606,
      "step": 6654
    },
    {
      "epoch": 0.7020042194092827,
      "grad_norm": 0.48302337527275085,
      "learning_rate": 0.0007567009705596673,
      "loss": 1.5549,
      "step": 6655
    },
    {
      "epoch": 0.7021097046413503,
      "grad_norm": 0.621795654296875,
      "learning_rate": 0.0007556514468789169,
      "loss": 1.5369,
      "step": 6656
    },
    {
      "epoch": 0.7022151898734177,
      "grad_norm": 0.48464787006378174,
      "learning_rate": 0.0007546033788589883,
      "loss": 1.5485,
      "step": 6657
    },
    {
      "epoch": 0.7023206751054852,
      "grad_norm": 0.5004467368125916,
      "learning_rate": 0.0007535567644809191,
      "loss": 1.5845,
      "step": 6658
    },
    {
      "epoch": 0.7024261603375528,
      "grad_norm": 0.4963955581188202,
      "learning_rate": 0.0007525116017285476,
      "loss": 1.5577,
      "step": 6659
    },
    {
      "epoch": 0.7025316455696202,
      "grad_norm": 0.5065571665763855,
      "learning_rate": 0.0007514678885885087,
      "loss": 1.5333,
      "step": 6660
    },
    {
      "epoch": 0.7026371308016878,
      "grad_norm": 0.5023692846298218,
      "learning_rate": 0.000750425623050229,
      "loss": 1.525,
      "step": 6661
    },
    {
      "epoch": 0.7027426160337553,
      "grad_norm": 0.5315647721290588,
      "learning_rate": 0.0007493848031059247,
      "loss": 1.5581,
      "step": 6662
    },
    {
      "epoch": 0.7028481012658228,
      "grad_norm": 0.4617088735103607,
      "learning_rate": 0.0007483454267505959,
      "loss": 1.5489,
      "step": 6663
    },
    {
      "epoch": 0.7029535864978903,
      "grad_norm": 0.4618825614452362,
      "learning_rate": 0.000747307491982024,
      "loss": 1.5284,
      "step": 6664
    },
    {
      "epoch": 0.7030590717299579,
      "grad_norm": 0.5144922137260437,
      "learning_rate": 0.0007462709968007675,
      "loss": 1.5541,
      "step": 6665
    },
    {
      "epoch": 0.7031645569620253,
      "grad_norm": 0.5299652218818665,
      "learning_rate": 0.0007452359392101578,
      "loss": 1.557,
      "step": 6666
    },
    {
      "epoch": 0.7032700421940928,
      "grad_norm": 0.5746594071388245,
      "learning_rate": 0.0007442023172162958,
      "loss": 1.5452,
      "step": 6667
    },
    {
      "epoch": 0.7033755274261604,
      "grad_norm": 0.4928874969482422,
      "learning_rate": 0.0007431701288280478,
      "loss": 1.5495,
      "step": 6668
    },
    {
      "epoch": 0.7034810126582278,
      "grad_norm": 0.6005679965019226,
      "learning_rate": 0.0007421393720570417,
      "loss": 1.5909,
      "step": 6669
    },
    {
      "epoch": 0.7035864978902954,
      "grad_norm": 0.5231202244758606,
      "learning_rate": 0.0007411100449176633,
      "loss": 1.5639,
      "step": 6670
    },
    {
      "epoch": 0.7036919831223629,
      "grad_norm": 0.5319112539291382,
      "learning_rate": 0.0007400821454270525,
      "loss": 1.5808,
      "step": 6671
    },
    {
      "epoch": 0.7037974683544304,
      "grad_norm": 0.7475935220718384,
      "learning_rate": 0.0007390556716050993,
      "loss": 1.5525,
      "step": 6672
    },
    {
      "epoch": 0.7039029535864979,
      "grad_norm": 0.5199599266052246,
      "learning_rate": 0.0007380306214744398,
      "loss": 1.5733,
      "step": 6673
    },
    {
      "epoch": 0.7040084388185655,
      "grad_norm": 0.656639814376831,
      "learning_rate": 0.000737006993060453,
      "loss": 1.5876,
      "step": 6674
    },
    {
      "epoch": 0.7041139240506329,
      "grad_norm": 0.48102647066116333,
      "learning_rate": 0.0007359847843912564,
      "loss": 1.5304,
      "step": 6675
    },
    {
      "epoch": 0.7042194092827004,
      "grad_norm": 0.6165723204612732,
      "learning_rate": 0.0007349639934977029,
      "loss": 1.5622,
      "step": 6676
    },
    {
      "epoch": 0.704324894514768,
      "grad_norm": 0.4821445047855377,
      "learning_rate": 0.0007339446184133759,
      "loss": 1.5614,
      "step": 6677
    },
    {
      "epoch": 0.7044303797468354,
      "grad_norm": 0.5459588170051575,
      "learning_rate": 0.0007329266571745864,
      "loss": 1.569,
      "step": 6678
    },
    {
      "epoch": 0.704535864978903,
      "grad_norm": 0.5332533717155457,
      "learning_rate": 0.0007319101078203694,
      "loss": 1.5115,
      "step": 6679
    },
    {
      "epoch": 0.7046413502109705,
      "grad_norm": 0.4783789813518524,
      "learning_rate": 0.0007308949683924791,
      "loss": 1.533,
      "step": 6680
    },
    {
      "epoch": 0.704746835443038,
      "grad_norm": 0.5693449378013611,
      "learning_rate": 0.0007298812369353862,
      "loss": 1.5556,
      "step": 6681
    },
    {
      "epoch": 0.7048523206751055,
      "grad_norm": 0.4317052662372589,
      "learning_rate": 0.0007288689114962731,
      "loss": 1.5578,
      "step": 6682
    },
    {
      "epoch": 0.7049578059071729,
      "grad_norm": 0.7175117135047913,
      "learning_rate": 0.0007278579901250316,
      "loss": 1.5323,
      "step": 6683
    },
    {
      "epoch": 0.7050632911392405,
      "grad_norm": 0.43420615792274475,
      "learning_rate": 0.0007268484708742574,
      "loss": 1.5126,
      "step": 6684
    },
    {
      "epoch": 0.705168776371308,
      "grad_norm": 0.5980640649795532,
      "learning_rate": 0.0007258403517992476,
      "loss": 1.5645,
      "step": 6685
    },
    {
      "epoch": 0.7052742616033755,
      "grad_norm": 0.5155321955680847,
      "learning_rate": 0.0007248336309579965,
      "loss": 1.5639,
      "step": 6686
    },
    {
      "epoch": 0.705379746835443,
      "grad_norm": 0.46816548705101013,
      "learning_rate": 0.0007238283064111917,
      "loss": 1.5626,
      "step": 6687
    },
    {
      "epoch": 0.7054852320675106,
      "grad_norm": 0.5139795541763306,
      "learning_rate": 0.0007228243762222109,
      "loss": 1.521,
      "step": 6688
    },
    {
      "epoch": 0.705590717299578,
      "grad_norm": 0.4960365891456604,
      "learning_rate": 0.0007218218384571178,
      "loss": 1.5857,
      "step": 6689
    },
    {
      "epoch": 0.7056962025316456,
      "grad_norm": 0.540518581867218,
      "learning_rate": 0.000720820691184658,
      "loss": 1.5441,
      "step": 6690
    },
    {
      "epoch": 0.7058016877637131,
      "grad_norm": 0.6019273996353149,
      "learning_rate": 0.0007198209324762563,
      "loss": 1.5547,
      "step": 6691
    },
    {
      "epoch": 0.7059071729957805,
      "grad_norm": 0.42266330122947693,
      "learning_rate": 0.0007188225604060119,
      "loss": 1.5589,
      "step": 6692
    },
    {
      "epoch": 0.7060126582278481,
      "grad_norm": 0.4965789020061493,
      "learning_rate": 0.0007178255730506955,
      "loss": 1.5742,
      "step": 6693
    },
    {
      "epoch": 0.7061181434599156,
      "grad_norm": 0.4750126004219055,
      "learning_rate": 0.0007168299684897451,
      "loss": 1.5545,
      "step": 6694
    },
    {
      "epoch": 0.7062236286919831,
      "grad_norm": 0.46209055185317993,
      "learning_rate": 0.0007158357448052624,
      "loss": 1.5373,
      "step": 6695
    },
    {
      "epoch": 0.7063291139240506,
      "grad_norm": 0.44234421849250793,
      "learning_rate": 0.0007148429000820094,
      "loss": 1.5492,
      "step": 6696
    },
    {
      "epoch": 0.7064345991561182,
      "grad_norm": 0.45361265540122986,
      "learning_rate": 0.0007138514324074042,
      "loss": 1.5253,
      "step": 6697
    },
    {
      "epoch": 0.7065400843881856,
      "grad_norm": 0.5187786221504211,
      "learning_rate": 0.0007128613398715179,
      "loss": 1.5796,
      "step": 6698
    },
    {
      "epoch": 0.7066455696202532,
      "grad_norm": 0.4797686040401459,
      "learning_rate": 0.0007118726205670703,
      "loss": 1.501,
      "step": 6699
    },
    {
      "epoch": 0.7067510548523207,
      "grad_norm": 0.4801947772502899,
      "learning_rate": 0.0007108852725894269,
      "loss": 1.544,
      "step": 6700
    },
    {
      "epoch": 0.7068565400843881,
      "grad_norm": 0.5309736728668213,
      "learning_rate": 0.0007098992940365947,
      "loss": 1.5786,
      "step": 6701
    },
    {
      "epoch": 0.7069620253164557,
      "grad_norm": 0.4449637532234192,
      "learning_rate": 0.0007089146830092185,
      "loss": 1.5686,
      "step": 6702
    },
    {
      "epoch": 0.7070675105485232,
      "grad_norm": 0.5048536658287048,
      "learning_rate": 0.0007079314376105778,
      "loss": 1.5446,
      "step": 6703
    },
    {
      "epoch": 0.7071729957805907,
      "grad_norm": 0.5020825266838074,
      "learning_rate": 0.0007069495559465826,
      "loss": 1.5345,
      "step": 6704
    },
    {
      "epoch": 0.7072784810126582,
      "grad_norm": 0.45239654183387756,
      "learning_rate": 0.0007059690361257701,
      "loss": 1.564,
      "step": 6705
    },
    {
      "epoch": 0.7073839662447258,
      "grad_norm": 0.4716244339942932,
      "learning_rate": 0.0007049898762593007,
      "loss": 1.5339,
      "step": 6706
    },
    {
      "epoch": 0.7074894514767932,
      "grad_norm": 0.4415547251701355,
      "learning_rate": 0.0007040120744609548,
      "loss": 1.5754,
      "step": 6707
    },
    {
      "epoch": 0.7075949367088608,
      "grad_norm": 0.5089999437332153,
      "learning_rate": 0.0007030356288471288,
      "loss": 1.533,
      "step": 6708
    },
    {
      "epoch": 0.7077004219409283,
      "grad_norm": 0.5045871138572693,
      "learning_rate": 0.0007020605375368316,
      "loss": 1.5544,
      "step": 6709
    },
    {
      "epoch": 0.7078059071729957,
      "grad_norm": 0.468222975730896,
      "learning_rate": 0.000701086798651681,
      "loss": 1.5268,
      "step": 6710
    },
    {
      "epoch": 0.7079113924050633,
      "grad_norm": 0.43993693590164185,
      "learning_rate": 0.0007001144103159,
      "loss": 1.5239,
      "step": 6711
    },
    {
      "epoch": 0.7080168776371308,
      "grad_norm": 0.4992451071739197,
      "learning_rate": 0.0006991433706563135,
      "loss": 1.5415,
      "step": 6712
    },
    {
      "epoch": 0.7081223628691983,
      "grad_norm": 0.4695892035961151,
      "learning_rate": 0.0006981736778023443,
      "loss": 1.5456,
      "step": 6713
    },
    {
      "epoch": 0.7082278481012658,
      "grad_norm": 0.46786677837371826,
      "learning_rate": 0.0006972053298860092,
      "loss": 1.5512,
      "step": 6714
    },
    {
      "epoch": 0.7083333333333334,
      "grad_norm": 0.5553925037384033,
      "learning_rate": 0.0006962383250419168,
      "loss": 1.5008,
      "step": 6715
    },
    {
      "epoch": 0.7084388185654008,
      "grad_norm": 0.4732776880264282,
      "learning_rate": 0.0006952726614072621,
      "loss": 1.5253,
      "step": 6716
    },
    {
      "epoch": 0.7085443037974684,
      "grad_norm": 0.7153058052062988,
      "learning_rate": 0.0006943083371218242,
      "loss": 1.575,
      "step": 6717
    },
    {
      "epoch": 0.7086497890295359,
      "grad_norm": 0.4462694227695465,
      "learning_rate": 0.0006933453503279619,
      "loss": 1.5211,
      "step": 6718
    },
    {
      "epoch": 0.7087552742616033,
      "grad_norm": 0.8087603449821472,
      "learning_rate": 0.000692383699170611,
      "loss": 1.5559,
      "step": 6719
    },
    {
      "epoch": 0.7088607594936709,
      "grad_norm": 0.4940249025821686,
      "learning_rate": 0.0006914233817972798,
      "loss": 1.5628,
      "step": 6720
    },
    {
      "epoch": 0.7089662447257384,
      "grad_norm": 0.6718744039535522,
      "learning_rate": 0.0006904643963580462,
      "loss": 1.5474,
      "step": 6721
    },
    {
      "epoch": 0.7090717299578059,
      "grad_norm": 0.4637923240661621,
      "learning_rate": 0.0006895067410055538,
      "loss": 1.5666,
      "step": 6722
    },
    {
      "epoch": 0.7091772151898734,
      "grad_norm": 0.6198895573616028,
      "learning_rate": 0.0006885504138950082,
      "loss": 1.5473,
      "step": 6723
    },
    {
      "epoch": 0.709282700421941,
      "grad_norm": 0.49211519956588745,
      "learning_rate": 0.0006875954131841743,
      "loss": 1.5466,
      "step": 6724
    },
    {
      "epoch": 0.7093881856540084,
      "grad_norm": 0.5213711857795715,
      "learning_rate": 0.0006866417370333716,
      "loss": 1.5409,
      "step": 6725
    },
    {
      "epoch": 0.709493670886076,
      "grad_norm": 0.4944036602973938,
      "learning_rate": 0.0006856893836054712,
      "loss": 1.5572,
      "step": 6726
    },
    {
      "epoch": 0.7095991561181435,
      "grad_norm": 0.4943283200263977,
      "learning_rate": 0.0006847383510658927,
      "loss": 1.5455,
      "step": 6727
    },
    {
      "epoch": 0.7097046413502109,
      "grad_norm": 0.4628939628601074,
      "learning_rate": 0.0006837886375825994,
      "loss": 1.572,
      "step": 6728
    },
    {
      "epoch": 0.7098101265822785,
      "grad_norm": 0.5107986927032471,
      "learning_rate": 0.0006828402413260966,
      "loss": 1.5497,
      "step": 6729
    },
    {
      "epoch": 0.709915611814346,
      "grad_norm": 0.5592716336250305,
      "learning_rate": 0.0006818931604694261,
      "loss": 1.538,
      "step": 6730
    },
    {
      "epoch": 0.7100210970464135,
      "grad_norm": 0.43582016229629517,
      "learning_rate": 0.0006809473931881645,
      "loss": 1.5641,
      "step": 6731
    },
    {
      "epoch": 0.710126582278481,
      "grad_norm": 0.5421050190925598,
      "learning_rate": 0.0006800029376604181,
      "loss": 1.5455,
      "step": 6732
    },
    {
      "epoch": 0.7102320675105486,
      "grad_norm": 0.49958208203315735,
      "learning_rate": 0.0006790597920668204,
      "loss": 1.5634,
      "step": 6733
    },
    {
      "epoch": 0.710337552742616,
      "grad_norm": 0.5179060101509094,
      "learning_rate": 0.0006781179545905287,
      "loss": 1.5224,
      "step": 6734
    },
    {
      "epoch": 0.7104430379746836,
      "grad_norm": 0.461525559425354,
      "learning_rate": 0.0006771774234172195,
      "loss": 1.5529,
      "step": 6735
    },
    {
      "epoch": 0.7105485232067511,
      "grad_norm": 0.48807981610298157,
      "learning_rate": 0.0006762381967350861,
      "loss": 1.5602,
      "step": 6736
    },
    {
      "epoch": 0.7106540084388185,
      "grad_norm": 0.433828741312027,
      "learning_rate": 0.0006753002727348349,
      "loss": 1.5287,
      "step": 6737
    },
    {
      "epoch": 0.7107594936708861,
      "grad_norm": 0.5105122923851013,
      "learning_rate": 0.0006743636496096813,
      "loss": 1.5554,
      "step": 6738
    },
    {
      "epoch": 0.7108649789029536,
      "grad_norm": 0.5092545747756958,
      "learning_rate": 0.0006734283255553471,
      "loss": 1.5854,
      "step": 6739
    },
    {
      "epoch": 0.7109704641350211,
      "grad_norm": 0.438706636428833,
      "learning_rate": 0.0006724942987700563,
      "loss": 1.539,
      "step": 6740
    },
    {
      "epoch": 0.7110759493670886,
      "grad_norm": 0.5282023549079895,
      "learning_rate": 0.0006715615674545319,
      "loss": 1.5423,
      "step": 6741
    },
    {
      "epoch": 0.7111814345991562,
      "grad_norm": 0.47605130076408386,
      "learning_rate": 0.0006706301298119925,
      "loss": 1.5498,
      "step": 6742
    },
    {
      "epoch": 0.7112869198312236,
      "grad_norm": 0.4943104088306427,
      "learning_rate": 0.0006696999840481491,
      "loss": 1.5332,
      "step": 6743
    },
    {
      "epoch": 0.7113924050632912,
      "grad_norm": 0.4608191251754761,
      "learning_rate": 0.0006687711283712009,
      "loss": 1.5101,
      "step": 6744
    },
    {
      "epoch": 0.7114978902953587,
      "grad_norm": 0.49573028087615967,
      "learning_rate": 0.0006678435609918323,
      "loss": 1.5847,
      "step": 6745
    },
    {
      "epoch": 0.7116033755274261,
      "grad_norm": 0.5111983418464661,
      "learning_rate": 0.0006669172801232098,
      "loss": 1.5657,
      "step": 6746
    },
    {
      "epoch": 0.7117088607594937,
      "grad_norm": 0.4408809244632721,
      "learning_rate": 0.0006659922839809779,
      "loss": 1.5509,
      "step": 6747
    },
    {
      "epoch": 0.7118143459915611,
      "grad_norm": 0.48499247431755066,
      "learning_rate": 0.0006650685707832559,
      "loss": 1.5468,
      "step": 6748
    },
    {
      "epoch": 0.7119198312236287,
      "grad_norm": 0.4597436487674713,
      "learning_rate": 0.0006641461387506347,
      "loss": 1.5816,
      "step": 6749
    },
    {
      "epoch": 0.7120253164556962,
      "grad_norm": 0.5312247276306152,
      "learning_rate": 0.0006632249861061732,
      "loss": 1.542,
      "step": 6750
    },
    {
      "epoch": 0.7121308016877637,
      "grad_norm": 0.41715967655181885,
      "learning_rate": 0.0006623051110753947,
      "loss": 1.5182,
      "step": 6751
    },
    {
      "epoch": 0.7122362869198312,
      "grad_norm": 0.47443118691444397,
      "learning_rate": 0.0006613865118862837,
      "loss": 1.5406,
      "step": 6752
    },
    {
      "epoch": 0.7123417721518988,
      "grad_norm": 0.5484854578971863,
      "learning_rate": 0.0006604691867692828,
      "loss": 1.5397,
      "step": 6753
    },
    {
      "epoch": 0.7124472573839662,
      "grad_norm": 0.43628114461898804,
      "learning_rate": 0.0006595531339572881,
      "loss": 1.5931,
      "step": 6754
    },
    {
      "epoch": 0.7125527426160337,
      "grad_norm": 0.5638396739959717,
      "learning_rate": 0.0006586383516856475,
      "loss": 1.5484,
      "step": 6755
    },
    {
      "epoch": 0.7126582278481013,
      "grad_norm": 0.4444471597671509,
      "learning_rate": 0.000657724838192156,
      "loss": 1.5138,
      "step": 6756
    },
    {
      "epoch": 0.7127637130801687,
      "grad_norm": 0.4277331531047821,
      "learning_rate": 0.0006568125917170526,
      "loss": 1.5556,
      "step": 6757
    },
    {
      "epoch": 0.7128691983122363,
      "grad_norm": 0.5183447003364563,
      "learning_rate": 0.0006559016105030176,
      "loss": 1.56,
      "step": 6758
    },
    {
      "epoch": 0.7129746835443038,
      "grad_norm": 0.5353155732154846,
      "learning_rate": 0.0006549918927951678,
      "loss": 1.5639,
      "step": 6759
    },
    {
      "epoch": 0.7130801687763713,
      "grad_norm": 0.4568440914154053,
      "learning_rate": 0.0006540834368410549,
      "loss": 1.5383,
      "step": 6760
    },
    {
      "epoch": 0.7131856540084388,
      "grad_norm": 0.5182054042816162,
      "learning_rate": 0.0006531762408906606,
      "loss": 1.5635,
      "step": 6761
    },
    {
      "epoch": 0.7132911392405064,
      "grad_norm": 0.5307048559188843,
      "learning_rate": 0.0006522703031963939,
      "loss": 1.5344,
      "step": 6762
    },
    {
      "epoch": 0.7133966244725738,
      "grad_norm": 0.4830094873905182,
      "learning_rate": 0.0006513656220130878,
      "loss": 1.5648,
      "step": 6763
    },
    {
      "epoch": 0.7135021097046413,
      "grad_norm": 0.5077260732650757,
      "learning_rate": 0.0006504621955979959,
      "loss": 1.5575,
      "step": 6764
    },
    {
      "epoch": 0.7136075949367089,
      "grad_norm": 0.5386867523193359,
      "learning_rate": 0.0006495600222107885,
      "loss": 1.5691,
      "step": 6765
    },
    {
      "epoch": 0.7137130801687763,
      "grad_norm": 0.4486021399497986,
      "learning_rate": 0.0006486591001135502,
      "loss": 1.5083,
      "step": 6766
    },
    {
      "epoch": 0.7138185654008439,
      "grad_norm": 0.5497661828994751,
      "learning_rate": 0.0006477594275707758,
      "loss": 1.5345,
      "step": 6767
    },
    {
      "epoch": 0.7139240506329114,
      "grad_norm": 0.45292559266090393,
      "learning_rate": 0.000646861002849367,
      "loss": 1.5525,
      "step": 6768
    },
    {
      "epoch": 0.7140295358649789,
      "grad_norm": 0.5148671269416809,
      "learning_rate": 0.0006459638242186297,
      "loss": 1.5703,
      "step": 6769
    },
    {
      "epoch": 0.7141350210970464,
      "grad_norm": 0.5652586817741394,
      "learning_rate": 0.0006450678899502701,
      "loss": 1.5541,
      "step": 6770
    },
    {
      "epoch": 0.714240506329114,
      "grad_norm": 0.47230979800224304,
      "learning_rate": 0.0006441731983183911,
      "loss": 1.5991,
      "step": 6771
    },
    {
      "epoch": 0.7143459915611814,
      "grad_norm": 0.4903254508972168,
      "learning_rate": 0.0006432797475994899,
      "loss": 1.5503,
      "step": 6772
    },
    {
      "epoch": 0.7144514767932489,
      "grad_norm": 0.508822500705719,
      "learning_rate": 0.0006423875360724538,
      "loss": 1.57,
      "step": 6773
    },
    {
      "epoch": 0.7145569620253165,
      "grad_norm": 0.48720335960388184,
      "learning_rate": 0.0006414965620185574,
      "loss": 1.51,
      "step": 6774
    },
    {
      "epoch": 0.7146624472573839,
      "grad_norm": 0.5062872767448425,
      "learning_rate": 0.0006406068237214593,
      "loss": 1.534,
      "step": 6775
    },
    {
      "epoch": 0.7147679324894515,
      "grad_norm": 0.5193167328834534,
      "learning_rate": 0.000639718319467198,
      "loss": 1.5091,
      "step": 6776
    },
    {
      "epoch": 0.714873417721519,
      "grad_norm": 0.5054948925971985,
      "learning_rate": 0.0006388310475441899,
      "loss": 1.5396,
      "step": 6777
    },
    {
      "epoch": 0.7149789029535865,
      "grad_norm": 0.538214921951294,
      "learning_rate": 0.0006379450062432248,
      "loss": 1.5514,
      "step": 6778
    },
    {
      "epoch": 0.715084388185654,
      "grad_norm": 0.5200361013412476,
      "learning_rate": 0.0006370601938574639,
      "loss": 1.5492,
      "step": 6779
    },
    {
      "epoch": 0.7151898734177216,
      "grad_norm": 0.5863214135169983,
      "learning_rate": 0.0006361766086824345,
      "loss": 1.5541,
      "step": 6780
    },
    {
      "epoch": 0.715295358649789,
      "grad_norm": 0.5286057591438293,
      "learning_rate": 0.0006352942490160293,
      "loss": 1.5844,
      "step": 6781
    },
    {
      "epoch": 0.7154008438818565,
      "grad_norm": 0.5346243977546692,
      "learning_rate": 0.0006344131131585007,
      "loss": 1.5468,
      "step": 6782
    },
    {
      "epoch": 0.7155063291139241,
      "grad_norm": 0.4713151752948761,
      "learning_rate": 0.0006335331994124592,
      "loss": 1.5269,
      "step": 6783
    },
    {
      "epoch": 0.7156118143459915,
      "grad_norm": 0.6031091809272766,
      "learning_rate": 0.0006326545060828696,
      "loss": 1.5856,
      "step": 6784
    },
    {
      "epoch": 0.7157172995780591,
      "grad_norm": 0.504538357257843,
      "learning_rate": 0.000631777031477047,
      "loss": 1.5647,
      "step": 6785
    },
    {
      "epoch": 0.7158227848101266,
      "grad_norm": 0.5394483804702759,
      "learning_rate": 0.0006309007739046551,
      "loss": 1.5109,
      "step": 6786
    },
    {
      "epoch": 0.7159282700421941,
      "grad_norm": 0.4800265431404114,
      "learning_rate": 0.0006300257316777014,
      "loss": 1.5292,
      "step": 6787
    },
    {
      "epoch": 0.7160337552742616,
      "grad_norm": 0.596564769744873,
      "learning_rate": 0.0006291519031105347,
      "loss": 1.5496,
      "step": 6788
    },
    {
      "epoch": 0.7161392405063292,
      "grad_norm": 0.5047953724861145,
      "learning_rate": 0.0006282792865198421,
      "loss": 1.5348,
      "step": 6789
    },
    {
      "epoch": 0.7162447257383966,
      "grad_norm": 0.5160011649131775,
      "learning_rate": 0.000627407880224645,
      "loss": 1.5847,
      "step": 6790
    },
    {
      "epoch": 0.7163502109704641,
      "grad_norm": 0.43249937891960144,
      "learning_rate": 0.0006265376825462964,
      "loss": 1.541,
      "step": 6791
    },
    {
      "epoch": 0.7164556962025317,
      "grad_norm": 0.4553639590740204,
      "learning_rate": 0.0006256686918084777,
      "loss": 1.5413,
      "step": 6792
    },
    {
      "epoch": 0.7165611814345991,
      "grad_norm": 0.48277103900909424,
      "learning_rate": 0.0006248009063371953,
      "loss": 1.5554,
      "step": 6793
    },
    {
      "epoch": 0.7166666666666667,
      "grad_norm": 0.4682178199291229,
      "learning_rate": 0.0006239343244607771,
      "loss": 1.5349,
      "step": 6794
    },
    {
      "epoch": 0.7167721518987342,
      "grad_norm": 0.458538293838501,
      "learning_rate": 0.0006230689445098697,
      "loss": 1.5283,
      "step": 6795
    },
    {
      "epoch": 0.7168776371308017,
      "grad_norm": 0.5126941204071045,
      "learning_rate": 0.0006222047648174351,
      "loss": 1.5005,
      "step": 6796
    },
    {
      "epoch": 0.7169831223628692,
      "grad_norm": 0.5203979015350342,
      "learning_rate": 0.0006213417837187475,
      "loss": 1.5298,
      "step": 6797
    },
    {
      "epoch": 0.7170886075949368,
      "grad_norm": 0.47017431259155273,
      "learning_rate": 0.00062047999955139,
      "loss": 1.5489,
      "step": 6798
    },
    {
      "epoch": 0.7171940928270042,
      "grad_norm": 0.5460437536239624,
      "learning_rate": 0.0006196194106552512,
      "loss": 1.5707,
      "step": 6799
    },
    {
      "epoch": 0.7172995780590717,
      "grad_norm": 0.4368559718132019,
      "learning_rate": 0.0006187600153725225,
      "loss": 1.5052,
      "step": 6800
    },
    {
      "epoch": 0.7174050632911393,
      "grad_norm": 0.44950324296951294,
      "learning_rate": 0.0006179018120476945,
      "loss": 1.5724,
      "step": 6801
    },
    {
      "epoch": 0.7175105485232067,
      "grad_norm": 0.48503077030181885,
      "learning_rate": 0.000617044799027554,
      "loss": 1.5275,
      "step": 6802
    },
    {
      "epoch": 0.7176160337552743,
      "grad_norm": 0.4647253751754761,
      "learning_rate": 0.0006161889746611808,
      "loss": 1.5534,
      "step": 6803
    },
    {
      "epoch": 0.7177215189873418,
      "grad_norm": 0.43151435256004333,
      "learning_rate": 0.0006153343372999444,
      "loss": 1.5005,
      "step": 6804
    },
    {
      "epoch": 0.7178270042194093,
      "grad_norm": 0.4642019271850586,
      "learning_rate": 0.0006144808852975009,
      "loss": 1.5747,
      "step": 6805
    },
    {
      "epoch": 0.7179324894514768,
      "grad_norm": 0.48096582293510437,
      "learning_rate": 0.00061362861700979,
      "loss": 1.5541,
      "step": 6806
    },
    {
      "epoch": 0.7180379746835444,
      "grad_norm": 0.4491911828517914,
      "learning_rate": 0.0006127775307950314,
      "loss": 1.5223,
      "step": 6807
    },
    {
      "epoch": 0.7181434599156118,
      "grad_norm": 0.4846862554550171,
      "learning_rate": 0.000611927625013722,
      "loss": 1.5723,
      "step": 6808
    },
    {
      "epoch": 0.7182489451476793,
      "grad_norm": 0.467145711183548,
      "learning_rate": 0.0006110788980286328,
      "loss": 1.5508,
      "step": 6809
    },
    {
      "epoch": 0.7183544303797469,
      "grad_norm": 0.5330292582511902,
      "learning_rate": 0.0006102313482048055,
      "loss": 1.4719,
      "step": 6810
    },
    {
      "epoch": 0.7184599156118143,
      "grad_norm": 0.4568015933036804,
      "learning_rate": 0.0006093849739095494,
      "loss": 1.5626,
      "step": 6811
    },
    {
      "epoch": 0.7185654008438819,
      "grad_norm": 0.4917406439781189,
      "learning_rate": 0.0006085397735124382,
      "loss": 1.5604,
      "step": 6812
    },
    {
      "epoch": 0.7186708860759494,
      "grad_norm": 0.47678980231285095,
      "learning_rate": 0.0006076957453853072,
      "loss": 1.5474,
      "step": 6813
    },
    {
      "epoch": 0.7187763713080169,
      "grad_norm": 0.5251429677009583,
      "learning_rate": 0.0006068528879022496,
      "loss": 1.5151,
      "step": 6814
    },
    {
      "epoch": 0.7188818565400844,
      "grad_norm": 0.4720397889614105,
      "learning_rate": 0.0006060111994396143,
      "loss": 1.5487,
      "step": 6815
    },
    {
      "epoch": 0.7189873417721518,
      "grad_norm": 0.49456480145454407,
      "learning_rate": 0.0006051706783760013,
      "loss": 1.5377,
      "step": 6816
    },
    {
      "epoch": 0.7190928270042194,
      "grad_norm": 0.4655502736568451,
      "learning_rate": 0.0006043313230922601,
      "loss": 1.5338,
      "step": 6817
    },
    {
      "epoch": 0.7191983122362869,
      "grad_norm": 0.44060802459716797,
      "learning_rate": 0.0006034931319714858,
      "loss": 1.5246,
      "step": 6818
    },
    {
      "epoch": 0.7193037974683544,
      "grad_norm": 0.4712231755256653,
      "learning_rate": 0.0006026561033990158,
      "loss": 1.5401,
      "step": 6819
    },
    {
      "epoch": 0.7194092827004219,
      "grad_norm": 0.48318424820899963,
      "learning_rate": 0.0006018202357624274,
      "loss": 1.5455,
      "step": 6820
    },
    {
      "epoch": 0.7195147679324895,
      "grad_norm": 0.5258792042732239,
      "learning_rate": 0.0006009855274515337,
      "loss": 1.5436,
      "step": 6821
    },
    {
      "epoch": 0.7196202531645569,
      "grad_norm": 0.4671013057231903,
      "learning_rate": 0.0006001519768583819,
      "loss": 1.5901,
      "step": 6822
    },
    {
      "epoch": 0.7197257383966245,
      "grad_norm": 0.4964064061641693,
      "learning_rate": 0.0005993195823772488,
      "loss": 1.4981,
      "step": 6823
    },
    {
      "epoch": 0.719831223628692,
      "grad_norm": 0.4290485382080078,
      "learning_rate": 0.0005984883424046384,
      "loss": 1.5417,
      "step": 6824
    },
    {
      "epoch": 0.7199367088607594,
      "grad_norm": 0.48920175433158875,
      "learning_rate": 0.0005976582553392788,
      "loss": 1.5413,
      "step": 6825
    },
    {
      "epoch": 0.720042194092827,
      "grad_norm": 0.4913826286792755,
      "learning_rate": 0.000596829319582119,
      "loss": 1.5136,
      "step": 6826
    },
    {
      "epoch": 0.7201476793248945,
      "grad_norm": 0.4746927320957184,
      "learning_rate": 0.0005960015335363258,
      "loss": 1.5734,
      "step": 6827
    },
    {
      "epoch": 0.720253164556962,
      "grad_norm": 0.4370986521244049,
      "learning_rate": 0.0005951748956072806,
      "loss": 1.5327,
      "step": 6828
    },
    {
      "epoch": 0.7203586497890295,
      "grad_norm": 0.4876527488231659,
      "learning_rate": 0.000594349404202577,
      "loss": 1.5264,
      "step": 6829
    },
    {
      "epoch": 0.7204641350210971,
      "grad_norm": 0.4784664809703827,
      "learning_rate": 0.0005935250577320168,
      "loss": 1.53,
      "step": 6830
    },
    {
      "epoch": 0.7205696202531645,
      "grad_norm": 0.51405268907547,
      "learning_rate": 0.0005927018546076072,
      "loss": 1.5459,
      "step": 6831
    },
    {
      "epoch": 0.7206751054852321,
      "grad_norm": 0.47052356600761414,
      "learning_rate": 0.0005918797932435585,
      "loss": 1.5165,
      "step": 6832
    },
    {
      "epoch": 0.7207805907172996,
      "grad_norm": 0.5199061036109924,
      "learning_rate": 0.0005910588720562799,
      "loss": 1.5278,
      "step": 6833
    },
    {
      "epoch": 0.720886075949367,
      "grad_norm": 0.46353626251220703,
      "learning_rate": 0.0005902390894643773,
      "loss": 1.5231,
      "step": 6834
    },
    {
      "epoch": 0.7209915611814346,
      "grad_norm": 0.49449896812438965,
      "learning_rate": 0.0005894204438886498,
      "loss": 1.5408,
      "step": 6835
    },
    {
      "epoch": 0.7210970464135021,
      "grad_norm": 0.48055365681648254,
      "learning_rate": 0.0005886029337520871,
      "loss": 1.5254,
      "step": 6836
    },
    {
      "epoch": 0.7212025316455696,
      "grad_norm": 0.465086430311203,
      "learning_rate": 0.0005877865574798655,
      "loss": 1.5349,
      "step": 6837
    },
    {
      "epoch": 0.7213080168776371,
      "grad_norm": 0.47648733854293823,
      "learning_rate": 0.0005869713134993463,
      "loss": 1.5483,
      "step": 6838
    },
    {
      "epoch": 0.7214135021097047,
      "grad_norm": 0.4783395528793335,
      "learning_rate": 0.0005861572002400716,
      "loss": 1.5161,
      "step": 6839
    },
    {
      "epoch": 0.7215189873417721,
      "grad_norm": 0.46843791007995605,
      "learning_rate": 0.0005853442161337618,
      "loss": 1.5241,
      "step": 6840
    },
    {
      "epoch": 0.7216244725738397,
      "grad_norm": 0.44689735770225525,
      "learning_rate": 0.0005845323596143124,
      "loss": 1.5196,
      "step": 6841
    },
    {
      "epoch": 0.7217299578059072,
      "grad_norm": 0.4600822627544403,
      "learning_rate": 0.0005837216291177911,
      "loss": 1.5461,
      "step": 6842
    },
    {
      "epoch": 0.7218354430379746,
      "grad_norm": 0.4526119530200958,
      "learning_rate": 0.0005829120230824345,
      "loss": 1.5256,
      "step": 6843
    },
    {
      "epoch": 0.7219409282700422,
      "grad_norm": 0.49139970541000366,
      "learning_rate": 0.0005821035399486458,
      "loss": 1.563,
      "step": 6844
    },
    {
      "epoch": 0.7220464135021097,
      "grad_norm": 0.4772544801235199,
      "learning_rate": 0.0005812961781589908,
      "loss": 1.5491,
      "step": 6845
    },
    {
      "epoch": 0.7221518987341772,
      "grad_norm": 0.4974606931209564,
      "learning_rate": 0.000580489936158196,
      "loss": 1.5521,
      "step": 6846
    },
    {
      "epoch": 0.7222573839662447,
      "grad_norm": 0.4158450663089752,
      "learning_rate": 0.0005796848123931443,
      "loss": 1.5462,
      "step": 6847
    },
    {
      "epoch": 0.7223628691983123,
      "grad_norm": 0.5039369463920593,
      "learning_rate": 0.0005788808053128734,
      "loss": 1.5191,
      "step": 6848
    },
    {
      "epoch": 0.7224683544303797,
      "grad_norm": 0.48272404074668884,
      "learning_rate": 0.0005780779133685717,
      "loss": 1.553,
      "step": 6849
    },
    {
      "epoch": 0.7225738396624473,
      "grad_norm": 0.47118261456489563,
      "learning_rate": 0.0005772761350135759,
      "loss": 1.5135,
      "step": 6850
    },
    {
      "epoch": 0.7226793248945148,
      "grad_norm": 0.47816649079322815,
      "learning_rate": 0.000576475468703368,
      "loss": 1.5575,
      "step": 6851
    },
    {
      "epoch": 0.7227848101265822,
      "grad_norm": 0.48343923687934875,
      "learning_rate": 0.0005756759128955721,
      "loss": 1.5462,
      "step": 6852
    },
    {
      "epoch": 0.7228902953586498,
      "grad_norm": 0.5833190679550171,
      "learning_rate": 0.0005748774660499515,
      "loss": 1.5293,
      "step": 6853
    },
    {
      "epoch": 0.7229957805907173,
      "grad_norm": 0.4382333755493164,
      "learning_rate": 0.0005740801266284058,
      "loss": 1.533,
      "step": 6854
    },
    {
      "epoch": 0.7231012658227848,
      "grad_norm": 0.5338608622550964,
      "learning_rate": 0.0005732838930949678,
      "loss": 1.4874,
      "step": 6855
    },
    {
      "epoch": 0.7232067510548523,
      "grad_norm": 0.5252207517623901,
      "learning_rate": 0.000572488763915801,
      "loss": 1.5237,
      "step": 6856
    },
    {
      "epoch": 0.7233122362869199,
      "grad_norm": 0.532107412815094,
      "learning_rate": 0.0005716947375591958,
      "loss": 1.5034,
      "step": 6857
    },
    {
      "epoch": 0.7234177215189873,
      "grad_norm": 0.5585890412330627,
      "learning_rate": 0.0005709018124955674,
      "loss": 1.5689,
      "step": 6858
    },
    {
      "epoch": 0.7235232067510549,
      "grad_norm": 0.4830462336540222,
      "learning_rate": 0.0005701099871974524,
      "loss": 1.5637,
      "step": 6859
    },
    {
      "epoch": 0.7236286919831224,
      "grad_norm": 0.5976365208625793,
      "learning_rate": 0.0005693192601395058,
      "loss": 1.5257,
      "step": 6860
    },
    {
      "epoch": 0.7237341772151898,
      "grad_norm": 0.5155830383300781,
      "learning_rate": 0.0005685296297984985,
      "loss": 1.5014,
      "step": 6861
    },
    {
      "epoch": 0.7238396624472574,
      "grad_norm": 0.4879033863544464,
      "learning_rate": 0.0005677410946533138,
      "loss": 1.5686,
      "step": 6862
    },
    {
      "epoch": 0.7239451476793249,
      "grad_norm": 0.611248791217804,
      "learning_rate": 0.0005669536531849449,
      "loss": 1.56,
      "step": 6863
    },
    {
      "epoch": 0.7240506329113924,
      "grad_norm": 0.4879395663738251,
      "learning_rate": 0.0005661673038764916,
      "loss": 1.5521,
      "step": 6864
    },
    {
      "epoch": 0.7241561181434599,
      "grad_norm": 0.5838509202003479,
      "learning_rate": 0.000565382045213158,
      "loss": 1.54,
      "step": 6865
    },
    {
      "epoch": 0.7242616033755275,
      "grad_norm": 0.4879651963710785,
      "learning_rate": 0.000564597875682249,
      "loss": 1.5229,
      "step": 6866
    },
    {
      "epoch": 0.7243670886075949,
      "grad_norm": 0.5476601719856262,
      "learning_rate": 0.0005638147937731673,
      "loss": 1.5195,
      "step": 6867
    },
    {
      "epoch": 0.7244725738396625,
      "grad_norm": 0.5659372806549072,
      "learning_rate": 0.0005630327979774111,
      "loss": 1.5301,
      "step": 6868
    },
    {
      "epoch": 0.72457805907173,
      "grad_norm": 0.5292827486991882,
      "learning_rate": 0.0005622518867885708,
      "loss": 1.5105,
      "step": 6869
    },
    {
      "epoch": 0.7246835443037974,
      "grad_norm": 0.5626305341720581,
      "learning_rate": 0.000561472058702326,
      "loss": 1.5146,
      "step": 6870
    },
    {
      "epoch": 0.724789029535865,
      "grad_norm": 0.5764346122741699,
      "learning_rate": 0.0005606933122164428,
      "loss": 1.5445,
      "step": 6871
    },
    {
      "epoch": 0.7248945147679325,
      "grad_norm": 0.598465085029602,
      "learning_rate": 0.000559915645830771,
      "loss": 1.5468,
      "step": 6872
    },
    {
      "epoch": 0.725,
      "grad_norm": 0.507113516330719,
      "learning_rate": 0.0005591390580472411,
      "loss": 1.5384,
      "step": 6873
    },
    {
      "epoch": 0.7251054852320675,
      "grad_norm": 0.5054839849472046,
      "learning_rate": 0.0005583635473698608,
      "loss": 1.539,
      "step": 6874
    },
    {
      "epoch": 0.7252109704641351,
      "grad_norm": 0.5646956562995911,
      "learning_rate": 0.0005575891123047136,
      "loss": 1.5366,
      "step": 6875
    },
    {
      "epoch": 0.7253164556962025,
      "grad_norm": 0.5350216031074524,
      "learning_rate": 0.0005568157513599543,
      "loss": 1.5278,
      "step": 6876
    },
    {
      "epoch": 0.7254219409282701,
      "grad_norm": 0.5685900449752808,
      "learning_rate": 0.0005560434630458071,
      "loss": 1.528,
      "step": 6877
    },
    {
      "epoch": 0.7255274261603376,
      "grad_norm": 0.5420767068862915,
      "learning_rate": 0.0005552722458745627,
      "loss": 1.5287,
      "step": 6878
    },
    {
      "epoch": 0.725632911392405,
      "grad_norm": 0.5841749906539917,
      "learning_rate": 0.0005545020983605748,
      "loss": 1.5748,
      "step": 6879
    },
    {
      "epoch": 0.7257383966244726,
      "grad_norm": 0.5920206308364868,
      "learning_rate": 0.000553733019020258,
      "loss": 1.5368,
      "step": 6880
    },
    {
      "epoch": 0.72584388185654,
      "grad_norm": 0.44279634952545166,
      "learning_rate": 0.0005529650063720844,
      "loss": 1.5332,
      "step": 6881
    },
    {
      "epoch": 0.7259493670886076,
      "grad_norm": 0.660584032535553,
      "learning_rate": 0.0005521980589365809,
      "loss": 1.5267,
      "step": 6882
    },
    {
      "epoch": 0.7260548523206751,
      "grad_norm": 0.572847843170166,
      "learning_rate": 0.0005514321752363265,
      "loss": 1.5362,
      "step": 6883
    },
    {
      "epoch": 0.7261603375527426,
      "grad_norm": 0.47688448429107666,
      "learning_rate": 0.0005506673537959495,
      "loss": 1.5292,
      "step": 6884
    },
    {
      "epoch": 0.7262658227848101,
      "grad_norm": 0.5942841172218323,
      "learning_rate": 0.0005499035931421242,
      "loss": 1.5507,
      "step": 6885
    },
    {
      "epoch": 0.7263713080168777,
      "grad_norm": 0.5383564829826355,
      "learning_rate": 0.0005491408918035683,
      "loss": 1.5042,
      "step": 6886
    },
    {
      "epoch": 0.7264767932489451,
      "grad_norm": 0.5472618937492371,
      "learning_rate": 0.0005483792483110407,
      "loss": 1.5407,
      "step": 6887
    },
    {
      "epoch": 0.7265822784810126,
      "grad_norm": 0.573840320110321,
      "learning_rate": 0.0005476186611973374,
      "loss": 1.5409,
      "step": 6888
    },
    {
      "epoch": 0.7266877637130802,
      "grad_norm": 0.5130054354667664,
      "learning_rate": 0.0005468591289972898,
      "loss": 1.5172,
      "step": 6889
    },
    {
      "epoch": 0.7267932489451476,
      "grad_norm": 0.5248138308525085,
      "learning_rate": 0.0005461006502477612,
      "loss": 1.5026,
      "step": 6890
    },
    {
      "epoch": 0.7268987341772152,
      "grad_norm": 0.4802354574203491,
      "learning_rate": 0.0005453432234876445,
      "loss": 1.5522,
      "step": 6891
    },
    {
      "epoch": 0.7270042194092827,
      "grad_norm": 0.47751444578170776,
      "learning_rate": 0.000544586847257859,
      "loss": 1.5515,
      "step": 6892
    },
    {
      "epoch": 0.7271097046413502,
      "grad_norm": 0.5061583518981934,
      "learning_rate": 0.0005438315201013477,
      "loss": 1.5465,
      "step": 6893
    },
    {
      "epoch": 0.7272151898734177,
      "grad_norm": 0.4960957467556,
      "learning_rate": 0.0005430772405630743,
      "loss": 1.5079,
      "step": 6894
    },
    {
      "epoch": 0.7273206751054853,
      "grad_norm": 0.5025561451911926,
      "learning_rate": 0.0005423240071900209,
      "loss": 1.5009,
      "step": 6895
    },
    {
      "epoch": 0.7274261603375527,
      "grad_norm": 0.5052862763404846,
      "learning_rate": 0.0005415718185311847,
      "loss": 1.5264,
      "step": 6896
    },
    {
      "epoch": 0.7275316455696202,
      "grad_norm": 0.5196452140808105,
      "learning_rate": 0.0005408206731375755,
      "loss": 1.5369,
      "step": 6897
    },
    {
      "epoch": 0.7276371308016878,
      "grad_norm": 0.5555493831634521,
      "learning_rate": 0.000540070569562213,
      "loss": 1.5405,
      "step": 6898
    },
    {
      "epoch": 0.7277426160337552,
      "grad_norm": 0.4979768693447113,
      "learning_rate": 0.0005393215063601232,
      "loss": 1.5278,
      "step": 6899
    },
    {
      "epoch": 0.7278481012658228,
      "grad_norm": 0.5646880865097046,
      "learning_rate": 0.0005385734820883369,
      "loss": 1.525,
      "step": 6900
    },
    {
      "epoch": 0.7279535864978903,
      "grad_norm": 0.5847175121307373,
      "learning_rate": 0.000537826495305886,
      "loss": 1.562,
      "step": 6901
    },
    {
      "epoch": 0.7280590717299578,
      "grad_norm": 0.5824342370033264,
      "learning_rate": 0.000537080544573801,
      "loss": 1.5441,
      "step": 6902
    },
    {
      "epoch": 0.7281645569620253,
      "grad_norm": 0.49832651019096375,
      "learning_rate": 0.000536335628455108,
      "loss": 1.5617,
      "step": 6903
    },
    {
      "epoch": 0.7282700421940929,
      "grad_norm": 0.5771176218986511,
      "learning_rate": 0.0005355917455148267,
      "loss": 1.5337,
      "step": 6904
    },
    {
      "epoch": 0.7283755274261603,
      "grad_norm": 0.5652545690536499,
      "learning_rate": 0.0005348488943199665,
      "loss": 1.5294,
      "step": 6905
    },
    {
      "epoch": 0.7284810126582278,
      "grad_norm": 0.6075310707092285,
      "learning_rate": 0.0005341070734395245,
      "loss": 1.5677,
      "step": 6906
    },
    {
      "epoch": 0.7285864978902954,
      "grad_norm": 0.5299625992774963,
      "learning_rate": 0.0005333662814444825,
      "loss": 1.5506,
      "step": 6907
    },
    {
      "epoch": 0.7286919831223628,
      "grad_norm": 0.5377302765846252,
      "learning_rate": 0.0005326265169078048,
      "loss": 1.5475,
      "step": 6908
    },
    {
      "epoch": 0.7287974683544304,
      "grad_norm": 0.6283926367759705,
      "learning_rate": 0.0005318877784044343,
      "loss": 1.605,
      "step": 6909
    },
    {
      "epoch": 0.7289029535864979,
      "grad_norm": 0.49732524156570435,
      "learning_rate": 0.0005311500645112907,
      "loss": 1.5416,
      "step": 6910
    },
    {
      "epoch": 0.7290084388185654,
      "grad_norm": 0.593315839767456,
      "learning_rate": 0.0005304133738072676,
      "loss": 1.5539,
      "step": 6911
    },
    {
      "epoch": 0.7291139240506329,
      "grad_norm": 0.5013739466667175,
      "learning_rate": 0.0005296777048732292,
      "loss": 1.4969,
      "step": 6912
    },
    {
      "epoch": 0.7292194092827005,
      "grad_norm": 0.5769602656364441,
      "learning_rate": 0.0005289430562920086,
      "loss": 1.5487,
      "step": 6913
    },
    {
      "epoch": 0.7293248945147679,
      "grad_norm": 0.6069418787956238,
      "learning_rate": 0.0005282094266484041,
      "loss": 1.519,
      "step": 6914
    },
    {
      "epoch": 0.7294303797468354,
      "grad_norm": 0.6460288166999817,
      "learning_rate": 0.0005274768145291769,
      "loss": 1.5384,
      "step": 6915
    },
    {
      "epoch": 0.729535864978903,
      "grad_norm": 0.5016931891441345,
      "learning_rate": 0.0005267452185230483,
      "loss": 1.5303,
      "step": 6916
    },
    {
      "epoch": 0.7296413502109704,
      "grad_norm": 0.557963490486145,
      "learning_rate": 0.000526014637220697,
      "loss": 1.5459,
      "step": 6917
    },
    {
      "epoch": 0.729746835443038,
      "grad_norm": 0.4795740842819214,
      "learning_rate": 0.0005252850692147567,
      "loss": 1.5441,
      "step": 6918
    },
    {
      "epoch": 0.7298523206751055,
      "grad_norm": 0.46857258677482605,
      "learning_rate": 0.0005245565130998126,
      "loss": 1.5326,
      "step": 6919
    },
    {
      "epoch": 0.729957805907173,
      "grad_norm": 0.4902258813381195,
      "learning_rate": 0.0005238289674723993,
      "loss": 1.53,
      "step": 6920
    },
    {
      "epoch": 0.7300632911392405,
      "grad_norm": 0.5052496790885925,
      "learning_rate": 0.0005231024309309981,
      "loss": 1.5458,
      "step": 6921
    },
    {
      "epoch": 0.7301687763713081,
      "grad_norm": 0.4708039164543152,
      "learning_rate": 0.0005223769020760345,
      "loss": 1.5223,
      "step": 6922
    },
    {
      "epoch": 0.7302742616033755,
      "grad_norm": 0.49621108174324036,
      "learning_rate": 0.0005216523795098743,
      "loss": 1.5444,
      "step": 6923
    },
    {
      "epoch": 0.730379746835443,
      "grad_norm": 0.5251426100730896,
      "learning_rate": 0.0005209288618368225,
      "loss": 1.5455,
      "step": 6924
    },
    {
      "epoch": 0.7304852320675106,
      "grad_norm": 0.45124509930610657,
      "learning_rate": 0.0005202063476631198,
      "loss": 1.5344,
      "step": 6925
    },
    {
      "epoch": 0.730590717299578,
      "grad_norm": 0.5204045176506042,
      "learning_rate": 0.0005194848355969396,
      "loss": 1.5593,
      "step": 6926
    },
    {
      "epoch": 0.7306962025316456,
      "grad_norm": 0.5032989382743835,
      "learning_rate": 0.0005187643242483862,
      "loss": 1.5308,
      "step": 6927
    },
    {
      "epoch": 0.7308016877637131,
      "grad_norm": 0.4412859380245209,
      "learning_rate": 0.0005180448122294913,
      "loss": 1.5187,
      "step": 6928
    },
    {
      "epoch": 0.7309071729957806,
      "grad_norm": 0.4908093214035034,
      "learning_rate": 0.000517326298154212,
      "loss": 1.5394,
      "step": 6929
    },
    {
      "epoch": 0.7310126582278481,
      "grad_norm": 0.5381917953491211,
      "learning_rate": 0.0005166087806384274,
      "loss": 1.5714,
      "step": 6930
    },
    {
      "epoch": 0.7311181434599157,
      "grad_norm": 0.43251463770866394,
      "learning_rate": 0.0005158922582999367,
      "loss": 1.5488,
      "step": 6931
    },
    {
      "epoch": 0.7312236286919831,
      "grad_norm": 0.5274459719657898,
      "learning_rate": 0.0005151767297584562,
      "loss": 1.5127,
      "step": 6932
    },
    {
      "epoch": 0.7313291139240506,
      "grad_norm": 0.575049638748169,
      "learning_rate": 0.0005144621936356161,
      "loss": 1.5735,
      "step": 6933
    },
    {
      "epoch": 0.7314345991561182,
      "grad_norm": 0.4349435865879059,
      "learning_rate": 0.000513748648554959,
      "loss": 1.5318,
      "step": 6934
    },
    {
      "epoch": 0.7315400843881856,
      "grad_norm": 0.5491678714752197,
      "learning_rate": 0.0005130360931419364,
      "loss": 1.5873,
      "step": 6935
    },
    {
      "epoch": 0.7316455696202532,
      "grad_norm": 0.5107643604278564,
      "learning_rate": 0.0005123245260239057,
      "loss": 1.4928,
      "step": 6936
    },
    {
      "epoch": 0.7317510548523207,
      "grad_norm": 0.48784181475639343,
      "learning_rate": 0.0005116139458301291,
      "loss": 1.5172,
      "step": 6937
    },
    {
      "epoch": 0.7318565400843882,
      "grad_norm": 0.5244420766830444,
      "learning_rate": 0.0005109043511917693,
      "loss": 1.5583,
      "step": 6938
    },
    {
      "epoch": 0.7319620253164557,
      "grad_norm": 0.49956828355789185,
      "learning_rate": 0.0005101957407418877,
      "loss": 1.5531,
      "step": 6939
    },
    {
      "epoch": 0.7320675105485233,
      "grad_norm": 0.5397307872772217,
      "learning_rate": 0.0005094881131154418,
      "loss": 1.5638,
      "step": 6940
    },
    {
      "epoch": 0.7321729957805907,
      "grad_norm": 0.49124979972839355,
      "learning_rate": 0.0005087814669492819,
      "loss": 1.5554,
      "step": 6941
    },
    {
      "epoch": 0.7322784810126582,
      "grad_norm": 0.5433766841888428,
      "learning_rate": 0.0005080758008821495,
      "loss": 1.5034,
      "step": 6942
    },
    {
      "epoch": 0.7323839662447258,
      "grad_norm": 0.5484342575073242,
      "learning_rate": 0.0005073711135546738,
      "loss": 1.5219,
      "step": 6943
    },
    {
      "epoch": 0.7324894514767932,
      "grad_norm": 0.4700629711151123,
      "learning_rate": 0.0005066674036093695,
      "loss": 1.5259,
      "step": 6944
    },
    {
      "epoch": 0.7325949367088608,
      "grad_norm": 0.7126783728599548,
      "learning_rate": 0.000505964669690634,
      "loss": 1.5439,
      "step": 6945
    },
    {
      "epoch": 0.7327004219409282,
      "grad_norm": 0.4676685929298401,
      "learning_rate": 0.0005052629104447452,
      "loss": 1.5116,
      "step": 6946
    },
    {
      "epoch": 0.7328059071729958,
      "grad_norm": 0.6898078918457031,
      "learning_rate": 0.0005045621245198582,
      "loss": 1.542,
      "step": 6947
    },
    {
      "epoch": 0.7329113924050633,
      "grad_norm": 0.4831641614437103,
      "learning_rate": 0.0005038623105660032,
      "loss": 1.5742,
      "step": 6948
    },
    {
      "epoch": 0.7330168776371307,
      "grad_norm": 0.60582035779953,
      "learning_rate": 0.0005031634672350829,
      "loss": 1.5797,
      "step": 6949
    },
    {
      "epoch": 0.7331223628691983,
      "grad_norm": 0.4761418402194977,
      "learning_rate": 0.0005024655931808696,
      "loss": 1.5374,
      "step": 6950
    },
    {
      "epoch": 0.7332278481012658,
      "grad_norm": 0.47191429138183594,
      "learning_rate": 0.0005017686870590029,
      "loss": 1.5326,
      "step": 6951
    },
    {
      "epoch": 0.7333333333333333,
      "grad_norm": 0.5319638252258301,
      "learning_rate": 0.0005010727475269868,
      "loss": 1.5304,
      "step": 6952
    },
    {
      "epoch": 0.7334388185654008,
      "grad_norm": 0.4492718577384949,
      "learning_rate": 0.0005003777732441875,
      "loss": 1.5262,
      "step": 6953
    },
    {
      "epoch": 0.7335443037974684,
      "grad_norm": 0.496799498796463,
      "learning_rate": 0.0004996837628718307,
      "loss": 1.5241,
      "step": 6954
    },
    {
      "epoch": 0.7336497890295358,
      "grad_norm": 0.5978667736053467,
      "learning_rate": 0.0004989907150729988,
      "loss": 1.5351,
      "step": 6955
    },
    {
      "epoch": 0.7337552742616034,
      "grad_norm": 0.45638683438301086,
      "learning_rate": 0.0004982986285126283,
      "loss": 1.5107,
      "step": 6956
    },
    {
      "epoch": 0.7338607594936709,
      "grad_norm": 0.5724701285362244,
      "learning_rate": 0.0004976075018575078,
      "loss": 1.5852,
      "step": 6957
    },
    {
      "epoch": 0.7339662447257383,
      "grad_norm": 0.5027989745140076,
      "learning_rate": 0.0004969173337762747,
      "loss": 1.4972,
      "step": 6958
    },
    {
      "epoch": 0.7340717299578059,
      "grad_norm": 0.4325365126132965,
      "learning_rate": 0.0004962281229394129,
      "loss": 1.4934,
      "step": 6959
    },
    {
      "epoch": 0.7341772151898734,
      "grad_norm": 0.5468115210533142,
      "learning_rate": 0.0004955398680192508,
      "loss": 1.5455,
      "step": 6960
    },
    {
      "epoch": 0.7342827004219409,
      "grad_norm": 0.49039095640182495,
      "learning_rate": 0.0004948525676899577,
      "loss": 1.5706,
      "step": 6961
    },
    {
      "epoch": 0.7343881856540084,
      "grad_norm": 0.48494425415992737,
      "learning_rate": 0.0004941662206275422,
      "loss": 1.5111,
      "step": 6962
    },
    {
      "epoch": 0.734493670886076,
      "grad_norm": 0.5570217967033386,
      "learning_rate": 0.0004934808255098488,
      "loss": 1.4724,
      "step": 6963
    },
    {
      "epoch": 0.7345991561181434,
      "grad_norm": 0.47119924426078796,
      "learning_rate": 0.000492796381016556,
      "loss": 1.5405,
      "step": 6964
    },
    {
      "epoch": 0.734704641350211,
      "grad_norm": 0.49364224076271057,
      "learning_rate": 0.0004921128858291739,
      "loss": 1.5082,
      "step": 6965
    },
    {
      "epoch": 0.7348101265822785,
      "grad_norm": 0.439891517162323,
      "learning_rate": 0.0004914303386310408,
      "loss": 1.5345,
      "step": 6966
    },
    {
      "epoch": 0.734915611814346,
      "grad_norm": 0.4645954668521881,
      "learning_rate": 0.0004907487381073215,
      "loss": 1.505,
      "step": 6967
    },
    {
      "epoch": 0.7350210970464135,
      "grad_norm": 0.4426006078720093,
      "learning_rate": 0.0004900680829450042,
      "loss": 1.5281,
      "step": 6968
    },
    {
      "epoch": 0.735126582278481,
      "grad_norm": 0.435122013092041,
      "learning_rate": 0.0004893883718328983,
      "loss": 1.5188,
      "step": 6969
    },
    {
      "epoch": 0.7352320675105485,
      "grad_norm": 0.46116241812705994,
      "learning_rate": 0.000488709603461632,
      "loss": 1.5286,
      "step": 6970
    },
    {
      "epoch": 0.735337552742616,
      "grad_norm": 0.4583013355731964,
      "learning_rate": 0.00048803177652364935,
      "loss": 1.5102,
      "step": 6971
    },
    {
      "epoch": 0.7354430379746836,
      "grad_norm": 0.4829154908657074,
      "learning_rate": 0.0004873548897132077,
      "loss": 1.5188,
      "step": 6972
    },
    {
      "epoch": 0.735548523206751,
      "grad_norm": 0.42616409063339233,
      "learning_rate": 0.000486678941726376,
      "loss": 1.5118,
      "step": 6973
    },
    {
      "epoch": 0.7356540084388186,
      "grad_norm": 0.4721052348613739,
      "learning_rate": 0.00048600393126103117,
      "loss": 1.5222,
      "step": 6974
    },
    {
      "epoch": 0.7357594936708861,
      "grad_norm": 0.47642770409584045,
      "learning_rate": 0.0004853298570168566,
      "loss": 1.5636,
      "step": 6975
    },
    {
      "epoch": 0.7358649789029535,
      "grad_norm": 0.47210177779197693,
      "learning_rate": 0.00048465671769533884,
      "loss": 1.5015,
      "step": 6976
    },
    {
      "epoch": 0.7359704641350211,
      "grad_norm": 0.4095679223537445,
      "learning_rate": 0.00048398451199976574,
      "loss": 1.5571,
      "step": 6977
    },
    {
      "epoch": 0.7360759493670886,
      "grad_norm": 0.453061580657959,
      "learning_rate": 0.0004833132386352233,
      "loss": 1.5364,
      "step": 6978
    },
    {
      "epoch": 0.7361814345991561,
      "grad_norm": 0.4312068223953247,
      "learning_rate": 0.0004826428963085938,
      "loss": 1.5308,
      "step": 6979
    },
    {
      "epoch": 0.7362869198312236,
      "grad_norm": 0.4213617146015167,
      "learning_rate": 0.000481973483728553,
      "loss": 1.5181,
      "step": 6980
    },
    {
      "epoch": 0.7363924050632912,
      "grad_norm": 0.46180662512779236,
      "learning_rate": 0.0004813049996055675,
      "loss": 1.5322,
      "step": 6981
    },
    {
      "epoch": 0.7364978902953586,
      "grad_norm": 0.44460228085517883,
      "learning_rate": 0.00048063744265189275,
      "loss": 1.5111,
      "step": 6982
    },
    {
      "epoch": 0.7366033755274262,
      "grad_norm": 0.4618487060070038,
      "learning_rate": 0.0004799708115815701,
      "loss": 1.4954,
      "step": 6983
    },
    {
      "epoch": 0.7367088607594937,
      "grad_norm": 0.4412814676761627,
      "learning_rate": 0.0004793051051104244,
      "loss": 1.5431,
      "step": 6984
    },
    {
      "epoch": 0.7368143459915611,
      "grad_norm": 0.5162926912307739,
      "learning_rate": 0.0004786403219560618,
      "loss": 1.5496,
      "step": 6985
    },
    {
      "epoch": 0.7369198312236287,
      "grad_norm": 0.42973792552948,
      "learning_rate": 0.000477976460837867,
      "loss": 1.4997,
      "step": 6986
    },
    {
      "epoch": 0.7370253164556962,
      "grad_norm": 0.47544416785240173,
      "learning_rate": 0.00047731352047700095,
      "loss": 1.492,
      "step": 6987
    },
    {
      "epoch": 0.7371308016877637,
      "grad_norm": 0.467846155166626,
      "learning_rate": 0.00047665149959639813,
      "loss": 1.5516,
      "step": 6988
    },
    {
      "epoch": 0.7372362869198312,
      "grad_norm": 0.478471040725708,
      "learning_rate": 0.00047599039692076457,
      "loss": 1.5659,
      "step": 6989
    },
    {
      "epoch": 0.7373417721518988,
      "grad_norm": 0.47790589928627014,
      "learning_rate": 0.0004753302111765748,
      "loss": 1.5022,
      "step": 6990
    },
    {
      "epoch": 0.7374472573839662,
      "grad_norm": 0.47267937660217285,
      "learning_rate": 0.00047467094109206984,
      "loss": 1.5571,
      "step": 6991
    },
    {
      "epoch": 0.7375527426160338,
      "grad_norm": 0.4550853669643402,
      "learning_rate": 0.0004740125853972546,
      "loss": 1.5711,
      "step": 6992
    },
    {
      "epoch": 0.7376582278481013,
      "grad_norm": 0.49018993973731995,
      "learning_rate": 0.00047335514282389557,
      "loss": 1.5107,
      "step": 6993
    },
    {
      "epoch": 0.7377637130801687,
      "grad_norm": 0.4644816517829895,
      "learning_rate": 0.0004726986121055179,
      "loss": 1.5286,
      "step": 6994
    },
    {
      "epoch": 0.7378691983122363,
      "grad_norm": 0.49155911803245544,
      "learning_rate": 0.00047204299197740364,
      "loss": 1.5393,
      "step": 6995
    },
    {
      "epoch": 0.7379746835443038,
      "grad_norm": 0.3947535753250122,
      "learning_rate": 0.0004713882811765889,
      "loss": 1.5335,
      "step": 6996
    },
    {
      "epoch": 0.7380801687763713,
      "grad_norm": 0.49089890718460083,
      "learning_rate": 0.0004707344784418611,
      "loss": 1.5233,
      "step": 6997
    },
    {
      "epoch": 0.7381856540084388,
      "grad_norm": 0.45698991417884827,
      "learning_rate": 0.0004700815825137577,
      "loss": 1.5086,
      "step": 6998
    },
    {
      "epoch": 0.7382911392405064,
      "grad_norm": 0.4739130437374115,
      "learning_rate": 0.0004694295921345622,
      "loss": 1.5711,
      "step": 6999
    },
    {
      "epoch": 0.7383966244725738,
      "grad_norm": 0.4922913610935211,
      "learning_rate": 0.0004687785060483032,
      "loss": 1.5269,
      "step": 7000
    },
    {
      "epoch": 0.7385021097046414,
      "grad_norm": 0.5071092844009399,
      "learning_rate": 0.0004681283230007507,
      "loss": 1.5113,
      "step": 7001
    },
    {
      "epoch": 0.7386075949367089,
      "grad_norm": 0.484287828207016,
      "learning_rate": 0.0004674790417394145,
      "loss": 1.5328,
      "step": 7002
    },
    {
      "epoch": 0.7387130801687763,
      "grad_norm": 0.5072734951972961,
      "learning_rate": 0.00046683066101354197,
      "loss": 1.5261,
      "step": 7003
    },
    {
      "epoch": 0.7388185654008439,
      "grad_norm": 0.4238506555557251,
      "learning_rate": 0.00046618317957411475,
      "loss": 1.4935,
      "step": 7004
    },
    {
      "epoch": 0.7389240506329114,
      "grad_norm": 0.45852428674697876,
      "learning_rate": 0.00046553659617384684,
      "loss": 1.4803,
      "step": 7005
    },
    {
      "epoch": 0.7390295358649789,
      "grad_norm": 0.5891818404197693,
      "learning_rate": 0.00046489090956718234,
      "loss": 1.4994,
      "step": 7006
    },
    {
      "epoch": 0.7391350210970464,
      "grad_norm": 0.5202815532684326,
      "learning_rate": 0.00046424611851029313,
      "loss": 1.5159,
      "step": 7007
    },
    {
      "epoch": 0.739240506329114,
      "grad_norm": 0.547779381275177,
      "learning_rate": 0.00046360222176107584,
      "loss": 1.4958,
      "step": 7008
    },
    {
      "epoch": 0.7393459915611814,
      "grad_norm": 0.4343096613883972,
      "learning_rate": 0.00046295921807915015,
      "loss": 1.5259,
      "step": 7009
    },
    {
      "epoch": 0.739451476793249,
      "grad_norm": 0.5305264592170715,
      "learning_rate": 0.0004623171062258558,
      "loss": 1.5139,
      "step": 7010
    },
    {
      "epoch": 0.7395569620253165,
      "grad_norm": 0.5664477348327637,
      "learning_rate": 0.00046167588496425074,
      "loss": 1.541,
      "step": 7011
    },
    {
      "epoch": 0.739662447257384,
      "grad_norm": 0.4491846263408661,
      "learning_rate": 0.0004610355530591087,
      "loss": 1.5137,
      "step": 7012
    },
    {
      "epoch": 0.7397679324894515,
      "grad_norm": 0.5544329881668091,
      "learning_rate": 0.0004603961092769163,
      "loss": 1.5303,
      "step": 7013
    },
    {
      "epoch": 0.7398734177215189,
      "grad_norm": 0.49536335468292236,
      "learning_rate": 0.0004597575523858713,
      "loss": 1.5008,
      "step": 7014
    },
    {
      "epoch": 0.7399789029535865,
      "grad_norm": 0.631169855594635,
      "learning_rate": 0.0004591198811558795,
      "loss": 1.5377,
      "step": 7015
    },
    {
      "epoch": 0.740084388185654,
      "grad_norm": 0.5485548377037048,
      "learning_rate": 0.0004584830943585531,
      "loss": 1.5411,
      "step": 7016
    },
    {
      "epoch": 0.7401898734177215,
      "grad_norm": 0.5110917091369629,
      "learning_rate": 0.0004578471907672084,
      "loss": 1.514,
      "step": 7017
    },
    {
      "epoch": 0.740295358649789,
      "grad_norm": 0.5429385304450989,
      "learning_rate": 0.0004572121691568625,
      "loss": 1.5476,
      "step": 7018
    },
    {
      "epoch": 0.7404008438818566,
      "grad_norm": 0.4515678584575653,
      "learning_rate": 0.00045657802830423164,
      "loss": 1.4886,
      "step": 7019
    },
    {
      "epoch": 0.740506329113924,
      "grad_norm": 0.5016543865203857,
      "learning_rate": 0.0004559447669877288,
      "loss": 1.54,
      "step": 7020
    },
    {
      "epoch": 0.7406118143459915,
      "grad_norm": 0.4992453455924988,
      "learning_rate": 0.00045531238398746133,
      "loss": 1.5308,
      "step": 7021
    },
    {
      "epoch": 0.7407172995780591,
      "grad_norm": 0.4468490481376648,
      "learning_rate": 0.0004546808780852286,
      "loss": 1.5373,
      "step": 7022
    },
    {
      "epoch": 0.7408227848101265,
      "grad_norm": 0.4730173945426941,
      "learning_rate": 0.0004540502480645194,
      "loss": 1.4811,
      "step": 7023
    },
    {
      "epoch": 0.7409282700421941,
      "grad_norm": 0.5012477040290833,
      "learning_rate": 0.0004534204927105097,
      "loss": 1.5054,
      "step": 7024
    },
    {
      "epoch": 0.7410337552742616,
      "grad_norm": 0.49354612827301025,
      "learning_rate": 0.0004527916108100607,
      "loss": 1.5146,
      "step": 7025
    },
    {
      "epoch": 0.7411392405063291,
      "grad_norm": 0.45262235403060913,
      "learning_rate": 0.00045216360115171613,
      "loss": 1.5229,
      "step": 7026
    },
    {
      "epoch": 0.7412447257383966,
      "grad_norm": 0.5362615585327148,
      "learning_rate": 0.00045153646252569976,
      "loss": 1.5665,
      "step": 7027
    },
    {
      "epoch": 0.7413502109704642,
      "grad_norm": 0.5683720707893372,
      "learning_rate": 0.00045091019372391354,
      "loss": 1.5533,
      "step": 7028
    },
    {
      "epoch": 0.7414556962025316,
      "grad_norm": 0.45193395018577576,
      "learning_rate": 0.00045028479353993473,
      "loss": 1.5484,
      "step": 7029
    },
    {
      "epoch": 0.7415611814345991,
      "grad_norm": 0.5412737131118774,
      "learning_rate": 0.00044966026076901413,
      "loss": 1.5052,
      "step": 7030
    },
    {
      "epoch": 0.7416666666666667,
      "grad_norm": 0.5181034803390503,
      "learning_rate": 0.00044903659420807347,
      "loss": 1.5274,
      "step": 7031
    },
    {
      "epoch": 0.7417721518987341,
      "grad_norm": 0.46654418110847473,
      "learning_rate": 0.000448413792655703,
      "loss": 1.5547,
      "step": 7032
    },
    {
      "epoch": 0.7418776371308017,
      "grad_norm": 0.5146116614341736,
      "learning_rate": 0.0004477918549121593,
      "loss": 1.5585,
      "step": 7033
    },
    {
      "epoch": 0.7419831223628692,
      "grad_norm": 0.5036954283714294,
      "learning_rate": 0.0004471707797793631,
      "loss": 1.5209,
      "step": 7034
    },
    {
      "epoch": 0.7420886075949367,
      "grad_norm": 0.4394099712371826,
      "learning_rate": 0.00044655056606089655,
      "loss": 1.5359,
      "step": 7035
    },
    {
      "epoch": 0.7421940928270042,
      "grad_norm": 0.5671720504760742,
      "learning_rate": 0.00044593121256200163,
      "loss": 1.5198,
      "step": 7036
    },
    {
      "epoch": 0.7422995780590718,
      "grad_norm": 0.5348916053771973,
      "learning_rate": 0.000445312718089577,
      "loss": 1.527,
      "step": 7037
    },
    {
      "epoch": 0.7424050632911392,
      "grad_norm": 0.43203607201576233,
      "learning_rate": 0.0004446950814521764,
      "loss": 1.5401,
      "step": 7038
    },
    {
      "epoch": 0.7425105485232067,
      "grad_norm": 0.5797217488288879,
      "learning_rate": 0.00044407830146000587,
      "loss": 1.5604,
      "step": 7039
    },
    {
      "epoch": 0.7426160337552743,
      "grad_norm": 0.47146809101104736,
      "learning_rate": 0.00044346237692492177,
      "loss": 1.5016,
      "step": 7040
    },
    {
      "epoch": 0.7427215189873417,
      "grad_norm": 0.5115475058555603,
      "learning_rate": 0.0004428473066604284,
      "loss": 1.4883,
      "step": 7041
    },
    {
      "epoch": 0.7428270042194093,
      "grad_norm": 0.48475152254104614,
      "learning_rate": 0.0004422330894816757,
      "loss": 1.5874,
      "step": 7042
    },
    {
      "epoch": 0.7429324894514768,
      "grad_norm": 0.5007703900337219,
      "learning_rate": 0.0004416197242054569,
      "loss": 1.5528,
      "step": 7043
    },
    {
      "epoch": 0.7430379746835443,
      "grad_norm": 0.48052096366882324,
      "learning_rate": 0.0004410072096502064,
      "loss": 1.4997,
      "step": 7044
    },
    {
      "epoch": 0.7431434599156118,
      "grad_norm": 0.4661366045475006,
      "learning_rate": 0.00044039554463599716,
      "loss": 1.5311,
      "step": 7045
    },
    {
      "epoch": 0.7432489451476794,
      "grad_norm": 0.4294412136077881,
      "learning_rate": 0.00043978472798453895,
      "loss": 1.533,
      "step": 7046
    },
    {
      "epoch": 0.7433544303797468,
      "grad_norm": 0.4827269911766052,
      "learning_rate": 0.0004391747585191759,
      "loss": 1.4939,
      "step": 7047
    },
    {
      "epoch": 0.7434599156118143,
      "grad_norm": 0.4447539746761322,
      "learning_rate": 0.0004385656350648835,
      "loss": 1.5213,
      "step": 7048
    },
    {
      "epoch": 0.7435654008438819,
      "grad_norm": 0.46135860681533813,
      "learning_rate": 0.0004379573564482676,
      "loss": 1.5203,
      "step": 7049
    },
    {
      "epoch": 0.7436708860759493,
      "grad_norm": 0.45153218507766724,
      "learning_rate": 0.0004373499214975615,
      "loss": 1.4899,
      "step": 7050
    },
    {
      "epoch": 0.7437763713080169,
      "grad_norm": 0.44256845116615295,
      "learning_rate": 0.0004367433290426232,
      "loss": 1.5204,
      "step": 7051
    },
    {
      "epoch": 0.7438818565400844,
      "grad_norm": 0.45578858256340027,
      "learning_rate": 0.0004361375779149342,
      "loss": 1.4973,
      "step": 7052
    },
    {
      "epoch": 0.7439873417721519,
      "grad_norm": 0.43108853697776794,
      "learning_rate": 0.0004355326669475963,
      "loss": 1.4828,
      "step": 7053
    },
    {
      "epoch": 0.7440928270042194,
      "grad_norm": 0.563687264919281,
      "learning_rate": 0.0004349285949753299,
      "loss": 1.5206,
      "step": 7054
    },
    {
      "epoch": 0.744198312236287,
      "grad_norm": 0.4552002549171448,
      "learning_rate": 0.0004343253608344718,
      "loss": 1.5621,
      "step": 7055
    },
    {
      "epoch": 0.7443037974683544,
      "grad_norm": 0.4878755807876587,
      "learning_rate": 0.0004337229633629726,
      "loss": 1.5592,
      "step": 7056
    },
    {
      "epoch": 0.744409282700422,
      "grad_norm": 0.49062827229499817,
      "learning_rate": 0.0004331214014003945,
      "loss": 1.5106,
      "step": 7057
    },
    {
      "epoch": 0.7445147679324895,
      "grad_norm": 0.4523552358150482,
      "learning_rate": 0.00043252067378790946,
      "loss": 1.5539,
      "step": 7058
    },
    {
      "epoch": 0.7446202531645569,
      "grad_norm": 0.4506322145462036,
      "learning_rate": 0.0004319207793682963,
      "loss": 1.5164,
      "step": 7059
    },
    {
      "epoch": 0.7447257383966245,
      "grad_norm": 0.5151163935661316,
      "learning_rate": 0.0004313217169859396,
      "loss": 1.538,
      "step": 7060
    },
    {
      "epoch": 0.744831223628692,
      "grad_norm": 0.49112361669540405,
      "learning_rate": 0.0004307234854868261,
      "loss": 1.5865,
      "step": 7061
    },
    {
      "epoch": 0.7449367088607595,
      "grad_norm": 0.5302866101264954,
      "learning_rate": 0.00043012608371854324,
      "loss": 1.5167,
      "step": 7062
    },
    {
      "epoch": 0.745042194092827,
      "grad_norm": 0.4698333740234375,
      "learning_rate": 0.00042952951053027696,
      "loss": 1.5089,
      "step": 7063
    },
    {
      "epoch": 0.7451476793248946,
      "grad_norm": 0.5170987844467163,
      "learning_rate": 0.0004289337647728092,
      "loss": 1.5267,
      "step": 7064
    },
    {
      "epoch": 0.745253164556962,
      "grad_norm": 0.5629542469978333,
      "learning_rate": 0.00042833884529851614,
      "loss": 1.5204,
      "step": 7065
    },
    {
      "epoch": 0.7453586497890295,
      "grad_norm": 0.46659767627716064,
      "learning_rate": 0.0004277447509613654,
      "loss": 1.5224,
      "step": 7066
    },
    {
      "epoch": 0.7454641350210971,
      "grad_norm": 0.5514228940010071,
      "learning_rate": 0.00042715148061691407,
      "loss": 1.5248,
      "step": 7067
    },
    {
      "epoch": 0.7455696202531645,
      "grad_norm": 0.5443357229232788,
      "learning_rate": 0.00042655903312230673,
      "loss": 1.5547,
      "step": 7068
    },
    {
      "epoch": 0.7456751054852321,
      "grad_norm": 0.4754493236541748,
      "learning_rate": 0.0004259674073362731,
      "loss": 1.5481,
      "step": 7069
    },
    {
      "epoch": 0.7457805907172996,
      "grad_norm": 0.4532555043697357,
      "learning_rate": 0.0004253766021191256,
      "loss": 1.5337,
      "step": 7070
    },
    {
      "epoch": 0.7458860759493671,
      "grad_norm": 0.52836012840271,
      "learning_rate": 0.0004247866163327576,
      "loss": 1.5501,
      "step": 7071
    },
    {
      "epoch": 0.7459915611814346,
      "grad_norm": 0.4554440975189209,
      "learning_rate": 0.00042419744884064083,
      "loss": 1.543,
      "step": 7072
    },
    {
      "epoch": 0.7460970464135022,
      "grad_norm": 0.5174999833106995,
      "learning_rate": 0.00042360909850782324,
      "loss": 1.5271,
      "step": 7073
    },
    {
      "epoch": 0.7462025316455696,
      "grad_norm": 0.4821670949459076,
      "learning_rate": 0.0004230215642009273,
      "loss": 1.5177,
      "step": 7074
    },
    {
      "epoch": 0.7463080168776371,
      "grad_norm": 0.4322897493839264,
      "learning_rate": 0.0004224348447881472,
      "loss": 1.5072,
      "step": 7075
    },
    {
      "epoch": 0.7464135021097047,
      "grad_norm": 0.4397633969783783,
      "learning_rate": 0.000421848939139247,
      "loss": 1.5055,
      "step": 7076
    },
    {
      "epoch": 0.7465189873417721,
      "grad_norm": 0.46378999948501587,
      "learning_rate": 0.0004212638461255582,
      "loss": 1.4972,
      "step": 7077
    },
    {
      "epoch": 0.7466244725738397,
      "grad_norm": 0.42818915843963623,
      "learning_rate": 0.0004206795646199778,
      "loss": 1.5395,
      "step": 7078
    },
    {
      "epoch": 0.7467299578059071,
      "grad_norm": 0.4345000088214874,
      "learning_rate": 0.00042009609349696626,
      "loss": 1.5503,
      "step": 7079
    },
    {
      "epoch": 0.7468354430379747,
      "grad_norm": 0.46376457810401917,
      "learning_rate": 0.00041951343163254497,
      "loss": 1.5453,
      "step": 7080
    },
    {
      "epoch": 0.7469409282700422,
      "grad_norm": 0.43095022439956665,
      "learning_rate": 0.0004189315779042942,
      "loss": 1.5034,
      "step": 7081
    },
    {
      "epoch": 0.7470464135021097,
      "grad_norm": 0.47321557998657227,
      "learning_rate": 0.00041835053119135095,
      "loss": 1.554,
      "step": 7082
    },
    {
      "epoch": 0.7471518987341772,
      "grad_norm": 0.43503642082214355,
      "learning_rate": 0.00041777029037440695,
      "loss": 1.5445,
      "step": 7083
    },
    {
      "epoch": 0.7472573839662447,
      "grad_norm": 0.5028491020202637,
      "learning_rate": 0.00041719085433570657,
      "loss": 1.5054,
      "step": 7084
    },
    {
      "epoch": 0.7473628691983122,
      "grad_norm": 0.49268436431884766,
      "learning_rate": 0.0004166122219590441,
      "loss": 1.5206,
      "step": 7085
    },
    {
      "epoch": 0.7474683544303797,
      "grad_norm": 0.43791013956069946,
      "learning_rate": 0.00041603439212976217,
      "loss": 1.5056,
      "step": 7086
    },
    {
      "epoch": 0.7475738396624473,
      "grad_norm": 0.5332858562469482,
      "learning_rate": 0.00041545736373474934,
      "loss": 1.5927,
      "step": 7087
    },
    {
      "epoch": 0.7476793248945147,
      "grad_norm": 0.5037378072738647,
      "learning_rate": 0.0004148811356624379,
      "loss": 1.5337,
      "step": 7088
    },
    {
      "epoch": 0.7477848101265823,
      "grad_norm": 0.48495596647262573,
      "learning_rate": 0.00041430570680280233,
      "loss": 1.545,
      "step": 7089
    },
    {
      "epoch": 0.7478902953586498,
      "grad_norm": 0.523064374923706,
      "learning_rate": 0.00041373107604735626,
      "loss": 1.4616,
      "step": 7090
    },
    {
      "epoch": 0.7479957805907173,
      "grad_norm": 0.49792978167533875,
      "learning_rate": 0.00041315724228915075,
      "loss": 1.5022,
      "step": 7091
    },
    {
      "epoch": 0.7481012658227848,
      "grad_norm": 0.4592820703983307,
      "learning_rate": 0.00041258420442277235,
      "loss": 1.4974,
      "step": 7092
    },
    {
      "epoch": 0.7482067510548523,
      "grad_norm": 0.4402250349521637,
      "learning_rate": 0.0004120119613443408,
      "loss": 1.5278,
      "step": 7093
    },
    {
      "epoch": 0.7483122362869198,
      "grad_norm": 0.4806707203388214,
      "learning_rate": 0.00041144051195150685,
      "loss": 1.5402,
      "step": 7094
    },
    {
      "epoch": 0.7484177215189873,
      "grad_norm": 0.5291142463684082,
      "learning_rate": 0.00041086985514345004,
      "loss": 1.5437,
      "step": 7095
    },
    {
      "epoch": 0.7485232067510549,
      "grad_norm": 0.4284331798553467,
      "learning_rate": 0.0004102999898208767,
      "loss": 1.5422,
      "step": 7096
    },
    {
      "epoch": 0.7486286919831223,
      "grad_norm": 0.48729121685028076,
      "learning_rate": 0.00040973091488601815,
      "loss": 1.52,
      "step": 7097
    },
    {
      "epoch": 0.7487341772151899,
      "grad_norm": 0.41935768723487854,
      "learning_rate": 0.0004091626292426282,
      "loss": 1.5477,
      "step": 7098
    },
    {
      "epoch": 0.7488396624472574,
      "grad_norm": 0.4817926287651062,
      "learning_rate": 0.0004085951317959809,
      "loss": 1.5047,
      "step": 7099
    },
    {
      "epoch": 0.7489451476793249,
      "grad_norm": 0.4516688585281372,
      "learning_rate": 0.0004080284214528687,
      "loss": 1.4943,
      "step": 7100
    },
    {
      "epoch": 0.7490506329113924,
      "grad_norm": 0.5066152215003967,
      "learning_rate": 0.00040746249712160065,
      "loss": 1.5581,
      "step": 7101
    },
    {
      "epoch": 0.74915611814346,
      "grad_norm": 0.4841637909412384,
      "learning_rate": 0.0004068973577119993,
      "loss": 1.5554,
      "step": 7102
    },
    {
      "epoch": 0.7492616033755274,
      "grad_norm": 0.4826940894126892,
      "learning_rate": 0.0004063330021354,
      "loss": 1.5217,
      "step": 7103
    },
    {
      "epoch": 0.7493670886075949,
      "grad_norm": 0.48390793800354004,
      "learning_rate": 0.0004057694293046476,
      "loss": 1.4964,
      "step": 7104
    },
    {
      "epoch": 0.7494725738396625,
      "grad_norm": 0.4363178014755249,
      "learning_rate": 0.00040520663813409474,
      "loss": 1.5488,
      "step": 7105
    },
    {
      "epoch": 0.7495780590717299,
      "grad_norm": 0.46231809258461,
      "learning_rate": 0.0004046446275396001,
      "loss": 1.5577,
      "step": 7106
    },
    {
      "epoch": 0.7496835443037975,
      "grad_norm": 0.5574545860290527,
      "learning_rate": 0.00040408339643852574,
      "loss": 1.5326,
      "step": 7107
    },
    {
      "epoch": 0.749789029535865,
      "grad_norm": 0.43836510181427,
      "learning_rate": 0.0004035229437497357,
      "loss": 1.5138,
      "step": 7108
    },
    {
      "epoch": 0.7498945147679325,
      "grad_norm": 0.5714179277420044,
      "learning_rate": 0.00040296326839359315,
      "loss": 1.5134,
      "step": 7109
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5309195518493652,
      "learning_rate": 0.0004024043692919589,
      "loss": 1.5416,
      "step": 7110
    },
    {
      "epoch": 0.7501054852320675,
      "grad_norm": 0.5411048531532288,
      "learning_rate": 0.000401846245368189,
      "loss": 1.5792,
      "step": 7111
    },
    {
      "epoch": 0.750210970464135,
      "grad_norm": 0.5172308087348938,
      "learning_rate": 0.00040128889554713273,
      "loss": 1.5278,
      "step": 7112
    },
    {
      "epoch": 0.7503164556962025,
      "grad_norm": 0.5390828847885132,
      "learning_rate": 0.0004007323187551308,
      "loss": 1.5863,
      "step": 7113
    },
    {
      "epoch": 0.7504219409282701,
      "grad_norm": 0.48591747879981995,
      "learning_rate": 0.0004001765139200129,
      "loss": 1.51,
      "step": 7114
    },
    {
      "epoch": 0.7505274261603375,
      "grad_norm": 0.4886431097984314,
      "learning_rate": 0.00039962147997109587,
      "loss": 1.5196,
      "step": 7115
    },
    {
      "epoch": 0.7506329113924051,
      "grad_norm": 0.4695136249065399,
      "learning_rate": 0.00039906721583918124,
      "loss": 1.4883,
      "step": 7116
    },
    {
      "epoch": 0.7507383966244726,
      "grad_norm": 0.47730475664138794,
      "learning_rate": 0.0003985137204565541,
      "loss": 1.5297,
      "step": 7117
    },
    {
      "epoch": 0.75084388185654,
      "grad_norm": 0.44424834847450256,
      "learning_rate": 0.00039796099275697986,
      "loss": 1.4922,
      "step": 7118
    },
    {
      "epoch": 0.7509493670886076,
      "grad_norm": 0.4683012068271637,
      "learning_rate": 0.000397409031675703,
      "loss": 1.4801,
      "step": 7119
    },
    {
      "epoch": 0.7510548523206751,
      "grad_norm": 0.4301441013813019,
      "learning_rate": 0.0003968578361494449,
      "loss": 1.5608,
      "step": 7120
    },
    {
      "epoch": 0.7511603375527426,
      "grad_norm": 0.47329288721084595,
      "learning_rate": 0.0003963074051164014,
      "loss": 1.5743,
      "step": 7121
    },
    {
      "epoch": 0.7512658227848101,
      "grad_norm": 0.4397728741168976,
      "learning_rate": 0.0003957577375162413,
      "loss": 1.5339,
      "step": 7122
    },
    {
      "epoch": 0.7513713080168777,
      "grad_norm": 0.4293994605541229,
      "learning_rate": 0.0003952088322901039,
      "loss": 1.531,
      "step": 7123
    },
    {
      "epoch": 0.7514767932489451,
      "grad_norm": 0.45915931463241577,
      "learning_rate": 0.0003946606883805972,
      "loss": 1.5436,
      "step": 7124
    },
    {
      "epoch": 0.7515822784810127,
      "grad_norm": 0.48117274045944214,
      "learning_rate": 0.0003941133047317957,
      "loss": 1.5483,
      "step": 7125
    },
    {
      "epoch": 0.7516877637130802,
      "grad_norm": 0.5470882058143616,
      "learning_rate": 0.0003935666802892382,
      "loss": 1.5483,
      "step": 7126
    },
    {
      "epoch": 0.7517932489451477,
      "grad_norm": 0.4597419202327728,
      "learning_rate": 0.00039302081399992676,
      "loss": 1.5455,
      "step": 7127
    },
    {
      "epoch": 0.7518987341772152,
      "grad_norm": 0.5148063898086548,
      "learning_rate": 0.0003924757048123232,
      "loss": 1.5311,
      "step": 7128
    },
    {
      "epoch": 0.7520042194092827,
      "grad_norm": 0.4838639199733734,
      "learning_rate": 0.00039193135167634786,
      "loss": 1.475,
      "step": 7129
    },
    {
      "epoch": 0.7521097046413502,
      "grad_norm": 0.4336302578449249,
      "learning_rate": 0.000391387753543378,
      "loss": 1.526,
      "step": 7130
    },
    {
      "epoch": 0.7522151898734177,
      "grad_norm": 0.4885268211364746,
      "learning_rate": 0.0003908449093662446,
      "loss": 1.5235,
      "step": 7131
    },
    {
      "epoch": 0.7523206751054853,
      "grad_norm": 0.4380110502243042,
      "learning_rate": 0.00039030281809923173,
      "loss": 1.5032,
      "step": 7132
    },
    {
      "epoch": 0.7524261603375527,
      "grad_norm": 0.40179675817489624,
      "learning_rate": 0.00038976147869807345,
      "loss": 1.5034,
      "step": 7133
    },
    {
      "epoch": 0.7525316455696203,
      "grad_norm": 0.5261813402175903,
      "learning_rate": 0.00038922089011995216,
      "loss": 1.4981,
      "step": 7134
    },
    {
      "epoch": 0.7526371308016878,
      "grad_norm": 0.448726624250412,
      "learning_rate": 0.0003886810513234966,
      "loss": 1.4806,
      "step": 7135
    },
    {
      "epoch": 0.7527426160337553,
      "grad_norm": 0.45645344257354736,
      "learning_rate": 0.0003881419612687803,
      "loss": 1.5357,
      "step": 7136
    },
    {
      "epoch": 0.7528481012658228,
      "grad_norm": 0.620891809463501,
      "learning_rate": 0.0003876036189173186,
      "loss": 1.5556,
      "step": 7137
    },
    {
      "epoch": 0.7529535864978903,
      "grad_norm": 0.4438353478908539,
      "learning_rate": 0.0003870660232320675,
      "loss": 1.5235,
      "step": 7138
    },
    {
      "epoch": 0.7530590717299578,
      "grad_norm": 0.4684808552265167,
      "learning_rate": 0.00038652917317742123,
      "loss": 1.5346,
      "step": 7139
    },
    {
      "epoch": 0.7531645569620253,
      "grad_norm": 0.4858201742172241,
      "learning_rate": 0.00038599306771921023,
      "loss": 1.5245,
      "step": 7140
    },
    {
      "epoch": 0.7532700421940929,
      "grad_norm": 0.43313518166542053,
      "learning_rate": 0.00038545770582469976,
      "loss": 1.5265,
      "step": 7141
    },
    {
      "epoch": 0.7533755274261603,
      "grad_norm": 0.49534353613853455,
      "learning_rate": 0.00038492308646258705,
      "loss": 1.5201,
      "step": 7142
    },
    {
      "epoch": 0.7534810126582279,
      "grad_norm": 0.450639545917511,
      "learning_rate": 0.0003843892086029999,
      "loss": 1.5105,
      "step": 7143
    },
    {
      "epoch": 0.7535864978902953,
      "grad_norm": 0.47175243496894836,
      "learning_rate": 0.0003838560712174944,
      "loss": 1.5325,
      "step": 7144
    },
    {
      "epoch": 0.7536919831223629,
      "grad_norm": 0.4821106195449829,
      "learning_rate": 0.00038332367327905293,
      "loss": 1.5106,
      "step": 7145
    },
    {
      "epoch": 0.7537974683544304,
      "grad_norm": 0.49153727293014526,
      "learning_rate": 0.00038279201376208285,
      "loss": 1.5349,
      "step": 7146
    },
    {
      "epoch": 0.7539029535864978,
      "grad_norm": 0.49545401334762573,
      "learning_rate": 0.00038226109164241355,
      "loss": 1.5253,
      "step": 7147
    },
    {
      "epoch": 0.7540084388185654,
      "grad_norm": 0.4788646399974823,
      "learning_rate": 0.000381730905897295,
      "loss": 1.4928,
      "step": 7148
    },
    {
      "epoch": 0.7541139240506329,
      "grad_norm": 0.5294283032417297,
      "learning_rate": 0.0003812014555053956,
      "loss": 1.4732,
      "step": 7149
    },
    {
      "epoch": 0.7542194092827004,
      "grad_norm": 0.5039865374565125,
      "learning_rate": 0.0003806727394468004,
      "loss": 1.5323,
      "step": 7150
    },
    {
      "epoch": 0.7543248945147679,
      "grad_norm": 0.473507821559906,
      "learning_rate": 0.00038014475670300935,
      "loss": 1.5052,
      "step": 7151
    },
    {
      "epoch": 0.7544303797468355,
      "grad_norm": 0.4828464984893799,
      "learning_rate": 0.0003796175062569345,
      "loss": 1.5139,
      "step": 7152
    },
    {
      "epoch": 0.7545358649789029,
      "grad_norm": 0.6287026405334473,
      "learning_rate": 0.0003790909870928989,
      "loss": 1.5082,
      "step": 7153
    },
    {
      "epoch": 0.7546413502109705,
      "grad_norm": 0.4311036765575409,
      "learning_rate": 0.0003785651981966342,
      "loss": 1.489,
      "step": 7154
    },
    {
      "epoch": 0.754746835443038,
      "grad_norm": 0.5604236721992493,
      "learning_rate": 0.00037804013855527886,
      "loss": 1.5698,
      "step": 7155
    },
    {
      "epoch": 0.7548523206751054,
      "grad_norm": 0.5759506821632385,
      "learning_rate": 0.0003775158071573762,
      "loss": 1.5462,
      "step": 7156
    },
    {
      "epoch": 0.754957805907173,
      "grad_norm": 0.48802420496940613,
      "learning_rate": 0.0003769922029928723,
      "loss": 1.4814,
      "step": 7157
    },
    {
      "epoch": 0.7550632911392405,
      "grad_norm": 0.5481204986572266,
      "learning_rate": 0.0003764693250531141,
      "loss": 1.5176,
      "step": 7158
    },
    {
      "epoch": 0.755168776371308,
      "grad_norm": 0.6281586289405823,
      "learning_rate": 0.00037594717233084774,
      "loss": 1.5191,
      "step": 7159
    },
    {
      "epoch": 0.7552742616033755,
      "grad_norm": 0.47377708554267883,
      "learning_rate": 0.0003754257438202162,
      "loss": 1.5617,
      "step": 7160
    },
    {
      "epoch": 0.7553797468354431,
      "grad_norm": 0.5991808176040649,
      "learning_rate": 0.0003749050385167578,
      "loss": 1.5371,
      "step": 7161
    },
    {
      "epoch": 0.7554852320675105,
      "grad_norm": 0.5201156139373779,
      "learning_rate": 0.00037438505541740366,
      "loss": 1.519,
      "step": 7162
    },
    {
      "epoch": 0.755590717299578,
      "grad_norm": 0.520068883895874,
      "learning_rate": 0.0003738657935204763,
      "loss": 1.5107,
      "step": 7163
    },
    {
      "epoch": 0.7556962025316456,
      "grad_norm": 0.6284279823303223,
      "learning_rate": 0.00037334725182568764,
      "loss": 1.5356,
      "step": 7164
    },
    {
      "epoch": 0.755801687763713,
      "grad_norm": 0.45455965399742126,
      "learning_rate": 0.00037282942933413685,
      "loss": 1.5394,
      "step": 7165
    },
    {
      "epoch": 0.7559071729957806,
      "grad_norm": 0.5479934215545654,
      "learning_rate": 0.00037231232504830866,
      "loss": 1.492,
      "step": 7166
    },
    {
      "epoch": 0.7560126582278481,
      "grad_norm": 0.6013789772987366,
      "learning_rate": 0.0003717959379720712,
      "loss": 1.5226,
      "step": 7167
    },
    {
      "epoch": 0.7561181434599156,
      "grad_norm": 0.46028947830200195,
      "learning_rate": 0.0003712802671106742,
      "loss": 1.5332,
      "step": 7168
    },
    {
      "epoch": 0.7562236286919831,
      "grad_norm": 0.6012073755264282,
      "learning_rate": 0.0003707653114707471,
      "loss": 1.5523,
      "step": 7169
    },
    {
      "epoch": 0.7563291139240507,
      "grad_norm": 0.5725463032722473,
      "learning_rate": 0.0003702510700602974,
      "loss": 1.5618,
      "step": 7170
    },
    {
      "epoch": 0.7564345991561181,
      "grad_norm": 0.4723758101463318,
      "learning_rate": 0.00036973754188870803,
      "loss": 1.513,
      "step": 7171
    },
    {
      "epoch": 0.7565400843881857,
      "grad_norm": 0.612431526184082,
      "learning_rate": 0.00036922472596673614,
      "loss": 1.5214,
      "step": 7172
    },
    {
      "epoch": 0.7566455696202532,
      "grad_norm": 0.5117740035057068,
      "learning_rate": 0.0003687126213065109,
      "loss": 1.5223,
      "step": 7173
    },
    {
      "epoch": 0.7567510548523206,
      "grad_norm": 0.6062926054000854,
      "learning_rate": 0.0003682012269215314,
      "loss": 1.4937,
      "step": 7174
    },
    {
      "epoch": 0.7568565400843882,
      "grad_norm": 0.5521675944328308,
      "learning_rate": 0.0003676905418266654,
      "loss": 1.5327,
      "step": 7175
    },
    {
      "epoch": 0.7569620253164557,
      "grad_norm": 0.4586500823497772,
      "learning_rate": 0.00036718056503814674,
      "loss": 1.5036,
      "step": 7176
    },
    {
      "epoch": 0.7570675105485232,
      "grad_norm": 0.5651522278785706,
      "learning_rate": 0.00036667129557357375,
      "loss": 1.525,
      "step": 7177
    },
    {
      "epoch": 0.7571729957805907,
      "grad_norm": 0.5301135182380676,
      "learning_rate": 0.0003661627324519073,
      "loss": 1.4717,
      "step": 7178
    },
    {
      "epoch": 0.7572784810126583,
      "grad_norm": 0.44709864258766174,
      "learning_rate": 0.00036565487469346906,
      "loss": 1.4948,
      "step": 7179
    },
    {
      "epoch": 0.7573839662447257,
      "grad_norm": 0.6086869835853577,
      "learning_rate": 0.0003651477213199393,
      "loss": 1.5189,
      "step": 7180
    },
    {
      "epoch": 0.7574894514767933,
      "grad_norm": 0.4575928747653961,
      "learning_rate": 0.0003646412713543554,
      "loss": 1.4973,
      "step": 7181
    },
    {
      "epoch": 0.7575949367088608,
      "grad_norm": 0.46181899309158325,
      "learning_rate": 0.0003641355238211096,
      "loss": 1.5286,
      "step": 7182
    },
    {
      "epoch": 0.7577004219409282,
      "grad_norm": 0.4870898127555847,
      "learning_rate": 0.0003636304777459472,
      "loss": 1.4782,
      "step": 7183
    },
    {
      "epoch": 0.7578059071729958,
      "grad_norm": 0.46541211009025574,
      "learning_rate": 0.0003631261321559652,
      "loss": 1.53,
      "step": 7184
    },
    {
      "epoch": 0.7579113924050633,
      "grad_norm": 0.5230966210365295,
      "learning_rate": 0.0003626224860796095,
      "loss": 1.5529,
      "step": 7185
    },
    {
      "epoch": 0.7580168776371308,
      "grad_norm": 0.47401633858680725,
      "learning_rate": 0.0003621195385466738,
      "loss": 1.4692,
      "step": 7186
    },
    {
      "epoch": 0.7581223628691983,
      "grad_norm": 0.5433784127235413,
      "learning_rate": 0.0003616172885882972,
      "loss": 1.541,
      "step": 7187
    },
    {
      "epoch": 0.7582278481012659,
      "grad_norm": 0.4681251347064972,
      "learning_rate": 0.0003611157352369628,
      "loss": 1.4892,
      "step": 7188
    },
    {
      "epoch": 0.7583333333333333,
      "grad_norm": 0.5100746750831604,
      "learning_rate": 0.0003606148775264958,
      "loss": 1.5083,
      "step": 7189
    },
    {
      "epoch": 0.7584388185654009,
      "grad_norm": 0.4604494869709015,
      "learning_rate": 0.000360114714492061,
      "loss": 1.4956,
      "step": 7190
    },
    {
      "epoch": 0.7585443037974684,
      "grad_norm": 0.4779258072376251,
      "learning_rate": 0.0003596152451701616,
      "loss": 1.5057,
      "step": 7191
    },
    {
      "epoch": 0.7586497890295358,
      "grad_norm": 0.4830056428909302,
      "learning_rate": 0.00035911646859863725,
      "loss": 1.5108,
      "step": 7192
    },
    {
      "epoch": 0.7587552742616034,
      "grad_norm": 0.4447198510169983,
      "learning_rate": 0.00035861838381666194,
      "loss": 1.5272,
      "step": 7193
    },
    {
      "epoch": 0.7588607594936709,
      "grad_norm": 0.5201935172080994,
      "learning_rate": 0.0003581209898647425,
      "loss": 1.4899,
      "step": 7194
    },
    {
      "epoch": 0.7589662447257384,
      "grad_norm": 0.45016908645629883,
      "learning_rate": 0.0003576242857847163,
      "loss": 1.5496,
      "step": 7195
    },
    {
      "epoch": 0.7590717299578059,
      "grad_norm": 0.5461925268173218,
      "learning_rate": 0.0003571282706197498,
      "loss": 1.5556,
      "step": 7196
    },
    {
      "epoch": 0.7591772151898735,
      "grad_norm": 0.44843021035194397,
      "learning_rate": 0.0003566329434143366,
      "loss": 1.4923,
      "step": 7197
    },
    {
      "epoch": 0.7592827004219409,
      "grad_norm": 0.5161300301551819,
      "learning_rate": 0.00035613830321429534,
      "loss": 1.5262,
      "step": 7198
    },
    {
      "epoch": 0.7593881856540085,
      "grad_norm": 0.4729826748371124,
      "learning_rate": 0.0003556443490667684,
      "loss": 1.5095,
      "step": 7199
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 0.45623841881752014,
      "learning_rate": 0.0003551510800202195,
      "loss": 1.497,
      "step": 7200
    },
    {
      "epoch": 0.7595991561181434,
      "grad_norm": 0.5028412938117981,
      "learning_rate": 0.0003546584951244323,
      "loss": 1.5002,
      "step": 7201
    },
    {
      "epoch": 0.759704641350211,
      "grad_norm": 0.4603593349456787,
      "learning_rate": 0.00035416659343050807,
      "loss": 1.508,
      "step": 7202
    },
    {
      "epoch": 0.7598101265822785,
      "grad_norm": 0.48026788234710693,
      "learning_rate": 0.0003536753739908646,
      "loss": 1.5279,
      "step": 7203
    },
    {
      "epoch": 0.759915611814346,
      "grad_norm": 0.43785303831100464,
      "learning_rate": 0.0003531848358592338,
      "loss": 1.5189,
      "step": 7204
    },
    {
      "epoch": 0.7600210970464135,
      "grad_norm": 0.4649600088596344,
      "learning_rate": 0.00035269497809065976,
      "loss": 1.5537,
      "step": 7205
    },
    {
      "epoch": 0.7601265822784811,
      "grad_norm": 0.4919481873512268,
      "learning_rate": 0.00035220579974149755,
      "loss": 1.5368,
      "step": 7206
    },
    {
      "epoch": 0.7602320675105485,
      "grad_norm": 0.4992193281650543,
      "learning_rate": 0.00035171729986941075,
      "loss": 1.5118,
      "step": 7207
    },
    {
      "epoch": 0.760337552742616,
      "grad_norm": 0.5554006099700928,
      "learning_rate": 0.00035122947753337037,
      "loss": 1.5393,
      "step": 7208
    },
    {
      "epoch": 0.7604430379746835,
      "grad_norm": 0.5136773586273193,
      "learning_rate": 0.0003507423317936521,
      "loss": 1.499,
      "step": 7209
    },
    {
      "epoch": 0.760548523206751,
      "grad_norm": 0.4227517247200012,
      "learning_rate": 0.0003502558617118353,
      "loss": 1.5237,
      "step": 7210
    },
    {
      "epoch": 0.7606540084388186,
      "grad_norm": 0.46549299359321594,
      "learning_rate": 0.0003497700663508009,
      "loss": 1.5517,
      "step": 7211
    },
    {
      "epoch": 0.760759493670886,
      "grad_norm": 0.5477752089500427,
      "learning_rate": 0.0003492849447747293,
      "loss": 1.5116,
      "step": 7212
    },
    {
      "epoch": 0.7608649789029536,
      "grad_norm": 0.5222634077072144,
      "learning_rate": 0.00034880049604909933,
      "loss": 1.5321,
      "step": 7213
    },
    {
      "epoch": 0.7609704641350211,
      "grad_norm": 0.5421080589294434,
      "learning_rate": 0.00034831671924068555,
      "loss": 1.5408,
      "step": 7214
    },
    {
      "epoch": 0.7610759493670886,
      "grad_norm": 0.5208792090415955,
      "learning_rate": 0.00034783361341755707,
      "loss": 1.5309,
      "step": 7215
    },
    {
      "epoch": 0.7611814345991561,
      "grad_norm": 0.49873366951942444,
      "learning_rate": 0.0003473511776490756,
      "loss": 1.5203,
      "step": 7216
    },
    {
      "epoch": 0.7612869198312237,
      "grad_norm": 0.5086596608161926,
      "learning_rate": 0.00034686941100589336,
      "loss": 1.5412,
      "step": 7217
    },
    {
      "epoch": 0.7613924050632911,
      "grad_norm": 0.47228971123695374,
      "learning_rate": 0.0003463883125599521,
      "loss": 1.5427,
      "step": 7218
    },
    {
      "epoch": 0.7614978902953586,
      "grad_norm": 0.4718664586544037,
      "learning_rate": 0.00034590788138448006,
      "loss": 1.5179,
      "step": 7219
    },
    {
      "epoch": 0.7616033755274262,
      "grad_norm": 0.48186349868774414,
      "learning_rate": 0.0003454281165539914,
      "loss": 1.5043,
      "step": 7220
    },
    {
      "epoch": 0.7617088607594936,
      "grad_norm": 0.4828650951385498,
      "learning_rate": 0.00034494901714428365,
      "loss": 1.536,
      "step": 7221
    },
    {
      "epoch": 0.7618143459915612,
      "grad_norm": 0.49547502398490906,
      "learning_rate": 0.0003444705822324364,
      "loss": 1.5191,
      "step": 7222
    },
    {
      "epoch": 0.7619198312236287,
      "grad_norm": 0.45523950457572937,
      "learning_rate": 0.0003439928108968091,
      "loss": 1.5047,
      "step": 7223
    },
    {
      "epoch": 0.7620253164556962,
      "grad_norm": 0.4441123306751251,
      "learning_rate": 0.0003435157022170396,
      "loss": 1.5123,
      "step": 7224
    },
    {
      "epoch": 0.7621308016877637,
      "grad_norm": 0.50005704164505,
      "learning_rate": 0.0003430392552740422,
      "loss": 1.5472,
      "step": 7225
    },
    {
      "epoch": 0.7622362869198313,
      "grad_norm": 0.4155592620372772,
      "learning_rate": 0.0003425634691500059,
      "loss": 1.527,
      "step": 7226
    },
    {
      "epoch": 0.7623417721518987,
      "grad_norm": 0.47661009430885315,
      "learning_rate": 0.0003420883429283929,
      "loss": 1.4991,
      "step": 7227
    },
    {
      "epoch": 0.7624472573839662,
      "grad_norm": 0.4774928689002991,
      "learning_rate": 0.00034161387569393647,
      "loss": 1.5245,
      "step": 7228
    },
    {
      "epoch": 0.7625527426160338,
      "grad_norm": 0.5262956619262695,
      "learning_rate": 0.0003411400665326393,
      "loss": 1.5734,
      "step": 7229
    },
    {
      "epoch": 0.7626582278481012,
      "grad_norm": 0.46142587065696716,
      "learning_rate": 0.00034066691453177176,
      "loss": 1.5297,
      "step": 7230
    },
    {
      "epoch": 0.7627637130801688,
      "grad_norm": 0.49738645553588867,
      "learning_rate": 0.00034019441877987015,
      "loss": 1.5514,
      "step": 7231
    },
    {
      "epoch": 0.7628691983122363,
      "grad_norm": 0.48326608538627625,
      "learning_rate": 0.00033972257836673513,
      "loss": 1.5221,
      "step": 7232
    },
    {
      "epoch": 0.7629746835443038,
      "grad_norm": 0.46682408452033997,
      "learning_rate": 0.00033925139238342956,
      "loss": 1.5107,
      "step": 7233
    },
    {
      "epoch": 0.7630801687763713,
      "grad_norm": 0.559844970703125,
      "learning_rate": 0.0003387808599222771,
      "loss": 1.5261,
      "step": 7234
    },
    {
      "epoch": 0.7631856540084389,
      "grad_norm": 0.586301863193512,
      "learning_rate": 0.0003383109800768603,
      "loss": 1.541,
      "step": 7235
    },
    {
      "epoch": 0.7632911392405063,
      "grad_norm": 0.4854789078235626,
      "learning_rate": 0.0003378417519420187,
      "loss": 1.5032,
      "step": 7236
    },
    {
      "epoch": 0.7633966244725738,
      "grad_norm": 0.5131997466087341,
      "learning_rate": 0.00033737317461384766,
      "loss": 1.556,
      "step": 7237
    },
    {
      "epoch": 0.7635021097046414,
      "grad_norm": 0.5481990575790405,
      "learning_rate": 0.00033690524718969593,
      "loss": 1.5492,
      "step": 7238
    },
    {
      "epoch": 0.7636075949367088,
      "grad_norm": 0.48803794384002686,
      "learning_rate": 0.00033643796876816424,
      "loss": 1.5082,
      "step": 7239
    },
    {
      "epoch": 0.7637130801687764,
      "grad_norm": 0.5214624404907227,
      "learning_rate": 0.0003359713384491037,
      "loss": 1.5122,
      "step": 7240
    },
    {
      "epoch": 0.7638185654008439,
      "grad_norm": 0.5683119893074036,
      "learning_rate": 0.00033550535533361366,
      "loss": 1.4833,
      "step": 7241
    },
    {
      "epoch": 0.7639240506329114,
      "grad_norm": 0.44384464621543884,
      "learning_rate": 0.0003350400185240405,
      "loss": 1.569,
      "step": 7242
    },
    {
      "epoch": 0.7640295358649789,
      "grad_norm": 0.48305878043174744,
      "learning_rate": 0.0003345753271239754,
      "loss": 1.5355,
      "step": 7243
    },
    {
      "epoch": 0.7641350210970465,
      "grad_norm": 0.565956711769104,
      "learning_rate": 0.00033411128023825296,
      "loss": 1.5162,
      "step": 7244
    },
    {
      "epoch": 0.7642405063291139,
      "grad_norm": 0.4637809991836548,
      "learning_rate": 0.0003336478769729492,
      "loss": 1.4923,
      "step": 7245
    },
    {
      "epoch": 0.7643459915611814,
      "grad_norm": 0.4785992205142975,
      "learning_rate": 0.0003331851164353802,
      "loss": 1.4947,
      "step": 7246
    },
    {
      "epoch": 0.764451476793249,
      "grad_norm": 0.4922090470790863,
      "learning_rate": 0.00033272299773410007,
      "loss": 1.5279,
      "step": 7247
    },
    {
      "epoch": 0.7645569620253164,
      "grad_norm": 0.43748775124549866,
      "learning_rate": 0.0003322615199788993,
      "loss": 1.494,
      "step": 7248
    },
    {
      "epoch": 0.764662447257384,
      "grad_norm": 0.46993544697761536,
      "learning_rate": 0.000331800682280803,
      "loss": 1.568,
      "step": 7249
    },
    {
      "epoch": 0.7647679324894515,
      "grad_norm": 0.5029410719871521,
      "learning_rate": 0.00033134048375206944,
      "loss": 1.5134,
      "step": 7250
    },
    {
      "epoch": 0.764873417721519,
      "grad_norm": 0.4553038775920868,
      "learning_rate": 0.0003308809235061881,
      "loss": 1.527,
      "step": 7251
    },
    {
      "epoch": 0.7649789029535865,
      "grad_norm": 0.4251484274864197,
      "learning_rate": 0.000330422000657878,
      "loss": 1.489,
      "step": 7252
    },
    {
      "epoch": 0.765084388185654,
      "grad_norm": 0.49786296486854553,
      "learning_rate": 0.00032996371432308605,
      "loss": 1.5262,
      "step": 7253
    },
    {
      "epoch": 0.7651898734177215,
      "grad_norm": 0.576773464679718,
      "learning_rate": 0.00032950606361898527,
      "loss": 1.5277,
      "step": 7254
    },
    {
      "epoch": 0.765295358649789,
      "grad_norm": 0.42370814085006714,
      "learning_rate": 0.0003290490476639731,
      "loss": 1.5282,
      "step": 7255
    },
    {
      "epoch": 0.7654008438818566,
      "grad_norm": 0.5374912023544312,
      "learning_rate": 0.00032859266557766996,
      "loss": 1.5236,
      "step": 7256
    },
    {
      "epoch": 0.765506329113924,
      "grad_norm": 0.627208948135376,
      "learning_rate": 0.000328136916480917,
      "loss": 1.5342,
      "step": 7257
    },
    {
      "epoch": 0.7656118143459916,
      "grad_norm": 0.4333391487598419,
      "learning_rate": 0.00032768179949577516,
      "loss": 1.5022,
      "step": 7258
    },
    {
      "epoch": 0.7657172995780591,
      "grad_norm": 0.5399424433708191,
      "learning_rate": 0.0003272273137455226,
      "loss": 1.5452,
      "step": 7259
    },
    {
      "epoch": 0.7658227848101266,
      "grad_norm": 0.4923236072063446,
      "learning_rate": 0.0003267734583546536,
      "loss": 1.4685,
      "step": 7260
    },
    {
      "epoch": 0.7659282700421941,
      "grad_norm": 0.4698296785354614,
      "learning_rate": 0.0003263202324488771,
      "loss": 1.4848,
      "step": 7261
    },
    {
      "epoch": 0.7660337552742617,
      "grad_norm": 0.4685630202293396,
      "learning_rate": 0.0003258676351551143,
      "loss": 1.4939,
      "step": 7262
    },
    {
      "epoch": 0.7661392405063291,
      "grad_norm": 0.43930912017822266,
      "learning_rate": 0.0003254156656014973,
      "loss": 1.512,
      "step": 7263
    },
    {
      "epoch": 0.7662447257383966,
      "grad_norm": 0.46170657873153687,
      "learning_rate": 0.0003249643229173677,
      "loss": 1.5429,
      "step": 7264
    },
    {
      "epoch": 0.7663502109704642,
      "grad_norm": 0.5445401072502136,
      "learning_rate": 0.0003245136062332745,
      "loss": 1.4954,
      "step": 7265
    },
    {
      "epoch": 0.7664556962025316,
      "grad_norm": 0.4706728458404541,
      "learning_rate": 0.0003240635146809727,
      "loss": 1.4842,
      "step": 7266
    },
    {
      "epoch": 0.7665611814345992,
      "grad_norm": 0.465275377035141,
      "learning_rate": 0.0003236140473934215,
      "loss": 1.5023,
      "step": 7267
    },
    {
      "epoch": 0.7666666666666667,
      "grad_norm": 0.4905329942703247,
      "learning_rate": 0.0003231652035047826,
      "loss": 1.5378,
      "step": 7268
    },
    {
      "epoch": 0.7667721518987342,
      "grad_norm": 0.433250367641449,
      "learning_rate": 0.00032271698215041863,
      "loss": 1.5278,
      "step": 7269
    },
    {
      "epoch": 0.7668776371308017,
      "grad_norm": 0.4492381811141968,
      "learning_rate": 0.00032226938246689157,
      "loss": 1.5558,
      "step": 7270
    },
    {
      "epoch": 0.7669831223628693,
      "grad_norm": 0.4395883083343506,
      "learning_rate": 0.00032182240359196083,
      "loss": 1.5235,
      "step": 7271
    },
    {
      "epoch": 0.7670886075949367,
      "grad_norm": 0.46264567971229553,
      "learning_rate": 0.0003213760446645818,
      "loss": 1.5227,
      "step": 7272
    },
    {
      "epoch": 0.7671940928270042,
      "grad_norm": 0.5490449070930481,
      "learning_rate": 0.00032093030482490396,
      "loss": 1.5246,
      "step": 7273
    },
    {
      "epoch": 0.7672995780590718,
      "grad_norm": 0.4366523325443268,
      "learning_rate": 0.00032048518321426946,
      "loss": 1.481,
      "step": 7274
    },
    {
      "epoch": 0.7674050632911392,
      "grad_norm": 0.4229549169540405,
      "learning_rate": 0.0003200406789752116,
      "loss": 1.5131,
      "step": 7275
    },
    {
      "epoch": 0.7675105485232068,
      "grad_norm": 0.514065146446228,
      "learning_rate": 0.00031959679125145277,
      "loss": 1.56,
      "step": 7276
    },
    {
      "epoch": 0.7676160337552742,
      "grad_norm": 0.5058407187461853,
      "learning_rate": 0.0003191535191879029,
      "loss": 1.4931,
      "step": 7277
    },
    {
      "epoch": 0.7677215189873418,
      "grad_norm": 0.4359470009803772,
      "learning_rate": 0.000318710861930658,
      "loss": 1.5412,
      "step": 7278
    },
    {
      "epoch": 0.7678270042194093,
      "grad_norm": 0.46236300468444824,
      "learning_rate": 0.0003182688186269984,
      "loss": 1.5219,
      "step": 7279
    },
    {
      "epoch": 0.7679324894514767,
      "grad_norm": 0.4985194206237793,
      "learning_rate": 0.0003178273884253874,
      "loss": 1.4942,
      "step": 7280
    },
    {
      "epoch": 0.7680379746835443,
      "grad_norm": 0.466630220413208,
      "learning_rate": 0.0003173865704754689,
      "loss": 1.5575,
      "step": 7281
    },
    {
      "epoch": 0.7681434599156118,
      "grad_norm": 0.5047515630722046,
      "learning_rate": 0.0003169463639280665,
      "loss": 1.5704,
      "step": 7282
    },
    {
      "epoch": 0.7682489451476793,
      "grad_norm": 0.4358955919742584,
      "learning_rate": 0.00031650676793518157,
      "loss": 1.5231,
      "step": 7283
    },
    {
      "epoch": 0.7683544303797468,
      "grad_norm": 0.4824666678905487,
      "learning_rate": 0.00031606778164999143,
      "loss": 1.5114,
      "step": 7284
    },
    {
      "epoch": 0.7684599156118144,
      "grad_norm": 0.4649924337863922,
      "learning_rate": 0.00031562940422684833,
      "loss": 1.5346,
      "step": 7285
    },
    {
      "epoch": 0.7685654008438818,
      "grad_norm": 0.5150036215782166,
      "learning_rate": 0.00031519163482127696,
      "loss": 1.5197,
      "step": 7286
    },
    {
      "epoch": 0.7686708860759494,
      "grad_norm": 0.49742093682289124,
      "learning_rate": 0.00031475447258997355,
      "loss": 1.5136,
      "step": 7287
    },
    {
      "epoch": 0.7687763713080169,
      "grad_norm": 0.48550671339035034,
      "learning_rate": 0.0003143179166908038,
      "loss": 1.5061,
      "step": 7288
    },
    {
      "epoch": 0.7688818565400843,
      "grad_norm": 0.5942526459693909,
      "learning_rate": 0.0003138819662828017,
      "loss": 1.5384,
      "step": 7289
    },
    {
      "epoch": 0.7689873417721519,
      "grad_norm": 0.463473916053772,
      "learning_rate": 0.0003134466205261674,
      "loss": 1.5046,
      "step": 7290
    },
    {
      "epoch": 0.7690928270042194,
      "grad_norm": 0.4554780125617981,
      "learning_rate": 0.0003130118785822658,
      "loss": 1.5911,
      "step": 7291
    },
    {
      "epoch": 0.7691983122362869,
      "grad_norm": 0.5002447366714478,
      "learning_rate": 0.0003125777396136251,
      "loss": 1.5398,
      "step": 7292
    },
    {
      "epoch": 0.7693037974683544,
      "grad_norm": 0.4926147758960724,
      "learning_rate": 0.00031214420278393487,
      "loss": 1.4932,
      "step": 7293
    },
    {
      "epoch": 0.769409282700422,
      "grad_norm": 0.49334704875946045,
      "learning_rate": 0.00031171126725804496,
      "loss": 1.4974,
      "step": 7294
    },
    {
      "epoch": 0.7695147679324894,
      "grad_norm": 0.545994222164154,
      "learning_rate": 0.0003112789322019633,
      "loss": 1.5533,
      "step": 7295
    },
    {
      "epoch": 0.769620253164557,
      "grad_norm": 0.5108402371406555,
      "learning_rate": 0.0003108471967828545,
      "loss": 1.5345,
      "step": 7296
    },
    {
      "epoch": 0.7697257383966245,
      "grad_norm": 0.4606148898601532,
      "learning_rate": 0.00031041606016903847,
      "loss": 1.517,
      "step": 7297
    },
    {
      "epoch": 0.7698312236286919,
      "grad_norm": 0.5326936841011047,
      "learning_rate": 0.00030998552152998834,
      "loss": 1.5076,
      "step": 7298
    },
    {
      "epoch": 0.7699367088607595,
      "grad_norm": 0.4955883324146271,
      "learning_rate": 0.00030955558003632966,
      "loss": 1.521,
      "step": 7299
    },
    {
      "epoch": 0.770042194092827,
      "grad_norm": 0.4963744878768921,
      "learning_rate": 0.0003091262348598378,
      "loss": 1.5626,
      "step": 7300
    },
    {
      "epoch": 0.7701476793248945,
      "grad_norm": 0.45918890833854675,
      "learning_rate": 0.00030869748517343705,
      "loss": 1.5241,
      "step": 7301
    },
    {
      "epoch": 0.770253164556962,
      "grad_norm": 0.5075101852416992,
      "learning_rate": 0.000308269330151199,
      "loss": 1.474,
      "step": 7302
    },
    {
      "epoch": 0.7703586497890296,
      "grad_norm": 0.48181527853012085,
      "learning_rate": 0.00030784176896834033,
      "loss": 1.5195,
      "step": 7303
    },
    {
      "epoch": 0.770464135021097,
      "grad_norm": 0.5091479420661926,
      "learning_rate": 0.0003074148008012223,
      "loss": 1.5256,
      "step": 7304
    },
    {
      "epoch": 0.7705696202531646,
      "grad_norm": 0.594851553440094,
      "learning_rate": 0.00030698842482734806,
      "loss": 1.4932,
      "step": 7305
    },
    {
      "epoch": 0.7706751054852321,
      "grad_norm": 0.42567554116249084,
      "learning_rate": 0.0003065626402253616,
      "loss": 1.5179,
      "step": 7306
    },
    {
      "epoch": 0.7707805907172995,
      "grad_norm": 0.4837222993373871,
      "learning_rate": 0.00030613744617504624,
      "loss": 1.5344,
      "step": 7307
    },
    {
      "epoch": 0.7708860759493671,
      "grad_norm": 0.6072279810905457,
      "learning_rate": 0.00030571284185732276,
      "loss": 1.5176,
      "step": 7308
    },
    {
      "epoch": 0.7709915611814346,
      "grad_norm": 0.48138466477394104,
      "learning_rate": 0.0003052888264542483,
      "loss": 1.5147,
      "step": 7309
    },
    {
      "epoch": 0.7710970464135021,
      "grad_norm": 0.45973870158195496,
      "learning_rate": 0.0003048653991490141,
      "loss": 1.503,
      "step": 7310
    },
    {
      "epoch": 0.7712025316455696,
      "grad_norm": 0.5439104437828064,
      "learning_rate": 0.0003044425591259445,
      "loss": 1.5084,
      "step": 7311
    },
    {
      "epoch": 0.7713080168776372,
      "grad_norm": 0.48476725816726685,
      "learning_rate": 0.0003040203055704949,
      "loss": 1.5319,
      "step": 7312
    },
    {
      "epoch": 0.7714135021097046,
      "grad_norm": 0.43819084763526917,
      "learning_rate": 0.000303598637669251,
      "loss": 1.5226,
      "step": 7313
    },
    {
      "epoch": 0.7715189873417722,
      "grad_norm": 0.4905666410923004,
      "learning_rate": 0.0003031775546099261,
      "loss": 1.4973,
      "step": 7314
    },
    {
      "epoch": 0.7716244725738397,
      "grad_norm": 0.44105851650238037,
      "learning_rate": 0.0003027570555813604,
      "loss": 1.4925,
      "step": 7315
    },
    {
      "epoch": 0.7717299578059071,
      "grad_norm": 0.5060027837753296,
      "learning_rate": 0.00030233713977351906,
      "loss": 1.5271,
      "step": 7316
    },
    {
      "epoch": 0.7718354430379747,
      "grad_norm": 0.462352454662323,
      "learning_rate": 0.00030191780637749084,
      "loss": 1.479,
      "step": 7317
    },
    {
      "epoch": 0.7719409282700422,
      "grad_norm": 0.4320354163646698,
      "learning_rate": 0.0003014990545854864,
      "loss": 1.4897,
      "step": 7318
    },
    {
      "epoch": 0.7720464135021097,
      "grad_norm": 0.5208417773246765,
      "learning_rate": 0.0003010808835908368,
      "loss": 1.528,
      "step": 7319
    },
    {
      "epoch": 0.7721518987341772,
      "grad_norm": 0.526212215423584,
      "learning_rate": 0.00030066329258799187,
      "loss": 1.5097,
      "step": 7320
    },
    {
      "epoch": 0.7722573839662448,
      "grad_norm": 0.4905802309513092,
      "learning_rate": 0.0003002462807725186,
      "loss": 1.5602,
      "step": 7321
    },
    {
      "epoch": 0.7723628691983122,
      "grad_norm": 0.5224194526672363,
      "learning_rate": 0.00029982984734109995,
      "loss": 1.4966,
      "step": 7322
    },
    {
      "epoch": 0.7724683544303798,
      "grad_norm": 0.41340774297714233,
      "learning_rate": 0.00029941399149153303,
      "loss": 1.4991,
      "step": 7323
    },
    {
      "epoch": 0.7725738396624473,
      "grad_norm": 0.45729508996009827,
      "learning_rate": 0.00029899871242272745,
      "loss": 1.5118,
      "step": 7324
    },
    {
      "epoch": 0.7726793248945147,
      "grad_norm": 0.5007904767990112,
      "learning_rate": 0.000298584009334704,
      "loss": 1.5371,
      "step": 7325
    },
    {
      "epoch": 0.7727848101265823,
      "grad_norm": 0.4312205910682678,
      "learning_rate": 0.00029816988142859286,
      "loss": 1.511,
      "step": 7326
    },
    {
      "epoch": 0.7728902953586498,
      "grad_norm": 0.419544517993927,
      "learning_rate": 0.0002977563279066324,
      "loss": 1.4808,
      "step": 7327
    },
    {
      "epoch": 0.7729957805907173,
      "grad_norm": 0.4282711148262024,
      "learning_rate": 0.0002973433479721675,
      "loss": 1.5094,
      "step": 7328
    },
    {
      "epoch": 0.7731012658227848,
      "grad_norm": 0.48061439394950867,
      "learning_rate": 0.00029693094082964785,
      "loss": 1.4829,
      "step": 7329
    },
    {
      "epoch": 0.7732067510548524,
      "grad_norm": 0.44154366850852966,
      "learning_rate": 0.0002965191056846266,
      "loss": 1.568,
      "step": 7330
    },
    {
      "epoch": 0.7733122362869198,
      "grad_norm": 0.4813012182712555,
      "learning_rate": 0.0002961078417437587,
      "loss": 1.4933,
      "step": 7331
    },
    {
      "epoch": 0.7734177215189874,
      "grad_norm": 0.480756551027298,
      "learning_rate": 0.0002956971482147996,
      "loss": 1.5202,
      "step": 7332
    },
    {
      "epoch": 0.7735232067510549,
      "grad_norm": 0.48635730147361755,
      "learning_rate": 0.0002952870243066035,
      "loss": 1.5246,
      "step": 7333
    },
    {
      "epoch": 0.7736286919831223,
      "grad_norm": 0.4382462799549103,
      "learning_rate": 0.00029487746922912173,
      "loss": 1.4973,
      "step": 7334
    },
    {
      "epoch": 0.7737341772151899,
      "grad_norm": 0.44947320222854614,
      "learning_rate": 0.00029446848219340173,
      "loss": 1.542,
      "step": 7335
    },
    {
      "epoch": 0.7738396624472574,
      "grad_norm": 0.4732278287410736,
      "learning_rate": 0.00029406006241158487,
      "loss": 1.4938,
      "step": 7336
    },
    {
      "epoch": 0.7739451476793249,
      "grad_norm": 0.43985113501548767,
      "learning_rate": 0.0002936522090969055,
      "loss": 1.4811,
      "step": 7337
    },
    {
      "epoch": 0.7740506329113924,
      "grad_norm": 0.4508075416088104,
      "learning_rate": 0.00029324492146368906,
      "loss": 1.5084,
      "step": 7338
    },
    {
      "epoch": 0.77415611814346,
      "grad_norm": 0.45869970321655273,
      "learning_rate": 0.0002928381987273508,
      "loss": 1.5337,
      "step": 7339
    },
    {
      "epoch": 0.7742616033755274,
      "grad_norm": 0.5038483738899231,
      "learning_rate": 0.000292432040104394,
      "loss": 1.505,
      "step": 7340
    },
    {
      "epoch": 0.774367088607595,
      "grad_norm": 0.46021801233291626,
      "learning_rate": 0.00029202644481240867,
      "loss": 1.4973,
      "step": 7341
    },
    {
      "epoch": 0.7744725738396624,
      "grad_norm": 0.4308036267757416,
      "learning_rate": 0.0002916214120700702,
      "loss": 1.5141,
      "step": 7342
    },
    {
      "epoch": 0.7745780590717299,
      "grad_norm": 0.533600926399231,
      "learning_rate": 0.00029121694109713757,
      "loss": 1.4802,
      "step": 7343
    },
    {
      "epoch": 0.7746835443037975,
      "grad_norm": 0.5042786598205566,
      "learning_rate": 0.0002908130311144518,
      "loss": 1.5374,
      "step": 7344
    },
    {
      "epoch": 0.7747890295358649,
      "grad_norm": 0.4555796980857849,
      "learning_rate": 0.0002904096813439346,
      "loss": 1.502,
      "step": 7345
    },
    {
      "epoch": 0.7748945147679325,
      "grad_norm": 0.5745311975479126,
      "learning_rate": 0.00029000689100858694,
      "loss": 1.4566,
      "step": 7346
    },
    {
      "epoch": 0.775,
      "grad_norm": 0.44467997550964355,
      "learning_rate": 0.0002896046593324875,
      "loss": 1.5242,
      "step": 7347
    },
    {
      "epoch": 0.7751054852320675,
      "grad_norm": 0.41884422302246094,
      "learning_rate": 0.00028920298554079113,
      "loss": 1.536,
      "step": 7348
    },
    {
      "epoch": 0.775210970464135,
      "grad_norm": 0.467204749584198,
      "learning_rate": 0.0002888018688597272,
      "loss": 1.5134,
      "step": 7349
    },
    {
      "epoch": 0.7753164556962026,
      "grad_norm": 0.5112577080726624,
      "learning_rate": 0.00028840130851659853,
      "loss": 1.481,
      "step": 7350
    },
    {
      "epoch": 0.77542194092827,
      "grad_norm": 0.4778899848461151,
      "learning_rate": 0.00028800130373977934,
      "loss": 1.4945,
      "step": 7351
    },
    {
      "epoch": 0.7755274261603375,
      "grad_norm": 0.4700300991535187,
      "learning_rate": 0.00028760185375871445,
      "loss": 1.5319,
      "step": 7352
    },
    {
      "epoch": 0.7756329113924051,
      "grad_norm": 0.49284234642982483,
      "learning_rate": 0.0002872029578039172,
      "loss": 1.5118,
      "step": 7353
    },
    {
      "epoch": 0.7757383966244725,
      "grad_norm": 0.4522155523300171,
      "learning_rate": 0.0002868046151069681,
      "loss": 1.4854,
      "step": 7354
    },
    {
      "epoch": 0.7758438818565401,
      "grad_norm": 0.4400084912776947,
      "learning_rate": 0.0002864068249005136,
      "loss": 1.5163,
      "step": 7355
    },
    {
      "epoch": 0.7759493670886076,
      "grad_norm": 0.49960556626319885,
      "learning_rate": 0.0002860095864182644,
      "loss": 1.5449,
      "step": 7356
    },
    {
      "epoch": 0.7760548523206751,
      "grad_norm": 0.46481797099113464,
      "learning_rate": 0.00028561289889499417,
      "loss": 1.519,
      "step": 7357
    },
    {
      "epoch": 0.7761603375527426,
      "grad_norm": 0.45330649614334106,
      "learning_rate": 0.00028521676156653756,
      "loss": 1.5286,
      "step": 7358
    },
    {
      "epoch": 0.7762658227848102,
      "grad_norm": 0.5569223761558533,
      "learning_rate": 0.0002848211736697894,
      "loss": 1.5152,
      "step": 7359
    },
    {
      "epoch": 0.7763713080168776,
      "grad_norm": 0.4300593435764313,
      "learning_rate": 0.0002844261344427028,
      "loss": 1.5043,
      "step": 7360
    },
    {
      "epoch": 0.7764767932489451,
      "grad_norm": 0.4349600076675415,
      "learning_rate": 0.000284031643124288,
      "loss": 1.5451,
      "step": 7361
    },
    {
      "epoch": 0.7765822784810127,
      "grad_norm": 0.5500975251197815,
      "learning_rate": 0.00028363769895461044,
      "loss": 1.546,
      "step": 7362
    },
    {
      "epoch": 0.7766877637130801,
      "grad_norm": 0.5433265566825867,
      "learning_rate": 0.00028324430117478974,
      "loss": 1.4898,
      "step": 7363
    },
    {
      "epoch": 0.7767932489451477,
      "grad_norm": 0.4187507629394531,
      "learning_rate": 0.0002828514490269979,
      "loss": 1.4834,
      "step": 7364
    },
    {
      "epoch": 0.7768987341772152,
      "grad_norm": 0.4652557075023651,
      "learning_rate": 0.0002824591417544582,
      "loss": 1.5001,
      "step": 7365
    },
    {
      "epoch": 0.7770042194092827,
      "grad_norm": 0.46672526001930237,
      "learning_rate": 0.0002820673786014436,
      "loss": 1.5035,
      "step": 7366
    },
    {
      "epoch": 0.7771097046413502,
      "grad_norm": 0.450577974319458,
      "learning_rate": 0.00028167615881327494,
      "loss": 1.5084,
      "step": 7367
    },
    {
      "epoch": 0.7772151898734178,
      "grad_norm": 0.4311329424381256,
      "learning_rate": 0.00028128548163632006,
      "loss": 1.4888,
      "step": 7368
    },
    {
      "epoch": 0.7773206751054852,
      "grad_norm": 0.45180225372314453,
      "learning_rate": 0.00028089534631799183,
      "loss": 1.5481,
      "step": 7369
    },
    {
      "epoch": 0.7774261603375527,
      "grad_norm": 0.4694453775882721,
      "learning_rate": 0.0002805057521067471,
      "loss": 1.5533,
      "step": 7370
    },
    {
      "epoch": 0.7775316455696203,
      "grad_norm": 0.49632954597473145,
      "learning_rate": 0.0002801166982520851,
      "loss": 1.5252,
      "step": 7371
    },
    {
      "epoch": 0.7776371308016877,
      "grad_norm": 0.4226900041103363,
      "learning_rate": 0.000279728184004546,
      "loss": 1.5397,
      "step": 7372
    },
    {
      "epoch": 0.7777426160337553,
      "grad_norm": 0.462501585483551,
      "learning_rate": 0.0002793402086157093,
      "loss": 1.4715,
      "step": 7373
    },
    {
      "epoch": 0.7778481012658228,
      "grad_norm": 0.45349717140197754,
      "learning_rate": 0.0002789527713381925,
      "loss": 1.5208,
      "step": 7374
    },
    {
      "epoch": 0.7779535864978903,
      "grad_norm": 0.4472450613975525,
      "learning_rate": 0.00027856587142565005,
      "loss": 1.5468,
      "step": 7375
    },
    {
      "epoch": 0.7780590717299578,
      "grad_norm": 0.43740612268447876,
      "learning_rate": 0.0002781795081327712,
      "loss": 1.4809,
      "step": 7376
    },
    {
      "epoch": 0.7781645569620254,
      "grad_norm": 0.4535679817199707,
      "learning_rate": 0.0002777936807152791,
      "loss": 1.49,
      "step": 7377
    },
    {
      "epoch": 0.7782700421940928,
      "grad_norm": 0.4833335280418396,
      "learning_rate": 0.0002774083884299292,
      "loss": 1.4678,
      "step": 7378
    },
    {
      "epoch": 0.7783755274261603,
      "grad_norm": 0.41335001587867737,
      "learning_rate": 0.0002770236305345076,
      "loss": 1.5071,
      "step": 7379
    },
    {
      "epoch": 0.7784810126582279,
      "grad_norm": 0.472217321395874,
      "learning_rate": 0.00027663940628783017,
      "loss": 1.5059,
      "step": 7380
    },
    {
      "epoch": 0.7785864978902953,
      "grad_norm": 0.4293840229511261,
      "learning_rate": 0.0002762557149497405,
      "loss": 1.5123,
      "step": 7381
    },
    {
      "epoch": 0.7786919831223629,
      "grad_norm": 0.44424277544021606,
      "learning_rate": 0.00027587255578110894,
      "loss": 1.5928,
      "step": 7382
    },
    {
      "epoch": 0.7787974683544304,
      "grad_norm": 0.4720429480075836,
      "learning_rate": 0.0002754899280438309,
      "loss": 1.5077,
      "step": 7383
    },
    {
      "epoch": 0.7789029535864979,
      "grad_norm": 0.4808584749698639,
      "learning_rate": 0.0002751078310008254,
      "loss": 1.5046,
      "step": 7384
    },
    {
      "epoch": 0.7790084388185654,
      "grad_norm": 0.5304023623466492,
      "learning_rate": 0.0002747262639160341,
      "loss": 1.5095,
      "step": 7385
    },
    {
      "epoch": 0.779113924050633,
      "grad_norm": 0.4539065659046173,
      "learning_rate": 0.0002743452260544193,
      "loss": 1.4633,
      "step": 7386
    },
    {
      "epoch": 0.7792194092827004,
      "grad_norm": 0.4765965938568115,
      "learning_rate": 0.0002739647166819628,
      "loss": 1.5318,
      "step": 7387
    },
    {
      "epoch": 0.7793248945147679,
      "grad_norm": 0.4670567810535431,
      "learning_rate": 0.0002735847350656645,
      "loss": 1.4936,
      "step": 7388
    },
    {
      "epoch": 0.7794303797468355,
      "grad_norm": 0.4970821738243103,
      "learning_rate": 0.00027320528047354093,
      "loss": 1.5027,
      "step": 7389
    },
    {
      "epoch": 0.7795358649789029,
      "grad_norm": 0.45334967970848083,
      "learning_rate": 0.00027282635217462393,
      "loss": 1.5042,
      "step": 7390
    },
    {
      "epoch": 0.7796413502109705,
      "grad_norm": 0.4550299644470215,
      "learning_rate": 0.0002724479494389592,
      "loss": 1.5229,
      "step": 7391
    },
    {
      "epoch": 0.779746835443038,
      "grad_norm": 0.4761160612106323,
      "learning_rate": 0.00027207007153760463,
      "loss": 1.5415,
      "step": 7392
    },
    {
      "epoch": 0.7798523206751055,
      "grad_norm": 0.45894306898117065,
      "learning_rate": 0.0002716927177426294,
      "loss": 1.5126,
      "step": 7393
    },
    {
      "epoch": 0.779957805907173,
      "grad_norm": 0.4720706343650818,
      "learning_rate": 0.0002713158873271122,
      "loss": 1.5146,
      "step": 7394
    },
    {
      "epoch": 0.7800632911392406,
      "grad_norm": 0.4470229744911194,
      "learning_rate": 0.00027093957956513985,
      "loss": 1.501,
      "step": 7395
    },
    {
      "epoch": 0.780168776371308,
      "grad_norm": 0.486998975276947,
      "learning_rate": 0.0002705637937318062,
      "loss": 1.5661,
      "step": 7396
    },
    {
      "epoch": 0.7802742616033755,
      "grad_norm": 0.46125513315200806,
      "learning_rate": 0.00027018852910321045,
      "loss": 1.5188,
      "step": 7397
    },
    {
      "epoch": 0.7803797468354431,
      "grad_norm": 0.4309431314468384,
      "learning_rate": 0.0002698137849564556,
      "loss": 1.4961,
      "step": 7398
    },
    {
      "epoch": 0.7804852320675105,
      "grad_norm": 0.477863073348999,
      "learning_rate": 0.00026943956056964773,
      "loss": 1.5175,
      "step": 7399
    },
    {
      "epoch": 0.7805907172995781,
      "grad_norm": 0.5207541584968567,
      "learning_rate": 0.0002690658552218937,
      "loss": 1.5303,
      "step": 7400
    },
    {
      "epoch": 0.7806962025316456,
      "grad_norm": 0.48380061984062195,
      "learning_rate": 0.0002686926681933006,
      "loss": 1.469,
      "step": 7401
    },
    {
      "epoch": 0.7808016877637131,
      "grad_norm": 0.5058482885360718,
      "learning_rate": 0.00026831999876497376,
      "loss": 1.5277,
      "step": 7402
    },
    {
      "epoch": 0.7809071729957806,
      "grad_norm": 0.4312707483768463,
      "learning_rate": 0.00026794784621901564,
      "loss": 1.4871,
      "step": 7403
    },
    {
      "epoch": 0.7810126582278482,
      "grad_norm": 0.4509865343570709,
      "learning_rate": 0.0002675762098385246,
      "loss": 1.5093,
      "step": 7404
    },
    {
      "epoch": 0.7811181434599156,
      "grad_norm": 0.496855765581131,
      "learning_rate": 0.000267205088907593,
      "loss": 1.4943,
      "step": 7405
    },
    {
      "epoch": 0.7812236286919831,
      "grad_norm": 0.45943617820739746,
      "learning_rate": 0.00026683448271130645,
      "loss": 1.4995,
      "step": 7406
    },
    {
      "epoch": 0.7813291139240506,
      "grad_norm": 0.42863786220550537,
      "learning_rate": 0.0002664643905357418,
      "loss": 1.5013,
      "step": 7407
    },
    {
      "epoch": 0.7814345991561181,
      "grad_norm": 0.4518774151802063,
      "learning_rate": 0.0002660948116679665,
      "loss": 1.5079,
      "step": 7408
    },
    {
      "epoch": 0.7815400843881857,
      "grad_norm": 0.4638252258300781,
      "learning_rate": 0.0002657257453960364,
      "loss": 1.5269,
      "step": 7409
    },
    {
      "epoch": 0.7816455696202531,
      "grad_norm": 0.43134576082229614,
      "learning_rate": 0.00026535719100899516,
      "loss": 1.4836,
      "step": 7410
    },
    {
      "epoch": 0.7817510548523207,
      "grad_norm": 0.4494491219520569,
      "learning_rate": 0.00026498914779687227,
      "loss": 1.5187,
      "step": 7411
    },
    {
      "epoch": 0.7818565400843882,
      "grad_norm": 0.4336400330066681,
      "learning_rate": 0.000264621615050682,
      "loss": 1.5312,
      "step": 7412
    },
    {
      "epoch": 0.7819620253164556,
      "grad_norm": 0.4288300573825836,
      "learning_rate": 0.0002642545920624219,
      "loss": 1.5182,
      "step": 7413
    },
    {
      "epoch": 0.7820675105485232,
      "grad_norm": 0.4808462858200073,
      "learning_rate": 0.0002638880781250718,
      "loss": 1.5173,
      "step": 7414
    },
    {
      "epoch": 0.7821729957805907,
      "grad_norm": 0.4376299977302551,
      "learning_rate": 0.00026352207253259167,
      "loss": 1.4805,
      "step": 7415
    },
    {
      "epoch": 0.7822784810126582,
      "grad_norm": 0.471394807100296,
      "learning_rate": 0.0002631565745799212,
      "loss": 1.4662,
      "step": 7416
    },
    {
      "epoch": 0.7823839662447257,
      "grad_norm": 0.4295235872268677,
      "learning_rate": 0.0002627915835629777,
      "loss": 1.4779,
      "step": 7417
    },
    {
      "epoch": 0.7824894514767933,
      "grad_norm": 0.4546544551849365,
      "learning_rate": 0.00026242709877865493,
      "loss": 1.5356,
      "step": 7418
    },
    {
      "epoch": 0.7825949367088607,
      "grad_norm": 0.46728840470314026,
      "learning_rate": 0.0002620631195248222,
      "loss": 1.5327,
      "step": 7419
    },
    {
      "epoch": 0.7827004219409283,
      "grad_norm": 0.41823697090148926,
      "learning_rate": 0.00026169964510032245,
      "loss": 1.498,
      "step": 7420
    },
    {
      "epoch": 0.7828059071729958,
      "grad_norm": 0.4348681569099426,
      "learning_rate": 0.0002613366748049711,
      "loss": 1.5058,
      "step": 7421
    },
    {
      "epoch": 0.7829113924050632,
      "grad_norm": 0.5143720507621765,
      "learning_rate": 0.0002609742079395546,
      "loss": 1.4959,
      "step": 7422
    },
    {
      "epoch": 0.7830168776371308,
      "grad_norm": 0.4234554171562195,
      "learning_rate": 0.0002606122438058295,
      "loss": 1.472,
      "step": 7423
    },
    {
      "epoch": 0.7831223628691983,
      "grad_norm": 0.4421462118625641,
      "learning_rate": 0.00026025078170652043,
      "loss": 1.523,
      "step": 7424
    },
    {
      "epoch": 0.7832278481012658,
      "grad_norm": 0.4718989431858063,
      "learning_rate": 0.00025988982094531945,
      "loss": 1.5049,
      "step": 7425
    },
    {
      "epoch": 0.7833333333333333,
      "grad_norm": 0.47068876028060913,
      "learning_rate": 0.00025952936082688415,
      "loss": 1.4611,
      "step": 7426
    },
    {
      "epoch": 0.7834388185654009,
      "grad_norm": 0.44434478878974915,
      "learning_rate": 0.00025916940065683655,
      "loss": 1.4884,
      "step": 7427
    },
    {
      "epoch": 0.7835443037974683,
      "grad_norm": 0.4254003167152405,
      "learning_rate": 0.00025880993974176204,
      "loss": 1.4767,
      "step": 7428
    },
    {
      "epoch": 0.7836497890295359,
      "grad_norm": 0.4874972403049469,
      "learning_rate": 0.00025845097738920735,
      "loss": 1.4991,
      "step": 7429
    },
    {
      "epoch": 0.7837552742616034,
      "grad_norm": 0.4271732270717621,
      "learning_rate": 0.00025809251290767984,
      "loss": 1.5084,
      "step": 7430
    },
    {
      "epoch": 0.7838607594936708,
      "grad_norm": 0.43641841411590576,
      "learning_rate": 0.000257734545606646,
      "loss": 1.4919,
      "step": 7431
    },
    {
      "epoch": 0.7839662447257384,
      "grad_norm": 0.44413816928863525,
      "learning_rate": 0.00025737707479652985,
      "loss": 1.4937,
      "step": 7432
    },
    {
      "epoch": 0.7840717299578059,
      "grad_norm": 0.4260719418525696,
      "learning_rate": 0.0002570200997887122,
      "loss": 1.4854,
      "step": 7433
    },
    {
      "epoch": 0.7841772151898734,
      "grad_norm": 0.45341113209724426,
      "learning_rate": 0.0002566636198955286,
      "loss": 1.514,
      "step": 7434
    },
    {
      "epoch": 0.7842827004219409,
      "grad_norm": 0.49296027421951294,
      "learning_rate": 0.0002563076344302685,
      "loss": 1.4864,
      "step": 7435
    },
    {
      "epoch": 0.7843881856540085,
      "grad_norm": 0.4983070194721222,
      "learning_rate": 0.00025595214270717387,
      "loss": 1.4561,
      "step": 7436
    },
    {
      "epoch": 0.7844936708860759,
      "grad_norm": 0.45008277893066406,
      "learning_rate": 0.00025559714404143767,
      "loss": 1.5255,
      "step": 7437
    },
    {
      "epoch": 0.7845991561181435,
      "grad_norm": 0.5047111511230469,
      "learning_rate": 0.0002552426377492028,
      "loss": 1.491,
      "step": 7438
    },
    {
      "epoch": 0.784704641350211,
      "grad_norm": 0.5234826803207397,
      "learning_rate": 0.0002548886231475606,
      "loss": 1.5264,
      "step": 7439
    },
    {
      "epoch": 0.7848101265822784,
      "grad_norm": 0.44678300619125366,
      "learning_rate": 0.00025453509955454957,
      "loss": 1.493,
      "step": 7440
    },
    {
      "epoch": 0.784915611814346,
      "grad_norm": 0.4589504897594452,
      "learning_rate": 0.00025418206628915406,
      "loss": 1.5097,
      "step": 7441
    },
    {
      "epoch": 0.7850210970464135,
      "grad_norm": 0.47607266902923584,
      "learning_rate": 0.00025382952267130306,
      "loss": 1.4914,
      "step": 7442
    },
    {
      "epoch": 0.785126582278481,
      "grad_norm": 0.4706583619117737,
      "learning_rate": 0.0002534774680218686,
      "loss": 1.4746,
      "step": 7443
    },
    {
      "epoch": 0.7852320675105485,
      "grad_norm": 0.4593980014324188,
      "learning_rate": 0.00025312590166266493,
      "loss": 1.4927,
      "step": 7444
    },
    {
      "epoch": 0.7853375527426161,
      "grad_norm": 0.500469446182251,
      "learning_rate": 0.00025277482291644667,
      "loss": 1.5049,
      "step": 7445
    },
    {
      "epoch": 0.7854430379746835,
      "grad_norm": 0.5348123908042908,
      "learning_rate": 0.00025242423110690787,
      "loss": 1.4939,
      "step": 7446
    },
    {
      "epoch": 0.7855485232067511,
      "grad_norm": 0.44165948033332825,
      "learning_rate": 0.0002520741255586806,
      "loss": 1.5236,
      "step": 7447
    },
    {
      "epoch": 0.7856540084388186,
      "grad_norm": 0.507296621799469,
      "learning_rate": 0.0002517245055973337,
      "loss": 1.4897,
      "step": 7448
    },
    {
      "epoch": 0.785759493670886,
      "grad_norm": 0.5057051777839661,
      "learning_rate": 0.0002513753705493713,
      "loss": 1.5185,
      "step": 7449
    },
    {
      "epoch": 0.7858649789029536,
      "grad_norm": 0.45048296451568604,
      "learning_rate": 0.00025102671974223175,
      "loss": 1.5152,
      "step": 7450
    },
    {
      "epoch": 0.7859704641350211,
      "grad_norm": 0.41650232672691345,
      "learning_rate": 0.0002506785525042861,
      "loss": 1.5169,
      "step": 7451
    },
    {
      "epoch": 0.7860759493670886,
      "grad_norm": 0.4856704771518707,
      "learning_rate": 0.0002503308681648371,
      "loss": 1.5135,
      "step": 7452
    },
    {
      "epoch": 0.7861814345991561,
      "grad_norm": 0.47491446137428284,
      "learning_rate": 0.0002499836660541177,
      "loss": 1.4863,
      "step": 7453
    },
    {
      "epoch": 0.7862869198312237,
      "grad_norm": 0.44051873683929443,
      "learning_rate": 0.00024963694550328967,
      "loss": 1.5167,
      "step": 7454
    },
    {
      "epoch": 0.7863924050632911,
      "grad_norm": 0.40391796827316284,
      "learning_rate": 0.0002492907058444425,
      "loss": 1.5129,
      "step": 7455
    },
    {
      "epoch": 0.7864978902953587,
      "grad_norm": 0.4652332067489624,
      "learning_rate": 0.00024894494641059217,
      "loss": 1.5171,
      "step": 7456
    },
    {
      "epoch": 0.7866033755274262,
      "grad_norm": 0.46692216396331787,
      "learning_rate": 0.00024859966653567963,
      "loss": 1.5061,
      "step": 7457
    },
    {
      "epoch": 0.7867088607594936,
      "grad_norm": 0.4094427824020386,
      "learning_rate": 0.00024825486555456975,
      "loss": 1.5116,
      "step": 7458
    },
    {
      "epoch": 0.7868143459915612,
      "grad_norm": 0.44194483757019043,
      "learning_rate": 0.0002479105428030497,
      "loss": 1.5216,
      "step": 7459
    },
    {
      "epoch": 0.7869198312236287,
      "grad_norm": 0.4444366991519928,
      "learning_rate": 0.00024756669761782815,
      "loss": 1.5073,
      "step": 7460
    },
    {
      "epoch": 0.7870253164556962,
      "grad_norm": 0.43794217705726624,
      "learning_rate": 0.00024722332933653344,
      "loss": 1.5121,
      "step": 7461
    },
    {
      "epoch": 0.7871308016877637,
      "grad_norm": 0.4627552330493927,
      "learning_rate": 0.000246880437297713,
      "loss": 1.5001,
      "step": 7462
    },
    {
      "epoch": 0.7872362869198313,
      "grad_norm": 0.4435422122478485,
      "learning_rate": 0.0002465380208408314,
      "loss": 1.5001,
      "step": 7463
    },
    {
      "epoch": 0.7873417721518987,
      "grad_norm": 0.43845805525779724,
      "learning_rate": 0.0002461960793062694,
      "loss": 1.4777,
      "step": 7464
    },
    {
      "epoch": 0.7874472573839663,
      "grad_norm": 0.43793636560440063,
      "learning_rate": 0.00024585461203532254,
      "loss": 1.5037,
      "step": 7465
    },
    {
      "epoch": 0.7875527426160338,
      "grad_norm": 0.4776095747947693,
      "learning_rate": 0.00024551361837020025,
      "loss": 1.5231,
      "step": 7466
    },
    {
      "epoch": 0.7876582278481012,
      "grad_norm": 0.5564210414886475,
      "learning_rate": 0.0002451730976540241,
      "loss": 1.4908,
      "step": 7467
    },
    {
      "epoch": 0.7877637130801688,
      "grad_norm": 0.4627194404602051,
      "learning_rate": 0.00024483304923082663,
      "loss": 1.5115,
      "step": 7468
    },
    {
      "epoch": 0.7878691983122363,
      "grad_norm": 0.5744985342025757,
      "learning_rate": 0.00024449347244555043,
      "loss": 1.476,
      "step": 7469
    },
    {
      "epoch": 0.7879746835443038,
      "grad_norm": 0.5017590522766113,
      "learning_rate": 0.0002441543666440464,
      "loss": 1.4964,
      "step": 7470
    },
    {
      "epoch": 0.7880801687763713,
      "grad_norm": 0.44695737957954407,
      "learning_rate": 0.00024381573117307302,
      "loss": 1.4929,
      "step": 7471
    },
    {
      "epoch": 0.7881856540084389,
      "grad_norm": 0.5167922973632812,
      "learning_rate": 0.00024347756538029453,
      "loss": 1.4742,
      "step": 7472
    },
    {
      "epoch": 0.7882911392405063,
      "grad_norm": 0.516939103603363,
      "learning_rate": 0.00024313986861428,
      "loss": 1.5389,
      "step": 7473
    },
    {
      "epoch": 0.7883966244725739,
      "grad_norm": 0.4884980618953705,
      "learning_rate": 0.00024280264022450215,
      "loss": 1.4988,
      "step": 7474
    },
    {
      "epoch": 0.7885021097046413,
      "grad_norm": 0.4783535599708557,
      "learning_rate": 0.00024246587956133572,
      "loss": 1.4991,
      "step": 7475
    },
    {
      "epoch": 0.7886075949367088,
      "grad_norm": 0.5935856699943542,
      "learning_rate": 0.0002421295859760568,
      "loss": 1.52,
      "step": 7476
    },
    {
      "epoch": 0.7887130801687764,
      "grad_norm": 0.4903963506221771,
      "learning_rate": 0.00024179375882084098,
      "loss": 1.5396,
      "step": 7477
    },
    {
      "epoch": 0.7888185654008438,
      "grad_norm": 0.42041078209877014,
      "learning_rate": 0.0002414583974487624,
      "loss": 1.5074,
      "step": 7478
    },
    {
      "epoch": 0.7889240506329114,
      "grad_norm": 0.46421265602111816,
      "learning_rate": 0.00024112350121379254,
      "loss": 1.465,
      "step": 7479
    },
    {
      "epoch": 0.7890295358649789,
      "grad_norm": 0.5674994587898254,
      "learning_rate": 0.00024078906947079878,
      "loss": 1.5258,
      "step": 7480
    },
    {
      "epoch": 0.7891350210970464,
      "grad_norm": 0.4770188629627228,
      "learning_rate": 0.00024045510157554356,
      "loss": 1.5213,
      "step": 7481
    },
    {
      "epoch": 0.7892405063291139,
      "grad_norm": 0.4871588945388794,
      "learning_rate": 0.00024012159688468254,
      "loss": 1.5423,
      "step": 7482
    },
    {
      "epoch": 0.7893459915611815,
      "grad_norm": 0.4368646442890167,
      "learning_rate": 0.00023978855475576384,
      "loss": 1.5012,
      "step": 7483
    },
    {
      "epoch": 0.7894514767932489,
      "grad_norm": 0.47414425015449524,
      "learning_rate": 0.00023945597454722657,
      "loss": 1.5298,
      "step": 7484
    },
    {
      "epoch": 0.7895569620253164,
      "grad_norm": 0.46037307381629944,
      "learning_rate": 0.00023912385561839983,
      "loss": 1.5251,
      "step": 7485
    },
    {
      "epoch": 0.789662447257384,
      "grad_norm": 0.44129765033721924,
      "learning_rate": 0.00023879219732950114,
      "loss": 1.4795,
      "step": 7486
    },
    {
      "epoch": 0.7897679324894514,
      "grad_norm": 0.4440038204193115,
      "learning_rate": 0.0002384609990416354,
      "loss": 1.5026,
      "step": 7487
    },
    {
      "epoch": 0.789873417721519,
      "grad_norm": 0.4725959002971649,
      "learning_rate": 0.00023813026011679372,
      "loss": 1.5209,
      "step": 7488
    },
    {
      "epoch": 0.7899789029535865,
      "grad_norm": 0.45418548583984375,
      "learning_rate": 0.000237799979917852,
      "loss": 1.5176,
      "step": 7489
    },
    {
      "epoch": 0.790084388185654,
      "grad_norm": 0.476900577545166,
      "learning_rate": 0.00023747015780857005,
      "loss": 1.5084,
      "step": 7490
    },
    {
      "epoch": 0.7901898734177215,
      "grad_norm": 0.44245731830596924,
      "learning_rate": 0.0002371407931535898,
      "loss": 1.4933,
      "step": 7491
    },
    {
      "epoch": 0.7902953586497891,
      "grad_norm": 0.5115064978599548,
      "learning_rate": 0.00023681188531843469,
      "loss": 1.5354,
      "step": 7492
    },
    {
      "epoch": 0.7904008438818565,
      "grad_norm": 0.4417183995246887,
      "learning_rate": 0.00023648343366950792,
      "loss": 1.5229,
      "step": 7493
    },
    {
      "epoch": 0.790506329113924,
      "grad_norm": 0.45005807280540466,
      "learning_rate": 0.0002361554375740916,
      "loss": 1.4984,
      "step": 7494
    },
    {
      "epoch": 0.7906118143459916,
      "grad_norm": 0.43941158056259155,
      "learning_rate": 0.00023582789640034545,
      "loss": 1.5512,
      "step": 7495
    },
    {
      "epoch": 0.790717299578059,
      "grad_norm": 0.42703139781951904,
      "learning_rate": 0.0002355008095173055,
      "loss": 1.4684,
      "step": 7496
    },
    {
      "epoch": 0.7908227848101266,
      "grad_norm": 0.42848125100135803,
      "learning_rate": 0.00023517417629488285,
      "loss": 1.487,
      "step": 7497
    },
    {
      "epoch": 0.7909282700421941,
      "grad_norm": 0.4281219244003296,
      "learning_rate": 0.0002348479961038625,
      "loss": 1.5075,
      "step": 7498
    },
    {
      "epoch": 0.7910337552742616,
      "grad_norm": 0.4721951186656952,
      "learning_rate": 0.00023452226831590227,
      "loss": 1.483,
      "step": 7499
    },
    {
      "epoch": 0.7911392405063291,
      "grad_norm": 0.42719757556915283,
      "learning_rate": 0.00023419699230353144,
      "loss": 1.5469,
      "step": 7500
    },
    {
      "epoch": 0.7912447257383967,
      "grad_norm": 0.5172522068023682,
      "learning_rate": 0.00023387216744014946,
      "loss": 1.5574,
      "step": 7501
    },
    {
      "epoch": 0.7913502109704641,
      "grad_norm": 0.4999002516269684,
      "learning_rate": 0.00023354779310002504,
      "loss": 1.5101,
      "step": 7502
    },
    {
      "epoch": 0.7914556962025316,
      "grad_norm": 0.45073187351226807,
      "learning_rate": 0.00023322386865829456,
      "loss": 1.5085,
      "step": 7503
    },
    {
      "epoch": 0.7915611814345992,
      "grad_norm": 0.46254685521125793,
      "learning_rate": 0.00023290039349096122,
      "loss": 1.5656,
      "step": 7504
    },
    {
      "epoch": 0.7916666666666666,
      "grad_norm": 0.5351743102073669,
      "learning_rate": 0.0002325773669748937,
      "loss": 1.4905,
      "step": 7505
    },
    {
      "epoch": 0.7917721518987342,
      "grad_norm": 0.5738751292228699,
      "learning_rate": 0.00023225478848782483,
      "loss": 1.5303,
      "step": 7506
    },
    {
      "epoch": 0.7918776371308017,
      "grad_norm": 0.4692639708518982,
      "learning_rate": 0.00023193265740835056,
      "loss": 1.5452,
      "step": 7507
    },
    {
      "epoch": 0.7919831223628692,
      "grad_norm": 0.688265323638916,
      "learning_rate": 0.00023161097311592867,
      "loss": 1.4936,
      "step": 7508
    },
    {
      "epoch": 0.7920886075949367,
      "grad_norm": 0.4623848497867584,
      "learning_rate": 0.00023128973499087779,
      "loss": 1.4539,
      "step": 7509
    },
    {
      "epoch": 0.7921940928270043,
      "grad_norm": 0.4545189142227173,
      "learning_rate": 0.00023096894241437586,
      "loss": 1.5028,
      "step": 7510
    },
    {
      "epoch": 0.7922995780590717,
      "grad_norm": 0.5511388778686523,
      "learning_rate": 0.00023064859476845908,
      "loss": 1.4953,
      "step": 7511
    },
    {
      "epoch": 0.7924050632911392,
      "grad_norm": 0.5001651644706726,
      "learning_rate": 0.00023032869143602085,
      "loss": 1.5024,
      "step": 7512
    },
    {
      "epoch": 0.7925105485232068,
      "grad_norm": 0.41897332668304443,
      "learning_rate": 0.00023000923180081047,
      "loss": 1.4915,
      "step": 7513
    },
    {
      "epoch": 0.7926160337552742,
      "grad_norm": 0.5125803351402283,
      "learning_rate": 0.00022969021524743197,
      "loss": 1.5093,
      "step": 7514
    },
    {
      "epoch": 0.7927215189873418,
      "grad_norm": 0.5241802930831909,
      "learning_rate": 0.00022937164116134282,
      "loss": 1.4723,
      "step": 7515
    },
    {
      "epoch": 0.7928270042194093,
      "grad_norm": 0.4525551199913025,
      "learning_rate": 0.00022905350892885298,
      "loss": 1.5446,
      "step": 7516
    },
    {
      "epoch": 0.7929324894514768,
      "grad_norm": 0.5055121183395386,
      "learning_rate": 0.0002287358179371235,
      "loss": 1.4842,
      "step": 7517
    },
    {
      "epoch": 0.7930379746835443,
      "grad_norm": 0.49243614077568054,
      "learning_rate": 0.00022841856757416538,
      "loss": 1.5157,
      "step": 7518
    },
    {
      "epoch": 0.7931434599156119,
      "grad_norm": 0.531338632106781,
      "learning_rate": 0.00022810175722883858,
      "loss": 1.5024,
      "step": 7519
    },
    {
      "epoch": 0.7932489451476793,
      "grad_norm": 0.5198273062705994,
      "learning_rate": 0.00022778538629085056,
      "loss": 1.486,
      "step": 7520
    },
    {
      "epoch": 0.7933544303797468,
      "grad_norm": 0.5028226375579834,
      "learning_rate": 0.0002274694541507553,
      "loss": 1.483,
      "step": 7521
    },
    {
      "epoch": 0.7934599156118144,
      "grad_norm": 0.4740949869155884,
      "learning_rate": 0.00022715396019995203,
      "loss": 1.5081,
      "step": 7522
    },
    {
      "epoch": 0.7935654008438818,
      "grad_norm": 0.4856537878513336,
      "learning_rate": 0.00022683890383068403,
      "loss": 1.4973,
      "step": 7523
    },
    {
      "epoch": 0.7936708860759494,
      "grad_norm": 0.5377978086471558,
      "learning_rate": 0.00022652428443603774,
      "loss": 1.5097,
      "step": 7524
    },
    {
      "epoch": 0.7937763713080169,
      "grad_norm": 0.4413415193557739,
      "learning_rate": 0.00022621010140994125,
      "loss": 1.5301,
      "step": 7525
    },
    {
      "epoch": 0.7938818565400844,
      "grad_norm": 0.4731031358242035,
      "learning_rate": 0.0002258963541471631,
      "loss": 1.4899,
      "step": 7526
    },
    {
      "epoch": 0.7939873417721519,
      "grad_norm": 0.550666093826294,
      "learning_rate": 0.00022558304204331147,
      "loss": 1.5005,
      "step": 7527
    },
    {
      "epoch": 0.7940928270042195,
      "grad_norm": 0.4401571750640869,
      "learning_rate": 0.0002252701644948328,
      "loss": 1.4955,
      "step": 7528
    },
    {
      "epoch": 0.7941983122362869,
      "grad_norm": 0.4887373149394989,
      "learning_rate": 0.0002249577208990106,
      "loss": 1.5072,
      "step": 7529
    },
    {
      "epoch": 0.7943037974683544,
      "grad_norm": 0.458577424287796,
      "learning_rate": 0.00022464571065396427,
      "loss": 1.5069,
      "step": 7530
    },
    {
      "epoch": 0.794409282700422,
      "grad_norm": 0.5035319924354553,
      "learning_rate": 0.0002243341331586481,
      "loss": 1.5273,
      "step": 7531
    },
    {
      "epoch": 0.7945147679324894,
      "grad_norm": 0.4222524166107178,
      "learning_rate": 0.0002240229878128499,
      "loss": 1.4949,
      "step": 7532
    },
    {
      "epoch": 0.794620253164557,
      "grad_norm": 0.46664515137672424,
      "learning_rate": 0.00022371227401719017,
      "loss": 1.523,
      "step": 7533
    },
    {
      "epoch": 0.7947257383966245,
      "grad_norm": 0.4458043575286865,
      "learning_rate": 0.00022340199117312052,
      "loss": 1.5176,
      "step": 7534
    },
    {
      "epoch": 0.794831223628692,
      "grad_norm": 0.44229182600975037,
      "learning_rate": 0.00022309213868292277,
      "loss": 1.4647,
      "step": 7535
    },
    {
      "epoch": 0.7949367088607595,
      "grad_norm": 0.4581355154514313,
      "learning_rate": 0.0002227827159497079,
      "loss": 1.4758,
      "step": 7536
    },
    {
      "epoch": 0.7950421940928271,
      "grad_norm": 0.5061090588569641,
      "learning_rate": 0.0002224737223774145,
      "loss": 1.5041,
      "step": 7537
    },
    {
      "epoch": 0.7951476793248945,
      "grad_norm": 0.43893036246299744,
      "learning_rate": 0.00022216515737080817,
      "loss": 1.5021,
      "step": 7538
    },
    {
      "epoch": 0.795253164556962,
      "grad_norm": 0.4617495536804199,
      "learning_rate": 0.00022185702033547996,
      "loss": 1.4818,
      "step": 7539
    },
    {
      "epoch": 0.7953586497890295,
      "grad_norm": 0.46008166670799255,
      "learning_rate": 0.00022154931067784521,
      "loss": 1.507,
      "step": 7540
    },
    {
      "epoch": 0.795464135021097,
      "grad_norm": 0.4736064672470093,
      "learning_rate": 0.0002212420278051428,
      "loss": 1.4692,
      "step": 7541
    },
    {
      "epoch": 0.7955696202531646,
      "grad_norm": 0.48402008414268494,
      "learning_rate": 0.0002209351711254335,
      "loss": 1.4942,
      "step": 7542
    },
    {
      "epoch": 0.795675105485232,
      "grad_norm": 0.4922538697719574,
      "learning_rate": 0.00022062874004759935,
      "loss": 1.5029,
      "step": 7543
    },
    {
      "epoch": 0.7957805907172996,
      "grad_norm": 0.4429817497730255,
      "learning_rate": 0.00022032273398134208,
      "loss": 1.4971,
      "step": 7544
    },
    {
      "epoch": 0.7958860759493671,
      "grad_norm": 0.5070194602012634,
      "learning_rate": 0.00022001715233718213,
      "loss": 1.5149,
      "step": 7545
    },
    {
      "epoch": 0.7959915611814345,
      "grad_norm": 0.47218137979507446,
      "learning_rate": 0.0002197119945264576,
      "loss": 1.5235,
      "step": 7546
    },
    {
      "epoch": 0.7960970464135021,
      "grad_norm": 0.4475003778934479,
      "learning_rate": 0.00021940725996132303,
      "loss": 1.476,
      "step": 7547
    },
    {
      "epoch": 0.7962025316455696,
      "grad_norm": 0.46315744519233704,
      "learning_rate": 0.00021910294805474833,
      "loss": 1.4808,
      "step": 7548
    },
    {
      "epoch": 0.7963080168776371,
      "grad_norm": 0.4656524360179901,
      "learning_rate": 0.00021879905822051756,
      "loss": 1.5053,
      "step": 7549
    },
    {
      "epoch": 0.7964135021097046,
      "grad_norm": 0.4421415328979492,
      "learning_rate": 0.00021849558987322782,
      "loss": 1.512,
      "step": 7550
    },
    {
      "epoch": 0.7965189873417722,
      "grad_norm": 0.447036474943161,
      "learning_rate": 0.00021819254242828816,
      "loss": 1.4906,
      "step": 7551
    },
    {
      "epoch": 0.7966244725738396,
      "grad_norm": 0.5037019848823547,
      "learning_rate": 0.0002178899153019185,
      "loss": 1.5291,
      "step": 7552
    },
    {
      "epoch": 0.7967299578059072,
      "grad_norm": 0.44761893153190613,
      "learning_rate": 0.00021758770791114845,
      "loss": 1.525,
      "step": 7553
    },
    {
      "epoch": 0.7968354430379747,
      "grad_norm": 0.49607813358306885,
      "learning_rate": 0.00021728591967381606,
      "loss": 1.4785,
      "step": 7554
    },
    {
      "epoch": 0.7969409282700421,
      "grad_norm": 0.5385924577713013,
      "learning_rate": 0.0002169845500085669,
      "loss": 1.436,
      "step": 7555
    },
    {
      "epoch": 0.7970464135021097,
      "grad_norm": 0.4618675708770752,
      "learning_rate": 0.00021668359833485287,
      "loss": 1.5542,
      "step": 7556
    },
    {
      "epoch": 0.7971518987341772,
      "grad_norm": 0.4341605007648468,
      "learning_rate": 0.00021638306407293116,
      "loss": 1.4901,
      "step": 7557
    },
    {
      "epoch": 0.7972573839662447,
      "grad_norm": 0.48466330766677856,
      "learning_rate": 0.0002160829466438629,
      "loss": 1.4762,
      "step": 7558
    },
    {
      "epoch": 0.7973628691983122,
      "grad_norm": 0.4358654022216797,
      "learning_rate": 0.00021578324546951222,
      "loss": 1.4916,
      "step": 7559
    },
    {
      "epoch": 0.7974683544303798,
      "grad_norm": 0.46213844418525696,
      "learning_rate": 0.0002154839599725452,
      "loss": 1.5033,
      "step": 7560
    },
    {
      "epoch": 0.7975738396624472,
      "grad_norm": 0.44458338618278503,
      "learning_rate": 0.0002151850895764285,
      "loss": 1.4937,
      "step": 7561
    },
    {
      "epoch": 0.7976793248945148,
      "grad_norm": 0.48640087246894836,
      "learning_rate": 0.00021488663370542862,
      "loss": 1.5282,
      "step": 7562
    },
    {
      "epoch": 0.7977848101265823,
      "grad_norm": 0.48494136333465576,
      "learning_rate": 0.00021458859178461048,
      "loss": 1.5035,
      "step": 7563
    },
    {
      "epoch": 0.7978902953586497,
      "grad_norm": 0.4744728207588196,
      "learning_rate": 0.00021429096323983645,
      "loss": 1.5076,
      "step": 7564
    },
    {
      "epoch": 0.7979957805907173,
      "grad_norm": 0.511477530002594,
      "learning_rate": 0.00021399374749776512,
      "loss": 1.5255,
      "step": 7565
    },
    {
      "epoch": 0.7981012658227848,
      "grad_norm": 0.4830210506916046,
      "learning_rate": 0.00021369694398585033,
      "loss": 1.4794,
      "step": 7566
    },
    {
      "epoch": 0.7982067510548523,
      "grad_norm": 0.48343178629875183,
      "learning_rate": 0.0002134005521323402,
      "loss": 1.528,
      "step": 7567
    },
    {
      "epoch": 0.7983122362869198,
      "grad_norm": 0.4899161756038666,
      "learning_rate": 0.00021310457136627562,
      "loss": 1.5103,
      "step": 7568
    },
    {
      "epoch": 0.7984177215189874,
      "grad_norm": 0.5383829474449158,
      "learning_rate": 0.00021280900111748948,
      "loss": 1.5332,
      "step": 7569
    },
    {
      "epoch": 0.7985232067510548,
      "grad_norm": 0.4811061918735504,
      "learning_rate": 0.00021251384081660544,
      "loss": 1.5007,
      "step": 7570
    },
    {
      "epoch": 0.7986286919831224,
      "grad_norm": 0.45103538036346436,
      "learning_rate": 0.00021221908989503698,
      "loss": 1.4703,
      "step": 7571
    },
    {
      "epoch": 0.7987341772151899,
      "grad_norm": 0.5614848732948303,
      "learning_rate": 0.00021192474778498606,
      "loss": 1.5151,
      "step": 7572
    },
    {
      "epoch": 0.7988396624472573,
      "grad_norm": 0.433655709028244,
      "learning_rate": 0.00021163081391944227,
      "loss": 1.5056,
      "step": 7573
    },
    {
      "epoch": 0.7989451476793249,
      "grad_norm": 0.4817821979522705,
      "learning_rate": 0.00021133728773218148,
      "loss": 1.524,
      "step": 7574
    },
    {
      "epoch": 0.7990506329113924,
      "grad_norm": 0.5563144087791443,
      "learning_rate": 0.00021104416865776502,
      "loss": 1.4911,
      "step": 7575
    },
    {
      "epoch": 0.7991561181434599,
      "grad_norm": 0.434097558259964,
      "learning_rate": 0.00021075145613153853,
      "loss": 1.5114,
      "step": 7576
    },
    {
      "epoch": 0.7992616033755274,
      "grad_norm": 0.40812230110168457,
      "learning_rate": 0.0002104591495896306,
      "loss": 1.5087,
      "step": 7577
    },
    {
      "epoch": 0.799367088607595,
      "grad_norm": 0.5394563674926758,
      "learning_rate": 0.00021016724846895213,
      "loss": 1.5171,
      "step": 7578
    },
    {
      "epoch": 0.7994725738396624,
      "grad_norm": 0.45321643352508545,
      "learning_rate": 0.00020987575220719483,
      "loss": 1.5285,
      "step": 7579
    },
    {
      "epoch": 0.79957805907173,
      "grad_norm": 0.4194830656051636,
      "learning_rate": 0.0002095846602428303,
      "loss": 1.4965,
      "step": 7580
    },
    {
      "epoch": 0.7996835443037975,
      "grad_norm": 0.5129013061523438,
      "learning_rate": 0.00020929397201510915,
      "loss": 1.482,
      "step": 7581
    },
    {
      "epoch": 0.799789029535865,
      "grad_norm": 0.45665669441223145,
      "learning_rate": 0.00020900368696405964,
      "loss": 1.5112,
      "step": 7582
    },
    {
      "epoch": 0.7998945147679325,
      "grad_norm": 0.4345030188560486,
      "learning_rate": 0.00020871380453048668,
      "loss": 1.5262,
      "step": 7583
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42984485626220703,
      "learning_rate": 0.00020842432415597067,
      "loss": 1.4959,
      "step": 7584
    },
    {
      "epoch": 0.8001054852320675,
      "grad_norm": 0.4645373225212097,
      "learning_rate": 0.0002081352452828667,
      "loss": 1.5096,
      "step": 7585
    },
    {
      "epoch": 0.800210970464135,
      "grad_norm": 0.44757378101348877,
      "learning_rate": 0.0002078465673543032,
      "loss": 1.4863,
      "step": 7586
    },
    {
      "epoch": 0.8003164556962026,
      "grad_norm": 0.4787604510784149,
      "learning_rate": 0.00020755828981418106,
      "loss": 1.4866,
      "step": 7587
    },
    {
      "epoch": 0.80042194092827,
      "grad_norm": 0.43615758419036865,
      "learning_rate": 0.00020727041210717235,
      "loss": 1.4999,
      "step": 7588
    },
    {
      "epoch": 0.8005274261603376,
      "grad_norm": 0.4338921308517456,
      "learning_rate": 0.00020698293367871933,
      "loss": 1.5085,
      "step": 7589
    },
    {
      "epoch": 0.8006329113924051,
      "grad_norm": 0.41479068994522095,
      "learning_rate": 0.00020669585397503358,
      "loss": 1.4724,
      "step": 7590
    },
    {
      "epoch": 0.8007383966244725,
      "grad_norm": 0.46705856919288635,
      "learning_rate": 0.0002064091724430947,
      "loss": 1.5077,
      "step": 7591
    },
    {
      "epoch": 0.8008438818565401,
      "grad_norm": 0.4474402070045471,
      "learning_rate": 0.00020612288853064925,
      "loss": 1.5107,
      "step": 7592
    },
    {
      "epoch": 0.8009493670886076,
      "grad_norm": 0.43865668773651123,
      "learning_rate": 0.00020583700168620985,
      "loss": 1.5285,
      "step": 7593
    },
    {
      "epoch": 0.8010548523206751,
      "grad_norm": 0.45596176385879517,
      "learning_rate": 0.00020555151135905384,
      "loss": 1.4841,
      "step": 7594
    },
    {
      "epoch": 0.8011603375527426,
      "grad_norm": 0.44505131244659424,
      "learning_rate": 0.00020526641699922268,
      "loss": 1.5454,
      "step": 7595
    },
    {
      "epoch": 0.8012658227848102,
      "grad_norm": 0.5033265352249146,
      "learning_rate": 0.00020498171805752037,
      "loss": 1.474,
      "step": 7596
    },
    {
      "epoch": 0.8013713080168776,
      "grad_norm": 0.4670073986053467,
      "learning_rate": 0.00020469741398551278,
      "loss": 1.4506,
      "step": 7597
    },
    {
      "epoch": 0.8014767932489452,
      "grad_norm": 0.49587923288345337,
      "learning_rate": 0.00020441350423552624,
      "loss": 1.4867,
      "step": 7598
    },
    {
      "epoch": 0.8015822784810127,
      "grad_norm": 0.47923460602760315,
      "learning_rate": 0.00020412998826064692,
      "loss": 1.5082,
      "step": 7599
    },
    {
      "epoch": 0.8016877637130801,
      "grad_norm": 0.4406692385673523,
      "learning_rate": 0.0002038468655147195,
      "loss": 1.4801,
      "step": 7600
    },
    {
      "epoch": 0.8017932489451477,
      "grad_norm": 0.4876795709133148,
      "learning_rate": 0.00020356413545234603,
      "loss": 1.5331,
      "step": 7601
    },
    {
      "epoch": 0.8018987341772152,
      "grad_norm": 0.5542328357696533,
      "learning_rate": 0.0002032817975288851,
      "loss": 1.4935,
      "step": 7602
    },
    {
      "epoch": 0.8020042194092827,
      "grad_norm": 0.45491883158683777,
      "learning_rate": 0.00020299985120045069,
      "loss": 1.5498,
      "step": 7603
    },
    {
      "epoch": 0.8021097046413502,
      "grad_norm": 0.4663879871368408,
      "learning_rate": 0.00020271829592391114,
      "loss": 1.5282,
      "step": 7604
    },
    {
      "epoch": 0.8022151898734177,
      "grad_norm": 0.4307539463043213,
      "learning_rate": 0.0002024371311568882,
      "loss": 1.4928,
      "step": 7605
    },
    {
      "epoch": 0.8023206751054852,
      "grad_norm": 0.4613824188709259,
      "learning_rate": 0.0002021563563577556,
      "loss": 1.5147,
      "step": 7606
    },
    {
      "epoch": 0.8024261603375528,
      "grad_norm": 0.45845553278923035,
      "learning_rate": 0.00020187597098563864,
      "loss": 1.4852,
      "step": 7607
    },
    {
      "epoch": 0.8025316455696202,
      "grad_norm": 0.45621466636657715,
      "learning_rate": 0.00020159597450041257,
      "loss": 1.4777,
      "step": 7608
    },
    {
      "epoch": 0.8026371308016877,
      "grad_norm": 0.48014456033706665,
      "learning_rate": 0.00020131636636270178,
      "loss": 1.5213,
      "step": 7609
    },
    {
      "epoch": 0.8027426160337553,
      "grad_norm": 0.5431879758834839,
      "learning_rate": 0.00020103714603387894,
      "loss": 1.5394,
      "step": 7610
    },
    {
      "epoch": 0.8028481012658227,
      "grad_norm": 0.48309117555618286,
      "learning_rate": 0.00020075831297606357,
      "loss": 1.4741,
      "step": 7611
    },
    {
      "epoch": 0.8029535864978903,
      "grad_norm": 0.46558430790901184,
      "learning_rate": 0.00020047986665212137,
      "loss": 1.5177,
      "step": 7612
    },
    {
      "epoch": 0.8030590717299578,
      "grad_norm": 0.4218272268772125,
      "learning_rate": 0.0002002018065256629,
      "loss": 1.4803,
      "step": 7613
    },
    {
      "epoch": 0.8031645569620253,
      "grad_norm": 0.45957159996032715,
      "learning_rate": 0.00019992413206104277,
      "loss": 1.4725,
      "step": 7614
    },
    {
      "epoch": 0.8032700421940928,
      "grad_norm": 0.4866376221179962,
      "learning_rate": 0.00019964684272335854,
      "loss": 1.5096,
      "step": 7615
    },
    {
      "epoch": 0.8033755274261604,
      "grad_norm": 0.420400470495224,
      "learning_rate": 0.0001993699379784496,
      "loss": 1.4884,
      "step": 7616
    },
    {
      "epoch": 0.8034810126582278,
      "grad_norm": 0.4081100821495056,
      "learning_rate": 0.0001990934172928962,
      "loss": 1.4676,
      "step": 7617
    },
    {
      "epoch": 0.8035864978902953,
      "grad_norm": 0.4189686179161072,
      "learning_rate": 0.00019881728013401842,
      "loss": 1.4769,
      "step": 7618
    },
    {
      "epoch": 0.8036919831223629,
      "grad_norm": 0.4664131700992584,
      "learning_rate": 0.00019854152596987523,
      "loss": 1.5003,
      "step": 7619
    },
    {
      "epoch": 0.8037974683544303,
      "grad_norm": 0.4068298935890198,
      "learning_rate": 0.00019826615426926338,
      "loss": 1.4765,
      "step": 7620
    },
    {
      "epoch": 0.8039029535864979,
      "grad_norm": 0.4602712392807007,
      "learning_rate": 0.00019799116450171624,
      "loss": 1.568,
      "step": 7621
    },
    {
      "epoch": 0.8040084388185654,
      "grad_norm": 0.46461352705955505,
      "learning_rate": 0.00019771655613750317,
      "loss": 1.4763,
      "step": 7622
    },
    {
      "epoch": 0.8041139240506329,
      "grad_norm": 0.501839816570282,
      "learning_rate": 0.0001974423286476279,
      "loss": 1.493,
      "step": 7623
    },
    {
      "epoch": 0.8042194092827004,
      "grad_norm": 0.49715879559516907,
      "learning_rate": 0.0001971684815038283,
      "loss": 1.5178,
      "step": 7624
    },
    {
      "epoch": 0.804324894514768,
      "grad_norm": 0.5090494155883789,
      "learning_rate": 0.00019689501417857462,
      "loss": 1.4937,
      "step": 7625
    },
    {
      "epoch": 0.8044303797468354,
      "grad_norm": 0.4455357491970062,
      "learning_rate": 0.00019662192614506883,
      "loss": 1.4912,
      "step": 7626
    },
    {
      "epoch": 0.804535864978903,
      "grad_norm": 0.4066576063632965,
      "learning_rate": 0.00019634921687724358,
      "loss": 1.5072,
      "step": 7627
    },
    {
      "epoch": 0.8046413502109705,
      "grad_norm": 0.43859365582466125,
      "learning_rate": 0.00019607688584976116,
      "loss": 1.5178,
      "step": 7628
    },
    {
      "epoch": 0.8047468354430379,
      "grad_norm": 0.567074716091156,
      "learning_rate": 0.00019580493253801255,
      "loss": 1.5171,
      "step": 7629
    },
    {
      "epoch": 0.8048523206751055,
      "grad_norm": 0.44157594442367554,
      "learning_rate": 0.00019553335641811625,
      "loss": 1.5401,
      "step": 7630
    },
    {
      "epoch": 0.804957805907173,
      "grad_norm": 0.4535594880580902,
      "learning_rate": 0.00019526215696691747,
      "loss": 1.5383,
      "step": 7631
    },
    {
      "epoch": 0.8050632911392405,
      "grad_norm": 0.4694213271141052,
      "learning_rate": 0.00019499133366198686,
      "loss": 1.4817,
      "step": 7632
    },
    {
      "epoch": 0.805168776371308,
      "grad_norm": 0.48461052775382996,
      "learning_rate": 0.00019472088598161984,
      "loss": 1.5058,
      "step": 7633
    },
    {
      "epoch": 0.8052742616033756,
      "grad_norm": 0.43508461117744446,
      "learning_rate": 0.00019445081340483536,
      "loss": 1.5178,
      "step": 7634
    },
    {
      "epoch": 0.805379746835443,
      "grad_norm": 0.46978190541267395,
      "learning_rate": 0.0001941811154113749,
      "loss": 1.5306,
      "step": 7635
    },
    {
      "epoch": 0.8054852320675105,
      "grad_norm": 0.4854903519153595,
      "learning_rate": 0.0001939117914817016,
      "loss": 1.4822,
      "step": 7636
    },
    {
      "epoch": 0.8055907172995781,
      "grad_norm": 0.4828737676143646,
      "learning_rate": 0.0001936428410969991,
      "loss": 1.4794,
      "step": 7637
    },
    {
      "epoch": 0.8056962025316455,
      "grad_norm": 0.44565919041633606,
      "learning_rate": 0.00019337426373917076,
      "loss": 1.4901,
      "step": 7638
    },
    {
      "epoch": 0.8058016877637131,
      "grad_norm": 0.45560306310653687,
      "learning_rate": 0.00019310605889083838,
      "loss": 1.5186,
      "step": 7639
    },
    {
      "epoch": 0.8059071729957806,
      "grad_norm": 0.4514120817184448,
      "learning_rate": 0.0001928382260353415,
      "loss": 1.477,
      "step": 7640
    },
    {
      "epoch": 0.8060126582278481,
      "grad_norm": 0.4615853726863861,
      "learning_rate": 0.00019257076465673605,
      "loss": 1.5186,
      "step": 7641
    },
    {
      "epoch": 0.8061181434599156,
      "grad_norm": 0.44695156812667847,
      "learning_rate": 0.00019230367423979372,
      "loss": 1.5419,
      "step": 7642
    },
    {
      "epoch": 0.8062236286919832,
      "grad_norm": 0.4508327841758728,
      "learning_rate": 0.0001920369542700008,
      "loss": 1.4762,
      "step": 7643
    },
    {
      "epoch": 0.8063291139240506,
      "grad_norm": 0.4132992625236511,
      "learning_rate": 0.00019177060423355714,
      "loss": 1.5286,
      "step": 7644
    },
    {
      "epoch": 0.8064345991561181,
      "grad_norm": 0.46180111169815063,
      "learning_rate": 0.00019150462361737527,
      "loss": 1.5159,
      "step": 7645
    },
    {
      "epoch": 0.8065400843881857,
      "grad_norm": 0.42987874150276184,
      "learning_rate": 0.00019123901190907928,
      "loss": 1.5419,
      "step": 7646
    },
    {
      "epoch": 0.8066455696202531,
      "grad_norm": 0.47122010588645935,
      "learning_rate": 0.00019097376859700393,
      "loss": 1.5015,
      "step": 7647
    },
    {
      "epoch": 0.8067510548523207,
      "grad_norm": 0.437377005815506,
      "learning_rate": 0.00019070889317019375,
      "loss": 1.4865,
      "step": 7648
    },
    {
      "epoch": 0.8068565400843882,
      "grad_norm": 0.4346117079257965,
      "learning_rate": 0.0001904443851184018,
      "loss": 1.5061,
      "step": 7649
    },
    {
      "epoch": 0.8069620253164557,
      "grad_norm": 0.45330169796943665,
      "learning_rate": 0.00019018024393208902,
      "loss": 1.5374,
      "step": 7650
    },
    {
      "epoch": 0.8070675105485232,
      "grad_norm": 0.4814563989639282,
      "learning_rate": 0.00018991646910242288,
      "loss": 1.5025,
      "step": 7651
    },
    {
      "epoch": 0.8071729957805908,
      "grad_norm": 0.4183913469314575,
      "learning_rate": 0.00018965306012127663,
      "loss": 1.4314,
      "step": 7652
    },
    {
      "epoch": 0.8072784810126582,
      "grad_norm": 0.4228047728538513,
      "learning_rate": 0.00018939001648122844,
      "loss": 1.5049,
      "step": 7653
    },
    {
      "epoch": 0.8073839662447257,
      "grad_norm": 0.49773094058036804,
      "learning_rate": 0.00018912733767556005,
      "loss": 1.5389,
      "step": 7654
    },
    {
      "epoch": 0.8074894514767933,
      "grad_norm": 0.4487854540348053,
      "learning_rate": 0.00018886502319825612,
      "loss": 1.4722,
      "step": 7655
    },
    {
      "epoch": 0.8075949367088607,
      "grad_norm": 0.48282310366630554,
      "learning_rate": 0.00018860307254400305,
      "loss": 1.4709,
      "step": 7656
    },
    {
      "epoch": 0.8077004219409283,
      "grad_norm": 0.4292629659175873,
      "learning_rate": 0.0001883414852081882,
      "loss": 1.5248,
      "step": 7657
    },
    {
      "epoch": 0.8078059071729958,
      "grad_norm": 0.46244215965270996,
      "learning_rate": 0.00018808026068689883,
      "loss": 1.5144,
      "step": 7658
    },
    {
      "epoch": 0.8079113924050633,
      "grad_norm": 0.4608635902404785,
      "learning_rate": 0.00018781939847692096,
      "loss": 1.4986,
      "step": 7659
    },
    {
      "epoch": 0.8080168776371308,
      "grad_norm": 0.4507381021976471,
      "learning_rate": 0.00018755889807573872,
      "loss": 1.4778,
      "step": 7660
    },
    {
      "epoch": 0.8081223628691984,
      "grad_norm": 0.4129912555217743,
      "learning_rate": 0.00018729875898153305,
      "loss": 1.4702,
      "step": 7661
    },
    {
      "epoch": 0.8082278481012658,
      "grad_norm": 0.4801464080810547,
      "learning_rate": 0.00018703898069318112,
      "loss": 1.5099,
      "step": 7662
    },
    {
      "epoch": 0.8083333333333333,
      "grad_norm": 0.4517633616924286,
      "learning_rate": 0.00018677956271025492,
      "loss": 1.5218,
      "step": 7663
    },
    {
      "epoch": 0.8084388185654009,
      "grad_norm": 0.49452513456344604,
      "learning_rate": 0.00018652050453302066,
      "loss": 1.4985,
      "step": 7664
    },
    {
      "epoch": 0.8085443037974683,
      "grad_norm": 0.4616701602935791,
      "learning_rate": 0.0001862618056624376,
      "loss": 1.5108,
      "step": 7665
    },
    {
      "epoch": 0.8086497890295359,
      "grad_norm": 0.4375028908252716,
      "learning_rate": 0.00018600346560015716,
      "loss": 1.455,
      "step": 7666
    },
    {
      "epoch": 0.8087552742616034,
      "grad_norm": 0.4428229033946991,
      "learning_rate": 0.00018574548384852206,
      "loss": 1.4951,
      "step": 7667
    },
    {
      "epoch": 0.8088607594936709,
      "grad_norm": 0.40699702501296997,
      "learning_rate": 0.00018548785991056508,
      "loss": 1.5383,
      "step": 7668
    },
    {
      "epoch": 0.8089662447257384,
      "grad_norm": 0.458650678396225,
      "learning_rate": 0.00018523059329000844,
      "loss": 1.4976,
      "step": 7669
    },
    {
      "epoch": 0.8090717299578059,
      "grad_norm": 0.43317022919654846,
      "learning_rate": 0.00018497368349126262,
      "loss": 1.5115,
      "step": 7670
    },
    {
      "epoch": 0.8091772151898734,
      "grad_norm": 0.4401693046092987,
      "learning_rate": 0.00018471713001942533,
      "loss": 1.51,
      "step": 7671
    },
    {
      "epoch": 0.809282700421941,
      "grad_norm": 0.42360663414001465,
      "learning_rate": 0.000184460932380281,
      "loss": 1.5366,
      "step": 7672
    },
    {
      "epoch": 0.8093881856540084,
      "grad_norm": 0.41593116521835327,
      "learning_rate": 0.00018420509008029931,
      "loss": 1.5379,
      "step": 7673
    },
    {
      "epoch": 0.8094936708860759,
      "grad_norm": 0.445113867521286,
      "learning_rate": 0.00018394960262663448,
      "loss": 1.5271,
      "step": 7674
    },
    {
      "epoch": 0.8095991561181435,
      "grad_norm": 0.45435038208961487,
      "learning_rate": 0.0001836944695271243,
      "loss": 1.472,
      "step": 7675
    },
    {
      "epoch": 0.8097046413502109,
      "grad_norm": 0.4509572386741638,
      "learning_rate": 0.00018343969029028915,
      "loss": 1.5007,
      "step": 7676
    },
    {
      "epoch": 0.8098101265822785,
      "grad_norm": 0.4851086139678955,
      "learning_rate": 0.00018318526442533123,
      "loss": 1.491,
      "step": 7677
    },
    {
      "epoch": 0.809915611814346,
      "grad_norm": 0.4428175389766693,
      "learning_rate": 0.00018293119144213328,
      "loss": 1.5429,
      "step": 7678
    },
    {
      "epoch": 0.8100210970464135,
      "grad_norm": 0.4549797475337982,
      "learning_rate": 0.0001826774708512579,
      "loss": 1.4834,
      "step": 7679
    },
    {
      "epoch": 0.810126582278481,
      "grad_norm": 0.4545847475528717,
      "learning_rate": 0.00018242410216394648,
      "loss": 1.5446,
      "step": 7680
    },
    {
      "epoch": 0.8102320675105485,
      "grad_norm": 0.40937238931655884,
      "learning_rate": 0.00018217108489211841,
      "loss": 1.5011,
      "step": 7681
    },
    {
      "epoch": 0.810337552742616,
      "grad_norm": 0.4805527329444885,
      "learning_rate": 0.00018191841854836994,
      "loss": 1.4952,
      "step": 7682
    },
    {
      "epoch": 0.8104430379746835,
      "grad_norm": 0.41970404982566833,
      "learning_rate": 0.00018166610264597332,
      "loss": 1.4804,
      "step": 7683
    },
    {
      "epoch": 0.8105485232067511,
      "grad_norm": 0.4368318021297455,
      "learning_rate": 0.00018141413669887598,
      "loss": 1.4863,
      "step": 7684
    },
    {
      "epoch": 0.8106540084388185,
      "grad_norm": 0.45059943199157715,
      "learning_rate": 0.00018116252022169936,
      "loss": 1.5212,
      "step": 7685
    },
    {
      "epoch": 0.8107594936708861,
      "grad_norm": 0.4365903437137604,
      "learning_rate": 0.00018091125272973825,
      "loss": 1.5077,
      "step": 7686
    },
    {
      "epoch": 0.8108649789029536,
      "grad_norm": 0.43383705615997314,
      "learning_rate": 0.0001806603337389596,
      "loss": 1.502,
      "step": 7687
    },
    {
      "epoch": 0.810970464135021,
      "grad_norm": 0.4276760518550873,
      "learning_rate": 0.00018040976276600176,
      "loss": 1.4718,
      "step": 7688
    },
    {
      "epoch": 0.8110759493670886,
      "grad_norm": 0.46179482340812683,
      "learning_rate": 0.00018015953932817348,
      "loss": 1.4835,
      "step": 7689
    },
    {
      "epoch": 0.8111814345991561,
      "grad_norm": 0.44500598311424255,
      "learning_rate": 0.0001799096629434529,
      "loss": 1.4523,
      "step": 7690
    },
    {
      "epoch": 0.8112869198312236,
      "grad_norm": 0.4377223253250122,
      "learning_rate": 0.00017966013313048688,
      "loss": 1.51,
      "step": 7691
    },
    {
      "epoch": 0.8113924050632911,
      "grad_norm": 0.44538792967796326,
      "learning_rate": 0.00017941094940858982,
      "loss": 1.5185,
      "step": 7692
    },
    {
      "epoch": 0.8114978902953587,
      "grad_norm": 0.4273275136947632,
      "learning_rate": 0.00017916211129774277,
      "loss": 1.4906,
      "step": 7693
    },
    {
      "epoch": 0.8116033755274261,
      "grad_norm": 0.4187488853931427,
      "learning_rate": 0.00017891361831859262,
      "loss": 1.4839,
      "step": 7694
    },
    {
      "epoch": 0.8117088607594937,
      "grad_norm": 0.4449957609176636,
      "learning_rate": 0.00017866546999245102,
      "loss": 1.4872,
      "step": 7695
    },
    {
      "epoch": 0.8118143459915612,
      "grad_norm": 0.4378944933414459,
      "learning_rate": 0.00017841766584129372,
      "loss": 1.5277,
      "step": 7696
    },
    {
      "epoch": 0.8119198312236287,
      "grad_norm": 0.46003258228302,
      "learning_rate": 0.00017817020538775933,
      "loss": 1.4837,
      "step": 7697
    },
    {
      "epoch": 0.8120253164556962,
      "grad_norm": 0.44949719309806824,
      "learning_rate": 0.00017792308815514854,
      "loss": 1.4938,
      "step": 7698
    },
    {
      "epoch": 0.8121308016877637,
      "grad_norm": 0.4335648715496063,
      "learning_rate": 0.00017767631366742332,
      "loss": 1.5005,
      "step": 7699
    },
    {
      "epoch": 0.8122362869198312,
      "grad_norm": 0.4382060766220093,
      "learning_rate": 0.00017742988144920578,
      "loss": 1.5221,
      "step": 7700
    },
    {
      "epoch": 0.8123417721518987,
      "grad_norm": 0.48175734281539917,
      "learning_rate": 0.00017718379102577746,
      "loss": 1.4764,
      "step": 7701
    },
    {
      "epoch": 0.8124472573839663,
      "grad_norm": 0.4919948875904083,
      "learning_rate": 0.00017693804192307826,
      "loss": 1.5199,
      "step": 7702
    },
    {
      "epoch": 0.8125527426160337,
      "grad_norm": 0.4192477762699127,
      "learning_rate": 0.0001766926336677056,
      "loss": 1.4669,
      "step": 7703
    },
    {
      "epoch": 0.8126582278481013,
      "grad_norm": 0.44176772236824036,
      "learning_rate": 0.00017644756578691345,
      "loss": 1.4892,
      "step": 7704
    },
    {
      "epoch": 0.8127637130801688,
      "grad_norm": 0.4153943359851837,
      "learning_rate": 0.0001762028378086116,
      "loss": 1.4875,
      "step": 7705
    },
    {
      "epoch": 0.8128691983122363,
      "grad_norm": 0.4569021165370941,
      "learning_rate": 0.00017595844926136456,
      "loss": 1.4811,
      "step": 7706
    },
    {
      "epoch": 0.8129746835443038,
      "grad_norm": 0.4806217551231384,
      "learning_rate": 0.0001757143996743906,
      "loss": 1.5111,
      "step": 7707
    },
    {
      "epoch": 0.8130801687763713,
      "grad_norm": 0.4286656379699707,
      "learning_rate": 0.00017547068857756104,
      "loss": 1.4941,
      "step": 7708
    },
    {
      "epoch": 0.8131856540084388,
      "grad_norm": 0.4367987811565399,
      "learning_rate": 0.00017522731550139922,
      "loss": 1.5301,
      "step": 7709
    },
    {
      "epoch": 0.8132911392405063,
      "grad_norm": 0.546692430973053,
      "learning_rate": 0.00017498427997707976,
      "loss": 1.4884,
      "step": 7710
    },
    {
      "epoch": 0.8133966244725739,
      "grad_norm": 0.4957149028778076,
      "learning_rate": 0.0001747415815364274,
      "loss": 1.5226,
      "step": 7711
    },
    {
      "epoch": 0.8135021097046413,
      "grad_norm": 0.4620266854763031,
      "learning_rate": 0.00017449921971191622,
      "loss": 1.4886,
      "step": 7712
    },
    {
      "epoch": 0.8136075949367089,
      "grad_norm": 0.4796673655509949,
      "learning_rate": 0.00017425719403666877,
      "loss": 1.4918,
      "step": 7713
    },
    {
      "epoch": 0.8137130801687764,
      "grad_norm": 0.5266776084899902,
      "learning_rate": 0.00017401550404445515,
      "loss": 1.5231,
      "step": 7714
    },
    {
      "epoch": 0.8138185654008439,
      "grad_norm": 0.4618789851665497,
      "learning_rate": 0.0001737741492696922,
      "loss": 1.4943,
      "step": 7715
    },
    {
      "epoch": 0.8139240506329114,
      "grad_norm": 0.4553345739841461,
      "learning_rate": 0.00017353312924744236,
      "loss": 1.5164,
      "step": 7716
    },
    {
      "epoch": 0.814029535864979,
      "grad_norm": 0.4804789423942566,
      "learning_rate": 0.000173292443513413,
      "loss": 1.5474,
      "step": 7717
    },
    {
      "epoch": 0.8141350210970464,
      "grad_norm": 0.4306790828704834,
      "learning_rate": 0.00017305209160395547,
      "loss": 1.4529,
      "step": 7718
    },
    {
      "epoch": 0.8142405063291139,
      "grad_norm": 0.4695904552936554,
      "learning_rate": 0.00017281207305606407,
      "loss": 1.4703,
      "step": 7719
    },
    {
      "epoch": 0.8143459915611815,
      "grad_norm": 0.49509137868881226,
      "learning_rate": 0.00017257238740737548,
      "loss": 1.499,
      "step": 7720
    },
    {
      "epoch": 0.8144514767932489,
      "grad_norm": 0.4523547887802124,
      "learning_rate": 0.0001723330341961675,
      "loss": 1.52,
      "step": 7721
    },
    {
      "epoch": 0.8145569620253165,
      "grad_norm": 0.4584343731403351,
      "learning_rate": 0.0001720940129613584,
      "loss": 1.4952,
      "step": 7722
    },
    {
      "epoch": 0.814662447257384,
      "grad_norm": 0.46166375279426575,
      "learning_rate": 0.0001718553232425059,
      "loss": 1.5241,
      "step": 7723
    },
    {
      "epoch": 0.8147679324894515,
      "grad_norm": 0.48565253615379333,
      "learning_rate": 0.00017161696457980641,
      "loss": 1.4841,
      "step": 7724
    },
    {
      "epoch": 0.814873417721519,
      "grad_norm": 0.454905241727829,
      "learning_rate": 0.00017137893651409406,
      "loss": 1.5503,
      "step": 7725
    },
    {
      "epoch": 0.8149789029535865,
      "grad_norm": 0.43393051624298096,
      "learning_rate": 0.0001711412385868398,
      "loss": 1.5087,
      "step": 7726
    },
    {
      "epoch": 0.815084388185654,
      "grad_norm": 0.4867404103279114,
      "learning_rate": 0.00017090387034015054,
      "loss": 1.488,
      "step": 7727
    },
    {
      "epoch": 0.8151898734177215,
      "grad_norm": 0.5071909427642822,
      "learning_rate": 0.00017066683131676825,
      "loss": 1.494,
      "step": 7728
    },
    {
      "epoch": 0.8152953586497891,
      "grad_norm": 0.43942975997924805,
      "learning_rate": 0.00017043012106006926,
      "loss": 1.4962,
      "step": 7729
    },
    {
      "epoch": 0.8154008438818565,
      "grad_norm": 0.46662601828575134,
      "learning_rate": 0.00017019373911406307,
      "loss": 1.5157,
      "step": 7730
    },
    {
      "epoch": 0.8155063291139241,
      "grad_norm": 0.43356215953826904,
      "learning_rate": 0.00016995768502339165,
      "loss": 1.4811,
      "step": 7731
    },
    {
      "epoch": 0.8156118143459916,
      "grad_norm": 0.45512300729751587,
      "learning_rate": 0.00016972195833332858,
      "loss": 1.5206,
      "step": 7732
    },
    {
      "epoch": 0.815717299578059,
      "grad_norm": 0.48683232069015503,
      "learning_rate": 0.00016948655858977808,
      "loss": 1.5083,
      "step": 7733
    },
    {
      "epoch": 0.8158227848101266,
      "grad_norm": 0.44006553292274475,
      "learning_rate": 0.00016925148533927429,
      "loss": 1.4814,
      "step": 7734
    },
    {
      "epoch": 0.8159282700421941,
      "grad_norm": 0.49457958340644836,
      "learning_rate": 0.00016901673812898022,
      "loss": 1.4982,
      "step": 7735
    },
    {
      "epoch": 0.8160337552742616,
      "grad_norm": 0.4615689814090729,
      "learning_rate": 0.0001687823165066869,
      "loss": 1.5066,
      "step": 7736
    },
    {
      "epoch": 0.8161392405063291,
      "grad_norm": 0.4143151044845581,
      "learning_rate": 0.00016854822002081266,
      "loss": 1.4655,
      "step": 7737
    },
    {
      "epoch": 0.8162447257383966,
      "grad_norm": 0.48641398549079895,
      "learning_rate": 0.00016831444822040207,
      "loss": 1.5175,
      "step": 7738
    },
    {
      "epoch": 0.8163502109704641,
      "grad_norm": 0.45126456022262573,
      "learning_rate": 0.00016808100065512528,
      "loss": 1.5488,
      "step": 7739
    },
    {
      "epoch": 0.8164556962025317,
      "grad_norm": 0.42679527401924133,
      "learning_rate": 0.000167847876875277,
      "loss": 1.5194,
      "step": 7740
    },
    {
      "epoch": 0.8165611814345991,
      "grad_norm": 0.44492048025131226,
      "learning_rate": 0.00016761507643177557,
      "loss": 1.4682,
      "step": 7741
    },
    {
      "epoch": 0.8166666666666667,
      "grad_norm": 0.46783220767974854,
      "learning_rate": 0.0001673825988761623,
      "loss": 1.4597,
      "step": 7742
    },
    {
      "epoch": 0.8167721518987342,
      "grad_norm": 0.4900056719779968,
      "learning_rate": 0.00016715044376060042,
      "loss": 1.5206,
      "step": 7743
    },
    {
      "epoch": 0.8168776371308016,
      "grad_norm": 0.5590119361877441,
      "learning_rate": 0.00016691861063787436,
      "loss": 1.5112,
      "step": 7744
    },
    {
      "epoch": 0.8169831223628692,
      "grad_norm": 0.49072030186653137,
      "learning_rate": 0.0001666870990613889,
      "loss": 1.4525,
      "step": 7745
    },
    {
      "epoch": 0.8170886075949367,
      "grad_norm": 0.4988759458065033,
      "learning_rate": 0.000166455908585168,
      "loss": 1.4933,
      "step": 7746
    },
    {
      "epoch": 0.8171940928270042,
      "grad_norm": 0.4702509045600891,
      "learning_rate": 0.00016622503876385437,
      "loss": 1.4759,
      "step": 7747
    },
    {
      "epoch": 0.8172995780590717,
      "grad_norm": 0.5422484278678894,
      "learning_rate": 0.00016599448915270845,
      "loss": 1.4785,
      "step": 7748
    },
    {
      "epoch": 0.8174050632911393,
      "grad_norm": 0.4553377330303192,
      "learning_rate": 0.00016576425930760734,
      "loss": 1.5389,
      "step": 7749
    },
    {
      "epoch": 0.8175105485232067,
      "grad_norm": 0.43447399139404297,
      "learning_rate": 0.00016553434878504428,
      "loss": 1.4951,
      "step": 7750
    },
    {
      "epoch": 0.8176160337552743,
      "grad_norm": 0.5603171586990356,
      "learning_rate": 0.00016530475714212755,
      "loss": 1.5108,
      "step": 7751
    },
    {
      "epoch": 0.8177215189873418,
      "grad_norm": 0.5328144431114197,
      "learning_rate": 0.00016507548393657973,
      "loss": 1.4536,
      "step": 7752
    },
    {
      "epoch": 0.8178270042194092,
      "grad_norm": 0.48087170720100403,
      "learning_rate": 0.00016484652872673694,
      "loss": 1.4849,
      "step": 7753
    },
    {
      "epoch": 0.8179324894514768,
      "grad_norm": 0.4372412860393524,
      "learning_rate": 0.00016461789107154767,
      "loss": 1.4906,
      "step": 7754
    },
    {
      "epoch": 0.8180379746835443,
      "grad_norm": 0.5650653839111328,
      "learning_rate": 0.00016438957053057236,
      "loss": 1.5046,
      "step": 7755
    },
    {
      "epoch": 0.8181434599156118,
      "grad_norm": 0.48459675908088684,
      "learning_rate": 0.0001641615666639821,
      "loss": 1.4975,
      "step": 7756
    },
    {
      "epoch": 0.8182489451476793,
      "grad_norm": 0.4763208329677582,
      "learning_rate": 0.00016393387903255815,
      "loss": 1.4954,
      "step": 7757
    },
    {
      "epoch": 0.8183544303797469,
      "grad_norm": 0.4780988097190857,
      "learning_rate": 0.000163706507197691,
      "loss": 1.4581,
      "step": 7758
    },
    {
      "epoch": 0.8184599156118143,
      "grad_norm": 0.4592362642288208,
      "learning_rate": 0.00016347945072137934,
      "loss": 1.5242,
      "step": 7759
    },
    {
      "epoch": 0.8185654008438819,
      "grad_norm": 0.47767072916030884,
      "learning_rate": 0.00016325270916622947,
      "loss": 1.5218,
      "step": 7760
    },
    {
      "epoch": 0.8186708860759494,
      "grad_norm": 0.4654233753681183,
      "learning_rate": 0.00016302628209545426,
      "loss": 1.4811,
      "step": 7761
    },
    {
      "epoch": 0.8187763713080168,
      "grad_norm": 0.47363775968551636,
      "learning_rate": 0.00016280016907287237,
      "loss": 1.468,
      "step": 7762
    },
    {
      "epoch": 0.8188818565400844,
      "grad_norm": 0.45746245980262756,
      "learning_rate": 0.00016257436966290764,
      "loss": 1.5372,
      "step": 7763
    },
    {
      "epoch": 0.8189873417721519,
      "grad_norm": 0.4786800444126129,
      "learning_rate": 0.0001623488834305878,
      "loss": 1.5259,
      "step": 7764
    },
    {
      "epoch": 0.8190928270042194,
      "grad_norm": 0.4710049033164978,
      "learning_rate": 0.000162123709941544,
      "loss": 1.4906,
      "step": 7765
    },
    {
      "epoch": 0.8191983122362869,
      "grad_norm": 0.44269460439682007,
      "learning_rate": 0.00016189884876200979,
      "loss": 1.5263,
      "step": 7766
    },
    {
      "epoch": 0.8193037974683545,
      "grad_norm": 0.5046398639678955,
      "learning_rate": 0.00016167429945882031,
      "loss": 1.5018,
      "step": 7767
    },
    {
      "epoch": 0.8194092827004219,
      "grad_norm": 0.5724698901176453,
      "learning_rate": 0.0001614500615994117,
      "loss": 1.5016,
      "step": 7768
    },
    {
      "epoch": 0.8195147679324895,
      "grad_norm": 0.44181177020072937,
      "learning_rate": 0.00016122613475181977,
      "loss": 1.5316,
      "step": 7769
    },
    {
      "epoch": 0.819620253164557,
      "grad_norm": 0.4330645799636841,
      "learning_rate": 0.00016100251848467966,
      "loss": 1.4414,
      "step": 7770
    },
    {
      "epoch": 0.8197257383966244,
      "grad_norm": 0.5350200533866882,
      "learning_rate": 0.00016077921236722464,
      "loss": 1.487,
      "step": 7771
    },
    {
      "epoch": 0.819831223628692,
      "grad_norm": 0.5268557071685791,
      "learning_rate": 0.00016055621596928563,
      "loss": 1.4756,
      "step": 7772
    },
    {
      "epoch": 0.8199367088607595,
      "grad_norm": 0.5045581459999084,
      "learning_rate": 0.00016033352886129,
      "loss": 1.4838,
      "step": 7773
    },
    {
      "epoch": 0.820042194092827,
      "grad_norm": 0.4686928987503052,
      "learning_rate": 0.00016011115061426103,
      "loss": 1.4462,
      "step": 7774
    },
    {
      "epoch": 0.8201476793248945,
      "grad_norm": 0.4847986698150635,
      "learning_rate": 0.00015988908079981698,
      "loss": 1.514,
      "step": 7775
    },
    {
      "epoch": 0.8202531645569621,
      "grad_norm": 0.443629652261734,
      "learning_rate": 0.00015966731899017014,
      "loss": 1.5264,
      "step": 7776
    },
    {
      "epoch": 0.8203586497890295,
      "grad_norm": 0.4528484046459198,
      "learning_rate": 0.00015944586475812633,
      "loss": 1.5244,
      "step": 7777
    },
    {
      "epoch": 0.820464135021097,
      "grad_norm": 0.4824484884738922,
      "learning_rate": 0.00015922471767708377,
      "loss": 1.4958,
      "step": 7778
    },
    {
      "epoch": 0.8205696202531646,
      "grad_norm": 0.5121501088142395,
      "learning_rate": 0.00015900387732103232,
      "loss": 1.4616,
      "step": 7779
    },
    {
      "epoch": 0.820675105485232,
      "grad_norm": 0.5574297308921814,
      "learning_rate": 0.0001587833432645528,
      "loss": 1.5236,
      "step": 7780
    },
    {
      "epoch": 0.8207805907172996,
      "grad_norm": 0.4380872845649719,
      "learning_rate": 0.00015856311508281594,
      "loss": 1.5378,
      "step": 7781
    },
    {
      "epoch": 0.8208860759493671,
      "grad_norm": 0.5012332797050476,
      "learning_rate": 0.00015834319235158187,
      "loss": 1.504,
      "step": 7782
    },
    {
      "epoch": 0.8209915611814346,
      "grad_norm": 0.5149298310279846,
      "learning_rate": 0.00015812357464719905,
      "loss": 1.5073,
      "step": 7783
    },
    {
      "epoch": 0.8210970464135021,
      "grad_norm": 0.4839160144329071,
      "learning_rate": 0.0001579042615466035,
      "loss": 1.527,
      "step": 7784
    },
    {
      "epoch": 0.8212025316455697,
      "grad_norm": 0.43355682492256165,
      "learning_rate": 0.00015768525262731804,
      "loss": 1.492,
      "step": 7785
    },
    {
      "epoch": 0.8213080168776371,
      "grad_norm": 0.44033581018447876,
      "learning_rate": 0.0001574665474674514,
      "loss": 1.5354,
      "step": 7786
    },
    {
      "epoch": 0.8214135021097047,
      "grad_norm": 0.4705195128917694,
      "learning_rate": 0.00015724814564569765,
      "loss": 1.5337,
      "step": 7787
    },
    {
      "epoch": 0.8215189873417722,
      "grad_norm": 0.4969990849494934,
      "learning_rate": 0.00015703004674133498,
      "loss": 1.4949,
      "step": 7788
    },
    {
      "epoch": 0.8216244725738396,
      "grad_norm": 0.4627792537212372,
      "learning_rate": 0.00015681225033422526,
      "loss": 1.4784,
      "step": 7789
    },
    {
      "epoch": 0.8217299578059072,
      "grad_norm": 0.41960787773132324,
      "learning_rate": 0.00015659475600481292,
      "loss": 1.4926,
      "step": 7790
    },
    {
      "epoch": 0.8218354430379747,
      "grad_norm": 0.48010876774787903,
      "learning_rate": 0.00015637756333412454,
      "loss": 1.5055,
      "step": 7791
    },
    {
      "epoch": 0.8219409282700422,
      "grad_norm": 0.4434579312801361,
      "learning_rate": 0.0001561606719037676,
      "loss": 1.544,
      "step": 7792
    },
    {
      "epoch": 0.8220464135021097,
      "grad_norm": 0.47622019052505493,
      "learning_rate": 0.00015594408129592993,
      "loss": 1.5105,
      "step": 7793
    },
    {
      "epoch": 0.8221518987341773,
      "grad_norm": 0.511711061000824,
      "learning_rate": 0.00015572779109337888,
      "loss": 1.4681,
      "step": 7794
    },
    {
      "epoch": 0.8222573839662447,
      "grad_norm": 0.43236175179481506,
      "learning_rate": 0.00015551180087946046,
      "loss": 1.4648,
      "step": 7795
    },
    {
      "epoch": 0.8223628691983123,
      "grad_norm": 0.39645713567733765,
      "learning_rate": 0.00015529611023809868,
      "loss": 1.5043,
      "step": 7796
    },
    {
      "epoch": 0.8224683544303798,
      "grad_norm": 0.45807018876075745,
      "learning_rate": 0.00015508071875379448,
      "loss": 1.4905,
      "step": 7797
    },
    {
      "epoch": 0.8225738396624472,
      "grad_norm": 0.44831138849258423,
      "learning_rate": 0.00015486562601162512,
      "loss": 1.4911,
      "step": 7798
    },
    {
      "epoch": 0.8226793248945148,
      "grad_norm": 0.45922088623046875,
      "learning_rate": 0.00015465083159724345,
      "loss": 1.5301,
      "step": 7799
    },
    {
      "epoch": 0.8227848101265823,
      "grad_norm": 0.42836442589759827,
      "learning_rate": 0.00015443633509687688,
      "loss": 1.4681,
      "step": 7800
    },
    {
      "epoch": 0.8228902953586498,
      "grad_norm": 0.45813196897506714,
      "learning_rate": 0.0001542221360973268,
      "loss": 1.5095,
      "step": 7801
    },
    {
      "epoch": 0.8229957805907173,
      "grad_norm": 0.44777148962020874,
      "learning_rate": 0.00015400823418596765,
      "loss": 1.4879,
      "step": 7802
    },
    {
      "epoch": 0.8231012658227848,
      "grad_norm": 0.48104265332221985,
      "learning_rate": 0.0001537946289507462,
      "loss": 1.4895,
      "step": 7803
    },
    {
      "epoch": 0.8232067510548523,
      "grad_norm": 0.46618038415908813,
      "learning_rate": 0.00015358131998018072,
      "loss": 1.5082,
      "step": 7804
    },
    {
      "epoch": 0.8233122362869199,
      "grad_norm": 0.42976656556129456,
      "learning_rate": 0.00015336830686336008,
      "loss": 1.466,
      "step": 7805
    },
    {
      "epoch": 0.8234177215189873,
      "grad_norm": 0.44920340180397034,
      "learning_rate": 0.00015315558918994331,
      "loss": 1.5368,
      "step": 7806
    },
    {
      "epoch": 0.8235232067510548,
      "grad_norm": 0.47994476556777954,
      "learning_rate": 0.0001529431665501584,
      "loss": 1.5114,
      "step": 7807
    },
    {
      "epoch": 0.8236286919831224,
      "grad_norm": 0.43936634063720703,
      "learning_rate": 0.0001527310385348017,
      "loss": 1.4638,
      "step": 7808
    },
    {
      "epoch": 0.8237341772151898,
      "grad_norm": 0.4491763412952423,
      "learning_rate": 0.0001525192047352371,
      "loss": 1.5098,
      "step": 7809
    },
    {
      "epoch": 0.8238396624472574,
      "grad_norm": 0.43240979313850403,
      "learning_rate": 0.00015230766474339536,
      "loss": 1.4836,
      "step": 7810
    },
    {
      "epoch": 0.8239451476793249,
      "grad_norm": 0.4447120428085327,
      "learning_rate": 0.0001520964181517731,
      "loss": 1.4735,
      "step": 7811
    },
    {
      "epoch": 0.8240506329113924,
      "grad_norm": 0.44551825523376465,
      "learning_rate": 0.00015188546455343228,
      "loss": 1.5006,
      "step": 7812
    },
    {
      "epoch": 0.8241561181434599,
      "grad_norm": 0.42466315627098083,
      "learning_rate": 0.00015167480354199908,
      "loss": 1.4889,
      "step": 7813
    },
    {
      "epoch": 0.8242616033755275,
      "grad_norm": 0.47459515929222107,
      "learning_rate": 0.00015146443471166345,
      "loss": 1.5504,
      "step": 7814
    },
    {
      "epoch": 0.8243670886075949,
      "grad_norm": 0.4613172113895416,
      "learning_rate": 0.00015125435765717815,
      "loss": 1.5016,
      "step": 7815
    },
    {
      "epoch": 0.8244725738396624,
      "grad_norm": 0.43431389331817627,
      "learning_rate": 0.000151044571973858,
      "loss": 1.468,
      "step": 7816
    },
    {
      "epoch": 0.82457805907173,
      "grad_norm": 0.44141438603401184,
      "learning_rate": 0.00015083507725757912,
      "loss": 1.4772,
      "step": 7817
    },
    {
      "epoch": 0.8246835443037974,
      "grad_norm": 0.4615953862667084,
      "learning_rate": 0.00015062587310477813,
      "loss": 1.5334,
      "step": 7818
    },
    {
      "epoch": 0.824789029535865,
      "grad_norm": 0.413897305727005,
      "learning_rate": 0.00015041695911245136,
      "loss": 1.511,
      "step": 7819
    },
    {
      "epoch": 0.8248945147679325,
      "grad_norm": 0.42081931233406067,
      "learning_rate": 0.00015020833487815416,
      "loss": 1.5277,
      "step": 7820
    },
    {
      "epoch": 0.825,
      "grad_norm": 0.46726006269454956,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.5148,
      "step": 7821
    },
    {
      "epoch": 0.8251054852320675,
      "grad_norm": 0.4627990424633026,
      "learning_rate": 0.00014979195407665975,
      "loss": 1.4826,
      "step": 7822
    },
    {
      "epoch": 0.825210970464135,
      "grad_norm": 0.4164053797721863,
      "learning_rate": 0.000149584196707361,
      "loss": 1.4979,
      "step": 7823
    },
    {
      "epoch": 0.8253164556962025,
      "grad_norm": 0.4177986681461334,
      "learning_rate": 0.00014937672749188704,
      "loss": 1.4729,
      "step": 7824
    },
    {
      "epoch": 0.82542194092827,
      "grad_norm": 0.4489454925060272,
      "learning_rate": 0.0001491695460305765,
      "loss": 1.4819,
      "step": 7825
    },
    {
      "epoch": 0.8255274261603376,
      "grad_norm": 0.45009520649909973,
      "learning_rate": 0.00014896265192432194,
      "loss": 1.4989,
      "step": 7826
    },
    {
      "epoch": 0.825632911392405,
      "grad_norm": 0.4520403742790222,
      "learning_rate": 0.00014875604477456987,
      "loss": 1.5242,
      "step": 7827
    },
    {
      "epoch": 0.8257383966244726,
      "grad_norm": 0.4293029010295868,
      "learning_rate": 0.00014854972418331948,
      "loss": 1.508,
      "step": 7828
    },
    {
      "epoch": 0.8258438818565401,
      "grad_norm": 0.4253372251987457,
      "learning_rate": 0.00014834368975312172,
      "loss": 1.5103,
      "step": 7829
    },
    {
      "epoch": 0.8259493670886076,
      "grad_norm": 0.4367186725139618,
      "learning_rate": 0.0001481379410870792,
      "loss": 1.4541,
      "step": 7830
    },
    {
      "epoch": 0.8260548523206751,
      "grad_norm": 0.4056040346622467,
      "learning_rate": 0.00014793247778884463,
      "loss": 1.5022,
      "step": 7831
    },
    {
      "epoch": 0.8261603375527427,
      "grad_norm": 0.43779054284095764,
      "learning_rate": 0.00014772729946262067,
      "loss": 1.4817,
      "step": 7832
    },
    {
      "epoch": 0.8262658227848101,
      "grad_norm": 0.4522009491920471,
      "learning_rate": 0.00014752240571315898,
      "loss": 1.4895,
      "step": 7833
    },
    {
      "epoch": 0.8263713080168776,
      "grad_norm": 0.4671722650527954,
      "learning_rate": 0.00014731779614575915,
      "loss": 1.4807,
      "step": 7834
    },
    {
      "epoch": 0.8264767932489452,
      "grad_norm": 0.45991694927215576,
      "learning_rate": 0.00014711347036626857,
      "loss": 1.4628,
      "step": 7835
    },
    {
      "epoch": 0.8265822784810126,
      "grad_norm": 0.4185478985309601,
      "learning_rate": 0.00014690942798108097,
      "loss": 1.4468,
      "step": 7836
    },
    {
      "epoch": 0.8266877637130802,
      "grad_norm": 0.4673352539539337,
      "learning_rate": 0.00014670566859713622,
      "loss": 1.5137,
      "step": 7837
    },
    {
      "epoch": 0.8267932489451477,
      "grad_norm": 0.5092229843139648,
      "learning_rate": 0.00014650219182191934,
      "loss": 1.5215,
      "step": 7838
    },
    {
      "epoch": 0.8268987341772152,
      "grad_norm": 0.4105064570903778,
      "learning_rate": 0.00014629899726345958,
      "loss": 1.4216,
      "step": 7839
    },
    {
      "epoch": 0.8270042194092827,
      "grad_norm": 0.47240203619003296,
      "learning_rate": 0.00014609608453033013,
      "loss": 1.4881,
      "step": 7840
    },
    {
      "epoch": 0.8271097046413503,
      "grad_norm": 0.4400210976600647,
      "learning_rate": 0.0001458934532316467,
      "loss": 1.5094,
      "step": 7841
    },
    {
      "epoch": 0.8272151898734177,
      "grad_norm": 0.45975586771965027,
      "learning_rate": 0.0001456911029770675,
      "loss": 1.4988,
      "step": 7842
    },
    {
      "epoch": 0.8273206751054852,
      "grad_norm": 0.4276962876319885,
      "learning_rate": 0.00014548903337679208,
      "loss": 1.5001,
      "step": 7843
    },
    {
      "epoch": 0.8274261603375528,
      "grad_norm": 0.47978460788726807,
      "learning_rate": 0.0001452872440415604,
      "loss": 1.4588,
      "step": 7844
    },
    {
      "epoch": 0.8275316455696202,
      "grad_norm": 0.49116069078445435,
      "learning_rate": 0.00014508573458265245,
      "loss": 1.4796,
      "step": 7845
    },
    {
      "epoch": 0.8276371308016878,
      "grad_norm": 0.4531501829624176,
      "learning_rate": 0.00014488450461188753,
      "loss": 1.496,
      "step": 7846
    },
    {
      "epoch": 0.8277426160337553,
      "grad_norm": 0.47048357129096985,
      "learning_rate": 0.000144683553741623,
      "loss": 1.5074,
      "step": 7847
    },
    {
      "epoch": 0.8278481012658228,
      "grad_norm": 0.5059932470321655,
      "learning_rate": 0.00014448288158475423,
      "loss": 1.5094,
      "step": 7848
    },
    {
      "epoch": 0.8279535864978903,
      "grad_norm": 0.4376550614833832,
      "learning_rate": 0.00014428248775471316,
      "loss": 1.4607,
      "step": 7849
    },
    {
      "epoch": 0.8280590717299579,
      "grad_norm": 0.4301762878894806,
      "learning_rate": 0.00014408237186546807,
      "loss": 1.4945,
      "step": 7850
    },
    {
      "epoch": 0.8281645569620253,
      "grad_norm": 0.421731561422348,
      "learning_rate": 0.00014388253353152277,
      "loss": 1.5044,
      "step": 7851
    },
    {
      "epoch": 0.8282700421940928,
      "grad_norm": 0.4436945915222168,
      "learning_rate": 0.00014368297236791545,
      "loss": 1.4603,
      "step": 7852
    },
    {
      "epoch": 0.8283755274261604,
      "grad_norm": 0.5222391486167908,
      "learning_rate": 0.00014348368799021845,
      "loss": 1.4921,
      "step": 7853
    },
    {
      "epoch": 0.8284810126582278,
      "grad_norm": 0.45426493883132935,
      "learning_rate": 0.00014328468001453717,
      "loss": 1.5011,
      "step": 7854
    },
    {
      "epoch": 0.8285864978902954,
      "grad_norm": 0.49810025095939636,
      "learning_rate": 0.00014308594805750958,
      "loss": 1.4833,
      "step": 7855
    },
    {
      "epoch": 0.8286919831223629,
      "grad_norm": 0.45834463834762573,
      "learning_rate": 0.0001428874917363054,
      "loss": 1.4696,
      "step": 7856
    },
    {
      "epoch": 0.8287974683544304,
      "grad_norm": 0.43906545639038086,
      "learning_rate": 0.000142689310668625,
      "loss": 1.4697,
      "step": 7857
    },
    {
      "epoch": 0.8289029535864979,
      "grad_norm": 0.5387390851974487,
      "learning_rate": 0.0001424914044726995,
      "loss": 1.4665,
      "step": 7858
    },
    {
      "epoch": 0.8290084388185655,
      "grad_norm": 0.45019060373306274,
      "learning_rate": 0.000142293772767289,
      "loss": 1.4622,
      "step": 7859
    },
    {
      "epoch": 0.8291139240506329,
      "grad_norm": 0.4874425232410431,
      "learning_rate": 0.00014209641517168273,
      "loss": 1.4639,
      "step": 7860
    },
    {
      "epoch": 0.8292194092827004,
      "grad_norm": 0.4442334473133087,
      "learning_rate": 0.0001418993313056979,
      "loss": 1.5061,
      "step": 7861
    },
    {
      "epoch": 0.829324894514768,
      "grad_norm": 0.4124840199947357,
      "learning_rate": 0.0001417025207896788,
      "loss": 1.5127,
      "step": 7862
    },
    {
      "epoch": 0.8294303797468354,
      "grad_norm": 0.4346408545970917,
      "learning_rate": 0.00014150598324449667,
      "loss": 1.4777,
      "step": 7863
    },
    {
      "epoch": 0.829535864978903,
      "grad_norm": 0.44085627794265747,
      "learning_rate": 0.0001413097182915482,
      "loss": 1.5012,
      "step": 7864
    },
    {
      "epoch": 0.8296413502109705,
      "grad_norm": 0.49189162254333496,
      "learning_rate": 0.0001411137255527554,
      "loss": 1.4575,
      "step": 7865
    },
    {
      "epoch": 0.829746835443038,
      "grad_norm": 0.42766907811164856,
      "learning_rate": 0.00014091800465056476,
      "loss": 1.5303,
      "step": 7866
    },
    {
      "epoch": 0.8298523206751055,
      "grad_norm": 0.4196573793888092,
      "learning_rate": 0.00014072255520794613,
      "loss": 1.4627,
      "step": 7867
    },
    {
      "epoch": 0.8299578059071729,
      "grad_norm": 0.49347805976867676,
      "learning_rate": 0.0001405273768483926,
      "loss": 1.4547,
      "step": 7868
    },
    {
      "epoch": 0.8300632911392405,
      "grad_norm": 0.4952002167701721,
      "learning_rate": 0.00014033246919591922,
      "loss": 1.5215,
      "step": 7869
    },
    {
      "epoch": 0.830168776371308,
      "grad_norm": 0.46306660771369934,
      "learning_rate": 0.00014013783187506265,
      "loss": 1.497,
      "step": 7870
    },
    {
      "epoch": 0.8302742616033755,
      "grad_norm": 0.5201023817062378,
      "learning_rate": 0.00013994346451088037,
      "loss": 1.5562,
      "step": 7871
    },
    {
      "epoch": 0.830379746835443,
      "grad_norm": 0.4886961579322815,
      "learning_rate": 0.00013974936672894972,
      "loss": 1.5095,
      "step": 7872
    },
    {
      "epoch": 0.8304852320675106,
      "grad_norm": 0.44482308626174927,
      "learning_rate": 0.0001395555381553675,
      "loss": 1.4824,
      "step": 7873
    },
    {
      "epoch": 0.830590717299578,
      "grad_norm": 0.43957915902137756,
      "learning_rate": 0.00013936197841674894,
      "loss": 1.5163,
      "step": 7874
    },
    {
      "epoch": 0.8306962025316456,
      "grad_norm": 0.42717117071151733,
      "learning_rate": 0.00013916868714022735,
      "loss": 1.4975,
      "step": 7875
    },
    {
      "epoch": 0.8308016877637131,
      "grad_norm": 0.4543043076992035,
      "learning_rate": 0.00013897566395345315,
      "loss": 1.5152,
      "step": 7876
    },
    {
      "epoch": 0.8309071729957805,
      "grad_norm": 0.47359344363212585,
      "learning_rate": 0.000138782908484593,
      "loss": 1.5193,
      "step": 7877
    },
    {
      "epoch": 0.8310126582278481,
      "grad_norm": 0.4322127401828766,
      "learning_rate": 0.0001385904203623296,
      "loss": 1.5129,
      "step": 7878
    },
    {
      "epoch": 0.8311181434599156,
      "grad_norm": 0.4316220283508301,
      "learning_rate": 0.00013839819921586025,
      "loss": 1.5055,
      "step": 7879
    },
    {
      "epoch": 0.8312236286919831,
      "grad_norm": 0.4473347067832947,
      "learning_rate": 0.00013820624467489697,
      "loss": 1.4959,
      "step": 7880
    },
    {
      "epoch": 0.8313291139240506,
      "grad_norm": 0.47121867537498474,
      "learning_rate": 0.00013801455636966518,
      "loss": 1.4996,
      "step": 7881
    },
    {
      "epoch": 0.8314345991561182,
      "grad_norm": 0.44001927971839905,
      "learning_rate": 0.00013782313393090301,
      "loss": 1.4859,
      "step": 7882
    },
    {
      "epoch": 0.8315400843881856,
      "grad_norm": 0.46005189418792725,
      "learning_rate": 0.00013763197698986107,
      "loss": 1.4819,
      "step": 7883
    },
    {
      "epoch": 0.8316455696202532,
      "grad_norm": 0.5089910626411438,
      "learning_rate": 0.00013744108517830105,
      "loss": 1.548,
      "step": 7884
    },
    {
      "epoch": 0.8317510548523207,
      "grad_norm": 0.4660554826259613,
      "learning_rate": 0.00013725045812849567,
      "loss": 1.5342,
      "step": 7885
    },
    {
      "epoch": 0.8318565400843881,
      "grad_norm": 0.4434336721897125,
      "learning_rate": 0.00013706009547322762,
      "loss": 1.5392,
      "step": 7886
    },
    {
      "epoch": 0.8319620253164557,
      "grad_norm": 0.4478782117366791,
      "learning_rate": 0.00013686999684578871,
      "loss": 1.5216,
      "step": 7887
    },
    {
      "epoch": 0.8320675105485232,
      "grad_norm": 0.4423481523990631,
      "learning_rate": 0.00013668016187997964,
      "loss": 1.4767,
      "step": 7888
    },
    {
      "epoch": 0.8321729957805907,
      "grad_norm": 0.4352787733078003,
      "learning_rate": 0.00013649059021010894,
      "loss": 1.4948,
      "step": 7889
    },
    {
      "epoch": 0.8322784810126582,
      "grad_norm": 0.4673626720905304,
      "learning_rate": 0.00013630128147099213,
      "loss": 1.5057,
      "step": 7890
    },
    {
      "epoch": 0.8323839662447258,
      "grad_norm": 0.42983680963516235,
      "learning_rate": 0.00013611223529795156,
      "loss": 1.4659,
      "step": 7891
    },
    {
      "epoch": 0.8324894514767932,
      "grad_norm": 0.41277554631233215,
      "learning_rate": 0.00013592345132681512,
      "loss": 1.4549,
      "step": 7892
    },
    {
      "epoch": 0.8325949367088608,
      "grad_norm": 0.41870588064193726,
      "learning_rate": 0.0001357349291939159,
      "loss": 1.4922,
      "step": 7893
    },
    {
      "epoch": 0.8327004219409283,
      "grad_norm": 0.4255370497703552,
      "learning_rate": 0.00013554666853609146,
      "loss": 1.469,
      "step": 7894
    },
    {
      "epoch": 0.8328059071729957,
      "grad_norm": 0.52479088306427,
      "learning_rate": 0.0001353586689906829,
      "loss": 1.5346,
      "step": 7895
    },
    {
      "epoch": 0.8329113924050633,
      "grad_norm": 0.4748759865760803,
      "learning_rate": 0.00013517093019553444,
      "loss": 1.4955,
      "step": 7896
    },
    {
      "epoch": 0.8330168776371308,
      "grad_norm": 0.44746047258377075,
      "learning_rate": 0.00013498345178899248,
      "loss": 1.4736,
      "step": 7897
    },
    {
      "epoch": 0.8331223628691983,
      "grad_norm": 0.466625452041626,
      "learning_rate": 0.0001347962334099052,
      "loss": 1.5018,
      "step": 7898
    },
    {
      "epoch": 0.8332278481012658,
      "grad_norm": 0.5671567320823669,
      "learning_rate": 0.00013460927469762155,
      "loss": 1.4827,
      "step": 7899
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.42769524455070496,
      "learning_rate": 0.00013442257529199068,
      "loss": 1.5205,
      "step": 7900
    },
    {
      "epoch": 0.8334388185654008,
      "grad_norm": 0.48645901679992676,
      "learning_rate": 0.00013423613483336144,
      "loss": 1.4899,
      "step": 7901
    },
    {
      "epoch": 0.8335443037974684,
      "grad_norm": 0.544087827205658,
      "learning_rate": 0.00013404995296258118,
      "loss": 1.5081,
      "step": 7902
    },
    {
      "epoch": 0.8336497890295359,
      "grad_norm": 0.5929046869277954,
      "learning_rate": 0.00013386402932099572,
      "loss": 1.456,
      "step": 7903
    },
    {
      "epoch": 0.8337552742616033,
      "grad_norm": 0.4338153004646301,
      "learning_rate": 0.0001336783635504482,
      "loss": 1.4701,
      "step": 7904
    },
    {
      "epoch": 0.8338607594936709,
      "grad_norm": 0.49112027883529663,
      "learning_rate": 0.00013349295529327843,
      "loss": 1.5106,
      "step": 7905
    },
    {
      "epoch": 0.8339662447257384,
      "grad_norm": 0.5577452778816223,
      "learning_rate": 0.00013330780419232245,
      "loss": 1.5185,
      "step": 7906
    },
    {
      "epoch": 0.8340717299578059,
      "grad_norm": 0.5064457654953003,
      "learning_rate": 0.0001331229098909114,
      "loss": 1.4973,
      "step": 7907
    },
    {
      "epoch": 0.8341772151898734,
      "grad_norm": 0.5183221697807312,
      "learning_rate": 0.00013293827203287141,
      "loss": 1.5074,
      "step": 7908
    },
    {
      "epoch": 0.834282700421941,
      "grad_norm": 0.5224149227142334,
      "learning_rate": 0.00013275389026252255,
      "loss": 1.4958,
      "step": 7909
    },
    {
      "epoch": 0.8343881856540084,
      "grad_norm": 0.5373936891555786,
      "learning_rate": 0.00013256976422467803,
      "loss": 1.5091,
      "step": 7910
    },
    {
      "epoch": 0.834493670886076,
      "grad_norm": 0.5740803480148315,
      "learning_rate": 0.00013238589356464394,
      "loss": 1.4995,
      "step": 7911
    },
    {
      "epoch": 0.8345991561181435,
      "grad_norm": 0.45745906233787537,
      "learning_rate": 0.00013220227792821806,
      "loss": 1.4902,
      "step": 7912
    },
    {
      "epoch": 0.8347046413502109,
      "grad_norm": 0.475676029920578,
      "learning_rate": 0.00013201891696168963,
      "loss": 1.5058,
      "step": 7913
    },
    {
      "epoch": 0.8348101265822785,
      "grad_norm": 0.46853286027908325,
      "learning_rate": 0.00013183581031183853,
      "loss": 1.4598,
      "step": 7914
    },
    {
      "epoch": 0.834915611814346,
      "grad_norm": 0.5400652289390564,
      "learning_rate": 0.00013165295762593426,
      "loss": 1.5235,
      "step": 7915
    },
    {
      "epoch": 0.8350210970464135,
      "grad_norm": 0.4536226987838745,
      "learning_rate": 0.0001314703585517359,
      "loss": 1.4968,
      "step": 7916
    },
    {
      "epoch": 0.835126582278481,
      "grad_norm": 0.42296475172042847,
      "learning_rate": 0.00013128801273749074,
      "loss": 1.4985,
      "step": 7917
    },
    {
      "epoch": 0.8352320675105486,
      "grad_norm": 0.4564037024974823,
      "learning_rate": 0.00013110591983193424,
      "loss": 1.4992,
      "step": 7918
    },
    {
      "epoch": 0.835337552742616,
      "grad_norm": 0.5482998490333557,
      "learning_rate": 0.0001309240794842889,
      "loss": 1.4901,
      "step": 7919
    },
    {
      "epoch": 0.8354430379746836,
      "grad_norm": 0.5197405815124512,
      "learning_rate": 0.00013074249134426366,
      "loss": 1.5032,
      "step": 7920
    },
    {
      "epoch": 0.8355485232067511,
      "grad_norm": 0.4468212127685547,
      "learning_rate": 0.00013056115506205357,
      "loss": 1.5176,
      "step": 7921
    },
    {
      "epoch": 0.8356540084388185,
      "grad_norm": 0.6020689010620117,
      "learning_rate": 0.00013038007028833853,
      "loss": 1.508,
      "step": 7922
    },
    {
      "epoch": 0.8357594936708861,
      "grad_norm": 0.5337532758712769,
      "learning_rate": 0.00013019923667428315,
      "loss": 1.5274,
      "step": 7923
    },
    {
      "epoch": 0.8358649789029536,
      "grad_norm": 0.5087836980819702,
      "learning_rate": 0.0001300186538715359,
      "loss": 1.5086,
      "step": 7924
    },
    {
      "epoch": 0.8359704641350211,
      "grad_norm": 0.4363085627555847,
      "learning_rate": 0.00012983832153222816,
      "loss": 1.4885,
      "step": 7925
    },
    {
      "epoch": 0.8360759493670886,
      "grad_norm": 0.5247862339019775,
      "learning_rate": 0.00012965823930897406,
      "loss": 1.5262,
      "step": 7926
    },
    {
      "epoch": 0.8361814345991562,
      "grad_norm": 0.4489036798477173,
      "learning_rate": 0.00012947840685486933,
      "loss": 1.4673,
      "step": 7927
    },
    {
      "epoch": 0.8362869198312236,
      "grad_norm": 0.4503890573978424,
      "learning_rate": 0.00012929882382349103,
      "loss": 1.5023,
      "step": 7928
    },
    {
      "epoch": 0.8363924050632912,
      "grad_norm": 0.4541943073272705,
      "learning_rate": 0.00012911948986889664,
      "loss": 1.5163,
      "step": 7929
    },
    {
      "epoch": 0.8364978902953587,
      "grad_norm": 0.46518415212631226,
      "learning_rate": 0.0001289404046456233,
      "loss": 1.4874,
      "step": 7930
    },
    {
      "epoch": 0.8366033755274261,
      "grad_norm": 0.5406955480575562,
      "learning_rate": 0.00012876156780868752,
      "loss": 1.5014,
      "step": 7931
    },
    {
      "epoch": 0.8367088607594937,
      "grad_norm": 0.4911135733127594,
      "learning_rate": 0.00012858297901358425,
      "loss": 1.4872,
      "step": 7932
    },
    {
      "epoch": 0.8368143459915611,
      "grad_norm": 0.44557490944862366,
      "learning_rate": 0.0001284046379162861,
      "loss": 1.5097,
      "step": 7933
    },
    {
      "epoch": 0.8369198312236287,
      "grad_norm": 0.47949543595314026,
      "learning_rate": 0.00012822654417324306,
      "loss": 1.499,
      "step": 7934
    },
    {
      "epoch": 0.8370253164556962,
      "grad_norm": 0.4387923777103424,
      "learning_rate": 0.00012804869744138136,
      "loss": 1.4771,
      "step": 7935
    },
    {
      "epoch": 0.8371308016877637,
      "grad_norm": 0.43199142813682556,
      "learning_rate": 0.0001278710973781033,
      "loss": 1.4906,
      "step": 7936
    },
    {
      "epoch": 0.8372362869198312,
      "grad_norm": 0.44868308305740356,
      "learning_rate": 0.00012769374364128628,
      "loss": 1.4728,
      "step": 7937
    },
    {
      "epoch": 0.8373417721518988,
      "grad_norm": 0.45661893486976624,
      "learning_rate": 0.0001275166358892821,
      "loss": 1.4951,
      "step": 7938
    },
    {
      "epoch": 0.8374472573839662,
      "grad_norm": 0.4271854758262634,
      "learning_rate": 0.00012733977378091664,
      "loss": 1.5025,
      "step": 7939
    },
    {
      "epoch": 0.8375527426160337,
      "grad_norm": 0.4594980478286743,
      "learning_rate": 0.0001271631569754887,
      "loss": 1.5093,
      "step": 7940
    },
    {
      "epoch": 0.8376582278481013,
      "grad_norm": 0.4268159866333008,
      "learning_rate": 0.00012698678513276985,
      "loss": 1.5065,
      "step": 7941
    },
    {
      "epoch": 0.8377637130801687,
      "grad_norm": 0.4283410608768463,
      "learning_rate": 0.00012681065791300352,
      "loss": 1.5137,
      "step": 7942
    },
    {
      "epoch": 0.8378691983122363,
      "grad_norm": 0.4050031006336212,
      "learning_rate": 0.00012663477497690422,
      "loss": 1.4993,
      "step": 7943
    },
    {
      "epoch": 0.8379746835443038,
      "grad_norm": 0.4224381446838379,
      "learning_rate": 0.0001264591359856572,
      "loss": 1.5073,
      "step": 7944
    },
    {
      "epoch": 0.8380801687763713,
      "grad_norm": 0.4567984938621521,
      "learning_rate": 0.00012628374060091757,
      "loss": 1.4819,
      "step": 7945
    },
    {
      "epoch": 0.8381856540084388,
      "grad_norm": 0.45892590284347534,
      "learning_rate": 0.00012610858848480968,
      "loss": 1.5071,
      "step": 7946
    },
    {
      "epoch": 0.8382911392405064,
      "grad_norm": 0.41472315788269043,
      "learning_rate": 0.00012593367929992667,
      "loss": 1.5063,
      "step": 7947
    },
    {
      "epoch": 0.8383966244725738,
      "grad_norm": 0.42245355248451233,
      "learning_rate": 0.00012575901270932944,
      "loss": 1.4663,
      "step": 7948
    },
    {
      "epoch": 0.8385021097046413,
      "grad_norm": 0.4199659526348114,
      "learning_rate": 0.00012558458837654633,
      "loss": 1.497,
      "step": 7949
    },
    {
      "epoch": 0.8386075949367089,
      "grad_norm": 0.43542659282684326,
      "learning_rate": 0.0001254104059655723,
      "loss": 1.5093,
      "step": 7950
    },
    {
      "epoch": 0.8387130801687763,
      "grad_norm": 0.4467887580394745,
      "learning_rate": 0.0001252364651408684,
      "loss": 1.5066,
      "step": 7951
    },
    {
      "epoch": 0.8388185654008439,
      "grad_norm": 0.42111241817474365,
      "learning_rate": 0.00012506276556736108,
      "loss": 1.4532,
      "step": 7952
    },
    {
      "epoch": 0.8389240506329114,
      "grad_norm": 0.42740586400032043,
      "learning_rate": 0.00012488930691044144,
      "loss": 1.4988,
      "step": 7953
    },
    {
      "epoch": 0.8390295358649789,
      "grad_norm": 0.3941822350025177,
      "learning_rate": 0.00012471608883596476,
      "loss": 1.4486,
      "step": 7954
    },
    {
      "epoch": 0.8391350210970464,
      "grad_norm": 0.41999900341033936,
      "learning_rate": 0.00012454311101024967,
      "loss": 1.466,
      "step": 7955
    },
    {
      "epoch": 0.839240506329114,
      "grad_norm": 0.4354103207588196,
      "learning_rate": 0.0001243703731000777,
      "loss": 1.4809,
      "step": 7956
    },
    {
      "epoch": 0.8393459915611814,
      "grad_norm": 0.4210897386074066,
      "learning_rate": 0.00012419787477269256,
      "loss": 1.4956,
      "step": 7957
    },
    {
      "epoch": 0.8394514767932489,
      "grad_norm": 0.43319812417030334,
      "learning_rate": 0.00012402561569579935,
      "loss": 1.5169,
      "step": 7958
    },
    {
      "epoch": 0.8395569620253165,
      "grad_norm": 0.42039015889167786,
      "learning_rate": 0.00012385359553756422,
      "loss": 1.5027,
      "step": 7959
    },
    {
      "epoch": 0.8396624472573839,
      "grad_norm": 0.41190609335899353,
      "learning_rate": 0.00012368181396661337,
      "loss": 1.4953,
      "step": 7960
    },
    {
      "epoch": 0.8397679324894515,
      "grad_norm": 0.45963868498802185,
      "learning_rate": 0.00012351027065203284,
      "loss": 1.459,
      "step": 7961
    },
    {
      "epoch": 0.839873417721519,
      "grad_norm": 0.4377284646034241,
      "learning_rate": 0.00012333896526336752,
      "loss": 1.4867,
      "step": 7962
    },
    {
      "epoch": 0.8399789029535865,
      "grad_norm": 0.4265303909778595,
      "learning_rate": 0.0001231678974706205,
      "loss": 1.4896,
      "step": 7963
    },
    {
      "epoch": 0.840084388185654,
      "grad_norm": 0.42678651213645935,
      "learning_rate": 0.00012299706694425286,
      "loss": 1.5024,
      "step": 7964
    },
    {
      "epoch": 0.8401898734177216,
      "grad_norm": 0.46492063999176025,
      "learning_rate": 0.00012282647335518248,
      "loss": 1.4917,
      "step": 7965
    },
    {
      "epoch": 0.840295358649789,
      "grad_norm": 0.4300294518470764,
      "learning_rate": 0.00012265611637478376,
      "loss": 1.4924,
      "step": 7966
    },
    {
      "epoch": 0.8404008438818565,
      "grad_norm": 0.4165033996105194,
      "learning_rate": 0.00012248599567488697,
      "loss": 1.4996,
      "step": 7967
    },
    {
      "epoch": 0.8405063291139241,
      "grad_norm": 0.47845935821533203,
      "learning_rate": 0.00012231611092777743,
      "loss": 1.5015,
      "step": 7968
    },
    {
      "epoch": 0.8406118143459915,
      "grad_norm": 0.414779931306839,
      "learning_rate": 0.0001221464618061951,
      "loss": 1.4643,
      "step": 7969
    },
    {
      "epoch": 0.8407172995780591,
      "grad_norm": 0.4585050940513611,
      "learning_rate": 0.00012197704798333364,
      "loss": 1.4804,
      "step": 7970
    },
    {
      "epoch": 0.8408227848101266,
      "grad_norm": 0.5141719579696655,
      "learning_rate": 0.00012180786913284024,
      "loss": 1.5097,
      "step": 7971
    },
    {
      "epoch": 0.8409282700421941,
      "grad_norm": 0.4613543152809143,
      "learning_rate": 0.00012163892492881463,
      "loss": 1.5212,
      "step": 7972
    },
    {
      "epoch": 0.8410337552742616,
      "grad_norm": 0.4344992935657501,
      "learning_rate": 0.00012147021504580839,
      "loss": 1.4745,
      "step": 7973
    },
    {
      "epoch": 0.8411392405063292,
      "grad_norm": 0.47057846188545227,
      "learning_rate": 0.00012130173915882475,
      "loss": 1.4612,
      "step": 7974
    },
    {
      "epoch": 0.8412447257383966,
      "grad_norm": 0.46060535311698914,
      "learning_rate": 0.00012113349694331764,
      "loss": 1.4989,
      "step": 7975
    },
    {
      "epoch": 0.8413502109704641,
      "grad_norm": 0.4393010139465332,
      "learning_rate": 0.0001209654880751909,
      "loss": 1.4976,
      "step": 7976
    },
    {
      "epoch": 0.8414556962025317,
      "grad_norm": 0.4202762246131897,
      "learning_rate": 0.00012079771223079825,
      "loss": 1.4643,
      "step": 7977
    },
    {
      "epoch": 0.8415611814345991,
      "grad_norm": 0.3892786204814911,
      "learning_rate": 0.00012063016908694192,
      "loss": 1.4769,
      "step": 7978
    },
    {
      "epoch": 0.8416666666666667,
      "grad_norm": 0.4344770908355713,
      "learning_rate": 0.0001204628583208727,
      "loss": 1.4901,
      "step": 7979
    },
    {
      "epoch": 0.8417721518987342,
      "grad_norm": 0.4278308153152466,
      "learning_rate": 0.00012029577961028894,
      "loss": 1.4604,
      "step": 7980
    },
    {
      "epoch": 0.8418776371308017,
      "grad_norm": 0.4394586980342865,
      "learning_rate": 0.00012012893263333586,
      "loss": 1.4804,
      "step": 7981
    },
    {
      "epoch": 0.8419831223628692,
      "grad_norm": 0.44682395458221436,
      "learning_rate": 0.00011996231706860537,
      "loss": 1.4723,
      "step": 7982
    },
    {
      "epoch": 0.8420886075949368,
      "grad_norm": 0.4192502498626709,
      "learning_rate": 0.00011979593259513487,
      "loss": 1.5238,
      "step": 7983
    },
    {
      "epoch": 0.8421940928270042,
      "grad_norm": 0.4484304189682007,
      "learning_rate": 0.00011962977889240712,
      "loss": 1.4605,
      "step": 7984
    },
    {
      "epoch": 0.8422995780590717,
      "grad_norm": 0.43339478969573975,
      "learning_rate": 0.00011946385564034942,
      "loss": 1.5264,
      "step": 7985
    },
    {
      "epoch": 0.8424050632911393,
      "grad_norm": 0.4308030605316162,
      "learning_rate": 0.00011929816251933285,
      "loss": 1.5039,
      "step": 7986
    },
    {
      "epoch": 0.8425105485232067,
      "grad_norm": 0.48961499333381653,
      "learning_rate": 0.00011913269921017203,
      "loss": 1.5141,
      "step": 7987
    },
    {
      "epoch": 0.8426160337552743,
      "grad_norm": 0.46717384457588196,
      "learning_rate": 0.00011896746539412405,
      "loss": 1.5326,
      "step": 7988
    },
    {
      "epoch": 0.8427215189873418,
      "grad_norm": 0.4560268521308899,
      "learning_rate": 0.00011880246075288824,
      "loss": 1.4736,
      "step": 7989
    },
    {
      "epoch": 0.8428270042194093,
      "grad_norm": 0.4474226236343384,
      "learning_rate": 0.00011863768496860542,
      "loss": 1.493,
      "step": 7990
    },
    {
      "epoch": 0.8429324894514768,
      "grad_norm": 0.42042869329452515,
      "learning_rate": 0.00011847313772385713,
      "loss": 1.4915,
      "step": 7991
    },
    {
      "epoch": 0.8430379746835444,
      "grad_norm": 0.434748113155365,
      "learning_rate": 0.00011830881870166533,
      "loss": 1.4714,
      "step": 7992
    },
    {
      "epoch": 0.8431434599156118,
      "grad_norm": 0.4294203221797943,
      "learning_rate": 0.00011814472758549143,
      "loss": 1.5106,
      "step": 7993
    },
    {
      "epoch": 0.8432489451476793,
      "grad_norm": 0.4985603094100952,
      "learning_rate": 0.00011798086405923605,
      "loss": 1.4789,
      "step": 7994
    },
    {
      "epoch": 0.8433544303797469,
      "grad_norm": 0.4327056407928467,
      "learning_rate": 0.0001178172278072382,
      "loss": 1.4834,
      "step": 7995
    },
    {
      "epoch": 0.8434599156118143,
      "grad_norm": 0.44394436478614807,
      "learning_rate": 0.00011765381851427454,
      "loss": 1.5142,
      "step": 7996
    },
    {
      "epoch": 0.8435654008438819,
      "grad_norm": 0.4221540093421936,
      "learning_rate": 0.00011749063586555921,
      "loss": 1.4887,
      "step": 7997
    },
    {
      "epoch": 0.8436708860759494,
      "grad_norm": 0.4521441161632538,
      "learning_rate": 0.00011732767954674264,
      "loss": 1.4921,
      "step": 7998
    },
    {
      "epoch": 0.8437763713080169,
      "grad_norm": 0.48578810691833496,
      "learning_rate": 0.00011716494924391148,
      "loss": 1.4957,
      "step": 7999
    },
    {
      "epoch": 0.8438818565400844,
      "grad_norm": 0.4415311813354492,
      "learning_rate": 0.00011700244464358777,
      "loss": 1.4783,
      "step": 8000
    },
    {
      "epoch": 0.8439873417721518,
      "grad_norm": 0.4834960401058197,
      "learning_rate": 0.00011684016543272816,
      "loss": 1.5062,
      "step": 8001
    },
    {
      "epoch": 0.8440928270042194,
      "grad_norm": 0.4440629482269287,
      "learning_rate": 0.00011667811129872368,
      "loss": 1.5352,
      "step": 8002
    },
    {
      "epoch": 0.8441983122362869,
      "grad_norm": 0.43222010135650635,
      "learning_rate": 0.00011651628192939872,
      "loss": 1.5149,
      "step": 8003
    },
    {
      "epoch": 0.8443037974683544,
      "grad_norm": 0.5237811207771301,
      "learning_rate": 0.0001163546770130109,
      "loss": 1.4665,
      "step": 8004
    },
    {
      "epoch": 0.8444092827004219,
      "grad_norm": 0.52041095495224,
      "learning_rate": 0.00011619329623825008,
      "loss": 1.4809,
      "step": 8005
    },
    {
      "epoch": 0.8445147679324895,
      "grad_norm": 0.4715472459793091,
      "learning_rate": 0.00011603213929423784,
      "loss": 1.5065,
      "step": 8006
    },
    {
      "epoch": 0.8446202531645569,
      "grad_norm": 0.4383639395236969,
      "learning_rate": 0.0001158712058705271,
      "loss": 1.5431,
      "step": 8007
    },
    {
      "epoch": 0.8447257383966245,
      "grad_norm": 0.5159831047058105,
      "learning_rate": 0.00011571049565710122,
      "loss": 1.4806,
      "step": 8008
    },
    {
      "epoch": 0.844831223628692,
      "grad_norm": 0.41650891304016113,
      "learning_rate": 0.00011555000834437363,
      "loss": 1.4621,
      "step": 8009
    },
    {
      "epoch": 0.8449367088607594,
      "grad_norm": 0.43178558349609375,
      "learning_rate": 0.00011538974362318715,
      "loss": 1.4795,
      "step": 8010
    },
    {
      "epoch": 0.845042194092827,
      "grad_norm": 0.40558838844299316,
      "learning_rate": 0.00011522970118481326,
      "loss": 1.4972,
      "step": 8011
    },
    {
      "epoch": 0.8451476793248945,
      "grad_norm": 0.41141316294670105,
      "learning_rate": 0.00011506988072095186,
      "loss": 1.4808,
      "step": 8012
    },
    {
      "epoch": 0.845253164556962,
      "grad_norm": 0.41347137093544006,
      "learning_rate": 0.00011491028192373022,
      "loss": 1.4857,
      "step": 8013
    },
    {
      "epoch": 0.8453586497890295,
      "grad_norm": 0.4615418016910553,
      "learning_rate": 0.00011475090448570282,
      "loss": 1.4939,
      "step": 8014
    },
    {
      "epoch": 0.8454641350210971,
      "grad_norm": 0.4312148988246918,
      "learning_rate": 0.00011459174809985047,
      "loss": 1.4876,
      "step": 8015
    },
    {
      "epoch": 0.8455696202531645,
      "grad_norm": 0.41781219840049744,
      "learning_rate": 0.00011443281245957975,
      "loss": 1.4842,
      "step": 8016
    },
    {
      "epoch": 0.8456751054852321,
      "grad_norm": 0.39488255977630615,
      "learning_rate": 0.00011427409725872264,
      "loss": 1.4404,
      "step": 8017
    },
    {
      "epoch": 0.8457805907172996,
      "grad_norm": 0.4170038402080536,
      "learning_rate": 0.00011411560219153552,
      "loss": 1.5243,
      "step": 8018
    },
    {
      "epoch": 0.845886075949367,
      "grad_norm": 0.4558310806751251,
      "learning_rate": 0.00011395732695269907,
      "loss": 1.5108,
      "step": 8019
    },
    {
      "epoch": 0.8459915611814346,
      "grad_norm": 0.4590003788471222,
      "learning_rate": 0.00011379927123731737,
      "loss": 1.4902,
      "step": 8020
    },
    {
      "epoch": 0.8460970464135021,
      "grad_norm": 0.41422438621520996,
      "learning_rate": 0.00011364143474091725,
      "loss": 1.5009,
      "step": 8021
    },
    {
      "epoch": 0.8462025316455696,
      "grad_norm": 0.41619187593460083,
      "learning_rate": 0.00011348381715944802,
      "loss": 1.4814,
      "step": 8022
    },
    {
      "epoch": 0.8463080168776371,
      "grad_norm": 0.42624762654304504,
      "learning_rate": 0.00011332641818928063,
      "loss": 1.4605,
      "step": 8023
    },
    {
      "epoch": 0.8464135021097047,
      "grad_norm": 0.4099412262439728,
      "learning_rate": 0.00011316923752720708,
      "loss": 1.5359,
      "step": 8024
    },
    {
      "epoch": 0.8465189873417721,
      "grad_norm": 0.4438430368900299,
      "learning_rate": 0.00011301227487044006,
      "loss": 1.5129,
      "step": 8025
    },
    {
      "epoch": 0.8466244725738397,
      "grad_norm": 0.4400119185447693,
      "learning_rate": 0.00011285552991661202,
      "loss": 1.4865,
      "step": 8026
    },
    {
      "epoch": 0.8467299578059072,
      "grad_norm": 0.4254751205444336,
      "learning_rate": 0.00011269900236377497,
      "loss": 1.4742,
      "step": 8027
    },
    {
      "epoch": 0.8468354430379746,
      "grad_norm": 0.453046590089798,
      "learning_rate": 0.0001125426919103997,
      "loss": 1.4691,
      "step": 8028
    },
    {
      "epoch": 0.8469409282700422,
      "grad_norm": 0.4189947545528412,
      "learning_rate": 0.00011238659825537507,
      "loss": 1.5128,
      "step": 8029
    },
    {
      "epoch": 0.8470464135021097,
      "grad_norm": 0.4589027464389801,
      "learning_rate": 0.0001122307210980077,
      "loss": 1.4782,
      "step": 8030
    },
    {
      "epoch": 0.8471518987341772,
      "grad_norm": 0.4292374849319458,
      "learning_rate": 0.00011207506013802118,
      "loss": 1.5112,
      "step": 8031
    },
    {
      "epoch": 0.8472573839662447,
      "grad_norm": 0.4513934254646301,
      "learning_rate": 0.00011191961507555564,
      "loss": 1.4644,
      "step": 8032
    },
    {
      "epoch": 0.8473628691983123,
      "grad_norm": 0.42146608233451843,
      "learning_rate": 0.00011176438561116717,
      "loss": 1.5161,
      "step": 8033
    },
    {
      "epoch": 0.8474683544303797,
      "grad_norm": 0.41826725006103516,
      "learning_rate": 0.00011160937144582693,
      "loss": 1.4671,
      "step": 8034
    },
    {
      "epoch": 0.8475738396624473,
      "grad_norm": 0.5036130547523499,
      "learning_rate": 0.00011145457228092119,
      "loss": 1.504,
      "step": 8035
    },
    {
      "epoch": 0.8476793248945148,
      "grad_norm": 0.43344515562057495,
      "learning_rate": 0.00011129998781824997,
      "loss": 1.5119,
      "step": 8036
    },
    {
      "epoch": 0.8477848101265822,
      "grad_norm": 0.44761455059051514,
      "learning_rate": 0.00011114561776002725,
      "loss": 1.4994,
      "step": 8037
    },
    {
      "epoch": 0.8478902953586498,
      "grad_norm": 0.44356489181518555,
      "learning_rate": 0.00011099146180887992,
      "loss": 1.5404,
      "step": 8038
    },
    {
      "epoch": 0.8479957805907173,
      "grad_norm": 0.4488217830657959,
      "learning_rate": 0.00011083751966784716,
      "loss": 1.5245,
      "step": 8039
    },
    {
      "epoch": 0.8481012658227848,
      "grad_norm": 0.40449029207229614,
      "learning_rate": 0.00011068379104038026,
      "loss": 1.5084,
      "step": 8040
    },
    {
      "epoch": 0.8482067510548523,
      "grad_norm": 0.4726657569408417,
      "learning_rate": 0.00011053027563034162,
      "loss": 1.4816,
      "step": 8041
    },
    {
      "epoch": 0.8483122362869199,
      "grad_norm": 0.45385703444480896,
      "learning_rate": 0.0001103769731420045,
      "loss": 1.4382,
      "step": 8042
    },
    {
      "epoch": 0.8484177215189873,
      "grad_norm": 0.469437837600708,
      "learning_rate": 0.00011022388328005234,
      "loss": 1.5195,
      "step": 8043
    },
    {
      "epoch": 0.8485232067510549,
      "grad_norm": 0.4331592619419098,
      "learning_rate": 0.00011007100574957802,
      "loss": 1.4807,
      "step": 8044
    },
    {
      "epoch": 0.8486286919831224,
      "grad_norm": 0.44067615270614624,
      "learning_rate": 0.00010991834025608366,
      "loss": 1.4719,
      "step": 8045
    },
    {
      "epoch": 0.8487341772151898,
      "grad_norm": 0.446806937456131,
      "learning_rate": 0.00010976588650547961,
      "loss": 1.4813,
      "step": 8046
    },
    {
      "epoch": 0.8488396624472574,
      "grad_norm": 0.4759073853492737,
      "learning_rate": 0.00010961364420408429,
      "loss": 1.4596,
      "step": 8047
    },
    {
      "epoch": 0.8489451476793249,
      "grad_norm": 0.4522542953491211,
      "learning_rate": 0.0001094616130586235,
      "loss": 1.4473,
      "step": 8048
    },
    {
      "epoch": 0.8490506329113924,
      "grad_norm": 0.45733293890953064,
      "learning_rate": 0.00010930979277622952,
      "loss": 1.541,
      "step": 8049
    },
    {
      "epoch": 0.8491561181434599,
      "grad_norm": 0.4479137361049652,
      "learning_rate": 0.00010915818306444116,
      "loss": 1.5032,
      "step": 8050
    },
    {
      "epoch": 0.8492616033755275,
      "grad_norm": 0.48625561594963074,
      "learning_rate": 0.00010900678363120256,
      "loss": 1.5059,
      "step": 8051
    },
    {
      "epoch": 0.8493670886075949,
      "grad_norm": 0.44374218583106995,
      "learning_rate": 0.00010885559418486318,
      "loss": 1.4674,
      "step": 8052
    },
    {
      "epoch": 0.8494725738396625,
      "grad_norm": 0.4311735928058624,
      "learning_rate": 0.00010870461443417695,
      "loss": 1.5055,
      "step": 8053
    },
    {
      "epoch": 0.84957805907173,
      "grad_norm": 0.42234835028648376,
      "learning_rate": 0.00010855384408830159,
      "loss": 1.4839,
      "step": 8054
    },
    {
      "epoch": 0.8496835443037974,
      "grad_norm": 0.44683894515037537,
      "learning_rate": 0.0001084032828567984,
      "loss": 1.509,
      "step": 8055
    },
    {
      "epoch": 0.849789029535865,
      "grad_norm": 0.4713289737701416,
      "learning_rate": 0.00010825293044963134,
      "loss": 1.4915,
      "step": 8056
    },
    {
      "epoch": 0.8498945147679325,
      "grad_norm": 0.42036235332489014,
      "learning_rate": 0.00010810278657716679,
      "loss": 1.4966,
      "step": 8057
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4391077756881714,
      "learning_rate": 0.00010795285095017282,
      "loss": 1.5194,
      "step": 8058
    },
    {
      "epoch": 0.8501054852320675,
      "grad_norm": 0.4358140826225281,
      "learning_rate": 0.00010780312327981853,
      "loss": 1.4412,
      "step": 8059
    },
    {
      "epoch": 0.8502109704641351,
      "grad_norm": 0.4823141098022461,
      "learning_rate": 0.00010765360327767384,
      "loss": 1.4924,
      "step": 8060
    },
    {
      "epoch": 0.8503164556962025,
      "grad_norm": 0.45511600375175476,
      "learning_rate": 0.00010750429065570842,
      "loss": 1.4973,
      "step": 8061
    },
    {
      "epoch": 0.8504219409282701,
      "grad_norm": 0.4123309552669525,
      "learning_rate": 0.0001073551851262917,
      "loss": 1.4953,
      "step": 8062
    },
    {
      "epoch": 0.8505274261603376,
      "grad_norm": 0.46833568811416626,
      "learning_rate": 0.000107206286402192,
      "loss": 1.4532,
      "step": 8063
    },
    {
      "epoch": 0.850632911392405,
      "grad_norm": 0.44981813430786133,
      "learning_rate": 0.00010705759419657583,
      "loss": 1.5022,
      "step": 8064
    },
    {
      "epoch": 0.8507383966244726,
      "grad_norm": 0.45205560326576233,
      "learning_rate": 0.00010690910822300775,
      "loss": 1.5429,
      "step": 8065
    },
    {
      "epoch": 0.85084388185654,
      "grad_norm": 0.4144775867462158,
      "learning_rate": 0.00010676082819544952,
      "loss": 1.4768,
      "step": 8066
    },
    {
      "epoch": 0.8509493670886076,
      "grad_norm": 0.4440843462944031,
      "learning_rate": 0.00010661275382825955,
      "loss": 1.4812,
      "step": 8067
    },
    {
      "epoch": 0.8510548523206751,
      "grad_norm": 0.4347103238105774,
      "learning_rate": 0.00010646488483619263,
      "loss": 1.5075,
      "step": 8068
    },
    {
      "epoch": 0.8511603375527426,
      "grad_norm": 0.42615094780921936,
      "learning_rate": 0.0001063172209343989,
      "loss": 1.5088,
      "step": 8069
    },
    {
      "epoch": 0.8512658227848101,
      "grad_norm": 0.42323145270347595,
      "learning_rate": 0.00010616976183842376,
      "loss": 1.4937,
      "step": 8070
    },
    {
      "epoch": 0.8513713080168777,
      "grad_norm": 0.4129416346549988,
      "learning_rate": 0.00010602250726420722,
      "loss": 1.4648,
      "step": 8071
    },
    {
      "epoch": 0.8514767932489451,
      "grad_norm": 0.4267805218696594,
      "learning_rate": 0.00010587545692808299,
      "loss": 1.4962,
      "step": 8072
    },
    {
      "epoch": 0.8515822784810126,
      "grad_norm": 0.44005367159843445,
      "learning_rate": 0.00010572861054677855,
      "loss": 1.4967,
      "step": 8073
    },
    {
      "epoch": 0.8516877637130802,
      "grad_norm": 0.43603193759918213,
      "learning_rate": 0.00010558196783741396,
      "loss": 1.5079,
      "step": 8074
    },
    {
      "epoch": 0.8517932489451476,
      "grad_norm": 0.42224976420402527,
      "learning_rate": 0.00010543552851750185,
      "loss": 1.5001,
      "step": 8075
    },
    {
      "epoch": 0.8518987341772152,
      "grad_norm": 0.44029930233955383,
      "learning_rate": 0.00010528929230494662,
      "loss": 1.5462,
      "step": 8076
    },
    {
      "epoch": 0.8520042194092827,
      "grad_norm": 0.4320455491542816,
      "learning_rate": 0.00010514325891804378,
      "loss": 1.4978,
      "step": 8077
    },
    {
      "epoch": 0.8521097046413502,
      "grad_norm": 0.4057701826095581,
      "learning_rate": 0.00010499742807547978,
      "loss": 1.4619,
      "step": 8078
    },
    {
      "epoch": 0.8522151898734177,
      "grad_norm": 0.4253707528114319,
      "learning_rate": 0.000104851799496331,
      "loss": 1.4994,
      "step": 8079
    },
    {
      "epoch": 0.8523206751054853,
      "grad_norm": 0.450650155544281,
      "learning_rate": 0.00010470637290006365,
      "loss": 1.5115,
      "step": 8080
    },
    {
      "epoch": 0.8524261603375527,
      "grad_norm": 0.42391568422317505,
      "learning_rate": 0.000104561148006533,
      "loss": 1.4405,
      "step": 8081
    },
    {
      "epoch": 0.8525316455696202,
      "grad_norm": 0.4349210858345032,
      "learning_rate": 0.00010441612453598276,
      "loss": 1.4248,
      "step": 8082
    },
    {
      "epoch": 0.8526371308016878,
      "grad_norm": 0.4500083327293396,
      "learning_rate": 0.0001042713022090448,
      "loss": 1.4697,
      "step": 8083
    },
    {
      "epoch": 0.8527426160337552,
      "grad_norm": 0.4362918734550476,
      "learning_rate": 0.00010412668074673832,
      "loss": 1.483,
      "step": 8084
    },
    {
      "epoch": 0.8528481012658228,
      "grad_norm": 0.4373531639575958,
      "learning_rate": 0.00010398225987046957,
      "loss": 1.4973,
      "step": 8085
    },
    {
      "epoch": 0.8529535864978903,
      "grad_norm": 0.4534114599227905,
      "learning_rate": 0.00010383803930203124,
      "loss": 1.4944,
      "step": 8086
    },
    {
      "epoch": 0.8530590717299578,
      "grad_norm": 0.43459632992744446,
      "learning_rate": 0.00010369401876360168,
      "loss": 1.4851,
      "step": 8087
    },
    {
      "epoch": 0.8531645569620253,
      "grad_norm": 0.4869888424873352,
      "learning_rate": 0.0001035501979777448,
      "loss": 1.508,
      "step": 8088
    },
    {
      "epoch": 0.8532700421940929,
      "grad_norm": 0.4642571806907654,
      "learning_rate": 0.00010340657666740917,
      "loss": 1.5368,
      "step": 8089
    },
    {
      "epoch": 0.8533755274261603,
      "grad_norm": 0.4289516806602478,
      "learning_rate": 0.00010326315455592764,
      "loss": 1.5203,
      "step": 8090
    },
    {
      "epoch": 0.8534810126582278,
      "grad_norm": 0.4206893742084503,
      "learning_rate": 0.00010311993136701693,
      "loss": 1.493,
      "step": 8091
    },
    {
      "epoch": 0.8535864978902954,
      "grad_norm": 0.44617852568626404,
      "learning_rate": 0.00010297690682477669,
      "loss": 1.4847,
      "step": 8092
    },
    {
      "epoch": 0.8536919831223628,
      "grad_norm": 0.4692865014076233,
      "learning_rate": 0.00010283408065368951,
      "loss": 1.4852,
      "step": 8093
    },
    {
      "epoch": 0.8537974683544304,
      "grad_norm": 0.40832483768463135,
      "learning_rate": 0.00010269145257861988,
      "loss": 1.4798,
      "step": 8094
    },
    {
      "epoch": 0.8539029535864979,
      "grad_norm": 0.42477303743362427,
      "learning_rate": 0.00010254902232481406,
      "loss": 1.4398,
      "step": 8095
    },
    {
      "epoch": 0.8540084388185654,
      "grad_norm": 0.4500899910926819,
      "learning_rate": 0.0001024067896178994,
      "loss": 1.4706,
      "step": 8096
    },
    {
      "epoch": 0.8541139240506329,
      "grad_norm": 0.42628321051597595,
      "learning_rate": 0.0001022647541838836,
      "loss": 1.5005,
      "step": 8097
    },
    {
      "epoch": 0.8542194092827005,
      "grad_norm": 0.4567301273345947,
      "learning_rate": 0.00010212291574915464,
      "loss": 1.4651,
      "step": 8098
    },
    {
      "epoch": 0.8543248945147679,
      "grad_norm": 0.466998428106308,
      "learning_rate": 0.00010198127404047976,
      "loss": 1.4847,
      "step": 8099
    },
    {
      "epoch": 0.8544303797468354,
      "grad_norm": 0.4482998847961426,
      "learning_rate": 0.0001018398287850053,
      "loss": 1.498,
      "step": 8100
    },
    {
      "epoch": 0.854535864978903,
      "grad_norm": 0.46892642974853516,
      "learning_rate": 0.00010169857971025608,
      "loss": 1.506,
      "step": 8101
    },
    {
      "epoch": 0.8546413502109704,
      "grad_norm": 0.43885427713394165,
      "learning_rate": 0.00010155752654413465,
      "loss": 1.5372,
      "step": 8102
    },
    {
      "epoch": 0.854746835443038,
      "grad_norm": 0.4367867112159729,
      "learning_rate": 0.00010141666901492119,
      "loss": 1.4863,
      "step": 8103
    },
    {
      "epoch": 0.8548523206751055,
      "grad_norm": 0.45450180768966675,
      "learning_rate": 0.00010127600685127249,
      "loss": 1.5218,
      "step": 8104
    },
    {
      "epoch": 0.854957805907173,
      "grad_norm": 0.40008068084716797,
      "learning_rate": 0.0001011355397822219,
      "loss": 1.5028,
      "step": 8105
    },
    {
      "epoch": 0.8550632911392405,
      "grad_norm": 0.4291958808898926,
      "learning_rate": 0.00010099526753717856,
      "loss": 1.4834,
      "step": 8106
    },
    {
      "epoch": 0.8551687763713081,
      "grad_norm": 0.43436047434806824,
      "learning_rate": 0.00010085518984592678,
      "loss": 1.4702,
      "step": 8107
    },
    {
      "epoch": 0.8552742616033755,
      "grad_norm": 0.48040902614593506,
      "learning_rate": 0.00010071530643862575,
      "loss": 1.4593,
      "step": 8108
    },
    {
      "epoch": 0.855379746835443,
      "grad_norm": 0.4501969516277313,
      "learning_rate": 0.00010057561704580898,
      "loss": 1.499,
      "step": 8109
    },
    {
      "epoch": 0.8554852320675106,
      "grad_norm": 0.4556061029434204,
      "learning_rate": 0.00010043612139838357,
      "loss": 1.5299,
      "step": 8110
    },
    {
      "epoch": 0.855590717299578,
      "grad_norm": 0.44409841299057007,
      "learning_rate": 0.00010029681922763002,
      "loss": 1.501,
      "step": 8111
    },
    {
      "epoch": 0.8556962025316456,
      "grad_norm": 0.4314638674259186,
      "learning_rate": 0.0001001577102652013,
      "loss": 1.5157,
      "step": 8112
    },
    {
      "epoch": 0.8558016877637131,
      "grad_norm": 0.45977160334587097,
      "learning_rate": 0.00010001879424312283,
      "loss": 1.445,
      "step": 8113
    },
    {
      "epoch": 0.8559071729957806,
      "grad_norm": 0.4589596092700958,
      "learning_rate": 9.988007089379161e-05,
      "loss": 1.5187,
      "step": 8114
    },
    {
      "epoch": 0.8560126582278481,
      "grad_norm": 0.42022794485092163,
      "learning_rate": 9.974153994997569e-05,
      "loss": 1.4916,
      "step": 8115
    },
    {
      "epoch": 0.8561181434599157,
      "grad_norm": 0.4470304548740387,
      "learning_rate": 9.960320114481391e-05,
      "loss": 1.4911,
      "step": 8116
    },
    {
      "epoch": 0.8562236286919831,
      "grad_norm": 0.4570592939853668,
      "learning_rate": 9.946505421181513e-05,
      "loss": 1.514,
      "step": 8117
    },
    {
      "epoch": 0.8563291139240506,
      "grad_norm": 0.41842663288116455,
      "learning_rate": 9.932709888485788e-05,
      "loss": 1.4606,
      "step": 8118
    },
    {
      "epoch": 0.8564345991561182,
      "grad_norm": 0.44179171323776245,
      "learning_rate": 9.918933489818986e-05,
      "loss": 1.4711,
      "step": 8119
    },
    {
      "epoch": 0.8565400843881856,
      "grad_norm": 0.4474397599697113,
      "learning_rate": 9.905176198642719e-05,
      "loss": 1.4614,
      "step": 8120
    },
    {
      "epoch": 0.8566455696202532,
      "grad_norm": 0.45716580748558044,
      "learning_rate": 9.891437988455427e-05,
      "loss": 1.4908,
      "step": 8121
    },
    {
      "epoch": 0.8567510548523207,
      "grad_norm": 0.4659733474254608,
      "learning_rate": 9.877718832792286e-05,
      "loss": 1.484,
      "step": 8122
    },
    {
      "epoch": 0.8568565400843882,
      "grad_norm": 0.4287756085395813,
      "learning_rate": 9.864018705225193e-05,
      "loss": 1.4818,
      "step": 8123
    },
    {
      "epoch": 0.8569620253164557,
      "grad_norm": 0.4370841085910797,
      "learning_rate": 9.850337579362703e-05,
      "loss": 1.4859,
      "step": 8124
    },
    {
      "epoch": 0.8570675105485233,
      "grad_norm": 0.43437787890434265,
      "learning_rate": 9.836675428849958e-05,
      "loss": 1.4864,
      "step": 8125
    },
    {
      "epoch": 0.8571729957805907,
      "grad_norm": 0.4429851174354553,
      "learning_rate": 9.823032227368671e-05,
      "loss": 1.4941,
      "step": 8126
    },
    {
      "epoch": 0.8572784810126582,
      "grad_norm": 0.4362345337867737,
      "learning_rate": 9.809407948637043e-05,
      "loss": 1.4925,
      "step": 8127
    },
    {
      "epoch": 0.8573839662447258,
      "grad_norm": 0.43401530385017395,
      "learning_rate": 9.79580256640974e-05,
      "loss": 1.4923,
      "step": 8128
    },
    {
      "epoch": 0.8574894514767932,
      "grad_norm": 0.4468132555484772,
      "learning_rate": 9.782216054477828e-05,
      "loss": 1.5269,
      "step": 8129
    },
    {
      "epoch": 0.8575949367088608,
      "grad_norm": 0.4374653398990631,
      "learning_rate": 9.76864838666871e-05,
      "loss": 1.4567,
      "step": 8130
    },
    {
      "epoch": 0.8577004219409282,
      "grad_norm": 0.4234811067581177,
      "learning_rate": 9.755099536846107e-05,
      "loss": 1.4524,
      "step": 8131
    },
    {
      "epoch": 0.8578059071729958,
      "grad_norm": 0.4564134180545807,
      "learning_rate": 9.741569478909979e-05,
      "loss": 1.4862,
      "step": 8132
    },
    {
      "epoch": 0.8579113924050633,
      "grad_norm": 0.44542422890663147,
      "learning_rate": 9.728058186796492e-05,
      "loss": 1.5007,
      "step": 8133
    },
    {
      "epoch": 0.8580168776371307,
      "grad_norm": 0.42091959714889526,
      "learning_rate": 9.714565634477964e-05,
      "loss": 1.4717,
      "step": 8134
    },
    {
      "epoch": 0.8581223628691983,
      "grad_norm": 0.4152686595916748,
      "learning_rate": 9.7010917959628e-05,
      "loss": 1.4923,
      "step": 8135
    },
    {
      "epoch": 0.8582278481012658,
      "grad_norm": 0.4614063799381256,
      "learning_rate": 9.687636645295472e-05,
      "loss": 1.5078,
      "step": 8136
    },
    {
      "epoch": 0.8583333333333333,
      "grad_norm": 0.42790308594703674,
      "learning_rate": 9.674200156556436e-05,
      "loss": 1.4799,
      "step": 8137
    },
    {
      "epoch": 0.8584388185654008,
      "grad_norm": 0.4314062297344208,
      "learning_rate": 9.660782303862107e-05,
      "loss": 1.4988,
      "step": 8138
    },
    {
      "epoch": 0.8585443037974684,
      "grad_norm": 0.46413227915763855,
      "learning_rate": 9.647383061364803e-05,
      "loss": 1.4954,
      "step": 8139
    },
    {
      "epoch": 0.8586497890295358,
      "grad_norm": 0.43041038513183594,
      "learning_rate": 9.634002403252676e-05,
      "loss": 1.4878,
      "step": 8140
    },
    {
      "epoch": 0.8587552742616034,
      "grad_norm": 0.41232773661613464,
      "learning_rate": 9.6206403037497e-05,
      "loss": 1.4551,
      "step": 8141
    },
    {
      "epoch": 0.8588607594936709,
      "grad_norm": 0.4198618233203888,
      "learning_rate": 9.60729673711558e-05,
      "loss": 1.4883,
      "step": 8142
    },
    {
      "epoch": 0.8589662447257383,
      "grad_norm": 0.4003661274909973,
      "learning_rate": 9.593971677645732e-05,
      "loss": 1.5032,
      "step": 8143
    },
    {
      "epoch": 0.8590717299578059,
      "grad_norm": 0.4171888530254364,
      "learning_rate": 9.58066509967123e-05,
      "loss": 1.47,
      "step": 8144
    },
    {
      "epoch": 0.8591772151898734,
      "grad_norm": 0.4533592164516449,
      "learning_rate": 9.56737697755873e-05,
      "loss": 1.4715,
      "step": 8145
    },
    {
      "epoch": 0.8592827004219409,
      "grad_norm": 0.4214538633823395,
      "learning_rate": 9.554107285710464e-05,
      "loss": 1.5216,
      "step": 8146
    },
    {
      "epoch": 0.8593881856540084,
      "grad_norm": 0.4249246418476105,
      "learning_rate": 9.540855998564147e-05,
      "loss": 1.4804,
      "step": 8147
    },
    {
      "epoch": 0.859493670886076,
      "grad_norm": 0.46606525778770447,
      "learning_rate": 9.527623090592962e-05,
      "loss": 1.4876,
      "step": 8148
    },
    {
      "epoch": 0.8595991561181434,
      "grad_norm": 0.4458865821361542,
      "learning_rate": 9.514408536305497e-05,
      "loss": 1.4607,
      "step": 8149
    },
    {
      "epoch": 0.859704641350211,
      "grad_norm": 0.42472967505455017,
      "learning_rate": 9.501212310245681e-05,
      "loss": 1.5226,
      "step": 8150
    },
    {
      "epoch": 0.8598101265822785,
      "grad_norm": 0.41407743096351624,
      "learning_rate": 9.488034386992769e-05,
      "loss": 1.4903,
      "step": 8151
    },
    {
      "epoch": 0.859915611814346,
      "grad_norm": 0.4198327362537384,
      "learning_rate": 9.474874741161267e-05,
      "loss": 1.4715,
      "step": 8152
    },
    {
      "epoch": 0.8600210970464135,
      "grad_norm": 0.4602448344230652,
      "learning_rate": 9.461733347400877e-05,
      "loss": 1.5297,
      "step": 8153
    },
    {
      "epoch": 0.860126582278481,
      "grad_norm": 0.44824647903442383,
      "learning_rate": 9.448610180396487e-05,
      "loss": 1.5045,
      "step": 8154
    },
    {
      "epoch": 0.8602320675105485,
      "grad_norm": 0.43564558029174805,
      "learning_rate": 9.435505214868068e-05,
      "loss": 1.5135,
      "step": 8155
    },
    {
      "epoch": 0.860337552742616,
      "grad_norm": 0.4624049961566925,
      "learning_rate": 9.422418425570673e-05,
      "loss": 1.516,
      "step": 8156
    },
    {
      "epoch": 0.8604430379746836,
      "grad_norm": 0.4127613604068756,
      "learning_rate": 9.409349787294371e-05,
      "loss": 1.4763,
      "step": 8157
    },
    {
      "epoch": 0.860548523206751,
      "grad_norm": 0.43876078724861145,
      "learning_rate": 9.396299274864176e-05,
      "loss": 1.4847,
      "step": 8158
    },
    {
      "epoch": 0.8606540084388186,
      "grad_norm": 0.40391695499420166,
      "learning_rate": 9.383266863140043e-05,
      "loss": 1.461,
      "step": 8159
    },
    {
      "epoch": 0.8607594936708861,
      "grad_norm": 0.4061271846294403,
      "learning_rate": 9.370252527016777e-05,
      "loss": 1.4961,
      "step": 8160
    },
    {
      "epoch": 0.8608649789029535,
      "grad_norm": 0.4176800847053528,
      "learning_rate": 9.357256241424012e-05,
      "loss": 1.4585,
      "step": 8161
    },
    {
      "epoch": 0.8609704641350211,
      "grad_norm": 0.4413154125213623,
      "learning_rate": 9.34427798132616e-05,
      "loss": 1.5116,
      "step": 8162
    },
    {
      "epoch": 0.8610759493670886,
      "grad_norm": 0.46794503927230835,
      "learning_rate": 9.331317721722338e-05,
      "loss": 1.5133,
      "step": 8163
    },
    {
      "epoch": 0.8611814345991561,
      "grad_norm": 0.4146137237548828,
      "learning_rate": 9.318375437646363e-05,
      "loss": 1.4636,
      "step": 8164
    },
    {
      "epoch": 0.8612869198312236,
      "grad_norm": 0.4689750075340271,
      "learning_rate": 9.305451104166652e-05,
      "loss": 1.4809,
      "step": 8165
    },
    {
      "epoch": 0.8613924050632912,
      "grad_norm": 0.3947790265083313,
      "learning_rate": 9.292544696386227e-05,
      "loss": 1.5177,
      "step": 8166
    },
    {
      "epoch": 0.8614978902953586,
      "grad_norm": 0.4780673682689667,
      "learning_rate": 9.279656189442629e-05,
      "loss": 1.494,
      "step": 8167
    },
    {
      "epoch": 0.8616033755274262,
      "grad_norm": 0.41865167021751404,
      "learning_rate": 9.266785558507876e-05,
      "loss": 1.5005,
      "step": 8168
    },
    {
      "epoch": 0.8617088607594937,
      "grad_norm": 0.43959128856658936,
      "learning_rate": 9.25393277878844e-05,
      "loss": 1.4673,
      "step": 8169
    },
    {
      "epoch": 0.8618143459915611,
      "grad_norm": 0.4763071537017822,
      "learning_rate": 9.241097825525163e-05,
      "loss": 1.4544,
      "step": 8170
    },
    {
      "epoch": 0.8619198312236287,
      "grad_norm": 0.4068686068058014,
      "learning_rate": 9.228280673993234e-05,
      "loss": 1.4956,
      "step": 8171
    },
    {
      "epoch": 0.8620253164556962,
      "grad_norm": 0.4224543273448944,
      "learning_rate": 9.215481299502145e-05,
      "loss": 1.4857,
      "step": 8172
    },
    {
      "epoch": 0.8621308016877637,
      "grad_norm": 0.4506922662258148,
      "learning_rate": 9.202699677395613e-05,
      "loss": 1.5292,
      "step": 8173
    },
    {
      "epoch": 0.8622362869198312,
      "grad_norm": 0.45534196496009827,
      "learning_rate": 9.189935783051572e-05,
      "loss": 1.4902,
      "step": 8174
    },
    {
      "epoch": 0.8623417721518988,
      "grad_norm": 0.48427584767341614,
      "learning_rate": 9.177189591882086e-05,
      "loss": 1.4493,
      "step": 8175
    },
    {
      "epoch": 0.8624472573839662,
      "grad_norm": 0.47470489144325256,
      "learning_rate": 9.164461079333342e-05,
      "loss": 1.4936,
      "step": 8176
    },
    {
      "epoch": 0.8625527426160338,
      "grad_norm": 0.4574161469936371,
      "learning_rate": 9.151750220885574e-05,
      "loss": 1.5146,
      "step": 8177
    },
    {
      "epoch": 0.8626582278481013,
      "grad_norm": 0.44946157932281494,
      "learning_rate": 9.139056992053017e-05,
      "loss": 1.4772,
      "step": 8178
    },
    {
      "epoch": 0.8627637130801687,
      "grad_norm": 0.4752751588821411,
      "learning_rate": 9.126381368383881e-05,
      "loss": 1.4814,
      "step": 8179
    },
    {
      "epoch": 0.8628691983122363,
      "grad_norm": 0.4545336067676544,
      "learning_rate": 9.113723325460276e-05,
      "loss": 1.5289,
      "step": 8180
    },
    {
      "epoch": 0.8629746835443038,
      "grad_norm": 0.46342095732688904,
      "learning_rate": 9.101082838898188e-05,
      "loss": 1.4866,
      "step": 8181
    },
    {
      "epoch": 0.8630801687763713,
      "grad_norm": 0.4117826223373413,
      "learning_rate": 9.088459884347427e-05,
      "loss": 1.4883,
      "step": 8182
    },
    {
      "epoch": 0.8631856540084388,
      "grad_norm": 0.4462929666042328,
      "learning_rate": 9.07585443749156e-05,
      "loss": 1.4792,
      "step": 8183
    },
    {
      "epoch": 0.8632911392405064,
      "grad_norm": 0.46005716919898987,
      "learning_rate": 9.063266474047899e-05,
      "loss": 1.5008,
      "step": 8184
    },
    {
      "epoch": 0.8633966244725738,
      "grad_norm": 0.42831310629844666,
      "learning_rate": 9.050695969767418e-05,
      "loss": 1.4929,
      "step": 8185
    },
    {
      "epoch": 0.8635021097046414,
      "grad_norm": 0.46223482489585876,
      "learning_rate": 9.038142900434736e-05,
      "loss": 1.4957,
      "step": 8186
    },
    {
      "epoch": 0.8636075949367089,
      "grad_norm": 0.4762667119503021,
      "learning_rate": 9.02560724186806e-05,
      "loss": 1.4786,
      "step": 8187
    },
    {
      "epoch": 0.8637130801687763,
      "grad_norm": 0.4461197853088379,
      "learning_rate": 9.01308896991912e-05,
      "loss": 1.4724,
      "step": 8188
    },
    {
      "epoch": 0.8638185654008439,
      "grad_norm": 0.4695635735988617,
      "learning_rate": 9.000588060473158e-05,
      "loss": 1.4956,
      "step": 8189
    },
    {
      "epoch": 0.8639240506329114,
      "grad_norm": 0.4212546646595001,
      "learning_rate": 8.988104489448849e-05,
      "loss": 1.48,
      "step": 8190
    },
    {
      "epoch": 0.8640295358649789,
      "grad_norm": 0.4889835715293884,
      "learning_rate": 8.975638232798275e-05,
      "loss": 1.4948,
      "step": 8191
    },
    {
      "epoch": 0.8641350210970464,
      "grad_norm": 0.42373472452163696,
      "learning_rate": 8.963189266506874e-05,
      "loss": 1.4802,
      "step": 8192
    },
    {
      "epoch": 0.864240506329114,
      "grad_norm": 0.43615368008613586,
      "learning_rate": 8.950757566593381e-05,
      "loss": 1.4774,
      "step": 8193
    },
    {
      "epoch": 0.8643459915611814,
      "grad_norm": 0.45238667726516724,
      "learning_rate": 8.938343109109802e-05,
      "loss": 1.5119,
      "step": 8194
    },
    {
      "epoch": 0.864451476793249,
      "grad_norm": 0.42367780208587646,
      "learning_rate": 8.925945870141361e-05,
      "loss": 1.4737,
      "step": 8195
    },
    {
      "epoch": 0.8645569620253165,
      "grad_norm": 0.4457396864891052,
      "learning_rate": 8.913565825806436e-05,
      "loss": 1.5324,
      "step": 8196
    },
    {
      "epoch": 0.864662447257384,
      "grad_norm": 0.4391903281211853,
      "learning_rate": 8.901202952256546e-05,
      "loss": 1.5069,
      "step": 8197
    },
    {
      "epoch": 0.8647679324894515,
      "grad_norm": 0.42450007796287537,
      "learning_rate": 8.88885722567627e-05,
      "loss": 1.4933,
      "step": 8198
    },
    {
      "epoch": 0.8648734177215189,
      "grad_norm": 0.40664932131767273,
      "learning_rate": 8.876528622283232e-05,
      "loss": 1.5231,
      "step": 8199
    },
    {
      "epoch": 0.8649789029535865,
      "grad_norm": 0.4372657835483551,
      "learning_rate": 8.864217118328042e-05,
      "loss": 1.4769,
      "step": 8200
    },
    {
      "epoch": 0.865084388185654,
      "grad_norm": 0.411454439163208,
      "learning_rate": 8.851922690094234e-05,
      "loss": 1.4916,
      "step": 8201
    },
    {
      "epoch": 0.8651898734177215,
      "grad_norm": 0.43713468313217163,
      "learning_rate": 8.839645313898257e-05,
      "loss": 1.5292,
      "step": 8202
    },
    {
      "epoch": 0.865295358649789,
      "grad_norm": 0.4268801510334015,
      "learning_rate": 8.827384966089387e-05,
      "loss": 1.4786,
      "step": 8203
    },
    {
      "epoch": 0.8654008438818566,
      "grad_norm": 0.45576122403144836,
      "learning_rate": 8.81514162304972e-05,
      "loss": 1.468,
      "step": 8204
    },
    {
      "epoch": 0.865506329113924,
      "grad_norm": 0.49695488810539246,
      "learning_rate": 8.802915261194108e-05,
      "loss": 1.4838,
      "step": 8205
    },
    {
      "epoch": 0.8656118143459915,
      "grad_norm": 0.45548149943351746,
      "learning_rate": 8.7907058569701e-05,
      "loss": 1.4867,
      "step": 8206
    },
    {
      "epoch": 0.8657172995780591,
      "grad_norm": 0.5141830444335938,
      "learning_rate": 8.778513386857931e-05,
      "loss": 1.5063,
      "step": 8207
    },
    {
      "epoch": 0.8658227848101265,
      "grad_norm": 0.4497213065624237,
      "learning_rate": 8.766337827370438e-05,
      "loss": 1.5157,
      "step": 8208
    },
    {
      "epoch": 0.8659282700421941,
      "grad_norm": 0.40176472067832947,
      "learning_rate": 8.754179155053052e-05,
      "loss": 1.461,
      "step": 8209
    },
    {
      "epoch": 0.8660337552742616,
      "grad_norm": 0.4418811500072479,
      "learning_rate": 8.742037346483729e-05,
      "loss": 1.477,
      "step": 8210
    },
    {
      "epoch": 0.8661392405063291,
      "grad_norm": 0.5150174498558044,
      "learning_rate": 8.7299123782729e-05,
      "loss": 1.4924,
      "step": 8211
    },
    {
      "epoch": 0.8662447257383966,
      "grad_norm": 0.45674073696136475,
      "learning_rate": 8.717804227063455e-05,
      "loss": 1.4728,
      "step": 8212
    },
    {
      "epoch": 0.8663502109704642,
      "grad_norm": 0.4809952974319458,
      "learning_rate": 8.705712869530661e-05,
      "loss": 1.4933,
      "step": 8213
    },
    {
      "epoch": 0.8664556962025316,
      "grad_norm": 0.4204850196838379,
      "learning_rate": 8.69363828238215e-05,
      "loss": 1.4935,
      "step": 8214
    },
    {
      "epoch": 0.8665611814345991,
      "grad_norm": 0.44213756918907166,
      "learning_rate": 8.681580442357857e-05,
      "loss": 1.5134,
      "step": 8215
    },
    {
      "epoch": 0.8666666666666667,
      "grad_norm": 0.43290165066719055,
      "learning_rate": 8.66953932622997e-05,
      "loss": 1.4845,
      "step": 8216
    },
    {
      "epoch": 0.8667721518987341,
      "grad_norm": 0.4767615497112274,
      "learning_rate": 8.657514910802906e-05,
      "loss": 1.5176,
      "step": 8217
    },
    {
      "epoch": 0.8668776371308017,
      "grad_norm": 0.45503509044647217,
      "learning_rate": 8.645507172913238e-05,
      "loss": 1.4719,
      "step": 8218
    },
    {
      "epoch": 0.8669831223628692,
      "grad_norm": 0.4896306097507477,
      "learning_rate": 8.63351608942968e-05,
      "loss": 1.5206,
      "step": 8219
    },
    {
      "epoch": 0.8670886075949367,
      "grad_norm": 0.471882164478302,
      "learning_rate": 8.62154163725303e-05,
      "loss": 1.5162,
      "step": 8220
    },
    {
      "epoch": 0.8671940928270042,
      "grad_norm": 0.44179245829582214,
      "learning_rate": 8.609583793316103e-05,
      "loss": 1.456,
      "step": 8221
    },
    {
      "epoch": 0.8672995780590718,
      "grad_norm": 0.42603135108947754,
      "learning_rate": 8.597642534583735e-05,
      "loss": 1.4733,
      "step": 8222
    },
    {
      "epoch": 0.8674050632911392,
      "grad_norm": 0.48473796248435974,
      "learning_rate": 8.585717838052689e-05,
      "loss": 1.5052,
      "step": 8223
    },
    {
      "epoch": 0.8675105485232067,
      "grad_norm": 0.41506513953208923,
      "learning_rate": 8.573809680751644e-05,
      "loss": 1.4988,
      "step": 8224
    },
    {
      "epoch": 0.8676160337552743,
      "grad_norm": 0.49285179376602173,
      "learning_rate": 8.561918039741144e-05,
      "loss": 1.524,
      "step": 8225
    },
    {
      "epoch": 0.8677215189873417,
      "grad_norm": 0.5024579167366028,
      "learning_rate": 8.550042892113534e-05,
      "loss": 1.4928,
      "step": 8226
    },
    {
      "epoch": 0.8678270042194093,
      "grad_norm": 0.4057333767414093,
      "learning_rate": 8.538184214992946e-05,
      "loss": 1.47,
      "step": 8227
    },
    {
      "epoch": 0.8679324894514768,
      "grad_norm": 0.42193830013275146,
      "learning_rate": 8.52634198553523e-05,
      "loss": 1.5053,
      "step": 8228
    },
    {
      "epoch": 0.8680379746835443,
      "grad_norm": 0.4237918257713318,
      "learning_rate": 8.514516180927926e-05,
      "loss": 1.4862,
      "step": 8229
    },
    {
      "epoch": 0.8681434599156118,
      "grad_norm": 0.45510023832321167,
      "learning_rate": 8.502706778390219e-05,
      "loss": 1.5171,
      "step": 8230
    },
    {
      "epoch": 0.8682489451476794,
      "grad_norm": 0.4293450713157654,
      "learning_rate": 8.490913755172874e-05,
      "loss": 1.4994,
      "step": 8231
    },
    {
      "epoch": 0.8683544303797468,
      "grad_norm": 0.42206162214279175,
      "learning_rate": 8.479137088558228e-05,
      "loss": 1.46,
      "step": 8232
    },
    {
      "epoch": 0.8684599156118143,
      "grad_norm": 0.41428276896476746,
      "learning_rate": 8.467376755860109e-05,
      "loss": 1.5131,
      "step": 8233
    },
    {
      "epoch": 0.8685654008438819,
      "grad_norm": 0.44789931178092957,
      "learning_rate": 8.455632734423823e-05,
      "loss": 1.4944,
      "step": 8234
    },
    {
      "epoch": 0.8686708860759493,
      "grad_norm": 0.42949697375297546,
      "learning_rate": 8.443905001626099e-05,
      "loss": 1.4808,
      "step": 8235
    },
    {
      "epoch": 0.8687763713080169,
      "grad_norm": 0.4153139293193817,
      "learning_rate": 8.432193534875027e-05,
      "loss": 1.4921,
      "step": 8236
    },
    {
      "epoch": 0.8688818565400844,
      "grad_norm": 0.4378165602684021,
      "learning_rate": 8.420498311610047e-05,
      "loss": 1.5136,
      "step": 8237
    },
    {
      "epoch": 0.8689873417721519,
      "grad_norm": 0.46940332651138306,
      "learning_rate": 8.408819309301891e-05,
      "loss": 1.5129,
      "step": 8238
    },
    {
      "epoch": 0.8690928270042194,
      "grad_norm": 0.4306110441684723,
      "learning_rate": 8.397156505452524e-05,
      "loss": 1.4427,
      "step": 8239
    },
    {
      "epoch": 0.869198312236287,
      "grad_norm": 0.4263593256473541,
      "learning_rate": 8.38550987759513e-05,
      "loss": 1.4869,
      "step": 8240
    },
    {
      "epoch": 0.8693037974683544,
      "grad_norm": 0.42645734548568726,
      "learning_rate": 8.373879403294042e-05,
      "loss": 1.5074,
      "step": 8241
    },
    {
      "epoch": 0.869409282700422,
      "grad_norm": 0.43293336033821106,
      "learning_rate": 8.36226506014472e-05,
      "loss": 1.4801,
      "step": 8242
    },
    {
      "epoch": 0.8695147679324895,
      "grad_norm": 0.4358377754688263,
      "learning_rate": 8.350666825773698e-05,
      "loss": 1.5189,
      "step": 8243
    },
    {
      "epoch": 0.8696202531645569,
      "grad_norm": 0.4333524703979492,
      "learning_rate": 8.339084677838533e-05,
      "loss": 1.4927,
      "step": 8244
    },
    {
      "epoch": 0.8697257383966245,
      "grad_norm": 0.44201257824897766,
      "learning_rate": 8.327518594027779e-05,
      "loss": 1.4753,
      "step": 8245
    },
    {
      "epoch": 0.869831223628692,
      "grad_norm": 0.39594489336013794,
      "learning_rate": 8.315968552060927e-05,
      "loss": 1.5101,
      "step": 8246
    },
    {
      "epoch": 0.8699367088607595,
      "grad_norm": 0.4450787305831909,
      "learning_rate": 8.304434529688379e-05,
      "loss": 1.5248,
      "step": 8247
    },
    {
      "epoch": 0.870042194092827,
      "grad_norm": 0.4281655251979828,
      "learning_rate": 8.292916504691398e-05,
      "loss": 1.5298,
      "step": 8248
    },
    {
      "epoch": 0.8701476793248946,
      "grad_norm": 0.4180588722229004,
      "learning_rate": 8.28141445488205e-05,
      "loss": 1.4985,
      "step": 8249
    },
    {
      "epoch": 0.870253164556962,
      "grad_norm": 0.43443799018859863,
      "learning_rate": 8.269928358103191e-05,
      "loss": 1.5182,
      "step": 8250
    },
    {
      "epoch": 0.8703586497890295,
      "grad_norm": 0.4392465353012085,
      "learning_rate": 8.258458192228395e-05,
      "loss": 1.5234,
      "step": 8251
    },
    {
      "epoch": 0.8704641350210971,
      "grad_norm": 0.42838016152381897,
      "learning_rate": 8.247003935161934e-05,
      "loss": 1.4707,
      "step": 8252
    },
    {
      "epoch": 0.8705696202531645,
      "grad_norm": 0.44824132323265076,
      "learning_rate": 8.235565564838727e-05,
      "loss": 1.4869,
      "step": 8253
    },
    {
      "epoch": 0.8706751054852321,
      "grad_norm": 0.44147592782974243,
      "learning_rate": 8.224143059224287e-05,
      "loss": 1.4861,
      "step": 8254
    },
    {
      "epoch": 0.8707805907172996,
      "grad_norm": 0.4234228730201721,
      "learning_rate": 8.2127363963147e-05,
      "loss": 1.5512,
      "step": 8255
    },
    {
      "epoch": 0.8708860759493671,
      "grad_norm": 0.4180000424385071,
      "learning_rate": 8.201345554136556e-05,
      "loss": 1.5041,
      "step": 8256
    },
    {
      "epoch": 0.8709915611814346,
      "grad_norm": 0.4617927074432373,
      "learning_rate": 8.189970510746936e-05,
      "loss": 1.4989,
      "step": 8257
    },
    {
      "epoch": 0.8710970464135022,
      "grad_norm": 0.4595535397529602,
      "learning_rate": 8.178611244233354e-05,
      "loss": 1.4776,
      "step": 8258
    },
    {
      "epoch": 0.8712025316455696,
      "grad_norm": 0.43817704916000366,
      "learning_rate": 8.167267732713705e-05,
      "loss": 1.4993,
      "step": 8259
    },
    {
      "epoch": 0.8713080168776371,
      "grad_norm": 0.449745237827301,
      "learning_rate": 8.155939954336243e-05,
      "loss": 1.4664,
      "step": 8260
    },
    {
      "epoch": 0.8714135021097047,
      "grad_norm": 0.417012482881546,
      "learning_rate": 8.144627887279526e-05,
      "loss": 1.4869,
      "step": 8261
    },
    {
      "epoch": 0.8715189873417721,
      "grad_norm": 0.46489375829696655,
      "learning_rate": 8.13333150975238e-05,
      "loss": 1.4831,
      "step": 8262
    },
    {
      "epoch": 0.8716244725738397,
      "grad_norm": 0.41472750902175903,
      "learning_rate": 8.122050799993858e-05,
      "loss": 1.5064,
      "step": 8263
    },
    {
      "epoch": 0.8717299578059071,
      "grad_norm": 0.4289701581001282,
      "learning_rate": 8.110785736273183e-05,
      "loss": 1.467,
      "step": 8264
    },
    {
      "epoch": 0.8718354430379747,
      "grad_norm": 0.4322354793548584,
      "learning_rate": 8.099536296889734e-05,
      "loss": 1.4674,
      "step": 8265
    },
    {
      "epoch": 0.8719409282700422,
      "grad_norm": 0.4393194615840912,
      "learning_rate": 8.08830246017297e-05,
      "loss": 1.4951,
      "step": 8266
    },
    {
      "epoch": 0.8720464135021097,
      "grad_norm": 0.4156050980091095,
      "learning_rate": 8.077084204482424e-05,
      "loss": 1.4897,
      "step": 8267
    },
    {
      "epoch": 0.8721518987341772,
      "grad_norm": 0.509539783000946,
      "learning_rate": 8.065881508207636e-05,
      "loss": 1.47,
      "step": 8268
    },
    {
      "epoch": 0.8722573839662447,
      "grad_norm": 0.4308638572692871,
      "learning_rate": 8.054694349768114e-05,
      "loss": 1.5139,
      "step": 8269
    },
    {
      "epoch": 0.8723628691983122,
      "grad_norm": 0.41226568818092346,
      "learning_rate": 8.043522707613312e-05,
      "loss": 1.5157,
      "step": 8270
    },
    {
      "epoch": 0.8724683544303797,
      "grad_norm": 0.4736034572124481,
      "learning_rate": 8.032366560222553e-05,
      "loss": 1.5354,
      "step": 8271
    },
    {
      "epoch": 0.8725738396624473,
      "grad_norm": 0.42093440890312195,
      "learning_rate": 8.021225886105027e-05,
      "loss": 1.4631,
      "step": 8272
    },
    {
      "epoch": 0.8726793248945147,
      "grad_norm": 0.42206841707229614,
      "learning_rate": 8.010100663799726e-05,
      "loss": 1.4761,
      "step": 8273
    },
    {
      "epoch": 0.8727848101265823,
      "grad_norm": 0.4045555591583252,
      "learning_rate": 7.998990871875402e-05,
      "loss": 1.475,
      "step": 8274
    },
    {
      "epoch": 0.8728902953586498,
      "grad_norm": 0.46745604276657104,
      "learning_rate": 7.987896488930541e-05,
      "loss": 1.4565,
      "step": 8275
    },
    {
      "epoch": 0.8729957805907173,
      "grad_norm": 0.4449433982372284,
      "learning_rate": 7.976817493593301e-05,
      "loss": 1.4939,
      "step": 8276
    },
    {
      "epoch": 0.8731012658227848,
      "grad_norm": 0.410609632730484,
      "learning_rate": 7.965753864521492e-05,
      "loss": 1.4703,
      "step": 8277
    },
    {
      "epoch": 0.8732067510548523,
      "grad_norm": 0.3928322494029999,
      "learning_rate": 7.954705580402525e-05,
      "loss": 1.4539,
      "step": 8278
    },
    {
      "epoch": 0.8733122362869198,
      "grad_norm": 0.45473843812942505,
      "learning_rate": 7.943672619953359e-05,
      "loss": 1.4915,
      "step": 8279
    },
    {
      "epoch": 0.8734177215189873,
      "grad_norm": 0.4489142596721649,
      "learning_rate": 7.932654961920486e-05,
      "loss": 1.445,
      "step": 8280
    },
    {
      "epoch": 0.8735232067510549,
      "grad_norm": 0.4001033902168274,
      "learning_rate": 7.921652585079873e-05,
      "loss": 1.477,
      "step": 8281
    },
    {
      "epoch": 0.8736286919831223,
      "grad_norm": 0.4195241928100586,
      "learning_rate": 7.910665468236916e-05,
      "loss": 1.5,
      "step": 8282
    },
    {
      "epoch": 0.8737341772151899,
      "grad_norm": 0.42359504103660583,
      "learning_rate": 7.899693590226418e-05,
      "loss": 1.494,
      "step": 8283
    },
    {
      "epoch": 0.8738396624472574,
      "grad_norm": 0.4651046097278595,
      "learning_rate": 7.888736929912525e-05,
      "loss": 1.4526,
      "step": 8284
    },
    {
      "epoch": 0.8739451476793249,
      "grad_norm": 0.4119933545589447,
      "learning_rate": 7.877795466188711e-05,
      "loss": 1.4588,
      "step": 8285
    },
    {
      "epoch": 0.8740506329113924,
      "grad_norm": 0.4278581738471985,
      "learning_rate": 7.866869177977722e-05,
      "loss": 1.4648,
      "step": 8286
    },
    {
      "epoch": 0.87415611814346,
      "grad_norm": 0.4105331599712372,
      "learning_rate": 7.855958044231526e-05,
      "loss": 1.5003,
      "step": 8287
    },
    {
      "epoch": 0.8742616033755274,
      "grad_norm": 0.41706085205078125,
      "learning_rate": 7.845062043931299e-05,
      "loss": 1.4973,
      "step": 8288
    },
    {
      "epoch": 0.8743670886075949,
      "grad_norm": 0.40420183539390564,
      "learning_rate": 7.834181156087357e-05,
      "loss": 1.4884,
      "step": 8289
    },
    {
      "epoch": 0.8744725738396625,
      "grad_norm": 0.421173632144928,
      "learning_rate": 7.823315359739135e-05,
      "loss": 1.4451,
      "step": 8290
    },
    {
      "epoch": 0.8745780590717299,
      "grad_norm": 0.4299216866493225,
      "learning_rate": 7.812464633955146e-05,
      "loss": 1.5048,
      "step": 8291
    },
    {
      "epoch": 0.8746835443037975,
      "grad_norm": 0.48581114411354065,
      "learning_rate": 7.801628957832916e-05,
      "loss": 1.5211,
      "step": 8292
    },
    {
      "epoch": 0.874789029535865,
      "grad_norm": 0.4901425242424011,
      "learning_rate": 7.790808310498984e-05,
      "loss": 1.4738,
      "step": 8293
    },
    {
      "epoch": 0.8748945147679325,
      "grad_norm": 0.4534350633621216,
      "learning_rate": 7.78000267110882e-05,
      "loss": 1.4988,
      "step": 8294
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.4104546010494232,
      "learning_rate": 7.769212018846815e-05,
      "loss": 1.5203,
      "step": 8295
    },
    {
      "epoch": 0.8751054852320675,
      "grad_norm": 0.3996504545211792,
      "learning_rate": 7.758436332926238e-05,
      "loss": 1.4781,
      "step": 8296
    },
    {
      "epoch": 0.875210970464135,
      "grad_norm": 0.41142135858535767,
      "learning_rate": 7.747675592589168e-05,
      "loss": 1.516,
      "step": 8297
    },
    {
      "epoch": 0.8753164556962025,
      "grad_norm": 0.4289790689945221,
      "learning_rate": 7.736929777106499e-05,
      "loss": 1.5019,
      "step": 8298
    },
    {
      "epoch": 0.8754219409282701,
      "grad_norm": 0.42494845390319824,
      "learning_rate": 7.726198865777852e-05,
      "loss": 1.5232,
      "step": 8299
    },
    {
      "epoch": 0.8755274261603375,
      "grad_norm": 0.40735816955566406,
      "learning_rate": 7.715482837931577e-05,
      "loss": 1.4852,
      "step": 8300
    },
    {
      "epoch": 0.8756329113924051,
      "grad_norm": 0.41134464740753174,
      "learning_rate": 7.704781672924692e-05,
      "loss": 1.4973,
      "step": 8301
    },
    {
      "epoch": 0.8757383966244726,
      "grad_norm": 0.46109405159950256,
      "learning_rate": 7.694095350142833e-05,
      "loss": 1.4769,
      "step": 8302
    },
    {
      "epoch": 0.87584388185654,
      "grad_norm": 0.44002947211265564,
      "learning_rate": 7.683423849000246e-05,
      "loss": 1.4615,
      "step": 8303
    },
    {
      "epoch": 0.8759493670886076,
      "grad_norm": 0.42934295535087585,
      "learning_rate": 7.672767148939714e-05,
      "loss": 1.4892,
      "step": 8304
    },
    {
      "epoch": 0.8760548523206751,
      "grad_norm": 0.49174800515174866,
      "learning_rate": 7.66212522943254e-05,
      "loss": 1.4637,
      "step": 8305
    },
    {
      "epoch": 0.8761603375527426,
      "grad_norm": 0.42006009817123413,
      "learning_rate": 7.651498069978505e-05,
      "loss": 1.4986,
      "step": 8306
    },
    {
      "epoch": 0.8762658227848101,
      "grad_norm": 0.4598025977611542,
      "learning_rate": 7.640885650105804e-05,
      "loss": 1.4671,
      "step": 8307
    },
    {
      "epoch": 0.8763713080168777,
      "grad_norm": 0.4459536373615265,
      "learning_rate": 7.630287949371051e-05,
      "loss": 1.5022,
      "step": 8308
    },
    {
      "epoch": 0.8764767932489451,
      "grad_norm": 0.4235796630382538,
      "learning_rate": 7.61970494735919e-05,
      "loss": 1.4598,
      "step": 8309
    },
    {
      "epoch": 0.8765822784810127,
      "grad_norm": 0.4373488128185272,
      "learning_rate": 7.6091366236835e-05,
      "loss": 1.4938,
      "step": 8310
    },
    {
      "epoch": 0.8766877637130802,
      "grad_norm": 0.4334496557712555,
      "learning_rate": 7.598582957985526e-05,
      "loss": 1.5341,
      "step": 8311
    },
    {
      "epoch": 0.8767932489451477,
      "grad_norm": 0.4387611448764801,
      "learning_rate": 7.588043929935049e-05,
      "loss": 1.5054,
      "step": 8312
    },
    {
      "epoch": 0.8768987341772152,
      "grad_norm": 0.45723050832748413,
      "learning_rate": 7.577519519230054e-05,
      "loss": 1.5161,
      "step": 8313
    },
    {
      "epoch": 0.8770042194092827,
      "grad_norm": 0.46606290340423584,
      "learning_rate": 7.567009705596672e-05,
      "loss": 1.4851,
      "step": 8314
    },
    {
      "epoch": 0.8771097046413502,
      "grad_norm": 0.474518358707428,
      "learning_rate": 7.556514468789169e-05,
      "loss": 1.4996,
      "step": 8315
    },
    {
      "epoch": 0.8772151898734177,
      "grad_norm": 0.4362832307815552,
      "learning_rate": 7.546033788589884e-05,
      "loss": 1.4971,
      "step": 8316
    },
    {
      "epoch": 0.8773206751054853,
      "grad_norm": 0.45857396721839905,
      "learning_rate": 7.53556764480919e-05,
      "loss": 1.4835,
      "step": 8317
    },
    {
      "epoch": 0.8774261603375527,
      "grad_norm": 0.39667078852653503,
      "learning_rate": 7.525116017285479e-05,
      "loss": 1.4921,
      "step": 8318
    },
    {
      "epoch": 0.8775316455696203,
      "grad_norm": 0.418235719203949,
      "learning_rate": 7.514678885885086e-05,
      "loss": 1.491,
      "step": 8319
    },
    {
      "epoch": 0.8776371308016878,
      "grad_norm": 0.4538114368915558,
      "learning_rate": 7.504256230502289e-05,
      "loss": 1.4799,
      "step": 8320
    },
    {
      "epoch": 0.8777426160337553,
      "grad_norm": 0.409012109041214,
      "learning_rate": 7.493848031059248e-05,
      "loss": 1.4632,
      "step": 8321
    },
    {
      "epoch": 0.8778481012658228,
      "grad_norm": 0.45271778106689453,
      "learning_rate": 7.483454267505959e-05,
      "loss": 1.5076,
      "step": 8322
    },
    {
      "epoch": 0.8779535864978903,
      "grad_norm": 0.4084085524082184,
      "learning_rate": 7.473074919820243e-05,
      "loss": 1.5085,
      "step": 8323
    },
    {
      "epoch": 0.8780590717299578,
      "grad_norm": 0.4420888125896454,
      "learning_rate": 7.462709968007676e-05,
      "loss": 1.5149,
      "step": 8324
    },
    {
      "epoch": 0.8781645569620253,
      "grad_norm": 0.4088989198207855,
      "learning_rate": 7.452359392101578e-05,
      "loss": 1.4962,
      "step": 8325
    },
    {
      "epoch": 0.8782700421940929,
      "grad_norm": 0.43204084038734436,
      "learning_rate": 7.442023172162959e-05,
      "loss": 1.4997,
      "step": 8326
    },
    {
      "epoch": 0.8783755274261603,
      "grad_norm": 0.4234655499458313,
      "learning_rate": 7.431701288280477e-05,
      "loss": 1.5129,
      "step": 8327
    },
    {
      "epoch": 0.8784810126582279,
      "grad_norm": 0.4055323600769043,
      "learning_rate": 7.421393720570416e-05,
      "loss": 1.4775,
      "step": 8328
    },
    {
      "epoch": 0.8785864978902953,
      "grad_norm": 0.4314819574356079,
      "learning_rate": 7.411100449176634e-05,
      "loss": 1.4602,
      "step": 8329
    },
    {
      "epoch": 0.8786919831223629,
      "grad_norm": 0.43039757013320923,
      "learning_rate": 7.400821454270524e-05,
      "loss": 1.497,
      "step": 8330
    },
    {
      "epoch": 0.8787974683544304,
      "grad_norm": 0.4349774420261383,
      "learning_rate": 7.390556716050994e-05,
      "loss": 1.5127,
      "step": 8331
    },
    {
      "epoch": 0.8789029535864978,
      "grad_norm": 0.42883288860321045,
      "learning_rate": 7.380306214744398e-05,
      "loss": 1.5239,
      "step": 8332
    },
    {
      "epoch": 0.8790084388185654,
      "grad_norm": 0.3942241370677948,
      "learning_rate": 7.370069930604528e-05,
      "loss": 1.4702,
      "step": 8333
    },
    {
      "epoch": 0.8791139240506329,
      "grad_norm": 0.41544026136398315,
      "learning_rate": 7.359847843912566e-05,
      "loss": 1.4862,
      "step": 8334
    },
    {
      "epoch": 0.8792194092827004,
      "grad_norm": 0.4057771861553192,
      "learning_rate": 7.349639934977028e-05,
      "loss": 1.5055,
      "step": 8335
    },
    {
      "epoch": 0.8793248945147679,
      "grad_norm": 0.40864747762680054,
      "learning_rate": 7.33944618413376e-05,
      "loss": 1.4702,
      "step": 8336
    },
    {
      "epoch": 0.8794303797468355,
      "grad_norm": 0.4636353552341461,
      "learning_rate": 7.329266571745865e-05,
      "loss": 1.4882,
      "step": 8337
    },
    {
      "epoch": 0.8795358649789029,
      "grad_norm": 0.4824613928794861,
      "learning_rate": 7.319101078203692e-05,
      "loss": 1.4888,
      "step": 8338
    },
    {
      "epoch": 0.8796413502109705,
      "grad_norm": 0.4449959993362427,
      "learning_rate": 7.308949683924792e-05,
      "loss": 1.4091,
      "step": 8339
    },
    {
      "epoch": 0.879746835443038,
      "grad_norm": 0.4504435658454895,
      "learning_rate": 7.29881236935386e-05,
      "loss": 1.4652,
      "step": 8340
    },
    {
      "epoch": 0.8798523206751054,
      "grad_norm": 0.40106186270713806,
      "learning_rate": 7.288689114962734e-05,
      "loss": 1.4853,
      "step": 8341
    },
    {
      "epoch": 0.879957805907173,
      "grad_norm": 0.41365739703178406,
      "learning_rate": 7.278579901250316e-05,
      "loss": 1.4698,
      "step": 8342
    },
    {
      "epoch": 0.8800632911392405,
      "grad_norm": 0.40213650465011597,
      "learning_rate": 7.268484708742574e-05,
      "loss": 1.4445,
      "step": 8343
    },
    {
      "epoch": 0.880168776371308,
      "grad_norm": 0.40378764271736145,
      "learning_rate": 7.258403517992476e-05,
      "loss": 1.4589,
      "step": 8344
    },
    {
      "epoch": 0.8802742616033755,
      "grad_norm": 0.4202226400375366,
      "learning_rate": 7.248336309579965e-05,
      "loss": 1.4998,
      "step": 8345
    },
    {
      "epoch": 0.8803797468354431,
      "grad_norm": 0.44546592235565186,
      "learning_rate": 7.238283064111919e-05,
      "loss": 1.4763,
      "step": 8346
    },
    {
      "epoch": 0.8804852320675105,
      "grad_norm": 0.45787084102630615,
      "learning_rate": 7.228243762222109e-05,
      "loss": 1.5235,
      "step": 8347
    },
    {
      "epoch": 0.880590717299578,
      "grad_norm": 0.46078041195869446,
      "learning_rate": 7.218218384571176e-05,
      "loss": 1.4947,
      "step": 8348
    },
    {
      "epoch": 0.8806962025316456,
      "grad_norm": 0.41745516657829285,
      "learning_rate": 7.208206911846581e-05,
      "loss": 1.4762,
      "step": 8349
    },
    {
      "epoch": 0.880801687763713,
      "grad_norm": 0.46912646293640137,
      "learning_rate": 7.198209324762562e-05,
      "loss": 1.5113,
      "step": 8350
    },
    {
      "epoch": 0.8809071729957806,
      "grad_norm": 0.41433611512184143,
      "learning_rate": 7.188225604060121e-05,
      "loss": 1.5035,
      "step": 8351
    },
    {
      "epoch": 0.8810126582278481,
      "grad_norm": 0.42304790019989014,
      "learning_rate": 7.178255730506956e-05,
      "loss": 1.4617,
      "step": 8352
    },
    {
      "epoch": 0.8811181434599156,
      "grad_norm": 0.41195395588874817,
      "learning_rate": 7.16829968489745e-05,
      "loss": 1.4362,
      "step": 8353
    },
    {
      "epoch": 0.8812236286919831,
      "grad_norm": 0.4199661314487457,
      "learning_rate": 7.158357448052624e-05,
      "loss": 1.4761,
      "step": 8354
    },
    {
      "epoch": 0.8813291139240507,
      "grad_norm": 0.4273452162742615,
      "learning_rate": 7.148429000820093e-05,
      "loss": 1.4561,
      "step": 8355
    },
    {
      "epoch": 0.8814345991561181,
      "grad_norm": 0.415198415517807,
      "learning_rate": 7.138514324074043e-05,
      "loss": 1.4841,
      "step": 8356
    },
    {
      "epoch": 0.8815400843881857,
      "grad_norm": 0.43732890486717224,
      "learning_rate": 7.128613398715179e-05,
      "loss": 1.5199,
      "step": 8357
    },
    {
      "epoch": 0.8816455696202532,
      "grad_norm": 0.4501116871833801,
      "learning_rate": 7.118726205670702e-05,
      "loss": 1.4847,
      "step": 8358
    },
    {
      "epoch": 0.8817510548523206,
      "grad_norm": 0.4412800669670105,
      "learning_rate": 7.10885272589427e-05,
      "loss": 1.5215,
      "step": 8359
    },
    {
      "epoch": 0.8818565400843882,
      "grad_norm": 0.43446093797683716,
      "learning_rate": 7.098992940365946e-05,
      "loss": 1.4707,
      "step": 8360
    },
    {
      "epoch": 0.8819620253164557,
      "grad_norm": 0.44392797350883484,
      "learning_rate": 7.089146830092187e-05,
      "loss": 1.4917,
      "step": 8361
    },
    {
      "epoch": 0.8820675105485232,
      "grad_norm": 0.4270941913127899,
      "learning_rate": 7.079314376105778e-05,
      "loss": 1.5309,
      "step": 8362
    },
    {
      "epoch": 0.8821729957805907,
      "grad_norm": 0.41732388734817505,
      "learning_rate": 7.069495559465825e-05,
      "loss": 1.4886,
      "step": 8363
    },
    {
      "epoch": 0.8822784810126583,
      "grad_norm": 0.4092230200767517,
      "learning_rate": 7.059690361257703e-05,
      "loss": 1.5017,
      "step": 8364
    },
    {
      "epoch": 0.8823839662447257,
      "grad_norm": 0.4176844656467438,
      "learning_rate": 7.049898762593007e-05,
      "loss": 1.5066,
      "step": 8365
    },
    {
      "epoch": 0.8824894514767933,
      "grad_norm": 0.47050392627716064,
      "learning_rate": 7.04012074460955e-05,
      "loss": 1.4678,
      "step": 8366
    },
    {
      "epoch": 0.8825949367088608,
      "grad_norm": 0.45269033312797546,
      "learning_rate": 7.030356288471289e-05,
      "loss": 1.4788,
      "step": 8367
    },
    {
      "epoch": 0.8827004219409282,
      "grad_norm": 0.4728701412677765,
      "learning_rate": 7.020605375368314e-05,
      "loss": 1.49,
      "step": 8368
    },
    {
      "epoch": 0.8828059071729958,
      "grad_norm": 0.4621485769748688,
      "learning_rate": 7.010867986516811e-05,
      "loss": 1.4744,
      "step": 8369
    },
    {
      "epoch": 0.8829113924050633,
      "grad_norm": 0.4135444462299347,
      "learning_rate": 7.001144103159e-05,
      "loss": 1.4441,
      "step": 8370
    },
    {
      "epoch": 0.8830168776371308,
      "grad_norm": 0.4157642424106598,
      "learning_rate": 6.991433706563134e-05,
      "loss": 1.4749,
      "step": 8371
    },
    {
      "epoch": 0.8831223628691983,
      "grad_norm": 0.45952558517456055,
      "learning_rate": 6.981736778023443e-05,
      "loss": 1.5109,
      "step": 8372
    },
    {
      "epoch": 0.8832278481012659,
      "grad_norm": 0.4747614562511444,
      "learning_rate": 6.972053298860092e-05,
      "loss": 1.4749,
      "step": 8373
    },
    {
      "epoch": 0.8833333333333333,
      "grad_norm": 0.4265487790107727,
      "learning_rate": 6.962383250419169e-05,
      "loss": 1.4799,
      "step": 8374
    },
    {
      "epoch": 0.8834388185654009,
      "grad_norm": 0.46043136715888977,
      "learning_rate": 6.952726614072621e-05,
      "loss": 1.5161,
      "step": 8375
    },
    {
      "epoch": 0.8835443037974684,
      "grad_norm": 0.44654718041419983,
      "learning_rate": 6.94308337121824e-05,
      "loss": 1.5292,
      "step": 8376
    },
    {
      "epoch": 0.8836497890295358,
      "grad_norm": 0.44159990549087524,
      "learning_rate": 6.93345350327962e-05,
      "loss": 1.4887,
      "step": 8377
    },
    {
      "epoch": 0.8837552742616034,
      "grad_norm": 0.44495388865470886,
      "learning_rate": 6.923836991706108e-05,
      "loss": 1.4829,
      "step": 8378
    },
    {
      "epoch": 0.8838607594936709,
      "grad_norm": 0.45411524176597595,
      "learning_rate": 6.914233817972799e-05,
      "loss": 1.476,
      "step": 8379
    },
    {
      "epoch": 0.8839662447257384,
      "grad_norm": 0.4120056927204132,
      "learning_rate": 6.904643963580461e-05,
      "loss": 1.5208,
      "step": 8380
    },
    {
      "epoch": 0.8840717299578059,
      "grad_norm": 0.43202728033065796,
      "learning_rate": 6.895067410055536e-05,
      "loss": 1.4677,
      "step": 8381
    },
    {
      "epoch": 0.8841772151898735,
      "grad_norm": 0.4222124516963959,
      "learning_rate": 6.885504138950084e-05,
      "loss": 1.5084,
      "step": 8382
    },
    {
      "epoch": 0.8842827004219409,
      "grad_norm": 0.48025941848754883,
      "learning_rate": 6.875954131841743e-05,
      "loss": 1.483,
      "step": 8383
    },
    {
      "epoch": 0.8843881856540085,
      "grad_norm": 0.4524250328540802,
      "learning_rate": 6.866417370333717e-05,
      "loss": 1.4878,
      "step": 8384
    },
    {
      "epoch": 0.884493670886076,
      "grad_norm": 0.41338518261909485,
      "learning_rate": 6.856893836054713e-05,
      "loss": 1.5055,
      "step": 8385
    },
    {
      "epoch": 0.8845991561181434,
      "grad_norm": 0.42971885204315186,
      "learning_rate": 6.847383510658925e-05,
      "loss": 1.4931,
      "step": 8386
    },
    {
      "epoch": 0.884704641350211,
      "grad_norm": 0.48985370993614197,
      "learning_rate": 6.837886375825995e-05,
      "loss": 1.4849,
      "step": 8387
    },
    {
      "epoch": 0.8848101265822785,
      "grad_norm": 0.42879849672317505,
      "learning_rate": 6.828402413260965e-05,
      "loss": 1.4597,
      "step": 8388
    },
    {
      "epoch": 0.884915611814346,
      "grad_norm": 0.39670246839523315,
      "learning_rate": 6.818931604694264e-05,
      "loss": 1.4705,
      "step": 8389
    },
    {
      "epoch": 0.8850210970464135,
      "grad_norm": 0.4451974630355835,
      "learning_rate": 6.809473931881644e-05,
      "loss": 1.5007,
      "step": 8390
    },
    {
      "epoch": 0.8851265822784811,
      "grad_norm": 0.4537569582462311,
      "learning_rate": 6.800029376604179e-05,
      "loss": 1.5034,
      "step": 8391
    },
    {
      "epoch": 0.8852320675105485,
      "grad_norm": 0.4212094843387604,
      "learning_rate": 6.790597920668206e-05,
      "loss": 1.4657,
      "step": 8392
    },
    {
      "epoch": 0.885337552742616,
      "grad_norm": 0.49739083647727966,
      "learning_rate": 6.781179545905287e-05,
      "loss": 1.4854,
      "step": 8393
    },
    {
      "epoch": 0.8854430379746835,
      "grad_norm": 0.4821615517139435,
      "learning_rate": 6.771774234172196e-05,
      "loss": 1.4712,
      "step": 8394
    },
    {
      "epoch": 0.885548523206751,
      "grad_norm": 0.42705926299095154,
      "learning_rate": 6.762381967350862e-05,
      "loss": 1.4806,
      "step": 8395
    },
    {
      "epoch": 0.8856540084388186,
      "grad_norm": 0.4525908827781677,
      "learning_rate": 6.753002727348348e-05,
      "loss": 1.4928,
      "step": 8396
    },
    {
      "epoch": 0.885759493670886,
      "grad_norm": 0.4396267235279083,
      "learning_rate": 6.743636496096815e-05,
      "loss": 1.4926,
      "step": 8397
    },
    {
      "epoch": 0.8858649789029536,
      "grad_norm": 0.4494251310825348,
      "learning_rate": 6.73428325555347e-05,
      "loss": 1.4733,
      "step": 8398
    },
    {
      "epoch": 0.8859704641350211,
      "grad_norm": 0.4896926283836365,
      "learning_rate": 6.724942987700563e-05,
      "loss": 1.5003,
      "step": 8399
    },
    {
      "epoch": 0.8860759493670886,
      "grad_norm": 0.407871276140213,
      "learning_rate": 6.71561567454532e-05,
      "loss": 1.4802,
      "step": 8400
    },
    {
      "epoch": 0.8861814345991561,
      "grad_norm": 0.4104156196117401,
      "learning_rate": 6.706301298119924e-05,
      "loss": 1.4989,
      "step": 8401
    },
    {
      "epoch": 0.8862869198312237,
      "grad_norm": 0.4375029504299164,
      "learning_rate": 6.696999840481492e-05,
      "loss": 1.4952,
      "step": 8402
    },
    {
      "epoch": 0.8863924050632911,
      "grad_norm": 0.41010478138923645,
      "learning_rate": 6.687711283712008e-05,
      "loss": 1.5152,
      "step": 8403
    },
    {
      "epoch": 0.8864978902953586,
      "grad_norm": 0.49851664900779724,
      "learning_rate": 6.678435609918325e-05,
      "loss": 1.5076,
      "step": 8404
    },
    {
      "epoch": 0.8866033755274262,
      "grad_norm": 0.532653272151947,
      "learning_rate": 6.669172801232099e-05,
      "loss": 1.5003,
      "step": 8405
    },
    {
      "epoch": 0.8867088607594936,
      "grad_norm": 0.4978204667568207,
      "learning_rate": 6.659922839809777e-05,
      "loss": 1.5218,
      "step": 8406
    },
    {
      "epoch": 0.8868143459915612,
      "grad_norm": 0.43870192766189575,
      "learning_rate": 6.65068570783256e-05,
      "loss": 1.5256,
      "step": 8407
    },
    {
      "epoch": 0.8869198312236287,
      "grad_norm": 0.45217710733413696,
      "learning_rate": 6.641461387506347e-05,
      "loss": 1.4964,
      "step": 8408
    },
    {
      "epoch": 0.8870253164556962,
      "grad_norm": 0.4979802668094635,
      "learning_rate": 6.632249861061733e-05,
      "loss": 1.5278,
      "step": 8409
    },
    {
      "epoch": 0.8871308016877637,
      "grad_norm": 0.5015259385108948,
      "learning_rate": 6.623051110753948e-05,
      "loss": 1.4785,
      "step": 8410
    },
    {
      "epoch": 0.8872362869198313,
      "grad_norm": 0.42942774295806885,
      "learning_rate": 6.613865118862837e-05,
      "loss": 1.5149,
      "step": 8411
    },
    {
      "epoch": 0.8873417721518987,
      "grad_norm": 0.433816522359848,
      "learning_rate": 6.604691867692828e-05,
      "loss": 1.4773,
      "step": 8412
    },
    {
      "epoch": 0.8874472573839662,
      "grad_norm": 0.43319758772850037,
      "learning_rate": 6.595531339572882e-05,
      "loss": 1.5179,
      "step": 8413
    },
    {
      "epoch": 0.8875527426160338,
      "grad_norm": 0.4250449538230896,
      "learning_rate": 6.586383516856473e-05,
      "loss": 1.4993,
      "step": 8414
    },
    {
      "epoch": 0.8876582278481012,
      "grad_norm": 0.4036279320716858,
      "learning_rate": 6.57724838192156e-05,
      "loss": 1.5048,
      "step": 8415
    },
    {
      "epoch": 0.8877637130801688,
      "grad_norm": 0.4795280992984772,
      "learning_rate": 6.568125917170527e-05,
      "loss": 1.455,
      "step": 8416
    },
    {
      "epoch": 0.8878691983122363,
      "grad_norm": 0.4198485314846039,
      "learning_rate": 6.559016105030177e-05,
      "loss": 1.4741,
      "step": 8417
    },
    {
      "epoch": 0.8879746835443038,
      "grad_norm": 0.42640748620033264,
      "learning_rate": 6.549918927951679e-05,
      "loss": 1.4831,
      "step": 8418
    },
    {
      "epoch": 0.8880801687763713,
      "grad_norm": 0.45151999592781067,
      "learning_rate": 6.540834368410549e-05,
      "loss": 1.4818,
      "step": 8419
    },
    {
      "epoch": 0.8881856540084389,
      "grad_norm": 0.42480534315109253,
      "learning_rate": 6.531762408906607e-05,
      "loss": 1.4962,
      "step": 8420
    },
    {
      "epoch": 0.8882911392405063,
      "grad_norm": 0.4287979006767273,
      "learning_rate": 6.522703031963938e-05,
      "loss": 1.4934,
      "step": 8421
    },
    {
      "epoch": 0.8883966244725738,
      "grad_norm": 0.4689832627773285,
      "learning_rate": 6.513656220130879e-05,
      "loss": 1.4902,
      "step": 8422
    },
    {
      "epoch": 0.8885021097046414,
      "grad_norm": 0.514316201210022,
      "learning_rate": 6.504621955979958e-05,
      "loss": 1.5132,
      "step": 8423
    },
    {
      "epoch": 0.8886075949367088,
      "grad_norm": 0.47200873494148254,
      "learning_rate": 6.495600222107884e-05,
      "loss": 1.456,
      "step": 8424
    },
    {
      "epoch": 0.8887130801687764,
      "grad_norm": 0.4109126329421997,
      "learning_rate": 6.486591001135503e-05,
      "loss": 1.4728,
      "step": 8425
    },
    {
      "epoch": 0.8888185654008439,
      "grad_norm": 0.41009843349456787,
      "learning_rate": 6.477594275707757e-05,
      "loss": 1.4522,
      "step": 8426
    },
    {
      "epoch": 0.8889240506329114,
      "grad_norm": 0.423771470785141,
      "learning_rate": 6.468610028493671e-05,
      "loss": 1.5272,
      "step": 8427
    },
    {
      "epoch": 0.8890295358649789,
      "grad_norm": 0.46336105465888977,
      "learning_rate": 6.459638242186298e-05,
      "loss": 1.4836,
      "step": 8428
    },
    {
      "epoch": 0.8891350210970465,
      "grad_norm": 0.4478496313095093,
      "learning_rate": 6.4506788995027e-05,
      "loss": 1.499,
      "step": 8429
    },
    {
      "epoch": 0.8892405063291139,
      "grad_norm": 0.42656466364860535,
      "learning_rate": 6.441731983183912e-05,
      "loss": 1.4779,
      "step": 8430
    },
    {
      "epoch": 0.8893459915611814,
      "grad_norm": 0.4402346909046173,
      "learning_rate": 6.432797475994898e-05,
      "loss": 1.4632,
      "step": 8431
    },
    {
      "epoch": 0.889451476793249,
      "grad_norm": 0.41152986884117126,
      "learning_rate": 6.42387536072454e-05,
      "loss": 1.4622,
      "step": 8432
    },
    {
      "epoch": 0.8895569620253164,
      "grad_norm": 0.4462791383266449,
      "learning_rate": 6.414965620185575e-05,
      "loss": 1.4554,
      "step": 8433
    },
    {
      "epoch": 0.889662447257384,
      "grad_norm": 0.4228445291519165,
      "learning_rate": 6.406068237214591e-05,
      "loss": 1.4586,
      "step": 8434
    },
    {
      "epoch": 0.8897679324894515,
      "grad_norm": 0.4302632510662079,
      "learning_rate": 6.39718319467198e-05,
      "loss": 1.5139,
      "step": 8435
    },
    {
      "epoch": 0.889873417721519,
      "grad_norm": 0.4118238389492035,
      "learning_rate": 6.388310475441898e-05,
      "loss": 1.5033,
      "step": 8436
    },
    {
      "epoch": 0.8899789029535865,
      "grad_norm": 0.4312010705471039,
      "learning_rate": 6.379450062432251e-05,
      "loss": 1.4702,
      "step": 8437
    },
    {
      "epoch": 0.890084388185654,
      "grad_norm": 0.4147666394710541,
      "learning_rate": 6.370601938574637e-05,
      "loss": 1.5071,
      "step": 8438
    },
    {
      "epoch": 0.8901898734177215,
      "grad_norm": 0.4334884583950043,
      "learning_rate": 6.361766086824344e-05,
      "loss": 1.4834,
      "step": 8439
    },
    {
      "epoch": 0.890295358649789,
      "grad_norm": 0.471153199672699,
      "learning_rate": 6.352942490160292e-05,
      "loss": 1.5024,
      "step": 8440
    },
    {
      "epoch": 0.8904008438818566,
      "grad_norm": 0.43951278924942017,
      "learning_rate": 6.344131131585007e-05,
      "loss": 1.4916,
      "step": 8441
    },
    {
      "epoch": 0.890506329113924,
      "grad_norm": 0.4353131055831909,
      "learning_rate": 6.335331994124594e-05,
      "loss": 1.4756,
      "step": 8442
    },
    {
      "epoch": 0.8906118143459916,
      "grad_norm": 0.4283924102783203,
      "learning_rate": 6.326545060828696e-05,
      "loss": 1.4942,
      "step": 8443
    },
    {
      "epoch": 0.8907172995780591,
      "grad_norm": 0.45377227663993835,
      "learning_rate": 6.31777031477047e-05,
      "loss": 1.4732,
      "step": 8444
    },
    {
      "epoch": 0.8908227848101266,
      "grad_norm": 0.4111506938934326,
      "learning_rate": 6.309007739046552e-05,
      "loss": 1.4686,
      "step": 8445
    },
    {
      "epoch": 0.8909282700421941,
      "grad_norm": 0.4356135129928589,
      "learning_rate": 6.300257316777014e-05,
      "loss": 1.4944,
      "step": 8446
    },
    {
      "epoch": 0.8910337552742617,
      "grad_norm": 0.4620656371116638,
      "learning_rate": 6.291519031105349e-05,
      "loss": 1.5234,
      "step": 8447
    },
    {
      "epoch": 0.8911392405063291,
      "grad_norm": 0.40962132811546326,
      "learning_rate": 6.282792865198421e-05,
      "loss": 1.5097,
      "step": 8448
    },
    {
      "epoch": 0.8912447257383966,
      "grad_norm": 0.44323867559432983,
      "learning_rate": 6.274078802246449e-05,
      "loss": 1.5173,
      "step": 8449
    },
    {
      "epoch": 0.8913502109704642,
      "grad_norm": 0.42913782596588135,
      "learning_rate": 6.265376825462966e-05,
      "loss": 1.4881,
      "step": 8450
    },
    {
      "epoch": 0.8914556962025316,
      "grad_norm": 0.4396832585334778,
      "learning_rate": 6.256686918084778e-05,
      "loss": 1.4764,
      "step": 8451
    },
    {
      "epoch": 0.8915611814345992,
      "grad_norm": 0.44532984495162964,
      "learning_rate": 6.248009063371955e-05,
      "loss": 1.4959,
      "step": 8452
    },
    {
      "epoch": 0.8916666666666667,
      "grad_norm": 0.41534358263015747,
      "learning_rate": 6.239343244607771e-05,
      "loss": 1.4803,
      "step": 8453
    },
    {
      "epoch": 0.8917721518987342,
      "grad_norm": 0.469448983669281,
      "learning_rate": 6.230689445098696e-05,
      "loss": 1.5099,
      "step": 8454
    },
    {
      "epoch": 0.8918776371308017,
      "grad_norm": 0.44022998213768005,
      "learning_rate": 6.222047648174353e-05,
      "loss": 1.451,
      "step": 8455
    },
    {
      "epoch": 0.8919831223628693,
      "grad_norm": 0.4374964237213135,
      "learning_rate": 6.213417837187475e-05,
      "loss": 1.5057,
      "step": 8456
    },
    {
      "epoch": 0.8920886075949367,
      "grad_norm": 0.41490861773490906,
      "learning_rate": 6.204799995513898e-05,
      "loss": 1.4671,
      "step": 8457
    },
    {
      "epoch": 0.8921940928270042,
      "grad_norm": 0.4097163677215576,
      "learning_rate": 6.196194106552512e-05,
      "loss": 1.4868,
      "step": 8458
    },
    {
      "epoch": 0.8922995780590718,
      "grad_norm": 0.4366304874420166,
      "learning_rate": 6.187600153725223e-05,
      "loss": 1.4949,
      "step": 8459
    },
    {
      "epoch": 0.8924050632911392,
      "grad_norm": 0.42559051513671875,
      "learning_rate": 6.179018120476945e-05,
      "loss": 1.5068,
      "step": 8460
    },
    {
      "epoch": 0.8925105485232068,
      "grad_norm": 0.526786208152771,
      "learning_rate": 6.17044799027554e-05,
      "loss": 1.5065,
      "step": 8461
    },
    {
      "epoch": 0.8926160337552742,
      "grad_norm": 0.41393354535102844,
      "learning_rate": 6.161889746611807e-05,
      "loss": 1.4841,
      "step": 8462
    },
    {
      "epoch": 0.8927215189873418,
      "grad_norm": 0.4025609493255615,
      "learning_rate": 6.153343372999445e-05,
      "loss": 1.4799,
      "step": 8463
    },
    {
      "epoch": 0.8928270042194093,
      "grad_norm": 0.4296261668205261,
      "learning_rate": 6.14480885297501e-05,
      "loss": 1.4796,
      "step": 8464
    },
    {
      "epoch": 0.8929324894514767,
      "grad_norm": 0.3971109092235565,
      "learning_rate": 6.1362861700979e-05,
      "loss": 1.4973,
      "step": 8465
    },
    {
      "epoch": 0.8930379746835443,
      "grad_norm": 0.44303449988365173,
      "learning_rate": 6.127775307950314e-05,
      "loss": 1.4709,
      "step": 8466
    },
    {
      "epoch": 0.8931434599156118,
      "grad_norm": 0.44857221841812134,
      "learning_rate": 6.119276250137219e-05,
      "loss": 1.4841,
      "step": 8467
    },
    {
      "epoch": 0.8932489451476793,
      "grad_norm": 0.4154420495033264,
      "learning_rate": 6.110788980286329e-05,
      "loss": 1.4679,
      "step": 8468
    },
    {
      "epoch": 0.8933544303797468,
      "grad_norm": 0.447130411863327,
      "learning_rate": 6.1023134820480546e-05,
      "loss": 1.4959,
      "step": 8469
    },
    {
      "epoch": 0.8934599156118144,
      "grad_norm": 0.416172593832016,
      "learning_rate": 6.0938497390954946e-05,
      "loss": 1.5046,
      "step": 8470
    },
    {
      "epoch": 0.8935654008438818,
      "grad_norm": 0.4664374887943268,
      "learning_rate": 6.0853977351243815e-05,
      "loss": 1.483,
      "step": 8471
    },
    {
      "epoch": 0.8936708860759494,
      "grad_norm": 0.423372358083725,
      "learning_rate": 6.0769574538530704e-05,
      "loss": 1.4736,
      "step": 8472
    },
    {
      "epoch": 0.8937763713080169,
      "grad_norm": 0.4114157259464264,
      "learning_rate": 6.0685288790224975e-05,
      "loss": 1.5107,
      "step": 8473
    },
    {
      "epoch": 0.8938818565400843,
      "grad_norm": 0.4139949381351471,
      "learning_rate": 6.0601119943961425e-05,
      "loss": 1.4657,
      "step": 8474
    },
    {
      "epoch": 0.8939873417721519,
      "grad_norm": 0.4176963269710541,
      "learning_rate": 6.0517067837600144e-05,
      "loss": 1.4962,
      "step": 8475
    },
    {
      "epoch": 0.8940928270042194,
      "grad_norm": 0.39363715052604675,
      "learning_rate": 6.0433132309226017e-05,
      "loss": 1.4312,
      "step": 8476
    },
    {
      "epoch": 0.8941983122362869,
      "grad_norm": 0.42532992362976074,
      "learning_rate": 6.034931319714857e-05,
      "loss": 1.4908,
      "step": 8477
    },
    {
      "epoch": 0.8943037974683544,
      "grad_norm": 0.41083085536956787,
      "learning_rate": 6.026561033990159e-05,
      "loss": 1.4956,
      "step": 8478
    },
    {
      "epoch": 0.894409282700422,
      "grad_norm": 0.4373190999031067,
      "learning_rate": 6.0182023576242725e-05,
      "loss": 1.4621,
      "step": 8479
    },
    {
      "epoch": 0.8945147679324894,
      "grad_norm": 0.42408502101898193,
      "learning_rate": 6.009855274515339e-05,
      "loss": 1.4893,
      "step": 8480
    },
    {
      "epoch": 0.894620253164557,
      "grad_norm": 0.42819565534591675,
      "learning_rate": 6.001519768583819e-05,
      "loss": 1.5027,
      "step": 8481
    },
    {
      "epoch": 0.8947257383966245,
      "grad_norm": 0.40804967284202576,
      "learning_rate": 5.993195823772487e-05,
      "loss": 1.5159,
      "step": 8482
    },
    {
      "epoch": 0.8948312236286919,
      "grad_norm": 0.41719532012939453,
      "learning_rate": 5.9848834240463846e-05,
      "loss": 1.4917,
      "step": 8483
    },
    {
      "epoch": 0.8949367088607595,
      "grad_norm": 0.43063557147979736,
      "learning_rate": 5.976582553392788e-05,
      "loss": 1.4739,
      "step": 8484
    },
    {
      "epoch": 0.895042194092827,
      "grad_norm": 0.4129980504512787,
      "learning_rate": 5.968293195821191e-05,
      "loss": 1.485,
      "step": 8485
    },
    {
      "epoch": 0.8951476793248945,
      "grad_norm": 0.4312863051891327,
      "learning_rate": 5.960015335363258e-05,
      "loss": 1.4964,
      "step": 8486
    },
    {
      "epoch": 0.895253164556962,
      "grad_norm": 0.44003424048423767,
      "learning_rate": 5.9517489560728056e-05,
      "loss": 1.5156,
      "step": 8487
    },
    {
      "epoch": 0.8953586497890296,
      "grad_norm": 0.4447747766971588,
      "learning_rate": 5.943494042025771e-05,
      "loss": 1.4593,
      "step": 8488
    },
    {
      "epoch": 0.895464135021097,
      "grad_norm": 0.4447772204875946,
      "learning_rate": 5.9352505773201664e-05,
      "loss": 1.4862,
      "step": 8489
    },
    {
      "epoch": 0.8955696202531646,
      "grad_norm": 0.41685861349105835,
      "learning_rate": 5.9270185460760735e-05,
      "loss": 1.4635,
      "step": 8490
    },
    {
      "epoch": 0.8956751054852321,
      "grad_norm": 0.4261850416660309,
      "learning_rate": 5.918797932435585e-05,
      "loss": 1.4678,
      "step": 8491
    },
    {
      "epoch": 0.8957805907172995,
      "grad_norm": 0.4361521303653717,
      "learning_rate": 5.9105887205627985e-05,
      "loss": 1.4565,
      "step": 8492
    },
    {
      "epoch": 0.8958860759493671,
      "grad_norm": 0.48076456785202026,
      "learning_rate": 5.9023908946437736e-05,
      "loss": 1.4374,
      "step": 8493
    },
    {
      "epoch": 0.8959915611814346,
      "grad_norm": 0.49267327785491943,
      "learning_rate": 5.894204438886499e-05,
      "loss": 1.4508,
      "step": 8494
    },
    {
      "epoch": 0.8960970464135021,
      "grad_norm": 0.41653257608413696,
      "learning_rate": 5.886029337520872e-05,
      "loss": 1.542,
      "step": 8495
    },
    {
      "epoch": 0.8962025316455696,
      "grad_norm": 0.424049437046051,
      "learning_rate": 5.877865574798656e-05,
      "loss": 1.4884,
      "step": 8496
    },
    {
      "epoch": 0.8963080168776372,
      "grad_norm": 0.4379669725894928,
      "learning_rate": 5.869713134993462e-05,
      "loss": 1.4796,
      "step": 8497
    },
    {
      "epoch": 0.8964135021097046,
      "grad_norm": 0.4263968765735626,
      "learning_rate": 5.8615720024007174e-05,
      "loss": 1.4667,
      "step": 8498
    },
    {
      "epoch": 0.8965189873417722,
      "grad_norm": 0.44898489117622375,
      "learning_rate": 5.8534421613376175e-05,
      "loss": 1.4899,
      "step": 8499
    },
    {
      "epoch": 0.8966244725738397,
      "grad_norm": 0.4628767669200897,
      "learning_rate": 5.8453235961431225e-05,
      "loss": 1.4843,
      "step": 8500
    },
    {
      "epoch": 0.8967299578059071,
      "grad_norm": 0.3954363465309143,
      "learning_rate": 5.837216291177911e-05,
      "loss": 1.496,
      "step": 8501
    },
    {
      "epoch": 0.8968354430379747,
      "grad_norm": 0.4127616286277771,
      "learning_rate": 5.829120230824344e-05,
      "loss": 1.5091,
      "step": 8502
    },
    {
      "epoch": 0.8969409282700422,
      "grad_norm": 0.42761513590812683,
      "learning_rate": 5.821035399486458e-05,
      "loss": 1.4671,
      "step": 8503
    },
    {
      "epoch": 0.8970464135021097,
      "grad_norm": 0.4368777573108673,
      "learning_rate": 5.8129617815899086e-05,
      "loss": 1.4995,
      "step": 8504
    },
    {
      "epoch": 0.8971518987341772,
      "grad_norm": 0.5013472437858582,
      "learning_rate": 5.8048993615819584e-05,
      "loss": 1.4686,
      "step": 8505
    },
    {
      "epoch": 0.8972573839662448,
      "grad_norm": 0.43895700573921204,
      "learning_rate": 5.7968481239314435e-05,
      "loss": 1.4609,
      "step": 8506
    },
    {
      "epoch": 0.8973628691983122,
      "grad_norm": 0.4145542085170746,
      "learning_rate": 5.788808053128733e-05,
      "loss": 1.4926,
      "step": 8507
    },
    {
      "epoch": 0.8974683544303798,
      "grad_norm": 0.4039081931114197,
      "learning_rate": 5.780779133685717e-05,
      "loss": 1.5155,
      "step": 8508
    },
    {
      "epoch": 0.8975738396624473,
      "grad_norm": 0.4022243022918701,
      "learning_rate": 5.772761350135759e-05,
      "loss": 1.505,
      "step": 8509
    },
    {
      "epoch": 0.8976793248945147,
      "grad_norm": 0.4137919247150421,
      "learning_rate": 5.764754687033678e-05,
      "loss": 1.4678,
      "step": 8510
    },
    {
      "epoch": 0.8977848101265823,
      "grad_norm": 0.42878103256225586,
      "learning_rate": 5.756759128955722e-05,
      "loss": 1.4717,
      "step": 8511
    },
    {
      "epoch": 0.8978902953586498,
      "grad_norm": 0.4243728518486023,
      "learning_rate": 5.748774660499514e-05,
      "loss": 1.4851,
      "step": 8512
    },
    {
      "epoch": 0.8979957805907173,
      "grad_norm": 0.43055763840675354,
      "learning_rate": 5.740801266284059e-05,
      "loss": 1.4735,
      "step": 8513
    },
    {
      "epoch": 0.8981012658227848,
      "grad_norm": 0.41846656799316406,
      "learning_rate": 5.732838930949679e-05,
      "loss": 1.4505,
      "step": 8514
    },
    {
      "epoch": 0.8982067510548524,
      "grad_norm": 0.44019097089767456,
      "learning_rate": 5.724887639158008e-05,
      "loss": 1.4933,
      "step": 8515
    },
    {
      "epoch": 0.8983122362869198,
      "grad_norm": 0.43868181109428406,
      "learning_rate": 5.716947375591959e-05,
      "loss": 1.5146,
      "step": 8516
    },
    {
      "epoch": 0.8984177215189874,
      "grad_norm": 0.42221599817276,
      "learning_rate": 5.709018124955674e-05,
      "loss": 1.4996,
      "step": 8517
    },
    {
      "epoch": 0.8985232067510549,
      "grad_norm": 0.4055478572845459,
      "learning_rate": 5.701099871974525e-05,
      "loss": 1.4902,
      "step": 8518
    },
    {
      "epoch": 0.8986286919831223,
      "grad_norm": 0.4379352629184723,
      "learning_rate": 5.6931926013950586e-05,
      "loss": 1.4605,
      "step": 8519
    },
    {
      "epoch": 0.8987341772151899,
      "grad_norm": 0.4605368971824646,
      "learning_rate": 5.6852962979849836e-05,
      "loss": 1.5155,
      "step": 8520
    },
    {
      "epoch": 0.8988396624472574,
      "grad_norm": 0.4232887923717499,
      "learning_rate": 5.677410946533138e-05,
      "loss": 1.4908,
      "step": 8521
    },
    {
      "epoch": 0.8989451476793249,
      "grad_norm": 0.42216604948043823,
      "learning_rate": 5.6695365318494475e-05,
      "loss": 1.5028,
      "step": 8522
    },
    {
      "epoch": 0.8990506329113924,
      "grad_norm": 0.4300329089164734,
      "learning_rate": 5.6616730387649173e-05,
      "loss": 1.5008,
      "step": 8523
    },
    {
      "epoch": 0.89915611814346,
      "grad_norm": 0.44915470480918884,
      "learning_rate": 5.6538204521315804e-05,
      "loss": 1.4754,
      "step": 8524
    },
    {
      "epoch": 0.8992616033755274,
      "grad_norm": 0.4172717332839966,
      "learning_rate": 5.6459787568224886e-05,
      "loss": 1.5211,
      "step": 8525
    },
    {
      "epoch": 0.899367088607595,
      "grad_norm": 0.4495176076889038,
      "learning_rate": 5.6381479377316726e-05,
      "loss": 1.5289,
      "step": 8526
    },
    {
      "epoch": 0.8994725738396624,
      "grad_norm": 0.39566850662231445,
      "learning_rate": 5.630327979774111e-05,
      "loss": 1.4936,
      "step": 8527
    },
    {
      "epoch": 0.8995780590717299,
      "grad_norm": 0.5286000967025757,
      "learning_rate": 5.6225188678857095e-05,
      "loss": 1.4968,
      "step": 8528
    },
    {
      "epoch": 0.8996835443037975,
      "grad_norm": 0.4285947382450104,
      "learning_rate": 5.61472058702326e-05,
      "loss": 1.4689,
      "step": 8529
    },
    {
      "epoch": 0.8997890295358649,
      "grad_norm": 0.42458072304725647,
      "learning_rate": 5.6069331221644284e-05,
      "loss": 1.5222,
      "step": 8530
    },
    {
      "epoch": 0.8998945147679325,
      "grad_norm": 0.44740793108940125,
      "learning_rate": 5.599156458307712e-05,
      "loss": 1.4906,
      "step": 8531
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.47773995995521545,
      "learning_rate": 5.5913905804724106e-05,
      "loss": 1.4882,
      "step": 8532
    },
    {
      "epoch": 0.9001054852320675,
      "grad_norm": 0.4227980375289917,
      "learning_rate": 5.58363547369861e-05,
      "loss": 1.4687,
      "step": 8533
    },
    {
      "epoch": 0.900210970464135,
      "grad_norm": 0.45700153708457947,
      "learning_rate": 5.575891123047136e-05,
      "loss": 1.5177,
      "step": 8534
    },
    {
      "epoch": 0.9003164556962026,
      "grad_norm": 0.412743479013443,
      "learning_rate": 5.568157513599542e-05,
      "loss": 1.4807,
      "step": 8535
    },
    {
      "epoch": 0.90042194092827,
      "grad_norm": 0.4132397174835205,
      "learning_rate": 5.5604346304580727e-05,
      "loss": 1.5148,
      "step": 8536
    },
    {
      "epoch": 0.9005274261603375,
      "grad_norm": 0.4457518756389618,
      "learning_rate": 5.552722458745626e-05,
      "loss": 1.5337,
      "step": 8537
    },
    {
      "epoch": 0.9006329113924051,
      "grad_norm": 0.4229176342487335,
      "learning_rate": 5.545020983605749e-05,
      "loss": 1.4872,
      "step": 8538
    },
    {
      "epoch": 0.9007383966244725,
      "grad_norm": 0.4493631422519684,
      "learning_rate": 5.53733019020258e-05,
      "loss": 1.4789,
      "step": 8539
    },
    {
      "epoch": 0.9008438818565401,
      "grad_norm": 0.4571084976196289,
      "learning_rate": 5.529650063720842e-05,
      "loss": 1.5118,
      "step": 8540
    },
    {
      "epoch": 0.9009493670886076,
      "grad_norm": 0.39605796337127686,
      "learning_rate": 5.52198058936581e-05,
      "loss": 1.4886,
      "step": 8541
    },
    {
      "epoch": 0.9010548523206751,
      "grad_norm": 0.4505995810031891,
      "learning_rate": 5.5143217523632655e-05,
      "loss": 1.4948,
      "step": 8542
    },
    {
      "epoch": 0.9011603375527426,
      "grad_norm": 0.41481825709342957,
      "learning_rate": 5.5066735379594944e-05,
      "loss": 1.4733,
      "step": 8543
    },
    {
      "epoch": 0.9012658227848102,
      "grad_norm": 0.4302605092525482,
      "learning_rate": 5.4990359314212424e-05,
      "loss": 1.514,
      "step": 8544
    },
    {
      "epoch": 0.9013713080168776,
      "grad_norm": 0.42191892862319946,
      "learning_rate": 5.491408918035683e-05,
      "loss": 1.4478,
      "step": 8545
    },
    {
      "epoch": 0.9014767932489451,
      "grad_norm": 0.451427698135376,
      "learning_rate": 5.483792483110408e-05,
      "loss": 1.4911,
      "step": 8546
    },
    {
      "epoch": 0.9015822784810127,
      "grad_norm": 0.43870601058006287,
      "learning_rate": 5.476186611973374e-05,
      "loss": 1.4691,
      "step": 8547
    },
    {
      "epoch": 0.9016877637130801,
      "grad_norm": 0.42102065682411194,
      "learning_rate": 5.4685912899728965e-05,
      "loss": 1.5076,
      "step": 8548
    },
    {
      "epoch": 0.9017932489451477,
      "grad_norm": 0.4193611741065979,
      "learning_rate": 5.4610065024776125e-05,
      "loss": 1.4602,
      "step": 8549
    },
    {
      "epoch": 0.9018987341772152,
      "grad_norm": 0.4292870759963989,
      "learning_rate": 5.453432234876445e-05,
      "loss": 1.5048,
      "step": 8550
    },
    {
      "epoch": 0.9020042194092827,
      "grad_norm": 0.4297439157962799,
      "learning_rate": 5.445868472578592e-05,
      "loss": 1.4623,
      "step": 8551
    },
    {
      "epoch": 0.9021097046413502,
      "grad_norm": 0.420657753944397,
      "learning_rate": 5.438315201013476e-05,
      "loss": 1.4664,
      "step": 8552
    },
    {
      "epoch": 0.9022151898734178,
      "grad_norm": 0.43493518233299255,
      "learning_rate": 5.430772405630742e-05,
      "loss": 1.4871,
      "step": 8553
    },
    {
      "epoch": 0.9023206751054852,
      "grad_norm": 0.42774251103401184,
      "learning_rate": 5.423240071900209e-05,
      "loss": 1.507,
      "step": 8554
    },
    {
      "epoch": 0.9024261603375527,
      "grad_norm": 0.4181724190711975,
      "learning_rate": 5.4157181853118464e-05,
      "loss": 1.4872,
      "step": 8555
    },
    {
      "epoch": 0.9025316455696203,
      "grad_norm": 0.4257463216781616,
      "learning_rate": 5.408206731375757e-05,
      "loss": 1.502,
      "step": 8556
    },
    {
      "epoch": 0.9026371308016877,
      "grad_norm": 0.41938287019729614,
      "learning_rate": 5.400705695622129e-05,
      "loss": 1.4289,
      "step": 8557
    },
    {
      "epoch": 0.9027426160337553,
      "grad_norm": 0.4359043836593628,
      "learning_rate": 5.39321506360123e-05,
      "loss": 1.4593,
      "step": 8558
    },
    {
      "epoch": 0.9028481012658228,
      "grad_norm": 0.406623899936676,
      "learning_rate": 5.38573482088337e-05,
      "loss": 1.4932,
      "step": 8559
    },
    {
      "epoch": 0.9029535864978903,
      "grad_norm": 0.430521696805954,
      "learning_rate": 5.37826495305886e-05,
      "loss": 1.4614,
      "step": 8560
    },
    {
      "epoch": 0.9030590717299578,
      "grad_norm": 0.4151771664619446,
      "learning_rate": 5.370805445738011e-05,
      "loss": 1.49,
      "step": 8561
    },
    {
      "epoch": 0.9031645569620254,
      "grad_norm": 0.42271867394447327,
      "learning_rate": 5.3633562845510806e-05,
      "loss": 1.5204,
      "step": 8562
    },
    {
      "epoch": 0.9032700421940928,
      "grad_norm": 0.45540252327919006,
      "learning_rate": 5.3559174551482656e-05,
      "loss": 1.4847,
      "step": 8563
    },
    {
      "epoch": 0.9033755274261603,
      "grad_norm": 0.4780273735523224,
      "learning_rate": 5.3484889431996646e-05,
      "loss": 1.4652,
      "step": 8564
    },
    {
      "epoch": 0.9034810126582279,
      "grad_norm": 0.4422934353351593,
      "learning_rate": 5.341070734395244e-05,
      "loss": 1.4931,
      "step": 8565
    },
    {
      "epoch": 0.9035864978902953,
      "grad_norm": 0.43156108260154724,
      "learning_rate": 5.3336628144448266e-05,
      "loss": 1.5259,
      "step": 8566
    },
    {
      "epoch": 0.9036919831223629,
      "grad_norm": 0.4315359890460968,
      "learning_rate": 5.326265169078048e-05,
      "loss": 1.498,
      "step": 8567
    },
    {
      "epoch": 0.9037974683544304,
      "grad_norm": 0.40850797295570374,
      "learning_rate": 5.318877784044342e-05,
      "loss": 1.4639,
      "step": 8568
    },
    {
      "epoch": 0.9039029535864979,
      "grad_norm": 0.42627614736557007,
      "learning_rate": 5.3115006451129075e-05,
      "loss": 1.4955,
      "step": 8569
    },
    {
      "epoch": 0.9040084388185654,
      "grad_norm": 0.45638468861579895,
      "learning_rate": 5.304133738072674e-05,
      "loss": 1.5209,
      "step": 8570
    },
    {
      "epoch": 0.904113924050633,
      "grad_norm": 0.44763416051864624,
      "learning_rate": 5.296777048732293e-05,
      "loss": 1.4884,
      "step": 8571
    },
    {
      "epoch": 0.9042194092827004,
      "grad_norm": 0.42213135957717896,
      "learning_rate": 5.289430562920086e-05,
      "loss": 1.4612,
      "step": 8572
    },
    {
      "epoch": 0.9043248945147679,
      "grad_norm": 0.41777390241622925,
      "learning_rate": 5.2820942664840405e-05,
      "loss": 1.4913,
      "step": 8573
    },
    {
      "epoch": 0.9044303797468355,
      "grad_norm": 0.4218050241470337,
      "learning_rate": 5.2747681452917697e-05,
      "loss": 1.5012,
      "step": 8574
    },
    {
      "epoch": 0.9045358649789029,
      "grad_norm": 0.4458419978618622,
      "learning_rate": 5.267452185230482e-05,
      "loss": 1.4912,
      "step": 8575
    },
    {
      "epoch": 0.9046413502109705,
      "grad_norm": 0.48722466826438904,
      "learning_rate": 5.260146372206972e-05,
      "loss": 1.4705,
      "step": 8576
    },
    {
      "epoch": 0.904746835443038,
      "grad_norm": 0.41805747151374817,
      "learning_rate": 5.2528506921475664e-05,
      "loss": 1.51,
      "step": 8577
    },
    {
      "epoch": 0.9048523206751055,
      "grad_norm": 0.43927815556526184,
      "learning_rate": 5.245565130998124e-05,
      "loss": 1.4641,
      "step": 8578
    },
    {
      "epoch": 0.904957805907173,
      "grad_norm": 0.4201597273349762,
      "learning_rate": 5.2382896747239935e-05,
      "loss": 1.5162,
      "step": 8579
    },
    {
      "epoch": 0.9050632911392406,
      "grad_norm": 0.4533957242965698,
      "learning_rate": 5.2310243093099814e-05,
      "loss": 1.4939,
      "step": 8580
    },
    {
      "epoch": 0.905168776371308,
      "grad_norm": 0.4219772219657898,
      "learning_rate": 5.223769020760346e-05,
      "loss": 1.4705,
      "step": 8581
    },
    {
      "epoch": 0.9052742616033755,
      "grad_norm": 0.42240914702415466,
      "learning_rate": 5.216523795098743e-05,
      "loss": 1.5021,
      "step": 8582
    },
    {
      "epoch": 0.9053797468354431,
      "grad_norm": 0.45295533537864685,
      "learning_rate": 5.209288618368225e-05,
      "loss": 1.4939,
      "step": 8583
    },
    {
      "epoch": 0.9054852320675105,
      "grad_norm": 0.3956073522567749,
      "learning_rate": 5.202063476631199e-05,
      "loss": 1.4733,
      "step": 8584
    },
    {
      "epoch": 0.9055907172995781,
      "grad_norm": 0.42122554779052734,
      "learning_rate": 5.194848355969396e-05,
      "loss": 1.4857,
      "step": 8585
    },
    {
      "epoch": 0.9056962025316456,
      "grad_norm": 0.4341195821762085,
      "learning_rate": 5.18764324248386e-05,
      "loss": 1.4678,
      "step": 8586
    },
    {
      "epoch": 0.9058016877637131,
      "grad_norm": 0.46259692311286926,
      "learning_rate": 5.180448122294913e-05,
      "loss": 1.496,
      "step": 8587
    },
    {
      "epoch": 0.9059071729957806,
      "grad_norm": 0.4379418194293976,
      "learning_rate": 5.173262981542119e-05,
      "loss": 1.4672,
      "step": 8588
    },
    {
      "epoch": 0.9060126582278482,
      "grad_norm": 0.4257470965385437,
      "learning_rate": 5.166087806384275e-05,
      "loss": 1.5093,
      "step": 8589
    },
    {
      "epoch": 0.9061181434599156,
      "grad_norm": 0.44887909293174744,
      "learning_rate": 5.158922582999368e-05,
      "loss": 1.526,
      "step": 8590
    },
    {
      "epoch": 0.9062236286919831,
      "grad_norm": 0.42257651686668396,
      "learning_rate": 5.1517672975845604e-05,
      "loss": 1.4833,
      "step": 8591
    },
    {
      "epoch": 0.9063291139240506,
      "grad_norm": 0.4242796301841736,
      "learning_rate": 5.144621936356162e-05,
      "loss": 1.486,
      "step": 8592
    },
    {
      "epoch": 0.9064345991561181,
      "grad_norm": 0.4501091539859772,
      "learning_rate": 5.1374864855495894e-05,
      "loss": 1.5073,
      "step": 8593
    },
    {
      "epoch": 0.9065400843881857,
      "grad_norm": 0.4101604223251343,
      "learning_rate": 5.130360931419364e-05,
      "loss": 1.4529,
      "step": 8594
    },
    {
      "epoch": 0.9066455696202531,
      "grad_norm": 0.4798033833503723,
      "learning_rate": 5.123245260239058e-05,
      "loss": 1.5441,
      "step": 8595
    },
    {
      "epoch": 0.9067510548523207,
      "grad_norm": 0.46724724769592285,
      "learning_rate": 5.1161394583012904e-05,
      "loss": 1.4261,
      "step": 8596
    },
    {
      "epoch": 0.9068565400843882,
      "grad_norm": 0.404538094997406,
      "learning_rate": 5.109043511917694e-05,
      "loss": 1.4516,
      "step": 8597
    },
    {
      "epoch": 0.9069620253164556,
      "grad_norm": 0.46111708879470825,
      "learning_rate": 5.101957407418877e-05,
      "loss": 1.5065,
      "step": 8598
    },
    {
      "epoch": 0.9070675105485232,
      "grad_norm": 0.43945568799972534,
      "learning_rate": 5.0948811311544186e-05,
      "loss": 1.4739,
      "step": 8599
    },
    {
      "epoch": 0.9071729957805907,
      "grad_norm": 0.45501163601875305,
      "learning_rate": 5.087814669492819e-05,
      "loss": 1.507,
      "step": 8600
    },
    {
      "epoch": 0.9072784810126582,
      "grad_norm": 0.4359736442565918,
      "learning_rate": 5.080758008821494e-05,
      "loss": 1.5128,
      "step": 8601
    },
    {
      "epoch": 0.9073839662447257,
      "grad_norm": 0.42234253883361816,
      "learning_rate": 5.073711135546738e-05,
      "loss": 1.4824,
      "step": 8602
    },
    {
      "epoch": 0.9074894514767933,
      "grad_norm": 0.41872018575668335,
      "learning_rate": 5.0666740360936944e-05,
      "loss": 1.4724,
      "step": 8603
    },
    {
      "epoch": 0.9075949367088607,
      "grad_norm": 0.4337361454963684,
      "learning_rate": 5.0596466969063415e-05,
      "loss": 1.5038,
      "step": 8604
    },
    {
      "epoch": 0.9077004219409283,
      "grad_norm": 0.4256981611251831,
      "learning_rate": 5.052629104447452e-05,
      "loss": 1.4745,
      "step": 8605
    },
    {
      "epoch": 0.9078059071729958,
      "grad_norm": 0.4462968707084656,
      "learning_rate": 5.0456212451985806e-05,
      "loss": 1.4767,
      "step": 8606
    },
    {
      "epoch": 0.9079113924050632,
      "grad_norm": 0.45027652382850647,
      "learning_rate": 5.038623105660033e-05,
      "loss": 1.4734,
      "step": 8607
    },
    {
      "epoch": 0.9080168776371308,
      "grad_norm": 0.47832170128822327,
      "learning_rate": 5.0316346723508287e-05,
      "loss": 1.4719,
      "step": 8608
    },
    {
      "epoch": 0.9081223628691983,
      "grad_norm": 0.4299798905849457,
      "learning_rate": 5.024655931808697e-05,
      "loss": 1.5033,
      "step": 8609
    },
    {
      "epoch": 0.9082278481012658,
      "grad_norm": 0.4214524030685425,
      "learning_rate": 5.017686870590028e-05,
      "loss": 1.4771,
      "step": 8610
    },
    {
      "epoch": 0.9083333333333333,
      "grad_norm": 0.4417315125465393,
      "learning_rate": 5.010727475269867e-05,
      "loss": 1.4945,
      "step": 8611
    },
    {
      "epoch": 0.9084388185654009,
      "grad_norm": 0.39711448550224304,
      "learning_rate": 5.0037777324418756e-05,
      "loss": 1.4856,
      "step": 8612
    },
    {
      "epoch": 0.9085443037974683,
      "grad_norm": 0.44426894187927246,
      "learning_rate": 4.9968376287183074e-05,
      "loss": 1.4691,
      "step": 8613
    },
    {
      "epoch": 0.9086497890295359,
      "grad_norm": 0.43709734082221985,
      "learning_rate": 4.989907150729989e-05,
      "loss": 1.5143,
      "step": 8614
    },
    {
      "epoch": 0.9087552742616034,
      "grad_norm": 0.44660085439682007,
      "learning_rate": 4.9829862851262845e-05,
      "loss": 1.4331,
      "step": 8615
    },
    {
      "epoch": 0.9088607594936708,
      "grad_norm": 0.4466531574726105,
      "learning_rate": 4.976075018575077e-05,
      "loss": 1.5082,
      "step": 8616
    },
    {
      "epoch": 0.9089662447257384,
      "grad_norm": 0.40991637110710144,
      "learning_rate": 4.9691733377627475e-05,
      "loss": 1.4561,
      "step": 8617
    },
    {
      "epoch": 0.9090717299578059,
      "grad_norm": 0.4238305389881134,
      "learning_rate": 4.962281229394129e-05,
      "loss": 1.5013,
      "step": 8618
    },
    {
      "epoch": 0.9091772151898734,
      "grad_norm": 0.45339417457580566,
      "learning_rate": 4.955398680192509e-05,
      "loss": 1.4504,
      "step": 8619
    },
    {
      "epoch": 0.9092827004219409,
      "grad_norm": 0.42653295397758484,
      "learning_rate": 4.948525676899577e-05,
      "loss": 1.5011,
      "step": 8620
    },
    {
      "epoch": 0.9093881856540085,
      "grad_norm": 0.42224910855293274,
      "learning_rate": 4.9416622062754195e-05,
      "loss": 1.4931,
      "step": 8621
    },
    {
      "epoch": 0.9094936708860759,
      "grad_norm": 0.45977023243904114,
      "learning_rate": 4.934808255098487e-05,
      "loss": 1.4436,
      "step": 8622
    },
    {
      "epoch": 0.9095991561181435,
      "grad_norm": 0.4222732484340668,
      "learning_rate": 4.92796381016556e-05,
      "loss": 1.4895,
      "step": 8623
    },
    {
      "epoch": 0.909704641350211,
      "grad_norm": 0.39933010935783386,
      "learning_rate": 4.9211288582917396e-05,
      "loss": 1.4771,
      "step": 8624
    },
    {
      "epoch": 0.9098101265822784,
      "grad_norm": 0.41332337260246277,
      "learning_rate": 4.9143033863104094e-05,
      "loss": 1.5038,
      "step": 8625
    },
    {
      "epoch": 0.909915611814346,
      "grad_norm": 0.416359543800354,
      "learning_rate": 4.907487381073214e-05,
      "loss": 1.5047,
      "step": 8626
    },
    {
      "epoch": 0.9100210970464135,
      "grad_norm": 0.43971553444862366,
      "learning_rate": 4.900680829450043e-05,
      "loss": 1.4957,
      "step": 8627
    },
    {
      "epoch": 0.910126582278481,
      "grad_norm": 0.45200228691101074,
      "learning_rate": 4.893883718328984e-05,
      "loss": 1.446,
      "step": 8628
    },
    {
      "epoch": 0.9102320675105485,
      "grad_norm": 0.4479728639125824,
      "learning_rate": 4.887096034616319e-05,
      "loss": 1.5115,
      "step": 8629
    },
    {
      "epoch": 0.9103375527426161,
      "grad_norm": 0.41752520203590393,
      "learning_rate": 4.880317765236493e-05,
      "loss": 1.5308,
      "step": 8630
    },
    {
      "epoch": 0.9104430379746835,
      "grad_norm": 0.4065489172935486,
      "learning_rate": 4.873548897132076e-05,
      "loss": 1.5043,
      "step": 8631
    },
    {
      "epoch": 0.9105485232067511,
      "grad_norm": 0.4325360357761383,
      "learning_rate": 4.8667894172637606e-05,
      "loss": 1.4526,
      "step": 8632
    },
    {
      "epoch": 0.9106540084388186,
      "grad_norm": 0.42594823241233826,
      "learning_rate": 4.860039312610312e-05,
      "loss": 1.4496,
      "step": 8633
    },
    {
      "epoch": 0.910759493670886,
      "grad_norm": 0.4176139831542969,
      "learning_rate": 4.8532985701685654e-05,
      "loss": 1.4972,
      "step": 8634
    },
    {
      "epoch": 0.9108649789029536,
      "grad_norm": 0.4402667284011841,
      "learning_rate": 4.846567176953389e-05,
      "loss": 1.4864,
      "step": 8635
    },
    {
      "epoch": 0.9109704641350211,
      "grad_norm": 0.4351549446582794,
      "learning_rate": 4.839845119997657e-05,
      "loss": 1.4483,
      "step": 8636
    },
    {
      "epoch": 0.9110759493670886,
      "grad_norm": 0.45790067315101624,
      "learning_rate": 4.833132386352234e-05,
      "loss": 1.5047,
      "step": 8637
    },
    {
      "epoch": 0.9111814345991561,
      "grad_norm": 0.4096372425556183,
      "learning_rate": 4.8264289630859386e-05,
      "loss": 1.4668,
      "step": 8638
    },
    {
      "epoch": 0.9112869198312237,
      "grad_norm": 0.41209712624549866,
      "learning_rate": 4.819734837285529e-05,
      "loss": 1.4655,
      "step": 8639
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 0.5742860436439514,
      "learning_rate": 4.8130499960556755e-05,
      "loss": 1.4536,
      "step": 8640
    },
    {
      "epoch": 0.9114978902953587,
      "grad_norm": 0.4653583765029907,
      "learning_rate": 4.806374426518927e-05,
      "loss": 1.465,
      "step": 8641
    },
    {
      "epoch": 0.9116033755274262,
      "grad_norm": 0.4464600682258606,
      "learning_rate": 4.799708115815702e-05,
      "loss": 1.4845,
      "step": 8642
    },
    {
      "epoch": 0.9117088607594936,
      "grad_norm": 0.43407562375068665,
      "learning_rate": 4.793051051104244e-05,
      "loss": 1.4728,
      "step": 8643
    },
    {
      "epoch": 0.9118143459915612,
      "grad_norm": 0.41169705986976624,
      "learning_rate": 4.786403219560617e-05,
      "loss": 1.4592,
      "step": 8644
    },
    {
      "epoch": 0.9119198312236287,
      "grad_norm": 0.4779900312423706,
      "learning_rate": 4.779764608378671e-05,
      "loss": 1.4238,
      "step": 8645
    },
    {
      "epoch": 0.9120253164556962,
      "grad_norm": 0.4812563955783844,
      "learning_rate": 4.7731352047700095e-05,
      "loss": 1.4934,
      "step": 8646
    },
    {
      "epoch": 0.9121308016877637,
      "grad_norm": 0.43489304184913635,
      "learning_rate": 4.7665149959639824e-05,
      "loss": 1.4639,
      "step": 8647
    },
    {
      "epoch": 0.9122362869198313,
      "grad_norm": 0.4822406470775604,
      "learning_rate": 4.759903969207646e-05,
      "loss": 1.5363,
      "step": 8648
    },
    {
      "epoch": 0.9123417721518987,
      "grad_norm": 0.40874624252319336,
      "learning_rate": 4.7533021117657475e-05,
      "loss": 1.4827,
      "step": 8649
    },
    {
      "epoch": 0.9124472573839663,
      "grad_norm": 0.41634196043014526,
      "learning_rate": 4.746709410920699e-05,
      "loss": 1.4977,
      "step": 8650
    },
    {
      "epoch": 0.9125527426160338,
      "grad_norm": 0.42765235900878906,
      "learning_rate": 4.740125853972546e-05,
      "loss": 1.4933,
      "step": 8651
    },
    {
      "epoch": 0.9126582278481012,
      "grad_norm": 0.4047209620475769,
      "learning_rate": 4.733551428238957e-05,
      "loss": 1.5106,
      "step": 8652
    },
    {
      "epoch": 0.9127637130801688,
      "grad_norm": 0.4538222551345825,
      "learning_rate": 4.726986121055179e-05,
      "loss": 1.486,
      "step": 8653
    },
    {
      "epoch": 0.9128691983122363,
      "grad_norm": 0.42175963521003723,
      "learning_rate": 4.720429919774036e-05,
      "loss": 1.5079,
      "step": 8654
    },
    {
      "epoch": 0.9129746835443038,
      "grad_norm": 0.4042593538761139,
      "learning_rate": 4.713882811765889e-05,
      "loss": 1.4582,
      "step": 8655
    },
    {
      "epoch": 0.9130801687763713,
      "grad_norm": 0.429984450340271,
      "learning_rate": 4.7073447844186114e-05,
      "loss": 1.4635,
      "step": 8656
    },
    {
      "epoch": 0.9131856540084389,
      "grad_norm": 0.42478910088539124,
      "learning_rate": 4.700815825137578e-05,
      "loss": 1.4412,
      "step": 8657
    },
    {
      "epoch": 0.9132911392405063,
      "grad_norm": 0.4158935844898224,
      "learning_rate": 4.694295921345623e-05,
      "loss": 1.4753,
      "step": 8658
    },
    {
      "epoch": 0.9133966244725739,
      "grad_norm": 0.43298426270484924,
      "learning_rate": 4.687785060483031e-05,
      "loss": 1.4619,
      "step": 8659
    },
    {
      "epoch": 0.9135021097046413,
      "grad_norm": 0.5064635276794434,
      "learning_rate": 4.681283230007507e-05,
      "loss": 1.5002,
      "step": 8660
    },
    {
      "epoch": 0.9136075949367088,
      "grad_norm": 0.42793628573417664,
      "learning_rate": 4.674790417394145e-05,
      "loss": 1.4471,
      "step": 8661
    },
    {
      "epoch": 0.9137130801687764,
      "grad_norm": 0.43282264471054077,
      "learning_rate": 4.6683066101354215e-05,
      "loss": 1.4773,
      "step": 8662
    },
    {
      "epoch": 0.9138185654008438,
      "grad_norm": 0.42674756050109863,
      "learning_rate": 4.661831795741148e-05,
      "loss": 1.5025,
      "step": 8663
    },
    {
      "epoch": 0.9139240506329114,
      "grad_norm": 0.42112410068511963,
      "learning_rate": 4.655365961738467e-05,
      "loss": 1.504,
      "step": 8664
    },
    {
      "epoch": 0.9140295358649789,
      "grad_norm": 0.4230802357196808,
      "learning_rate": 4.648909095671825e-05,
      "loss": 1.4939,
      "step": 8665
    },
    {
      "epoch": 0.9141350210970464,
      "grad_norm": 0.4045351445674896,
      "learning_rate": 4.6424611851029316e-05,
      "loss": 1.4667,
      "step": 8666
    },
    {
      "epoch": 0.9142405063291139,
      "grad_norm": 0.4341583847999573,
      "learning_rate": 4.63602221761076e-05,
      "loss": 1.4771,
      "step": 8667
    },
    {
      "epoch": 0.9143459915611815,
      "grad_norm": 0.4149828255176544,
      "learning_rate": 4.629592180791501e-05,
      "loss": 1.4638,
      "step": 8668
    },
    {
      "epoch": 0.9144514767932489,
      "grad_norm": 0.4330116808414459,
      "learning_rate": 4.623171062258557e-05,
      "loss": 1.4888,
      "step": 8669
    },
    {
      "epoch": 0.9145569620253164,
      "grad_norm": 0.4128780961036682,
      "learning_rate": 4.616758849642509e-05,
      "loss": 1.5071,
      "step": 8670
    },
    {
      "epoch": 0.914662447257384,
      "grad_norm": 0.4493173360824585,
      "learning_rate": 4.610355530591087e-05,
      "loss": 1.5305,
      "step": 8671
    },
    {
      "epoch": 0.9147679324894514,
      "grad_norm": 0.4235368072986603,
      "learning_rate": 4.6039610927691646e-05,
      "loss": 1.5272,
      "step": 8672
    },
    {
      "epoch": 0.914873417721519,
      "grad_norm": 0.38943812251091003,
      "learning_rate": 4.597575523858712e-05,
      "loss": 1.4731,
      "step": 8673
    },
    {
      "epoch": 0.9149789029535865,
      "grad_norm": 0.38808494806289673,
      "learning_rate": 4.5911988115587936e-05,
      "loss": 1.4741,
      "step": 8674
    },
    {
      "epoch": 0.915084388185654,
      "grad_norm": 0.4508820176124573,
      "learning_rate": 4.584830943585533e-05,
      "loss": 1.4989,
      "step": 8675
    },
    {
      "epoch": 0.9151898734177215,
      "grad_norm": 0.42681100964546204,
      "learning_rate": 4.5784719076720844e-05,
      "loss": 1.4887,
      "step": 8676
    },
    {
      "epoch": 0.9152953586497891,
      "grad_norm": 0.46018165349960327,
      "learning_rate": 4.572121691568624e-05,
      "loss": 1.475,
      "step": 8677
    },
    {
      "epoch": 0.9154008438818565,
      "grad_norm": 0.38980650901794434,
      "learning_rate": 4.565780283042316e-05,
      "loss": 1.5026,
      "step": 8678
    },
    {
      "epoch": 0.915506329113924,
      "grad_norm": 0.4348743259906769,
      "learning_rate": 4.559447669877288e-05,
      "loss": 1.5131,
      "step": 8679
    },
    {
      "epoch": 0.9156118143459916,
      "grad_norm": 0.4234529435634613,
      "learning_rate": 4.553123839874615e-05,
      "loss": 1.4619,
      "step": 8680
    },
    {
      "epoch": 0.915717299578059,
      "grad_norm": 0.49747994542121887,
      "learning_rate": 4.546808780852286e-05,
      "loss": 1.525,
      "step": 8681
    },
    {
      "epoch": 0.9158227848101266,
      "grad_norm": 0.4310455322265625,
      "learning_rate": 4.5405024806451926e-05,
      "loss": 1.4902,
      "step": 8682
    },
    {
      "epoch": 0.9159282700421941,
      "grad_norm": 0.45674219727516174,
      "learning_rate": 4.534204927105098e-05,
      "loss": 1.4646,
      "step": 8683
    },
    {
      "epoch": 0.9160337552742616,
      "grad_norm": 0.4182247817516327,
      "learning_rate": 4.5279161081006076e-05,
      "loss": 1.4861,
      "step": 8684
    },
    {
      "epoch": 0.9161392405063291,
      "grad_norm": 0.44030308723449707,
      "learning_rate": 4.521636011517162e-05,
      "loss": 1.4806,
      "step": 8685
    },
    {
      "epoch": 0.9162447257383967,
      "grad_norm": 0.4233447313308716,
      "learning_rate": 4.515364625256998e-05,
      "loss": 1.4938,
      "step": 8686
    },
    {
      "epoch": 0.9163502109704641,
      "grad_norm": 0.4525858461856842,
      "learning_rate": 4.5091019372391345e-05,
      "loss": 1.4813,
      "step": 8687
    },
    {
      "epoch": 0.9164556962025316,
      "grad_norm": 0.4176623225212097,
      "learning_rate": 4.502847935399348e-05,
      "loss": 1.4965,
      "step": 8688
    },
    {
      "epoch": 0.9165611814345992,
      "grad_norm": 0.4262511134147644,
      "learning_rate": 4.496602607690141e-05,
      "loss": 1.5199,
      "step": 8689
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.4176427721977234,
      "learning_rate": 4.490365942080736e-05,
      "loss": 1.4616,
      "step": 8690
    },
    {
      "epoch": 0.9167721518987342,
      "grad_norm": 0.4170861542224884,
      "learning_rate": 4.48413792655703e-05,
      "loss": 1.5228,
      "step": 8691
    },
    {
      "epoch": 0.9168776371308017,
      "grad_norm": 0.42994213104248047,
      "learning_rate": 4.4779185491215926e-05,
      "loss": 1.4898,
      "step": 8692
    },
    {
      "epoch": 0.9169831223628692,
      "grad_norm": 0.42551156878471375,
      "learning_rate": 4.471707797793631e-05,
      "loss": 1.4855,
      "step": 8693
    },
    {
      "epoch": 0.9170886075949367,
      "grad_norm": 0.41395485401153564,
      "learning_rate": 4.465505660608965e-05,
      "loss": 1.5009,
      "step": 8694
    },
    {
      "epoch": 0.9171940928270043,
      "grad_norm": 0.4158569574356079,
      "learning_rate": 4.459312125620017e-05,
      "loss": 1.4634,
      "step": 8695
    },
    {
      "epoch": 0.9172995780590717,
      "grad_norm": 0.4420301616191864,
      "learning_rate": 4.4531271808957704e-05,
      "loss": 1.4883,
      "step": 8696
    },
    {
      "epoch": 0.9174050632911392,
      "grad_norm": 0.44408783316612244,
      "learning_rate": 4.4469508145217626e-05,
      "loss": 1.473,
      "step": 8697
    },
    {
      "epoch": 0.9175105485232068,
      "grad_norm": 0.44045335054397583,
      "learning_rate": 4.440783014600059e-05,
      "loss": 1.4707,
      "step": 8698
    },
    {
      "epoch": 0.9176160337552742,
      "grad_norm": 0.4094163477420807,
      "learning_rate": 4.434623769249217e-05,
      "loss": 1.5064,
      "step": 8699
    },
    {
      "epoch": 0.9177215189873418,
      "grad_norm": 0.4268554151058197,
      "learning_rate": 4.428473066604285e-05,
      "loss": 1.4774,
      "step": 8700
    },
    {
      "epoch": 0.9178270042194093,
      "grad_norm": 0.45362943410873413,
      "learning_rate": 4.422330894816757e-05,
      "loss": 1.4788,
      "step": 8701
    },
    {
      "epoch": 0.9179324894514768,
      "grad_norm": 0.4464951753616333,
      "learning_rate": 4.4161972420545684e-05,
      "loss": 1.4517,
      "step": 8702
    },
    {
      "epoch": 0.9180379746835443,
      "grad_norm": 0.44934725761413574,
      "learning_rate": 4.410072096502064e-05,
      "loss": 1.5036,
      "step": 8703
    },
    {
      "epoch": 0.9181434599156119,
      "grad_norm": 0.4298765957355499,
      "learning_rate": 4.403955446359971e-05,
      "loss": 1.4768,
      "step": 8704
    },
    {
      "epoch": 0.9182489451476793,
      "grad_norm": 0.42397987842559814,
      "learning_rate": 4.397847279845391e-05,
      "loss": 1.5047,
      "step": 8705
    },
    {
      "epoch": 0.9183544303797468,
      "grad_norm": 0.4378174841403961,
      "learning_rate": 4.391747585191758e-05,
      "loss": 1.4753,
      "step": 8706
    },
    {
      "epoch": 0.9184599156118144,
      "grad_norm": 0.4254377782344818,
      "learning_rate": 4.385656350648834e-05,
      "loss": 1.494,
      "step": 8707
    },
    {
      "epoch": 0.9185654008438818,
      "grad_norm": 0.40909865498542786,
      "learning_rate": 4.3795735644826776e-05,
      "loss": 1.4708,
      "step": 8708
    },
    {
      "epoch": 0.9186708860759494,
      "grad_norm": 0.41189640760421753,
      "learning_rate": 4.373499214975615e-05,
      "loss": 1.4893,
      "step": 8709
    },
    {
      "epoch": 0.9187763713080169,
      "grad_norm": 0.4533466696739197,
      "learning_rate": 4.367433290426233e-05,
      "loss": 1.5144,
      "step": 8710
    },
    {
      "epoch": 0.9188818565400844,
      "grad_norm": 0.49745699763298035,
      "learning_rate": 4.361375779149342e-05,
      "loss": 1.4996,
      "step": 8711
    },
    {
      "epoch": 0.9189873417721519,
      "grad_norm": 0.44249609112739563,
      "learning_rate": 4.3553266694759614e-05,
      "loss": 1.4717,
      "step": 8712
    },
    {
      "epoch": 0.9190928270042195,
      "grad_norm": 0.4108584523200989,
      "learning_rate": 4.3492859497533e-05,
      "loss": 1.4927,
      "step": 8713
    },
    {
      "epoch": 0.9191983122362869,
      "grad_norm": 0.4718115031719208,
      "learning_rate": 4.343253608344718e-05,
      "loss": 1.4801,
      "step": 8714
    },
    {
      "epoch": 0.9193037974683544,
      "grad_norm": 0.42481932044029236,
      "learning_rate": 4.337229633629727e-05,
      "loss": 1.4924,
      "step": 8715
    },
    {
      "epoch": 0.919409282700422,
      "grad_norm": 0.45272520184516907,
      "learning_rate": 4.3312140140039447e-05,
      "loss": 1.4589,
      "step": 8716
    },
    {
      "epoch": 0.9195147679324894,
      "grad_norm": 0.4103776514530182,
      "learning_rate": 4.3252067378790934e-05,
      "loss": 1.4301,
      "step": 8717
    },
    {
      "epoch": 0.919620253164557,
      "grad_norm": 0.4505520761013031,
      "learning_rate": 4.319207793682965e-05,
      "loss": 1.5018,
      "step": 8718
    },
    {
      "epoch": 0.9197257383966245,
      "grad_norm": 0.4533357322216034,
      "learning_rate": 4.313217169859397e-05,
      "loss": 1.4859,
      "step": 8719
    },
    {
      "epoch": 0.919831223628692,
      "grad_norm": 0.4405952990055084,
      "learning_rate": 4.3072348548682595e-05,
      "loss": 1.4656,
      "step": 8720
    },
    {
      "epoch": 0.9199367088607595,
      "grad_norm": 0.43280160427093506,
      "learning_rate": 4.3012608371854326e-05,
      "loss": 1.4976,
      "step": 8721
    },
    {
      "epoch": 0.9200421940928271,
      "grad_norm": 0.4532991051673889,
      "learning_rate": 4.2952951053027684e-05,
      "loss": 1.4918,
      "step": 8722
    },
    {
      "epoch": 0.9201476793248945,
      "grad_norm": 0.41342782974243164,
      "learning_rate": 4.2893376477280934e-05,
      "loss": 1.5023,
      "step": 8723
    },
    {
      "epoch": 0.920253164556962,
      "grad_norm": 0.4268062710762024,
      "learning_rate": 4.283388452985162e-05,
      "loss": 1.5335,
      "step": 8724
    },
    {
      "epoch": 0.9203586497890295,
      "grad_norm": 0.4370148181915283,
      "learning_rate": 4.2774475096136525e-05,
      "loss": 1.4778,
      "step": 8725
    },
    {
      "epoch": 0.920464135021097,
      "grad_norm": 0.4215529263019562,
      "learning_rate": 4.271514806169141e-05,
      "loss": 1.442,
      "step": 8726
    },
    {
      "epoch": 0.9205696202531646,
      "grad_norm": 0.43270763754844666,
      "learning_rate": 4.265590331223067e-05,
      "loss": 1.4888,
      "step": 8727
    },
    {
      "epoch": 0.920675105485232,
      "grad_norm": 0.41159388422966003,
      "learning_rate": 4.259674073362732e-05,
      "loss": 1.4481,
      "step": 8728
    },
    {
      "epoch": 0.9207805907172996,
      "grad_norm": 0.4133281111717224,
      "learning_rate": 4.253766021191256e-05,
      "loss": 1.54,
      "step": 8729
    },
    {
      "epoch": 0.9208860759493671,
      "grad_norm": 0.42678168416023254,
      "learning_rate": 4.247866163327575e-05,
      "loss": 1.4846,
      "step": 8730
    },
    {
      "epoch": 0.9209915611814345,
      "grad_norm": 0.41132545471191406,
      "learning_rate": 4.241974488406408e-05,
      "loss": 1.4704,
      "step": 8731
    },
    {
      "epoch": 0.9210970464135021,
      "grad_norm": 0.41572389006614685,
      "learning_rate": 4.236090985078232e-05,
      "loss": 1.4853,
      "step": 8732
    },
    {
      "epoch": 0.9212025316455696,
      "grad_norm": 0.4238037168979645,
      "learning_rate": 4.230215642009275e-05,
      "loss": 1.5133,
      "step": 8733
    },
    {
      "epoch": 0.9213080168776371,
      "grad_norm": 0.43294909596443176,
      "learning_rate": 4.224348447881473e-05,
      "loss": 1.4609,
      "step": 8734
    },
    {
      "epoch": 0.9214135021097046,
      "grad_norm": 0.4385758638381958,
      "learning_rate": 4.218489391392469e-05,
      "loss": 1.466,
      "step": 8735
    },
    {
      "epoch": 0.9215189873417722,
      "grad_norm": 0.4836699664592743,
      "learning_rate": 4.212638461255582e-05,
      "loss": 1.4657,
      "step": 8736
    },
    {
      "epoch": 0.9216244725738396,
      "grad_norm": 0.45895034074783325,
      "learning_rate": 4.206795646199778e-05,
      "loss": 1.4668,
      "step": 8737
    },
    {
      "epoch": 0.9217299578059072,
      "grad_norm": 0.4387611150741577,
      "learning_rate": 4.200960934969664e-05,
      "loss": 1.5008,
      "step": 8738
    },
    {
      "epoch": 0.9218354430379747,
      "grad_norm": 0.3952721655368805,
      "learning_rate": 4.19513431632545e-05,
      "loss": 1.4963,
      "step": 8739
    },
    {
      "epoch": 0.9219409282700421,
      "grad_norm": 0.44427233934402466,
      "learning_rate": 4.1893157790429404e-05,
      "loss": 1.4801,
      "step": 8740
    },
    {
      "epoch": 0.9220464135021097,
      "grad_norm": 0.47141897678375244,
      "learning_rate": 4.1835053119135096e-05,
      "loss": 1.4709,
      "step": 8741
    },
    {
      "epoch": 0.9221518987341772,
      "grad_norm": 0.40252095460891724,
      "learning_rate": 4.17770290374407e-05,
      "loss": 1.4659,
      "step": 8742
    },
    {
      "epoch": 0.9222573839662447,
      "grad_norm": 0.42490851879119873,
      "learning_rate": 4.171908543357067e-05,
      "loss": 1.423,
      "step": 8743
    },
    {
      "epoch": 0.9223628691983122,
      "grad_norm": 0.43121129274368286,
      "learning_rate": 4.166122219590441e-05,
      "loss": 1.4882,
      "step": 8744
    },
    {
      "epoch": 0.9224683544303798,
      "grad_norm": 0.4186294972896576,
      "learning_rate": 4.1603439212976205e-05,
      "loss": 1.5205,
      "step": 8745
    },
    {
      "epoch": 0.9225738396624472,
      "grad_norm": 0.4393500089645386,
      "learning_rate": 4.1545736373474935e-05,
      "loss": 1.4874,
      "step": 8746
    },
    {
      "epoch": 0.9226793248945148,
      "grad_norm": 0.4342838227748871,
      "learning_rate": 4.148811356624379e-05,
      "loss": 1.477,
      "step": 8747
    },
    {
      "epoch": 0.9227848101265823,
      "grad_norm": 0.4205479323863983,
      "learning_rate": 4.143057068028024e-05,
      "loss": 1.4729,
      "step": 8748
    },
    {
      "epoch": 0.9228902953586497,
      "grad_norm": 0.44601762294769287,
      "learning_rate": 4.1373107604735626e-05,
      "loss": 1.4914,
      "step": 8749
    },
    {
      "epoch": 0.9229957805907173,
      "grad_norm": 0.4681358337402344,
      "learning_rate": 4.1315724228915066e-05,
      "loss": 1.5047,
      "step": 8750
    },
    {
      "epoch": 0.9231012658227848,
      "grad_norm": 0.4017626643180847,
      "learning_rate": 4.125842044227725e-05,
      "loss": 1.4495,
      "step": 8751
    },
    {
      "epoch": 0.9232067510548523,
      "grad_norm": 0.4187672436237335,
      "learning_rate": 4.120119613443409e-05,
      "loss": 1.5042,
      "step": 8752
    },
    {
      "epoch": 0.9233122362869198,
      "grad_norm": 0.4254954159259796,
      "learning_rate": 4.114405119515069e-05,
      "loss": 1.4548,
      "step": 8753
    },
    {
      "epoch": 0.9234177215189874,
      "grad_norm": 0.43452268838882446,
      "learning_rate": 4.1086985514344996e-05,
      "loss": 1.4782,
      "step": 8754
    },
    {
      "epoch": 0.9235232067510548,
      "grad_norm": 0.4373135566711426,
      "learning_rate": 4.102999898208766e-05,
      "loss": 1.4828,
      "step": 8755
    },
    {
      "epoch": 0.9236286919831224,
      "grad_norm": 0.40655672550201416,
      "learning_rate": 4.0973091488601826e-05,
      "loss": 1.4741,
      "step": 8756
    },
    {
      "epoch": 0.9237341772151899,
      "grad_norm": 0.4433337450027466,
      "learning_rate": 4.091626292426282e-05,
      "loss": 1.4806,
      "step": 8757
    },
    {
      "epoch": 0.9238396624472573,
      "grad_norm": 0.4077359735965729,
      "learning_rate": 4.0859513179598096e-05,
      "loss": 1.4127,
      "step": 8758
    },
    {
      "epoch": 0.9239451476793249,
      "grad_norm": 0.4369218945503235,
      "learning_rate": 4.0802842145286876e-05,
      "loss": 1.4951,
      "step": 8759
    },
    {
      "epoch": 0.9240506329113924,
      "grad_norm": 0.45306119322776794,
      "learning_rate": 4.074624971216005e-05,
      "loss": 1.4595,
      "step": 8760
    },
    {
      "epoch": 0.9241561181434599,
      "grad_norm": 0.43634605407714844,
      "learning_rate": 4.0689735771199944e-05,
      "loss": 1.4991,
      "step": 8761
    },
    {
      "epoch": 0.9242616033755274,
      "grad_norm": 0.39773571491241455,
      "learning_rate": 4.0633300213540004e-05,
      "loss": 1.4482,
      "step": 8762
    },
    {
      "epoch": 0.924367088607595,
      "grad_norm": 0.44319018721580505,
      "learning_rate": 4.057694293046475e-05,
      "loss": 1.5368,
      "step": 8763
    },
    {
      "epoch": 0.9244725738396624,
      "grad_norm": 0.4130871593952179,
      "learning_rate": 4.052066381340948e-05,
      "loss": 1.4888,
      "step": 8764
    },
    {
      "epoch": 0.92457805907173,
      "grad_norm": 0.43742549419403076,
      "learning_rate": 4.0464462753960006e-05,
      "loss": 1.4546,
      "step": 8765
    },
    {
      "epoch": 0.9246835443037975,
      "grad_norm": 0.41864845156669617,
      "learning_rate": 4.040833964385259e-05,
      "loss": 1.4767,
      "step": 8766
    },
    {
      "epoch": 0.924789029535865,
      "grad_norm": 0.4739699959754944,
      "learning_rate": 4.035229437497357e-05,
      "loss": 1.4836,
      "step": 8767
    },
    {
      "epoch": 0.9248945147679325,
      "grad_norm": 0.40845295786857605,
      "learning_rate": 4.02963268393593e-05,
      "loss": 1.4727,
      "step": 8768
    },
    {
      "epoch": 0.925,
      "grad_norm": 0.4212695360183716,
      "learning_rate": 4.024043692919589e-05,
      "loss": 1.4748,
      "step": 8769
    },
    {
      "epoch": 0.9251054852320675,
      "grad_norm": 0.4217427372932434,
      "learning_rate": 4.018462453681889e-05,
      "loss": 1.4581,
      "step": 8770
    },
    {
      "epoch": 0.925210970464135,
      "grad_norm": 0.40990760922431946,
      "learning_rate": 4.0128889554713276e-05,
      "loss": 1.5143,
      "step": 8771
    },
    {
      "epoch": 0.9253164556962026,
      "grad_norm": 0.45956286787986755,
      "learning_rate": 4.007323187551308e-05,
      "loss": 1.4487,
      "step": 8772
    },
    {
      "epoch": 0.92542194092827,
      "grad_norm": 0.45178842544555664,
      "learning_rate": 4.0017651392001285e-05,
      "loss": 1.5012,
      "step": 8773
    },
    {
      "epoch": 0.9255274261603376,
      "grad_norm": 0.42147213220596313,
      "learning_rate": 3.9962147997109584e-05,
      "loss": 1.4642,
      "step": 8774
    },
    {
      "epoch": 0.9256329113924051,
      "grad_norm": 0.44957101345062256,
      "learning_rate": 3.990672158391812e-05,
      "loss": 1.497,
      "step": 8775
    },
    {
      "epoch": 0.9257383966244725,
      "grad_norm": 0.4100128710269928,
      "learning_rate": 3.9851372045655414e-05,
      "loss": 1.4838,
      "step": 8776
    },
    {
      "epoch": 0.9258438818565401,
      "grad_norm": 0.4358738660812378,
      "learning_rate": 3.979609927569798e-05,
      "loss": 1.4819,
      "step": 8777
    },
    {
      "epoch": 0.9259493670886076,
      "grad_norm": 0.4237467646598816,
      "learning_rate": 3.974090316757029e-05,
      "loss": 1.4763,
      "step": 8778
    },
    {
      "epoch": 0.9260548523206751,
      "grad_norm": 0.4317324757575989,
      "learning_rate": 3.968578361494449e-05,
      "loss": 1.5093,
      "step": 8779
    },
    {
      "epoch": 0.9261603375527426,
      "grad_norm": 0.42482826113700867,
      "learning_rate": 3.963074051164014e-05,
      "loss": 1.4984,
      "step": 8780
    },
    {
      "epoch": 0.9262658227848102,
      "grad_norm": 0.44853195548057556,
      "learning_rate": 3.957577375162414e-05,
      "loss": 1.4992,
      "step": 8781
    },
    {
      "epoch": 0.9263713080168776,
      "grad_norm": 0.4298540949821472,
      "learning_rate": 3.952088322901039e-05,
      "loss": 1.521,
      "step": 8782
    },
    {
      "epoch": 0.9264767932489452,
      "grad_norm": 0.4193238615989685,
      "learning_rate": 3.946606883805971e-05,
      "loss": 1.4506,
      "step": 8783
    },
    {
      "epoch": 0.9265822784810127,
      "grad_norm": 0.432110995054245,
      "learning_rate": 3.941133047317956e-05,
      "loss": 1.4679,
      "step": 8784
    },
    {
      "epoch": 0.9266877637130801,
      "grad_norm": 0.4807945191860199,
      "learning_rate": 3.9356668028923825e-05,
      "loss": 1.4871,
      "step": 8785
    },
    {
      "epoch": 0.9267932489451477,
      "grad_norm": 0.4376818835735321,
      "learning_rate": 3.930208139999269e-05,
      "loss": 1.4673,
      "step": 8786
    },
    {
      "epoch": 0.9268987341772152,
      "grad_norm": 0.40703439712524414,
      "learning_rate": 3.9247570481232314e-05,
      "loss": 1.4521,
      "step": 8787
    },
    {
      "epoch": 0.9270042194092827,
      "grad_norm": 0.41416510939598083,
      "learning_rate": 3.919313516763478e-05,
      "loss": 1.4776,
      "step": 8788
    },
    {
      "epoch": 0.9271097046413502,
      "grad_norm": 0.40741485357284546,
      "learning_rate": 3.91387753543378e-05,
      "loss": 1.4904,
      "step": 8789
    },
    {
      "epoch": 0.9272151898734177,
      "grad_norm": 0.4166701138019562,
      "learning_rate": 3.908449093662446e-05,
      "loss": 1.493,
      "step": 8790
    },
    {
      "epoch": 0.9273206751054852,
      "grad_norm": 0.40980982780456543,
      "learning_rate": 3.9030281809923186e-05,
      "loss": 1.5011,
      "step": 8791
    },
    {
      "epoch": 0.9274261603375528,
      "grad_norm": 0.4167083203792572,
      "learning_rate": 3.897614786980734e-05,
      "loss": 1.4579,
      "step": 8792
    },
    {
      "epoch": 0.9275316455696202,
      "grad_norm": 0.4176901876926422,
      "learning_rate": 3.892208901199521e-05,
      "loss": 1.4854,
      "step": 8793
    },
    {
      "epoch": 0.9276371308016877,
      "grad_norm": 0.41362348198890686,
      "learning_rate": 3.886810513234967e-05,
      "loss": 1.4591,
      "step": 8794
    },
    {
      "epoch": 0.9277426160337553,
      "grad_norm": 0.42431047558784485,
      "learning_rate": 3.881419612687803e-05,
      "loss": 1.47,
      "step": 8795
    },
    {
      "epoch": 0.9278481012658227,
      "grad_norm": 0.42078936100006104,
      "learning_rate": 3.8760361891731874e-05,
      "loss": 1.4951,
      "step": 8796
    },
    {
      "epoch": 0.9279535864978903,
      "grad_norm": 0.43294990062713623,
      "learning_rate": 3.870660232320675e-05,
      "loss": 1.4807,
      "step": 8797
    },
    {
      "epoch": 0.9280590717299578,
      "grad_norm": 0.4440297782421112,
      "learning_rate": 3.8652917317742106e-05,
      "loss": 1.4889,
      "step": 8798
    },
    {
      "epoch": 0.9281645569620253,
      "grad_norm": 0.42233535647392273,
      "learning_rate": 3.859930677192103e-05,
      "loss": 1.4586,
      "step": 8799
    },
    {
      "epoch": 0.9282700421940928,
      "grad_norm": 0.40886858105659485,
      "learning_rate": 3.854577058246998e-05,
      "loss": 1.4785,
      "step": 8800
    },
    {
      "epoch": 0.9283755274261604,
      "grad_norm": 0.43115973472595215,
      "learning_rate": 3.8492308646258714e-05,
      "loss": 1.5125,
      "step": 8801
    },
    {
      "epoch": 0.9284810126582278,
      "grad_norm": 0.44896188378334045,
      "learning_rate": 3.843892086029999e-05,
      "loss": 1.4777,
      "step": 8802
    },
    {
      "epoch": 0.9285864978902953,
      "grad_norm": 0.42077797651290894,
      "learning_rate": 3.8385607121749426e-05,
      "loss": 1.4761,
      "step": 8803
    },
    {
      "epoch": 0.9286919831223629,
      "grad_norm": 0.42118996381759644,
      "learning_rate": 3.83323673279053e-05,
      "loss": 1.4781,
      "step": 8804
    },
    {
      "epoch": 0.9287974683544303,
      "grad_norm": 0.4214320480823517,
      "learning_rate": 3.827920137620828e-05,
      "loss": 1.483,
      "step": 8805
    },
    {
      "epoch": 0.9289029535864979,
      "grad_norm": 0.4134511947631836,
      "learning_rate": 3.822610916424134e-05,
      "loss": 1.4853,
      "step": 8806
    },
    {
      "epoch": 0.9290084388185654,
      "grad_norm": 0.42818230390548706,
      "learning_rate": 3.81730905897295e-05,
      "loss": 1.4904,
      "step": 8807
    },
    {
      "epoch": 0.9291139240506329,
      "grad_norm": 0.44591575860977173,
      "learning_rate": 3.812014555053955e-05,
      "loss": 1.5103,
      "step": 8808
    },
    {
      "epoch": 0.9292194092827004,
      "grad_norm": 0.4346730411052704,
      "learning_rate": 3.806727394468005e-05,
      "loss": 1.518,
      "step": 8809
    },
    {
      "epoch": 0.929324894514768,
      "grad_norm": 0.4165668785572052,
      "learning_rate": 3.801447567030094e-05,
      "loss": 1.5227,
      "step": 8810
    },
    {
      "epoch": 0.9294303797468354,
      "grad_norm": 0.4201684892177582,
      "learning_rate": 3.796175062569344e-05,
      "loss": 1.4928,
      "step": 8811
    },
    {
      "epoch": 0.929535864978903,
      "grad_norm": 0.4057765305042267,
      "learning_rate": 3.790909870928989e-05,
      "loss": 1.4675,
      "step": 8812
    },
    {
      "epoch": 0.9296413502109705,
      "grad_norm": 0.4186115562915802,
      "learning_rate": 3.785651981966341e-05,
      "loss": 1.4885,
      "step": 8813
    },
    {
      "epoch": 0.9297468354430379,
      "grad_norm": 0.4474346339702606,
      "learning_rate": 3.7804013855527896e-05,
      "loss": 1.5056,
      "step": 8814
    },
    {
      "epoch": 0.9298523206751055,
      "grad_norm": 0.3840869963169098,
      "learning_rate": 3.7751580715737614e-05,
      "loss": 1.4743,
      "step": 8815
    },
    {
      "epoch": 0.929957805907173,
      "grad_norm": 0.4033017158508301,
      "learning_rate": 3.7699220299287214e-05,
      "loss": 1.4991,
      "step": 8816
    },
    {
      "epoch": 0.9300632911392405,
      "grad_norm": 0.40306559205055237,
      "learning_rate": 3.764693250531141e-05,
      "loss": 1.4951,
      "step": 8817
    },
    {
      "epoch": 0.930168776371308,
      "grad_norm": 0.42635712027549744,
      "learning_rate": 3.759471723308477e-05,
      "loss": 1.4819,
      "step": 8818
    },
    {
      "epoch": 0.9302742616033756,
      "grad_norm": 0.4280884265899658,
      "learning_rate": 3.7542574382021635e-05,
      "loss": 1.4658,
      "step": 8819
    },
    {
      "epoch": 0.930379746835443,
      "grad_norm": 0.4127960503101349,
      "learning_rate": 3.7490503851675777e-05,
      "loss": 1.4662,
      "step": 8820
    },
    {
      "epoch": 0.9304852320675105,
      "grad_norm": 0.4023503065109253,
      "learning_rate": 3.7438505541740356e-05,
      "loss": 1.4964,
      "step": 8821
    },
    {
      "epoch": 0.9305907172995781,
      "grad_norm": 0.41875559091567993,
      "learning_rate": 3.738657935204763e-05,
      "loss": 1.5565,
      "step": 8822
    },
    {
      "epoch": 0.9306962025316455,
      "grad_norm": 0.4306015968322754,
      "learning_rate": 3.733472518256876e-05,
      "loss": 1.4737,
      "step": 8823
    },
    {
      "epoch": 0.9308016877637131,
      "grad_norm": 0.41580015420913696,
      "learning_rate": 3.7282942933413696e-05,
      "loss": 1.5252,
      "step": 8824
    },
    {
      "epoch": 0.9309071729957806,
      "grad_norm": 0.41895830631256104,
      "learning_rate": 3.723123250483086e-05,
      "loss": 1.4942,
      "step": 8825
    },
    {
      "epoch": 0.9310126582278481,
      "grad_norm": 0.4369781017303467,
      "learning_rate": 3.717959379720711e-05,
      "loss": 1.4983,
      "step": 8826
    },
    {
      "epoch": 0.9311181434599156,
      "grad_norm": 0.3920938968658447,
      "learning_rate": 3.712802671106742e-05,
      "loss": 1.4917,
      "step": 8827
    },
    {
      "epoch": 0.9312236286919832,
      "grad_norm": 0.42213407158851624,
      "learning_rate": 3.707653114707471e-05,
      "loss": 1.5117,
      "step": 8828
    },
    {
      "epoch": 0.9313291139240506,
      "grad_norm": 0.458345502614975,
      "learning_rate": 3.702510700602975e-05,
      "loss": 1.4798,
      "step": 8829
    },
    {
      "epoch": 0.9314345991561181,
      "grad_norm": 0.4231816232204437,
      "learning_rate": 3.6973754188870806e-05,
      "loss": 1.4565,
      "step": 8830
    },
    {
      "epoch": 0.9315400843881857,
      "grad_norm": 0.40538638830184937,
      "learning_rate": 3.692247259667361e-05,
      "loss": 1.4443,
      "step": 8831
    },
    {
      "epoch": 0.9316455696202531,
      "grad_norm": 0.3995458781719208,
      "learning_rate": 3.687126213065109e-05,
      "loss": 1.5157,
      "step": 8832
    },
    {
      "epoch": 0.9317510548523207,
      "grad_norm": 0.4182884693145752,
      "learning_rate": 3.682012269215314e-05,
      "loss": 1.4712,
      "step": 8833
    },
    {
      "epoch": 0.9318565400843882,
      "grad_norm": 0.40237295627593994,
      "learning_rate": 3.676905418266655e-05,
      "loss": 1.4699,
      "step": 8834
    },
    {
      "epoch": 0.9319620253164557,
      "grad_norm": 0.4152362048625946,
      "learning_rate": 3.671805650381468e-05,
      "loss": 1.4907,
      "step": 8835
    },
    {
      "epoch": 0.9320675105485232,
      "grad_norm": 0.44561782479286194,
      "learning_rate": 3.666712955735737e-05,
      "loss": 1.5013,
      "step": 8836
    },
    {
      "epoch": 0.9321729957805908,
      "grad_norm": 0.4342091679573059,
      "learning_rate": 3.661627324519074e-05,
      "loss": 1.508,
      "step": 8837
    },
    {
      "epoch": 0.9322784810126582,
      "grad_norm": 0.39658495783805847,
      "learning_rate": 3.6565487469346904e-05,
      "loss": 1.501,
      "step": 8838
    },
    {
      "epoch": 0.9323839662447257,
      "grad_norm": 0.4108422100543976,
      "learning_rate": 3.651477213199394e-05,
      "loss": 1.4541,
      "step": 8839
    },
    {
      "epoch": 0.9324894514767933,
      "grad_norm": 0.4589267671108246,
      "learning_rate": 3.6464127135435536e-05,
      "loss": 1.4936,
      "step": 8840
    },
    {
      "epoch": 0.9325949367088607,
      "grad_norm": 0.3822910189628601,
      "learning_rate": 3.641355238211095e-05,
      "loss": 1.462,
      "step": 8841
    },
    {
      "epoch": 0.9327004219409283,
      "grad_norm": 0.41139668226242065,
      "learning_rate": 3.6363047774594736e-05,
      "loss": 1.5275,
      "step": 8842
    },
    {
      "epoch": 0.9328059071729958,
      "grad_norm": 0.3933762311935425,
      "learning_rate": 3.631261321559652e-05,
      "loss": 1.4796,
      "step": 8843
    },
    {
      "epoch": 0.9329113924050633,
      "grad_norm": 0.4006390869617462,
      "learning_rate": 3.626224860796096e-05,
      "loss": 1.4826,
      "step": 8844
    },
    {
      "epoch": 0.9330168776371308,
      "grad_norm": 0.4192161560058594,
      "learning_rate": 3.6211953854667373e-05,
      "loss": 1.4709,
      "step": 8845
    },
    {
      "epoch": 0.9331223628691984,
      "grad_norm": 0.4311361610889435,
      "learning_rate": 3.616172885882972e-05,
      "loss": 1.4792,
      "step": 8846
    },
    {
      "epoch": 0.9332278481012658,
      "grad_norm": 0.3900858461856842,
      "learning_rate": 3.6111573523696295e-05,
      "loss": 1.436,
      "step": 8847
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 0.40646398067474365,
      "learning_rate": 3.606148775264958e-05,
      "loss": 1.4986,
      "step": 8848
    },
    {
      "epoch": 0.9334388185654009,
      "grad_norm": 0.4317643642425537,
      "learning_rate": 3.601147144920609e-05,
      "loss": 1.5078,
      "step": 8849
    },
    {
      "epoch": 0.9335443037974683,
      "grad_norm": 0.46351680159568787,
      "learning_rate": 3.596152451701616e-05,
      "loss": 1.4602,
      "step": 8850
    },
    {
      "epoch": 0.9336497890295359,
      "grad_norm": 0.45119708776474,
      "learning_rate": 3.591164685986372e-05,
      "loss": 1.4958,
      "step": 8851
    },
    {
      "epoch": 0.9337552742616034,
      "grad_norm": 0.4067215323448181,
      "learning_rate": 3.58618383816662e-05,
      "loss": 1.4883,
      "step": 8852
    },
    {
      "epoch": 0.9338607594936709,
      "grad_norm": 0.4363328516483307,
      "learning_rate": 3.581209898647425e-05,
      "loss": 1.4718,
      "step": 8853
    },
    {
      "epoch": 0.9339662447257384,
      "grad_norm": 0.4221624433994293,
      "learning_rate": 3.576242857847162e-05,
      "loss": 1.4761,
      "step": 8854
    },
    {
      "epoch": 0.9340717299578059,
      "grad_norm": 0.4309861958026886,
      "learning_rate": 3.5712827061974984e-05,
      "loss": 1.5065,
      "step": 8855
    },
    {
      "epoch": 0.9341772151898734,
      "grad_norm": 0.44510766863822937,
      "learning_rate": 3.566329434143365e-05,
      "loss": 1.4844,
      "step": 8856
    },
    {
      "epoch": 0.934282700421941,
      "grad_norm": 0.411089688539505,
      "learning_rate": 3.5613830321429545e-05,
      "loss": 1.4546,
      "step": 8857
    },
    {
      "epoch": 0.9343881856540084,
      "grad_norm": 0.41197821497917175,
      "learning_rate": 3.5564434906676834e-05,
      "loss": 1.5001,
      "step": 8858
    },
    {
      "epoch": 0.9344936708860759,
      "grad_norm": 0.4134894013404846,
      "learning_rate": 3.5515108002021946e-05,
      "loss": 1.4967,
      "step": 8859
    },
    {
      "epoch": 0.9345991561181435,
      "grad_norm": 0.4444829523563385,
      "learning_rate": 3.5465849512443226e-05,
      "loss": 1.4623,
      "step": 8860
    },
    {
      "epoch": 0.9347046413502109,
      "grad_norm": 0.4189072251319885,
      "learning_rate": 3.541665934305081e-05,
      "loss": 1.4498,
      "step": 8861
    },
    {
      "epoch": 0.9348101265822785,
      "grad_norm": 0.3960464894771576,
      "learning_rate": 3.5367537399086476e-05,
      "loss": 1.5137,
      "step": 8862
    },
    {
      "epoch": 0.934915611814346,
      "grad_norm": 0.4477297067642212,
      "learning_rate": 3.531848358592338e-05,
      "loss": 1.4866,
      "step": 8863
    },
    {
      "epoch": 0.9350210970464135,
      "grad_norm": 0.40495413541793823,
      "learning_rate": 3.5269497809065966e-05,
      "loss": 1.4682,
      "step": 8864
    },
    {
      "epoch": 0.935126582278481,
      "grad_norm": 0.4138062298297882,
      "learning_rate": 3.522057997414975e-05,
      "loss": 1.5021,
      "step": 8865
    },
    {
      "epoch": 0.9352320675105485,
      "grad_norm": 0.4065333306789398,
      "learning_rate": 3.517172998694108e-05,
      "loss": 1.4902,
      "step": 8866
    },
    {
      "epoch": 0.935337552742616,
      "grad_norm": 0.42695364356040955,
      "learning_rate": 3.512294775333705e-05,
      "loss": 1.5104,
      "step": 8867
    },
    {
      "epoch": 0.9354430379746835,
      "grad_norm": 0.41356247663497925,
      "learning_rate": 3.507423317936521e-05,
      "loss": 1.4824,
      "step": 8868
    },
    {
      "epoch": 0.9355485232067511,
      "grad_norm": 0.44853776693344116,
      "learning_rate": 3.502558617118352e-05,
      "loss": 1.4828,
      "step": 8869
    },
    {
      "epoch": 0.9356540084388185,
      "grad_norm": 0.4080412983894348,
      "learning_rate": 3.4977006635080086e-05,
      "loss": 1.448,
      "step": 8870
    },
    {
      "epoch": 0.9357594936708861,
      "grad_norm": 0.40535593032836914,
      "learning_rate": 3.4928494477472926e-05,
      "loss": 1.5186,
      "step": 8871
    },
    {
      "epoch": 0.9358649789029536,
      "grad_norm": 0.4081481099128723,
      "learning_rate": 3.488004960490994e-05,
      "loss": 1.4428,
      "step": 8872
    },
    {
      "epoch": 0.935970464135021,
      "grad_norm": 0.4423657953739166,
      "learning_rate": 3.4831671924068555e-05,
      "loss": 1.5068,
      "step": 8873
    },
    {
      "epoch": 0.9360759493670886,
      "grad_norm": 0.4110405147075653,
      "learning_rate": 3.47833613417557e-05,
      "loss": 1.4691,
      "step": 8874
    },
    {
      "epoch": 0.9361814345991561,
      "grad_norm": 0.4406619966030121,
      "learning_rate": 3.473511776490756e-05,
      "loss": 1.4696,
      "step": 8875
    },
    {
      "epoch": 0.9362869198312236,
      "grad_norm": 0.4179662764072418,
      "learning_rate": 3.4686941100589344e-05,
      "loss": 1.4982,
      "step": 8876
    },
    {
      "epoch": 0.9363924050632911,
      "grad_norm": 0.3849756419658661,
      "learning_rate": 3.463883125599521e-05,
      "loss": 1.4852,
      "step": 8877
    },
    {
      "epoch": 0.9364978902953587,
      "grad_norm": 0.43558844923973083,
      "learning_rate": 3.4590788138448004e-05,
      "loss": 1.5128,
      "step": 8878
    },
    {
      "epoch": 0.9366033755274261,
      "grad_norm": 0.4147240221500397,
      "learning_rate": 3.454281165539913e-05,
      "loss": 1.5067,
      "step": 8879
    },
    {
      "epoch": 0.9367088607594937,
      "grad_norm": 0.4129604995250702,
      "learning_rate": 3.449490171442838e-05,
      "loss": 1.4887,
      "step": 8880
    },
    {
      "epoch": 0.9368143459915612,
      "grad_norm": 0.4376642107963562,
      "learning_rate": 3.444705822324364e-05,
      "loss": 1.4898,
      "step": 8881
    },
    {
      "epoch": 0.9369198312236287,
      "grad_norm": 0.4201657176017761,
      "learning_rate": 3.4399281089680924e-05,
      "loss": 1.5274,
      "step": 8882
    },
    {
      "epoch": 0.9370253164556962,
      "grad_norm": 0.46530795097351074,
      "learning_rate": 3.435157022170396e-05,
      "loss": 1.4604,
      "step": 8883
    },
    {
      "epoch": 0.9371308016877637,
      "grad_norm": 0.42180606722831726,
      "learning_rate": 3.430392552740421e-05,
      "loss": 1.4986,
      "step": 8884
    },
    {
      "epoch": 0.9372362869198312,
      "grad_norm": 0.4004462659358978,
      "learning_rate": 3.42563469150006e-05,
      "loss": 1.4714,
      "step": 8885
    },
    {
      "epoch": 0.9373417721518987,
      "grad_norm": 0.43905842304229736,
      "learning_rate": 3.42088342928393e-05,
      "loss": 1.5073,
      "step": 8886
    },
    {
      "epoch": 0.9374472573839663,
      "grad_norm": 0.43236976861953735,
      "learning_rate": 3.416138756939366e-05,
      "loss": 1.4783,
      "step": 8887
    },
    {
      "epoch": 0.9375527426160337,
      "grad_norm": 0.4080219864845276,
      "learning_rate": 3.411400665326393e-05,
      "loss": 1.4877,
      "step": 8888
    },
    {
      "epoch": 0.9376582278481013,
      "grad_norm": 0.43416228890419006,
      "learning_rate": 3.406669145317717e-05,
      "loss": 1.4884,
      "step": 8889
    },
    {
      "epoch": 0.9377637130801688,
      "grad_norm": 0.4268539547920227,
      "learning_rate": 3.401944187798702e-05,
      "loss": 1.4628,
      "step": 8890
    },
    {
      "epoch": 0.9378691983122363,
      "grad_norm": 0.4611803889274597,
      "learning_rate": 3.397225783667351e-05,
      "loss": 1.4823,
      "step": 8891
    },
    {
      "epoch": 0.9379746835443038,
      "grad_norm": 0.42448803782463074,
      "learning_rate": 3.3925139238342954e-05,
      "loss": 1.4854,
      "step": 8892
    },
    {
      "epoch": 0.9380801687763713,
      "grad_norm": 0.4257838726043701,
      "learning_rate": 3.387808599222771e-05,
      "loss": 1.4608,
      "step": 8893
    },
    {
      "epoch": 0.9381856540084388,
      "grad_norm": 0.41752883791923523,
      "learning_rate": 3.383109800768602e-05,
      "loss": 1.4622,
      "step": 8894
    },
    {
      "epoch": 0.9382911392405063,
      "grad_norm": 0.42213907837867737,
      "learning_rate": 3.378417519420188e-05,
      "loss": 1.4974,
      "step": 8895
    },
    {
      "epoch": 0.9383966244725739,
      "grad_norm": 0.42600783705711365,
      "learning_rate": 3.373731746138477e-05,
      "loss": 1.4779,
      "step": 8896
    },
    {
      "epoch": 0.9385021097046413,
      "grad_norm": 0.40594252943992615,
      "learning_rate": 3.3690524718969586e-05,
      "loss": 1.4885,
      "step": 8897
    },
    {
      "epoch": 0.9386075949367089,
      "grad_norm": 0.43008822202682495,
      "learning_rate": 3.364379687681642e-05,
      "loss": 1.4475,
      "step": 8898
    },
    {
      "epoch": 0.9387130801687764,
      "grad_norm": 0.4326474070549011,
      "learning_rate": 3.359713384491036e-05,
      "loss": 1.5037,
      "step": 8899
    },
    {
      "epoch": 0.9388185654008439,
      "grad_norm": 0.4208463430404663,
      "learning_rate": 3.355053553336137e-05,
      "loss": 1.4792,
      "step": 8900
    },
    {
      "epoch": 0.9389240506329114,
      "grad_norm": 0.40826287865638733,
      "learning_rate": 3.350400185240404e-05,
      "loss": 1.4914,
      "step": 8901
    },
    {
      "epoch": 0.939029535864979,
      "grad_norm": 0.41179195046424866,
      "learning_rate": 3.345753271239753e-05,
      "loss": 1.46,
      "step": 8902
    },
    {
      "epoch": 0.9391350210970464,
      "grad_norm": 0.40700075030326843,
      "learning_rate": 3.3411128023825295e-05,
      "loss": 1.4839,
      "step": 8903
    },
    {
      "epoch": 0.9392405063291139,
      "grad_norm": 0.3948523700237274,
      "learning_rate": 3.336478769729492e-05,
      "loss": 1.4832,
      "step": 8904
    },
    {
      "epoch": 0.9393459915611815,
      "grad_norm": 0.45551279187202454,
      "learning_rate": 3.331851164353803e-05,
      "loss": 1.4862,
      "step": 8905
    },
    {
      "epoch": 0.9394514767932489,
      "grad_norm": 0.4131926894187927,
      "learning_rate": 3.327229977341001e-05,
      "loss": 1.5137,
      "step": 8906
    },
    {
      "epoch": 0.9395569620253165,
      "grad_norm": 0.4327172040939331,
      "learning_rate": 3.322615199788992e-05,
      "loss": 1.472,
      "step": 8907
    },
    {
      "epoch": 0.939662447257384,
      "grad_norm": 0.4147562086582184,
      "learning_rate": 3.31800682280803e-05,
      "loss": 1.4664,
      "step": 8908
    },
    {
      "epoch": 0.9397679324894515,
      "grad_norm": 0.4207040071487427,
      "learning_rate": 3.313404837520694e-05,
      "loss": 1.4617,
      "step": 8909
    },
    {
      "epoch": 0.939873417721519,
      "grad_norm": 0.5038466453552246,
      "learning_rate": 3.308809235061882e-05,
      "loss": 1.4536,
      "step": 8910
    },
    {
      "epoch": 0.9399789029535865,
      "grad_norm": 0.4140454828739166,
      "learning_rate": 3.30422000657878e-05,
      "loss": 1.4652,
      "step": 8911
    },
    {
      "epoch": 0.940084388185654,
      "grad_norm": 0.3985019028186798,
      "learning_rate": 3.2996371432308596e-05,
      "loss": 1.5085,
      "step": 8912
    },
    {
      "epoch": 0.9401898734177215,
      "grad_norm": 0.39875873923301697,
      "learning_rate": 3.295060636189853e-05,
      "loss": 1.4854,
      "step": 8913
    },
    {
      "epoch": 0.9402953586497891,
      "grad_norm": 0.4082636535167694,
      "learning_rate": 3.290490476639731e-05,
      "loss": 1.4904,
      "step": 8914
    },
    {
      "epoch": 0.9404008438818565,
      "grad_norm": 0.428849995136261,
      "learning_rate": 3.2859266557767e-05,
      "loss": 1.4726,
      "step": 8915
    },
    {
      "epoch": 0.9405063291139241,
      "grad_norm": 0.4269358515739441,
      "learning_rate": 3.2813691648091704e-05,
      "loss": 1.5335,
      "step": 8916
    },
    {
      "epoch": 0.9406118143459916,
      "grad_norm": 0.4042153060436249,
      "learning_rate": 3.2768179949577505e-05,
      "loss": 1.5043,
      "step": 8917
    },
    {
      "epoch": 0.940717299578059,
      "grad_norm": 0.4402705430984497,
      "learning_rate": 3.272273137455225e-05,
      "loss": 1.4492,
      "step": 8918
    },
    {
      "epoch": 0.9408227848101266,
      "grad_norm": 0.4324682950973511,
      "learning_rate": 3.267734583546536e-05,
      "loss": 1.4817,
      "step": 8919
    },
    {
      "epoch": 0.9409282700421941,
      "grad_norm": 0.4331611394882202,
      "learning_rate": 3.263202324488772e-05,
      "loss": 1.4556,
      "step": 8920
    },
    {
      "epoch": 0.9410337552742616,
      "grad_norm": 0.43215611577033997,
      "learning_rate": 3.258676351551143e-05,
      "loss": 1.4953,
      "step": 8921
    },
    {
      "epoch": 0.9411392405063291,
      "grad_norm": 0.4027799069881439,
      "learning_rate": 3.2541566560149726e-05,
      "loss": 1.4736,
      "step": 8922
    },
    {
      "epoch": 0.9412447257383966,
      "grad_norm": 0.4355621933937073,
      "learning_rate": 3.249643229173678e-05,
      "loss": 1.4826,
      "step": 8923
    },
    {
      "epoch": 0.9413502109704641,
      "grad_norm": 0.40698695182800293,
      "learning_rate": 3.245136062332745e-05,
      "loss": 1.4803,
      "step": 8924
    },
    {
      "epoch": 0.9414556962025317,
      "grad_norm": 0.4355679154396057,
      "learning_rate": 3.240635146809728e-05,
      "loss": 1.5119,
      "step": 8925
    },
    {
      "epoch": 0.9415611814345991,
      "grad_norm": 0.40811678767204285,
      "learning_rate": 3.236140473934215e-05,
      "loss": 1.46,
      "step": 8926
    },
    {
      "epoch": 0.9416666666666667,
      "grad_norm": 0.4243764281272888,
      "learning_rate": 3.231652035047825e-05,
      "loss": 1.5004,
      "step": 8927
    },
    {
      "epoch": 0.9417721518987342,
      "grad_norm": 0.4042734205722809,
      "learning_rate": 3.227169821504187e-05,
      "loss": 1.4992,
      "step": 8928
    },
    {
      "epoch": 0.9418776371308016,
      "grad_norm": 0.3997011184692383,
      "learning_rate": 3.222693824668916e-05,
      "loss": 1.4497,
      "step": 8929
    },
    {
      "epoch": 0.9419831223628692,
      "grad_norm": 0.4170098900794983,
      "learning_rate": 3.218224035919609e-05,
      "loss": 1.4648,
      "step": 8930
    },
    {
      "epoch": 0.9420886075949367,
      "grad_norm": 0.4268629252910614,
      "learning_rate": 3.213760446645818e-05,
      "loss": 1.4642,
      "step": 8931
    },
    {
      "epoch": 0.9421940928270042,
      "grad_norm": 0.4398791193962097,
      "learning_rate": 3.2093030482490385e-05,
      "loss": 1.4695,
      "step": 8932
    },
    {
      "epoch": 0.9422995780590717,
      "grad_norm": 0.38858598470687866,
      "learning_rate": 3.204851832142696e-05,
      "loss": 1.4788,
      "step": 8933
    },
    {
      "epoch": 0.9424050632911393,
      "grad_norm": 0.4406694173812866,
      "learning_rate": 3.200406789752116e-05,
      "loss": 1.4962,
      "step": 8934
    },
    {
      "epoch": 0.9425105485232067,
      "grad_norm": 0.4146156907081604,
      "learning_rate": 3.195967912514527e-05,
      "loss": 1.4952,
      "step": 8935
    },
    {
      "epoch": 0.9426160337552743,
      "grad_norm": 0.41857171058654785,
      "learning_rate": 3.191535191879029e-05,
      "loss": 1.4725,
      "step": 8936
    },
    {
      "epoch": 0.9427215189873418,
      "grad_norm": 0.4443456828594208,
      "learning_rate": 3.1871086193065794e-05,
      "loss": 1.4938,
      "step": 8937
    },
    {
      "epoch": 0.9428270042194092,
      "grad_norm": 0.4142570495605469,
      "learning_rate": 3.182688186269985e-05,
      "loss": 1.4546,
      "step": 8938
    },
    {
      "epoch": 0.9429324894514768,
      "grad_norm": 0.43589162826538086,
      "learning_rate": 3.178273884253874e-05,
      "loss": 1.5063,
      "step": 8939
    },
    {
      "epoch": 0.9430379746835443,
      "grad_norm": 0.4269370436668396,
      "learning_rate": 3.173865704754688e-05,
      "loss": 1.4894,
      "step": 8940
    },
    {
      "epoch": 0.9431434599156118,
      "grad_norm": 0.41896170377731323,
      "learning_rate": 3.169463639280665e-05,
      "loss": 1.4516,
      "step": 8941
    },
    {
      "epoch": 0.9432489451476793,
      "grad_norm": 0.42555418610572815,
      "learning_rate": 3.1650676793518144e-05,
      "loss": 1.4785,
      "step": 8942
    },
    {
      "epoch": 0.9433544303797469,
      "grad_norm": 0.4105265140533447,
      "learning_rate": 3.1606778164999155e-05,
      "loss": 1.4828,
      "step": 8943
    },
    {
      "epoch": 0.9434599156118143,
      "grad_norm": 0.42442068457603455,
      "learning_rate": 3.156294042268483e-05,
      "loss": 1.5177,
      "step": 8944
    },
    {
      "epoch": 0.9435654008438819,
      "grad_norm": 0.4226602613925934,
      "learning_rate": 3.151916348212769e-05,
      "loss": 1.4704,
      "step": 8945
    },
    {
      "epoch": 0.9436708860759494,
      "grad_norm": 0.4202861785888672,
      "learning_rate": 3.147544725899736e-05,
      "loss": 1.4867,
      "step": 8946
    },
    {
      "epoch": 0.9437763713080168,
      "grad_norm": 0.4213012456893921,
      "learning_rate": 3.1431791669080386e-05,
      "loss": 1.4731,
      "step": 8947
    },
    {
      "epoch": 0.9438818565400844,
      "grad_norm": 0.41682150959968567,
      "learning_rate": 3.138819662828018e-05,
      "loss": 1.4751,
      "step": 8948
    },
    {
      "epoch": 0.9439873417721519,
      "grad_norm": 0.4381452202796936,
      "learning_rate": 3.134466205261674e-05,
      "loss": 1.5183,
      "step": 8949
    },
    {
      "epoch": 0.9440928270042194,
      "grad_norm": 0.4284520745277405,
      "learning_rate": 3.130118785822657e-05,
      "loss": 1.4833,
      "step": 8950
    },
    {
      "epoch": 0.9441983122362869,
      "grad_norm": 0.40535178780555725,
      "learning_rate": 3.125777396136251e-05,
      "loss": 1.4758,
      "step": 8951
    },
    {
      "epoch": 0.9443037974683545,
      "grad_norm": 0.4109024107456207,
      "learning_rate": 3.121442027839349e-05,
      "loss": 1.4847,
      "step": 8952
    },
    {
      "epoch": 0.9444092827004219,
      "grad_norm": 0.4157865643501282,
      "learning_rate": 3.117112672580451e-05,
      "loss": 1.507,
      "step": 8953
    },
    {
      "epoch": 0.9445147679324895,
      "grad_norm": 0.4004267156124115,
      "learning_rate": 3.112789322019633e-05,
      "loss": 1.4884,
      "step": 8954
    },
    {
      "epoch": 0.944620253164557,
      "grad_norm": 0.41594013571739197,
      "learning_rate": 3.108471967828545e-05,
      "loss": 1.5182,
      "step": 8955
    },
    {
      "epoch": 0.9447257383966244,
      "grad_norm": 0.40083497762680054,
      "learning_rate": 3.1041606016903844e-05,
      "loss": 1.4747,
      "step": 8956
    },
    {
      "epoch": 0.944831223628692,
      "grad_norm": 0.4216148257255554,
      "learning_rate": 3.0998552152998837e-05,
      "loss": 1.5007,
      "step": 8957
    },
    {
      "epoch": 0.9449367088607595,
      "grad_norm": 0.41635462641716003,
      "learning_rate": 3.095555800363297e-05,
      "loss": 1.4557,
      "step": 8958
    },
    {
      "epoch": 0.945042194092827,
      "grad_norm": 0.39775124192237854,
      "learning_rate": 3.0912623485983774e-05,
      "loss": 1.4672,
      "step": 8959
    },
    {
      "epoch": 0.9451476793248945,
      "grad_norm": 0.42075133323669434,
      "learning_rate": 3.08697485173437e-05,
      "loss": 1.498,
      "step": 8960
    },
    {
      "epoch": 0.9452531645569621,
      "grad_norm": 0.40941378474235535,
      "learning_rate": 3.0826933015119895e-05,
      "loss": 1.4668,
      "step": 8961
    },
    {
      "epoch": 0.9453586497890295,
      "grad_norm": 0.4217377305030823,
      "learning_rate": 3.0784176896834036e-05,
      "loss": 1.4781,
      "step": 8962
    },
    {
      "epoch": 0.945464135021097,
      "grad_norm": 0.41346275806427,
      "learning_rate": 3.074148008012224e-05,
      "loss": 1.4892,
      "step": 8963
    },
    {
      "epoch": 0.9455696202531646,
      "grad_norm": 0.4290733337402344,
      "learning_rate": 3.06988424827348e-05,
      "loss": 1.4734,
      "step": 8964
    },
    {
      "epoch": 0.945675105485232,
      "grad_norm": 0.45879629254341125,
      "learning_rate": 3.0656264022536146e-05,
      "loss": 1.4996,
      "step": 8965
    },
    {
      "epoch": 0.9457805907172996,
      "grad_norm": 0.41879481077194214,
      "learning_rate": 3.0613744617504625e-05,
      "loss": 1.5236,
      "step": 8966
    },
    {
      "epoch": 0.9458860759493671,
      "grad_norm": 0.40473487973213196,
      "learning_rate": 3.0571284185732275e-05,
      "loss": 1.4391,
      "step": 8967
    },
    {
      "epoch": 0.9459915611814346,
      "grad_norm": 0.44342854619026184,
      "learning_rate": 3.052888264542484e-05,
      "loss": 1.4955,
      "step": 8968
    },
    {
      "epoch": 0.9460970464135021,
      "grad_norm": 0.43697628378868103,
      "learning_rate": 3.048653991490141e-05,
      "loss": 1.5029,
      "step": 8969
    },
    {
      "epoch": 0.9462025316455697,
      "grad_norm": 0.4347820281982422,
      "learning_rate": 3.0444255912594442e-05,
      "loss": 1.4782,
      "step": 8970
    },
    {
      "epoch": 0.9463080168776371,
      "grad_norm": 0.40996524691581726,
      "learning_rate": 3.0402030557049503e-05,
      "loss": 1.4923,
      "step": 8971
    },
    {
      "epoch": 0.9464135021097047,
      "grad_norm": 0.4250170886516571,
      "learning_rate": 3.0359863766925097e-05,
      "loss": 1.4683,
      "step": 8972
    },
    {
      "epoch": 0.9465189873417722,
      "grad_norm": 0.3988959491252899,
      "learning_rate": 3.0317755460992616e-05,
      "loss": 1.4424,
      "step": 8973
    },
    {
      "epoch": 0.9466244725738396,
      "grad_norm": 0.4415169954299927,
      "learning_rate": 3.027570555813604e-05,
      "loss": 1.4726,
      "step": 8974
    },
    {
      "epoch": 0.9467299578059072,
      "grad_norm": 0.4174901843070984,
      "learning_rate": 3.0233713977351904e-05,
      "loss": 1.4975,
      "step": 8975
    },
    {
      "epoch": 0.9468354430379747,
      "grad_norm": 0.42810869216918945,
      "learning_rate": 3.0191780637749097e-05,
      "loss": 1.5003,
      "step": 8976
    },
    {
      "epoch": 0.9469409282700422,
      "grad_norm": 0.4096430242061615,
      "learning_rate": 3.0149905458548646e-05,
      "loss": 1.4803,
      "step": 8977
    },
    {
      "epoch": 0.9470464135021097,
      "grad_norm": 0.41747745871543884,
      "learning_rate": 3.0108088359083675e-05,
      "loss": 1.5027,
      "step": 8978
    },
    {
      "epoch": 0.9471518987341773,
      "grad_norm": 0.40123745799064636,
      "learning_rate": 3.0066329258799184e-05,
      "loss": 1.454,
      "step": 8979
    },
    {
      "epoch": 0.9472573839662447,
      "grad_norm": 0.42217034101486206,
      "learning_rate": 3.002462807725185e-05,
      "loss": 1.4994,
      "step": 8980
    },
    {
      "epoch": 0.9473628691983123,
      "grad_norm": 0.4400993883609772,
      "learning_rate": 2.9982984734110005e-05,
      "loss": 1.4885,
      "step": 8981
    },
    {
      "epoch": 0.9474683544303798,
      "grad_norm": 0.4110012650489807,
      "learning_rate": 2.9941399149153305e-05,
      "loss": 1.4778,
      "step": 8982
    },
    {
      "epoch": 0.9475738396624472,
      "grad_norm": 0.4226163923740387,
      "learning_rate": 2.9899871242272736e-05,
      "loss": 1.5008,
      "step": 8983
    },
    {
      "epoch": 0.9476793248945148,
      "grad_norm": 0.39635103940963745,
      "learning_rate": 2.9858400933470395e-05,
      "loss": 1.506,
      "step": 8984
    },
    {
      "epoch": 0.9477848101265823,
      "grad_norm": 0.42404916882514954,
      "learning_rate": 2.9816988142859272e-05,
      "loss": 1.4987,
      "step": 8985
    },
    {
      "epoch": 0.9478902953586498,
      "grad_norm": 0.42395398020744324,
      "learning_rate": 2.9775632790663244e-05,
      "loss": 1.4889,
      "step": 8986
    },
    {
      "epoch": 0.9479957805907173,
      "grad_norm": 0.4184955060482025,
      "learning_rate": 2.973433479721675e-05,
      "loss": 1.4723,
      "step": 8987
    },
    {
      "epoch": 0.9481012658227848,
      "grad_norm": 0.42186737060546875,
      "learning_rate": 2.9693094082964775e-05,
      "loss": 1.4778,
      "step": 8988
    },
    {
      "epoch": 0.9482067510548523,
      "grad_norm": 0.409239798784256,
      "learning_rate": 2.965191056846266e-05,
      "loss": 1.4953,
      "step": 8989
    },
    {
      "epoch": 0.9483122362869199,
      "grad_norm": 0.4182921051979065,
      "learning_rate": 2.9610784174375868e-05,
      "loss": 1.4708,
      "step": 8990
    },
    {
      "epoch": 0.9484177215189873,
      "grad_norm": 0.41539087891578674,
      "learning_rate": 2.9569714821479966e-05,
      "loss": 1.4463,
      "step": 8991
    },
    {
      "epoch": 0.9485232067510548,
      "grad_norm": 0.4016892910003662,
      "learning_rate": 2.9528702430660346e-05,
      "loss": 1.4675,
      "step": 8992
    },
    {
      "epoch": 0.9486286919831224,
      "grad_norm": 0.4415346682071686,
      "learning_rate": 2.948774692291217e-05,
      "loss": 1.4691,
      "step": 8993
    },
    {
      "epoch": 0.9487341772151898,
      "grad_norm": 0.4102402329444885,
      "learning_rate": 2.9446848219340173e-05,
      "loss": 1.4967,
      "step": 8994
    },
    {
      "epoch": 0.9488396624472574,
      "grad_norm": 0.4085753858089447,
      "learning_rate": 2.9406006241158485e-05,
      "loss": 1.5326,
      "step": 8995
    },
    {
      "epoch": 0.9489451476793249,
      "grad_norm": 0.4016677439212799,
      "learning_rate": 2.936522090969056e-05,
      "loss": 1.4653,
      "step": 8996
    },
    {
      "epoch": 0.9490506329113924,
      "grad_norm": 0.4078262150287628,
      "learning_rate": 2.9324492146368908e-05,
      "loss": 1.4843,
      "step": 8997
    },
    {
      "epoch": 0.9491561181434599,
      "grad_norm": 0.40966683626174927,
      "learning_rate": 2.928381987273507e-05,
      "loss": 1.5045,
      "step": 8998
    },
    {
      "epoch": 0.9492616033755275,
      "grad_norm": 0.43814384937286377,
      "learning_rate": 2.9243204010439396e-05,
      "loss": 1.4812,
      "step": 8999
    },
    {
      "epoch": 0.9493670886075949,
      "grad_norm": 0.4208926260471344,
      "learning_rate": 2.920264448124087e-05,
      "loss": 1.4672,
      "step": 9000
    },
    {
      "epoch": 0.9494725738396624,
      "grad_norm": 0.4082159101963043,
      "learning_rate": 2.916214120700703e-05,
      "loss": 1.4853,
      "step": 9001
    },
    {
      "epoch": 0.94957805907173,
      "grad_norm": 0.4235600531101227,
      "learning_rate": 2.9121694109713758e-05,
      "loss": 1.4884,
      "step": 9002
    },
    {
      "epoch": 0.9496835443037974,
      "grad_norm": 0.44142910838127136,
      "learning_rate": 2.9081303111445172e-05,
      "loss": 1.4962,
      "step": 9003
    },
    {
      "epoch": 0.949789029535865,
      "grad_norm": 0.42570793628692627,
      "learning_rate": 2.904096813439346e-05,
      "loss": 1.4594,
      "step": 9004
    },
    {
      "epoch": 0.9498945147679325,
      "grad_norm": 0.41383957862854004,
      "learning_rate": 2.9000689100858695e-05,
      "loss": 1.4459,
      "step": 9005
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4112379252910614,
      "learning_rate": 2.896046593324876e-05,
      "loss": 1.4787,
      "step": 9006
    },
    {
      "epoch": 0.9501054852320675,
      "grad_norm": 0.43490731716156006,
      "learning_rate": 2.8920298554079114e-05,
      "loss": 1.5016,
      "step": 9007
    },
    {
      "epoch": 0.950210970464135,
      "grad_norm": 0.4120459258556366,
      "learning_rate": 2.8880186885972716e-05,
      "loss": 1.4997,
      "step": 9008
    },
    {
      "epoch": 0.9503164556962025,
      "grad_norm": 0.4352714419364929,
      "learning_rate": 2.8840130851659852e-05,
      "loss": 1.4638,
      "step": 9009
    },
    {
      "epoch": 0.95042194092827,
      "grad_norm": 0.4593077301979065,
      "learning_rate": 2.8800130373977936e-05,
      "loss": 1.476,
      "step": 9010
    },
    {
      "epoch": 0.9505274261603376,
      "grad_norm": 0.4143889546394348,
      "learning_rate": 2.876018537587146e-05,
      "loss": 1.4503,
      "step": 9011
    },
    {
      "epoch": 0.950632911392405,
      "grad_norm": 0.4272405207157135,
      "learning_rate": 2.8720295780391722e-05,
      "loss": 1.4982,
      "step": 9012
    },
    {
      "epoch": 0.9507383966244726,
      "grad_norm": 0.43704649806022644,
      "learning_rate": 2.8680461510696802e-05,
      "loss": 1.4944,
      "step": 9013
    },
    {
      "epoch": 0.9508438818565401,
      "grad_norm": 0.43519794940948486,
      "learning_rate": 2.8640682490051365e-05,
      "loss": 1.488,
      "step": 9014
    },
    {
      "epoch": 0.9509493670886076,
      "grad_norm": 0.4351371228694916,
      "learning_rate": 2.8600958641826447e-05,
      "loss": 1.4774,
      "step": 9015
    },
    {
      "epoch": 0.9510548523206751,
      "grad_norm": 0.40209436416625977,
      "learning_rate": 2.8561289889499422e-05,
      "loss": 1.4337,
      "step": 9016
    },
    {
      "epoch": 0.9511603375527427,
      "grad_norm": 0.4071151912212372,
      "learning_rate": 2.8521676156653756e-05,
      "loss": 1.459,
      "step": 9017
    },
    {
      "epoch": 0.9512658227848101,
      "grad_norm": 0.42927947640419006,
      "learning_rate": 2.8482117366978935e-05,
      "loss": 1.4955,
      "step": 9018
    },
    {
      "epoch": 0.9513713080168776,
      "grad_norm": 0.40375077724456787,
      "learning_rate": 2.844261344427029e-05,
      "loss": 1.4769,
      "step": 9019
    },
    {
      "epoch": 0.9514767932489452,
      "grad_norm": 0.4263576567173004,
      "learning_rate": 2.84031643124288e-05,
      "loss": 1.5201,
      "step": 9020
    },
    {
      "epoch": 0.9515822784810126,
      "grad_norm": 0.428631067276001,
      "learning_rate": 2.8363769895461053e-05,
      "loss": 1.4924,
      "step": 9021
    },
    {
      "epoch": 0.9516877637130802,
      "grad_norm": 0.4250580370426178,
      "learning_rate": 2.8324430117478972e-05,
      "loss": 1.5346,
      "step": 9022
    },
    {
      "epoch": 0.9517932489451477,
      "grad_norm": 0.42467814683914185,
      "learning_rate": 2.8285144902699785e-05,
      "loss": 1.5153,
      "step": 9023
    },
    {
      "epoch": 0.9518987341772152,
      "grad_norm": 0.39394938945770264,
      "learning_rate": 2.824591417544583e-05,
      "loss": 1.4603,
      "step": 9024
    },
    {
      "epoch": 0.9520042194092827,
      "grad_norm": 0.4156087338924408,
      "learning_rate": 2.820673786014436e-05,
      "loss": 1.4977,
      "step": 9025
    },
    {
      "epoch": 0.9521097046413503,
      "grad_norm": 0.42903465032577515,
      "learning_rate": 2.816761588132749e-05,
      "loss": 1.5351,
      "step": 9026
    },
    {
      "epoch": 0.9522151898734177,
      "grad_norm": 0.4756835103034973,
      "learning_rate": 2.8128548163632005e-05,
      "loss": 1.4729,
      "step": 9027
    },
    {
      "epoch": 0.9523206751054852,
      "grad_norm": 0.4078097343444824,
      "learning_rate": 2.808953463179918e-05,
      "loss": 1.49,
      "step": 9028
    },
    {
      "epoch": 0.9524261603375528,
      "grad_norm": 0.466543585062027,
      "learning_rate": 2.805057521067472e-05,
      "loss": 1.5201,
      "step": 9029
    },
    {
      "epoch": 0.9525316455696202,
      "grad_norm": 0.4056161046028137,
      "learning_rate": 2.8011669825208517e-05,
      "loss": 1.4754,
      "step": 9030
    },
    {
      "epoch": 0.9526371308016878,
      "grad_norm": 0.4038311541080475,
      "learning_rate": 2.7972818400454596e-05,
      "loss": 1.4466,
      "step": 9031
    },
    {
      "epoch": 0.9527426160337553,
      "grad_norm": 0.42786046862602234,
      "learning_rate": 2.7934020861570928e-05,
      "loss": 1.5102,
      "step": 9032
    },
    {
      "epoch": 0.9528481012658228,
      "grad_norm": 0.44332414865493774,
      "learning_rate": 2.789527713381925e-05,
      "loss": 1.4857,
      "step": 9033
    },
    {
      "epoch": 0.9529535864978903,
      "grad_norm": 0.4434492886066437,
      "learning_rate": 2.7856587142565008e-05,
      "loss": 1.518,
      "step": 9034
    },
    {
      "epoch": 0.9530590717299579,
      "grad_norm": 0.4417775571346283,
      "learning_rate": 2.781795081327712e-05,
      "loss": 1.4519,
      "step": 9035
    },
    {
      "epoch": 0.9531645569620253,
      "grad_norm": 0.41022711992263794,
      "learning_rate": 2.7779368071527907e-05,
      "loss": 1.4438,
      "step": 9036
    },
    {
      "epoch": 0.9532700421940928,
      "grad_norm": 0.4475588798522949,
      "learning_rate": 2.7740838842992916e-05,
      "loss": 1.4384,
      "step": 9037
    },
    {
      "epoch": 0.9533755274261604,
      "grad_norm": 0.4204328656196594,
      "learning_rate": 2.770236305345076e-05,
      "loss": 1.4726,
      "step": 9038
    },
    {
      "epoch": 0.9534810126582278,
      "grad_norm": 0.46305549144744873,
      "learning_rate": 2.766394062878302e-05,
      "loss": 1.4689,
      "step": 9039
    },
    {
      "epoch": 0.9535864978902954,
      "grad_norm": 0.4167904257774353,
      "learning_rate": 2.762557149497405e-05,
      "loss": 1.4952,
      "step": 9040
    },
    {
      "epoch": 0.9536919831223629,
      "grad_norm": 0.435484915971756,
      "learning_rate": 2.758725557811089e-05,
      "loss": 1.4985,
      "step": 9041
    },
    {
      "epoch": 0.9537974683544304,
      "grad_norm": 0.40769943594932556,
      "learning_rate": 2.754899280438309e-05,
      "loss": 1.4905,
      "step": 9042
    },
    {
      "epoch": 0.9539029535864979,
      "grad_norm": 0.4437706470489502,
      "learning_rate": 2.7510783100082544e-05,
      "loss": 1.4612,
      "step": 9043
    },
    {
      "epoch": 0.9540084388185655,
      "grad_norm": 0.40064650774002075,
      "learning_rate": 2.747262639160342e-05,
      "loss": 1.4796,
      "step": 9044
    },
    {
      "epoch": 0.9541139240506329,
      "grad_norm": 0.41875338554382324,
      "learning_rate": 2.743452260544193e-05,
      "loss": 1.4933,
      "step": 9045
    },
    {
      "epoch": 0.9542194092827004,
      "grad_norm": 0.4265892803668976,
      "learning_rate": 2.7396471668196274e-05,
      "loss": 1.4777,
      "step": 9046
    },
    {
      "epoch": 0.954324894514768,
      "grad_norm": 0.43435630202293396,
      "learning_rate": 2.7358473506566453e-05,
      "loss": 1.5325,
      "step": 9047
    },
    {
      "epoch": 0.9544303797468354,
      "grad_norm": 0.41065502166748047,
      "learning_rate": 2.732052804735409e-05,
      "loss": 1.4744,
      "step": 9048
    },
    {
      "epoch": 0.954535864978903,
      "grad_norm": 0.4011315405368805,
      "learning_rate": 2.7282635217462405e-05,
      "loss": 1.4712,
      "step": 9049
    },
    {
      "epoch": 0.9546413502109705,
      "grad_norm": 0.41347742080688477,
      "learning_rate": 2.724479494389592e-05,
      "loss": 1.4808,
      "step": 9050
    },
    {
      "epoch": 0.954746835443038,
      "grad_norm": 0.43332159519195557,
      "learning_rate": 2.7207007153760456e-05,
      "loss": 1.5044,
      "step": 9051
    },
    {
      "epoch": 0.9548523206751055,
      "grad_norm": 0.4061329662799835,
      "learning_rate": 2.7169271774262942e-05,
      "loss": 1.4843,
      "step": 9052
    },
    {
      "epoch": 0.9549578059071729,
      "grad_norm": 0.41654402017593384,
      "learning_rate": 2.7131588732711214e-05,
      "loss": 1.4729,
      "step": 9053
    },
    {
      "epoch": 0.9550632911392405,
      "grad_norm": 0.42718738317489624,
      "learning_rate": 2.7093957956513993e-05,
      "loss": 1.5102,
      "step": 9054
    },
    {
      "epoch": 0.955168776371308,
      "grad_norm": 0.4063757658004761,
      "learning_rate": 2.7056379373180626e-05,
      "loss": 1.4599,
      "step": 9055
    },
    {
      "epoch": 0.9552742616033755,
      "grad_norm": 0.40532445907592773,
      "learning_rate": 2.701885291032104e-05,
      "loss": 1.5087,
      "step": 9056
    },
    {
      "epoch": 0.955379746835443,
      "grad_norm": 0.41733431816101074,
      "learning_rate": 2.6981378495645566e-05,
      "loss": 1.4763,
      "step": 9057
    },
    {
      "epoch": 0.9554852320675106,
      "grad_norm": 0.4174603819847107,
      "learning_rate": 2.6943956056964773e-05,
      "loss": 1.5041,
      "step": 9058
    },
    {
      "epoch": 0.955590717299578,
      "grad_norm": 0.4342232644557953,
      "learning_rate": 2.6906585522189378e-05,
      "loss": 1.4641,
      "step": 9059
    },
    {
      "epoch": 0.9556962025316456,
      "grad_norm": 0.3987276256084442,
      "learning_rate": 2.6869266819330058e-05,
      "loss": 1.477,
      "step": 9060
    },
    {
      "epoch": 0.9558016877637131,
      "grad_norm": 0.40920597314834595,
      "learning_rate": 2.6831999876497372e-05,
      "loss": 1.5008,
      "step": 9061
    },
    {
      "epoch": 0.9559071729957805,
      "grad_norm": 0.4030047059059143,
      "learning_rate": 2.6794784621901574e-05,
      "loss": 1.5007,
      "step": 9062
    },
    {
      "epoch": 0.9560126582278481,
      "grad_norm": 0.4014549255371094,
      "learning_rate": 2.675762098385246e-05,
      "loss": 1.4896,
      "step": 9063
    },
    {
      "epoch": 0.9561181434599156,
      "grad_norm": 0.42015427350997925,
      "learning_rate": 2.672050889075931e-05,
      "loss": 1.4725,
      "step": 9064
    },
    {
      "epoch": 0.9562236286919831,
      "grad_norm": 0.41310104727745056,
      "learning_rate": 2.6683448271130646e-05,
      "loss": 1.5127,
      "step": 9065
    },
    {
      "epoch": 0.9563291139240506,
      "grad_norm": 0.4185239374637604,
      "learning_rate": 2.6646439053574176e-05,
      "loss": 1.4699,
      "step": 9066
    },
    {
      "epoch": 0.9564345991561182,
      "grad_norm": 0.4112013578414917,
      "learning_rate": 2.6609481166796652e-05,
      "loss": 1.5027,
      "step": 9067
    },
    {
      "epoch": 0.9565400843881856,
      "grad_norm": 0.40110740065574646,
      "learning_rate": 2.6572574539603643e-05,
      "loss": 1.4796,
      "step": 9068
    },
    {
      "epoch": 0.9566455696202532,
      "grad_norm": 0.43752819299697876,
      "learning_rate": 2.653571910089951e-05,
      "loss": 1.476,
      "step": 9069
    },
    {
      "epoch": 0.9567510548523207,
      "grad_norm": 0.42355623841285706,
      "learning_rate": 2.6498914779687228e-05,
      "loss": 1.453,
      "step": 9070
    },
    {
      "epoch": 0.9568565400843881,
      "grad_norm": 0.39908015727996826,
      "learning_rate": 2.646216150506819e-05,
      "loss": 1.5225,
      "step": 9071
    },
    {
      "epoch": 0.9569620253164557,
      "grad_norm": 0.4024796783924103,
      "learning_rate": 2.6425459206242196e-05,
      "loss": 1.463,
      "step": 9072
    },
    {
      "epoch": 0.9570675105485232,
      "grad_norm": 0.4407555162906647,
      "learning_rate": 2.6388807812507172e-05,
      "loss": 1.4992,
      "step": 9073
    },
    {
      "epoch": 0.9571729957805907,
      "grad_norm": 0.4400232434272766,
      "learning_rate": 2.6352207253259166e-05,
      "loss": 1.5028,
      "step": 9074
    },
    {
      "epoch": 0.9572784810126582,
      "grad_norm": 0.4058392643928528,
      "learning_rate": 2.6315657457992123e-05,
      "loss": 1.463,
      "step": 9075
    },
    {
      "epoch": 0.9573839662447258,
      "grad_norm": 0.4155997335910797,
      "learning_rate": 2.627915835629776e-05,
      "loss": 1.4509,
      "step": 9076
    },
    {
      "epoch": 0.9574894514767932,
      "grad_norm": 0.4103969931602478,
      "learning_rate": 2.62427098778655e-05,
      "loss": 1.5034,
      "step": 9077
    },
    {
      "epoch": 0.9575949367088608,
      "grad_norm": 0.4225928485393524,
      "learning_rate": 2.6206311952482224e-05,
      "loss": 1.5109,
      "step": 9078
    },
    {
      "epoch": 0.9577004219409283,
      "grad_norm": 0.4252573549747467,
      "learning_rate": 2.6169964510032243e-05,
      "loss": 1.4454,
      "step": 9079
    },
    {
      "epoch": 0.9578059071729957,
      "grad_norm": 0.42723262310028076,
      "learning_rate": 2.6133667480497115e-05,
      "loss": 1.5067,
      "step": 9080
    },
    {
      "epoch": 0.9579113924050633,
      "grad_norm": 0.42512160539627075,
      "learning_rate": 2.6097420793955464e-05,
      "loss": 1.5057,
      "step": 9081
    },
    {
      "epoch": 0.9580168776371308,
      "grad_norm": 0.41484829783439636,
      "learning_rate": 2.6061224380582955e-05,
      "loss": 1.499,
      "step": 9082
    },
    {
      "epoch": 0.9581223628691983,
      "grad_norm": 0.4189888536930084,
      "learning_rate": 2.6025078170652045e-05,
      "loss": 1.4917,
      "step": 9083
    },
    {
      "epoch": 0.9582278481012658,
      "grad_norm": 0.42506399750709534,
      "learning_rate": 2.5988982094531942e-05,
      "loss": 1.4828,
      "step": 9084
    },
    {
      "epoch": 0.9583333333333334,
      "grad_norm": 0.42952707409858704,
      "learning_rate": 2.595293608268842e-05,
      "loss": 1.4544,
      "step": 9085
    },
    {
      "epoch": 0.9584388185654008,
      "grad_norm": 0.3887086510658264,
      "learning_rate": 2.591694006568366e-05,
      "loss": 1.4844,
      "step": 9086
    },
    {
      "epoch": 0.9585443037974684,
      "grad_norm": 0.45235297083854675,
      "learning_rate": 2.588099397417621e-05,
      "loss": 1.47,
      "step": 9087
    },
    {
      "epoch": 0.9586497890295359,
      "grad_norm": 0.44235000014305115,
      "learning_rate": 2.584509773892073e-05,
      "loss": 1.5028,
      "step": 9088
    },
    {
      "epoch": 0.9587552742616033,
      "grad_norm": 0.4243945777416229,
      "learning_rate": 2.580925129076798e-05,
      "loss": 1.4854,
      "step": 9089
    },
    {
      "epoch": 0.9588607594936709,
      "grad_norm": 0.43124672770500183,
      "learning_rate": 2.5773454560664597e-05,
      "loss": 1.5133,
      "step": 9090
    },
    {
      "epoch": 0.9589662447257384,
      "grad_norm": 0.408216655254364,
      "learning_rate": 2.5737707479652988e-05,
      "loss": 1.4438,
      "step": 9091
    },
    {
      "epoch": 0.9590717299578059,
      "grad_norm": 0.41597023606300354,
      "learning_rate": 2.5702009978871223e-05,
      "loss": 1.4667,
      "step": 9092
    },
    {
      "epoch": 0.9591772151898734,
      "grad_norm": 0.4255770146846771,
      "learning_rate": 2.566636198955286e-05,
      "loss": 1.4763,
      "step": 9093
    },
    {
      "epoch": 0.959282700421941,
      "grad_norm": 0.40157386660575867,
      "learning_rate": 2.5630763443026845e-05,
      "loss": 1.4559,
      "step": 9094
    },
    {
      "epoch": 0.9593881856540084,
      "grad_norm": 0.40931519865989685,
      "learning_rate": 2.5595214270717388e-05,
      "loss": 1.463,
      "step": 9095
    },
    {
      "epoch": 0.959493670886076,
      "grad_norm": 0.4324822425842285,
      "learning_rate": 2.5559714404143766e-05,
      "loss": 1.4965,
      "step": 9096
    },
    {
      "epoch": 0.9595991561181435,
      "grad_norm": 0.42928382754325867,
      "learning_rate": 2.5524263774920287e-05,
      "loss": 1.5136,
      "step": 9097
    },
    {
      "epoch": 0.9597046413502109,
      "grad_norm": 0.41698092222213745,
      "learning_rate": 2.5488862314756066e-05,
      "loss": 1.4392,
      "step": 9098
    },
    {
      "epoch": 0.9598101265822785,
      "grad_norm": 0.4160037338733673,
      "learning_rate": 2.5453509955454954e-05,
      "loss": 1.4255,
      "step": 9099
    },
    {
      "epoch": 0.959915611814346,
      "grad_norm": 0.3954348862171173,
      "learning_rate": 2.541820662891541e-05,
      "loss": 1.4704,
      "step": 9100
    },
    {
      "epoch": 0.9600210970464135,
      "grad_norm": 0.4444388449192047,
      "learning_rate": 2.53829522671303e-05,
      "loss": 1.5075,
      "step": 9101
    },
    {
      "epoch": 0.960126582278481,
      "grad_norm": 0.4075108468532562,
      "learning_rate": 2.5347746802186866e-05,
      "loss": 1.4699,
      "step": 9102
    },
    {
      "epoch": 0.9602320675105486,
      "grad_norm": 0.41993606090545654,
      "learning_rate": 2.531259016626649e-05,
      "loss": 1.4717,
      "step": 9103
    },
    {
      "epoch": 0.960337552742616,
      "grad_norm": 0.41762638092041016,
      "learning_rate": 2.5277482291644662e-05,
      "loss": 1.4883,
      "step": 9104
    },
    {
      "epoch": 0.9604430379746836,
      "grad_norm": 0.4408147633075714,
      "learning_rate": 2.524242311069079e-05,
      "loss": 1.4797,
      "step": 9105
    },
    {
      "epoch": 0.9605485232067511,
      "grad_norm": 0.42754238843917847,
      "learning_rate": 2.520741255586806e-05,
      "loss": 1.4971,
      "step": 9106
    },
    {
      "epoch": 0.9606540084388185,
      "grad_norm": 0.40313130617141724,
      "learning_rate": 2.5172450559733375e-05,
      "loss": 1.4655,
      "step": 9107
    },
    {
      "epoch": 0.9607594936708861,
      "grad_norm": 0.4181784689426422,
      "learning_rate": 2.513753705493713e-05,
      "loss": 1.4592,
      "step": 9108
    },
    {
      "epoch": 0.9608649789029536,
      "grad_norm": 0.40294092893600464,
      "learning_rate": 2.510267197422317e-05,
      "loss": 1.5101,
      "step": 9109
    },
    {
      "epoch": 0.9609704641350211,
      "grad_norm": 0.4241686165332794,
      "learning_rate": 2.5067855250428616e-05,
      "loss": 1.4662,
      "step": 9110
    },
    {
      "epoch": 0.9610759493670886,
      "grad_norm": 0.43026596307754517,
      "learning_rate": 2.5033086816483705e-05,
      "loss": 1.4781,
      "step": 9111
    },
    {
      "epoch": 0.9611814345991562,
      "grad_norm": 0.40692874789237976,
      "learning_rate": 2.499836660541176e-05,
      "loss": 1.4924,
      "step": 9112
    },
    {
      "epoch": 0.9612869198312236,
      "grad_norm": 0.4048171043395996,
      "learning_rate": 2.4963694550328964e-05,
      "loss": 1.4767,
      "step": 9113
    },
    {
      "epoch": 0.9613924050632912,
      "grad_norm": 0.40509337186813354,
      "learning_rate": 2.492907058444425e-05,
      "loss": 1.4785,
      "step": 9114
    },
    {
      "epoch": 0.9614978902953587,
      "grad_norm": 0.39835599064826965,
      "learning_rate": 2.489449464105922e-05,
      "loss": 1.4829,
      "step": 9115
    },
    {
      "epoch": 0.9616033755274261,
      "grad_norm": 0.4123097062110901,
      "learning_rate": 2.4859966653567965e-05,
      "loss": 1.5266,
      "step": 9116
    },
    {
      "epoch": 0.9617088607594937,
      "grad_norm": 0.43954089283943176,
      "learning_rate": 2.482548655545697e-05,
      "loss": 1.4887,
      "step": 9117
    },
    {
      "epoch": 0.9618143459915611,
      "grad_norm": 0.4100317656993866,
      "learning_rate": 2.4791054280304972e-05,
      "loss": 1.5334,
      "step": 9118
    },
    {
      "epoch": 0.9619198312236287,
      "grad_norm": 0.45626506209373474,
      "learning_rate": 2.4756669761782806e-05,
      "loss": 1.5027,
      "step": 9119
    },
    {
      "epoch": 0.9620253164556962,
      "grad_norm": 0.41467225551605225,
      "learning_rate": 2.472233293365335e-05,
      "loss": 1.4324,
      "step": 9120
    },
    {
      "epoch": 0.9621308016877637,
      "grad_norm": 0.4484040439128876,
      "learning_rate": 2.4688043729771304e-05,
      "loss": 1.4928,
      "step": 9121
    },
    {
      "epoch": 0.9622362869198312,
      "grad_norm": 0.4317052662372589,
      "learning_rate": 2.4653802084083134e-05,
      "loss": 1.4589,
      "step": 9122
    },
    {
      "epoch": 0.9623417721518988,
      "grad_norm": 0.4185953438282013,
      "learning_rate": 2.4619607930626937e-05,
      "loss": 1.4897,
      "step": 9123
    },
    {
      "epoch": 0.9624472573839662,
      "grad_norm": 0.3883945941925049,
      "learning_rate": 2.4585461203532253e-05,
      "loss": 1.4608,
      "step": 9124
    },
    {
      "epoch": 0.9625527426160337,
      "grad_norm": 0.39219507575035095,
      "learning_rate": 2.4551361837020032e-05,
      "loss": 1.4543,
      "step": 9125
    },
    {
      "epoch": 0.9626582278481013,
      "grad_norm": 0.4329569339752197,
      "learning_rate": 2.4517309765402408e-05,
      "loss": 1.4565,
      "step": 9126
    },
    {
      "epoch": 0.9627637130801687,
      "grad_norm": 0.4056680202484131,
      "learning_rate": 2.448330492308266e-05,
      "loss": 1.4935,
      "step": 9127
    },
    {
      "epoch": 0.9628691983122363,
      "grad_norm": 0.3911163806915283,
      "learning_rate": 2.4449347244555043e-05,
      "loss": 1.4487,
      "step": 9128
    },
    {
      "epoch": 0.9629746835443038,
      "grad_norm": 0.4146103262901306,
      "learning_rate": 2.4415436664404643e-05,
      "loss": 1.4401,
      "step": 9129
    },
    {
      "epoch": 0.9630801687763713,
      "grad_norm": 0.44063320755958557,
      "learning_rate": 2.4381573117307307e-05,
      "loss": 1.5436,
      "step": 9130
    },
    {
      "epoch": 0.9631856540084388,
      "grad_norm": 0.40029656887054443,
      "learning_rate": 2.4347756538029454e-05,
      "loss": 1.501,
      "step": 9131
    },
    {
      "epoch": 0.9632911392405064,
      "grad_norm": 0.41834521293640137,
      "learning_rate": 2.4313986861427997e-05,
      "loss": 1.5183,
      "step": 9132
    },
    {
      "epoch": 0.9633966244725738,
      "grad_norm": 0.437883198261261,
      "learning_rate": 2.4280264022450212e-05,
      "loss": 1.5414,
      "step": 9133
    },
    {
      "epoch": 0.9635021097046413,
      "grad_norm": 0.4124026894569397,
      "learning_rate": 2.4246587956133573e-05,
      "loss": 1.4557,
      "step": 9134
    },
    {
      "epoch": 0.9636075949367089,
      "grad_norm": 0.435684472322464,
      "learning_rate": 2.421295859760569e-05,
      "loss": 1.4734,
      "step": 9135
    },
    {
      "epoch": 0.9637130801687763,
      "grad_norm": 0.4155442416667938,
      "learning_rate": 2.4179375882084098e-05,
      "loss": 1.4934,
      "step": 9136
    },
    {
      "epoch": 0.9638185654008439,
      "grad_norm": 0.39959219098091125,
      "learning_rate": 2.4145839744876232e-05,
      "loss": 1.4883,
      "step": 9137
    },
    {
      "epoch": 0.9639240506329114,
      "grad_norm": 0.43970921635627747,
      "learning_rate": 2.4112350121379255e-05,
      "loss": 1.494,
      "step": 9138
    },
    {
      "epoch": 0.9640295358649789,
      "grad_norm": 0.42591482400894165,
      "learning_rate": 2.4078906947079882e-05,
      "loss": 1.4931,
      "step": 9139
    },
    {
      "epoch": 0.9641350210970464,
      "grad_norm": 0.4545920193195343,
      "learning_rate": 2.4045510157554362e-05,
      "loss": 1.4886,
      "step": 9140
    },
    {
      "epoch": 0.964240506329114,
      "grad_norm": 0.45448213815689087,
      "learning_rate": 2.4012159688468252e-05,
      "loss": 1.4651,
      "step": 9141
    },
    {
      "epoch": 0.9643459915611814,
      "grad_norm": 0.42389097809791565,
      "learning_rate": 2.397885547557638e-05,
      "loss": 1.4738,
      "step": 9142
    },
    {
      "epoch": 0.9644514767932489,
      "grad_norm": 0.45470306277275085,
      "learning_rate": 2.3945597454722663e-05,
      "loss": 1.4846,
      "step": 9143
    },
    {
      "epoch": 0.9645569620253165,
      "grad_norm": 0.38892287015914917,
      "learning_rate": 2.3912385561839984e-05,
      "loss": 1.4856,
      "step": 9144
    },
    {
      "epoch": 0.9646624472573839,
      "grad_norm": 0.41130340099334717,
      "learning_rate": 2.3879219732950117e-05,
      "loss": 1.4719,
      "step": 9145
    },
    {
      "epoch": 0.9647679324894515,
      "grad_norm": 0.40683552622795105,
      "learning_rate": 2.384609990416354e-05,
      "loss": 1.5036,
      "step": 9146
    },
    {
      "epoch": 0.964873417721519,
      "grad_norm": 0.4692808985710144,
      "learning_rate": 2.3813026011679366e-05,
      "loss": 1.4981,
      "step": 9147
    },
    {
      "epoch": 0.9649789029535865,
      "grad_norm": 0.4820142686367035,
      "learning_rate": 2.3779997991785207e-05,
      "loss": 1.4675,
      "step": 9148
    },
    {
      "epoch": 0.965084388185654,
      "grad_norm": 0.4010757505893707,
      "learning_rate": 2.3747015780857007e-05,
      "loss": 1.4877,
      "step": 9149
    },
    {
      "epoch": 0.9651898734177216,
      "grad_norm": 0.4162488579750061,
      "learning_rate": 2.3714079315358985e-05,
      "loss": 1.4712,
      "step": 9150
    },
    {
      "epoch": 0.965295358649789,
      "grad_norm": 0.42586034536361694,
      "learning_rate": 2.3681188531843466e-05,
      "loss": 1.4496,
      "step": 9151
    },
    {
      "epoch": 0.9654008438818565,
      "grad_norm": 0.4219142198562622,
      "learning_rate": 2.3648343366950783e-05,
      "loss": 1.5054,
      "step": 9152
    },
    {
      "epoch": 0.9655063291139241,
      "grad_norm": 0.4600399136543274,
      "learning_rate": 2.3615543757409166e-05,
      "loss": 1.5034,
      "step": 9153
    },
    {
      "epoch": 0.9656118143459915,
      "grad_norm": 0.408123642206192,
      "learning_rate": 2.3582789640034548e-05,
      "loss": 1.4917,
      "step": 9154
    },
    {
      "epoch": 0.9657172995780591,
      "grad_norm": 0.3981526792049408,
      "learning_rate": 2.3550080951730548e-05,
      "loss": 1.4988,
      "step": 9155
    },
    {
      "epoch": 0.9658227848101266,
      "grad_norm": 0.43337729573249817,
      "learning_rate": 2.3517417629488286e-05,
      "loss": 1.4862,
      "step": 9156
    },
    {
      "epoch": 0.9659282700421941,
      "grad_norm": 0.39067745208740234,
      "learning_rate": 2.3484799610386246e-05,
      "loss": 1.4731,
      "step": 9157
    },
    {
      "epoch": 0.9660337552742616,
      "grad_norm": 0.4386807680130005,
      "learning_rate": 2.3452226831590232e-05,
      "loss": 1.5109,
      "step": 9158
    },
    {
      "epoch": 0.9661392405063292,
      "grad_norm": 0.485840767621994,
      "learning_rate": 2.3419699230353144e-05,
      "loss": 1.4642,
      "step": 9159
    },
    {
      "epoch": 0.9662447257383966,
      "grad_norm": 0.4368363618850708,
      "learning_rate": 2.338721674401494e-05,
      "loss": 1.5036,
      "step": 9160
    },
    {
      "epoch": 0.9663502109704641,
      "grad_norm": 0.4158051609992981,
      "learning_rate": 2.3354779310002503e-05,
      "loss": 1.4417,
      "step": 9161
    },
    {
      "epoch": 0.9664556962025317,
      "grad_norm": 0.4055517017841339,
      "learning_rate": 2.3322386865829453e-05,
      "loss": 1.476,
      "step": 9162
    },
    {
      "epoch": 0.9665611814345991,
      "grad_norm": 0.41120645403862,
      "learning_rate": 2.3290039349096127e-05,
      "loss": 1.4507,
      "step": 9163
    },
    {
      "epoch": 0.9666666666666667,
      "grad_norm": 0.4314862787723541,
      "learning_rate": 2.325773669748937e-05,
      "loss": 1.5242,
      "step": 9164
    },
    {
      "epoch": 0.9667721518987342,
      "grad_norm": 0.4585617184638977,
      "learning_rate": 2.3225478848782476e-05,
      "loss": 1.5275,
      "step": 9165
    },
    {
      "epoch": 0.9668776371308017,
      "grad_norm": 0.4475068747997284,
      "learning_rate": 2.3193265740835058e-05,
      "loss": 1.5003,
      "step": 9166
    },
    {
      "epoch": 0.9669831223628692,
      "grad_norm": 0.416963666677475,
      "learning_rate": 2.3161097311592866e-05,
      "loss": 1.4908,
      "step": 9167
    },
    {
      "epoch": 0.9670886075949368,
      "grad_norm": 0.42864346504211426,
      "learning_rate": 2.3128973499087785e-05,
      "loss": 1.4747,
      "step": 9168
    },
    {
      "epoch": 0.9671940928270042,
      "grad_norm": 0.45106953382492065,
      "learning_rate": 2.3096894241437583e-05,
      "loss": 1.5217,
      "step": 9169
    },
    {
      "epoch": 0.9672995780590717,
      "grad_norm": 0.47146642208099365,
      "learning_rate": 2.30648594768459e-05,
      "loss": 1.4461,
      "step": 9170
    },
    {
      "epoch": 0.9674050632911393,
      "grad_norm": 0.40260064601898193,
      "learning_rate": 2.3032869143602086e-05,
      "loss": 1.4385,
      "step": 9171
    },
    {
      "epoch": 0.9675105485232067,
      "grad_norm": 0.43034082651138306,
      "learning_rate": 2.3000923180081046e-05,
      "loss": 1.5207,
      "step": 9172
    },
    {
      "epoch": 0.9676160337552743,
      "grad_norm": 0.41302481293678284,
      "learning_rate": 2.29690215247432e-05,
      "loss": 1.5081,
      "step": 9173
    },
    {
      "epoch": 0.9677215189873418,
      "grad_norm": 0.429710328578949,
      "learning_rate": 2.293716411613428e-05,
      "loss": 1.4861,
      "step": 9174
    },
    {
      "epoch": 0.9678270042194093,
      "grad_norm": 0.42314034700393677,
      "learning_rate": 2.2905350892885293e-05,
      "loss": 1.4748,
      "step": 9175
    },
    {
      "epoch": 0.9679324894514768,
      "grad_norm": 0.44195985794067383,
      "learning_rate": 2.287358179371235e-05,
      "loss": 1.4618,
      "step": 9176
    },
    {
      "epoch": 0.9680379746835444,
      "grad_norm": 0.4341367781162262,
      "learning_rate": 2.2841856757416538e-05,
      "loss": 1.481,
      "step": 9177
    },
    {
      "epoch": 0.9681434599156118,
      "grad_norm": 0.4109688699245453,
      "learning_rate": 2.2810175722883866e-05,
      "loss": 1.5089,
      "step": 9178
    },
    {
      "epoch": 0.9682489451476793,
      "grad_norm": 0.41590961813926697,
      "learning_rate": 2.2778538629085057e-05,
      "loss": 1.4911,
      "step": 9179
    },
    {
      "epoch": 0.9683544303797469,
      "grad_norm": 0.4081091284751892,
      "learning_rate": 2.2746945415075523e-05,
      "loss": 1.4514,
      "step": 9180
    },
    {
      "epoch": 0.9684599156118143,
      "grad_norm": 0.4389456510543823,
      "learning_rate": 2.27153960199952e-05,
      "loss": 1.4929,
      "step": 9181
    },
    {
      "epoch": 0.9685654008438819,
      "grad_norm": 0.4213710129261017,
      "learning_rate": 2.26838903830684e-05,
      "loss": 1.4786,
      "step": 9182
    },
    {
      "epoch": 0.9686708860759494,
      "grad_norm": 0.41330716013908386,
      "learning_rate": 2.2652428443603782e-05,
      "loss": 1.4048,
      "step": 9183
    },
    {
      "epoch": 0.9687763713080169,
      "grad_norm": 0.43445727229118347,
      "learning_rate": 2.2621010140994126e-05,
      "loss": 1.4949,
      "step": 9184
    },
    {
      "epoch": 0.9688818565400844,
      "grad_norm": 0.41455334424972534,
      "learning_rate": 2.2589635414716307e-05,
      "loss": 1.4997,
      "step": 9185
    },
    {
      "epoch": 0.9689873417721518,
      "grad_norm": 0.41250765323638916,
      "learning_rate": 2.2558304204331152e-05,
      "loss": 1.478,
      "step": 9186
    },
    {
      "epoch": 0.9690928270042194,
      "grad_norm": 0.46336549520492554,
      "learning_rate": 2.2527016449483282e-05,
      "loss": 1.4975,
      "step": 9187
    },
    {
      "epoch": 0.9691983122362869,
      "grad_norm": 0.4073362648487091,
      "learning_rate": 2.2495772089901067e-05,
      "loss": 1.4822,
      "step": 9188
    },
    {
      "epoch": 0.9693037974683544,
      "grad_norm": 0.45580729842185974,
      "learning_rate": 2.2464571065396428e-05,
      "loss": 1.5042,
      "step": 9189
    },
    {
      "epoch": 0.9694092827004219,
      "grad_norm": 0.4042230546474457,
      "learning_rate": 2.2433413315864803e-05,
      "loss": 1.458,
      "step": 9190
    },
    {
      "epoch": 0.9695147679324895,
      "grad_norm": 0.4477214813232422,
      "learning_rate": 2.2402298781284995e-05,
      "loss": 1.5238,
      "step": 9191
    },
    {
      "epoch": 0.9696202531645569,
      "grad_norm": 0.38857710361480713,
      "learning_rate": 2.2371227401719017e-05,
      "loss": 1.4755,
      "step": 9192
    },
    {
      "epoch": 0.9697257383966245,
      "grad_norm": 0.39348363876342773,
      "learning_rate": 2.2340199117312058e-05,
      "loss": 1.4535,
      "step": 9193
    },
    {
      "epoch": 0.969831223628692,
      "grad_norm": 0.4211290776729584,
      "learning_rate": 2.2309213868292277e-05,
      "loss": 1.4945,
      "step": 9194
    },
    {
      "epoch": 0.9699367088607594,
      "grad_norm": 0.4405970573425293,
      "learning_rate": 2.2278271594970783e-05,
      "loss": 1.4782,
      "step": 9195
    },
    {
      "epoch": 0.970042194092827,
      "grad_norm": 0.4098755419254303,
      "learning_rate": 2.2247372237741457e-05,
      "loss": 1.4882,
      "step": 9196
    },
    {
      "epoch": 0.9701476793248945,
      "grad_norm": 0.4063092768192291,
      "learning_rate": 2.2216515737080818e-05,
      "loss": 1.4987,
      "step": 9197
    },
    {
      "epoch": 0.970253164556962,
      "grad_norm": 0.4036597013473511,
      "learning_rate": 2.218570203354799e-05,
      "loss": 1.4521,
      "step": 9198
    },
    {
      "epoch": 0.9703586497890295,
      "grad_norm": 0.4226246476173401,
      "learning_rate": 2.2154931067784525e-05,
      "loss": 1.517,
      "step": 9199
    },
    {
      "epoch": 0.9704641350210971,
      "grad_norm": 0.4177335500717163,
      "learning_rate": 2.2124202780514277e-05,
      "loss": 1.4491,
      "step": 9200
    },
    {
      "epoch": 0.9705696202531645,
      "grad_norm": 0.42933499813079834,
      "learning_rate": 2.2093517112543358e-05,
      "loss": 1.4871,
      "step": 9201
    },
    {
      "epoch": 0.9706751054852321,
      "grad_norm": 0.4181468188762665,
      "learning_rate": 2.2062874004759936e-05,
      "loss": 1.497,
      "step": 9202
    },
    {
      "epoch": 0.9707805907172996,
      "grad_norm": 0.43489357829093933,
      "learning_rate": 2.20322733981342e-05,
      "loss": 1.4567,
      "step": 9203
    },
    {
      "epoch": 0.970886075949367,
      "grad_norm": 0.4201034903526306,
      "learning_rate": 2.200171523371821e-05,
      "loss": 1.4985,
      "step": 9204
    },
    {
      "epoch": 0.9709915611814346,
      "grad_norm": 0.42879852652549744,
      "learning_rate": 2.197119945264576e-05,
      "loss": 1.5035,
      "step": 9205
    },
    {
      "epoch": 0.9710970464135021,
      "grad_norm": 0.4183104634284973,
      "learning_rate": 2.1940725996132308e-05,
      "loss": 1.4483,
      "step": 9206
    },
    {
      "epoch": 0.9712025316455696,
      "grad_norm": 0.47237175703048706,
      "learning_rate": 2.1910294805474834e-05,
      "loss": 1.5052,
      "step": 9207
    },
    {
      "epoch": 0.9713080168776371,
      "grad_norm": 0.4298136532306671,
      "learning_rate": 2.187990582205175e-05,
      "loss": 1.4729,
      "step": 9208
    },
    {
      "epoch": 0.9714135021097047,
      "grad_norm": 0.3972509801387787,
      "learning_rate": 2.1849558987322783e-05,
      "loss": 1.4684,
      "step": 9209
    },
    {
      "epoch": 0.9715189873417721,
      "grad_norm": 0.4220639765262604,
      "learning_rate": 2.1819254242828815e-05,
      "loss": 1.4796,
      "step": 9210
    },
    {
      "epoch": 0.9716244725738397,
      "grad_norm": 0.4244382679462433,
      "learning_rate": 2.1788991530191857e-05,
      "loss": 1.4894,
      "step": 9211
    },
    {
      "epoch": 0.9717299578059072,
      "grad_norm": 0.4105396866798401,
      "learning_rate": 2.1758770791114845e-05,
      "loss": 1.5095,
      "step": 9212
    },
    {
      "epoch": 0.9718354430379746,
      "grad_norm": 0.4359987676143646,
      "learning_rate": 2.17285919673816e-05,
      "loss": 1.5202,
      "step": 9213
    },
    {
      "epoch": 0.9719409282700422,
      "grad_norm": 0.44183844327926636,
      "learning_rate": 2.1698455000856692e-05,
      "loss": 1.4652,
      "step": 9214
    },
    {
      "epoch": 0.9720464135021097,
      "grad_norm": 0.4069547951221466,
      "learning_rate": 2.1668359833485287e-05,
      "loss": 1.4815,
      "step": 9215
    },
    {
      "epoch": 0.9721518987341772,
      "grad_norm": 0.431631863117218,
      "learning_rate": 2.1638306407293122e-05,
      "loss": 1.4504,
      "step": 9216
    },
    {
      "epoch": 0.9722573839662447,
      "grad_norm": 0.43690598011016846,
      "learning_rate": 2.160829466438629e-05,
      "loss": 1.4899,
      "step": 9217
    },
    {
      "epoch": 0.9723628691983123,
      "grad_norm": 0.4441690444946289,
      "learning_rate": 2.157832454695122e-05,
      "loss": 1.4484,
      "step": 9218
    },
    {
      "epoch": 0.9724683544303797,
      "grad_norm": 0.45397254824638367,
      "learning_rate": 2.1548395997254516e-05,
      "loss": 1.4878,
      "step": 9219
    },
    {
      "epoch": 0.9725738396624473,
      "grad_norm": 0.4372881054878235,
      "learning_rate": 2.151850895764285e-05,
      "loss": 1.4812,
      "step": 9220
    },
    {
      "epoch": 0.9726793248945148,
      "grad_norm": 0.4144532084465027,
      "learning_rate": 2.148866337054287e-05,
      "loss": 1.495,
      "step": 9221
    },
    {
      "epoch": 0.9727848101265822,
      "grad_norm": 0.41240862011909485,
      "learning_rate": 2.145885917846105e-05,
      "loss": 1.448,
      "step": 9222
    },
    {
      "epoch": 0.9728902953586498,
      "grad_norm": 0.4212949275970459,
      "learning_rate": 2.1429096323983638e-05,
      "loss": 1.4764,
      "step": 9223
    },
    {
      "epoch": 0.9729957805907173,
      "grad_norm": 0.4755083918571472,
      "learning_rate": 2.1399374749776512e-05,
      "loss": 1.4609,
      "step": 9224
    },
    {
      "epoch": 0.9731012658227848,
      "grad_norm": 0.40488022565841675,
      "learning_rate": 2.1369694398585035e-05,
      "loss": 1.4677,
      "step": 9225
    },
    {
      "epoch": 0.9732067510548523,
      "grad_norm": 0.41448238492012024,
      "learning_rate": 2.1340055213234025e-05,
      "loss": 1.4872,
      "step": 9226
    },
    {
      "epoch": 0.9733122362869199,
      "grad_norm": 0.4279082715511322,
      "learning_rate": 2.131045713662756e-05,
      "loss": 1.4441,
      "step": 9227
    },
    {
      "epoch": 0.9734177215189873,
      "grad_norm": 0.40274637937545776,
      "learning_rate": 2.1280900111748943e-05,
      "loss": 1.4822,
      "step": 9228
    },
    {
      "epoch": 0.9735232067510549,
      "grad_norm": 0.40205472707748413,
      "learning_rate": 2.1251384081660546e-05,
      "loss": 1.4564,
      "step": 9229
    },
    {
      "epoch": 0.9736286919831224,
      "grad_norm": 0.40804171562194824,
      "learning_rate": 2.12219089895037e-05,
      "loss": 1.4977,
      "step": 9230
    },
    {
      "epoch": 0.9737341772151898,
      "grad_norm": 0.41353917121887207,
      "learning_rate": 2.1192474778498613e-05,
      "loss": 1.4942,
      "step": 9231
    },
    {
      "epoch": 0.9738396624472574,
      "grad_norm": 0.41624319553375244,
      "learning_rate": 2.1163081391944224e-05,
      "loss": 1.4342,
      "step": 9232
    },
    {
      "epoch": 0.9739451476793249,
      "grad_norm": 0.4061141908168793,
      "learning_rate": 2.1133728773218143e-05,
      "loss": 1.4569,
      "step": 9233
    },
    {
      "epoch": 0.9740506329113924,
      "grad_norm": 0.39513930678367615,
      "learning_rate": 2.1104416865776507e-05,
      "loss": 1.4791,
      "step": 9234
    },
    {
      "epoch": 0.9741561181434599,
      "grad_norm": 0.43303805589675903,
      "learning_rate": 2.1075145613153853e-05,
      "loss": 1.4755,
      "step": 9235
    },
    {
      "epoch": 0.9742616033755275,
      "grad_norm": 0.40141043066978455,
      "learning_rate": 2.104591495896307e-05,
      "loss": 1.4624,
      "step": 9236
    },
    {
      "epoch": 0.9743670886075949,
      "grad_norm": 0.4169008731842041,
      "learning_rate": 2.1016724846895213e-05,
      "loss": 1.4707,
      "step": 9237
    },
    {
      "epoch": 0.9744725738396625,
      "grad_norm": 0.40253081917762756,
      "learning_rate": 2.0987575220719476e-05,
      "loss": 1.4834,
      "step": 9238
    },
    {
      "epoch": 0.97457805907173,
      "grad_norm": 0.4180942475795746,
      "learning_rate": 2.0958466024283035e-05,
      "loss": 1.4337,
      "step": 9239
    },
    {
      "epoch": 0.9746835443037974,
      "grad_norm": 0.3958808481693268,
      "learning_rate": 2.092939720151092e-05,
      "loss": 1.4722,
      "step": 9240
    },
    {
      "epoch": 0.974789029535865,
      "grad_norm": 0.4027206003665924,
      "learning_rate": 2.090036869640596e-05,
      "loss": 1.4484,
      "step": 9241
    },
    {
      "epoch": 0.9748945147679325,
      "grad_norm": 0.39618241786956787,
      "learning_rate": 2.0871380453048667e-05,
      "loss": 1.5062,
      "step": 9242
    },
    {
      "epoch": 0.975,
      "grad_norm": 0.3900398015975952,
      "learning_rate": 2.0842432415597064e-05,
      "loss": 1.4529,
      "step": 9243
    },
    {
      "epoch": 0.9751054852320675,
      "grad_norm": 0.4288707673549652,
      "learning_rate": 2.0813524528286672e-05,
      "loss": 1.499,
      "step": 9244
    },
    {
      "epoch": 0.9752109704641351,
      "grad_norm": 0.42837774753570557,
      "learning_rate": 2.0784656735430323e-05,
      "loss": 1.4378,
      "step": 9245
    },
    {
      "epoch": 0.9753164556962025,
      "grad_norm": 0.42409801483154297,
      "learning_rate": 2.07558289814181e-05,
      "loss": 1.4858,
      "step": 9246
    },
    {
      "epoch": 0.9754219409282701,
      "grad_norm": 0.4169222414493561,
      "learning_rate": 2.0727041210717232e-05,
      "loss": 1.5086,
      "step": 9247
    },
    {
      "epoch": 0.9755274261603376,
      "grad_norm": 0.40484586358070374,
      "learning_rate": 2.069829336787193e-05,
      "loss": 1.4701,
      "step": 9248
    },
    {
      "epoch": 0.975632911392405,
      "grad_norm": 0.40863052010536194,
      "learning_rate": 2.0669585397503362e-05,
      "loss": 1.504,
      "step": 9249
    },
    {
      "epoch": 0.9757383966244726,
      "grad_norm": 0.4081283509731293,
      "learning_rate": 2.064091724430947e-05,
      "loss": 1.4696,
      "step": 9250
    },
    {
      "epoch": 0.97584388185654,
      "grad_norm": 0.4262232482433319,
      "learning_rate": 2.061228885306492e-05,
      "loss": 1.4609,
      "step": 9251
    },
    {
      "epoch": 0.9759493670886076,
      "grad_norm": 0.4494875967502594,
      "learning_rate": 2.0583700168620984e-05,
      "loss": 1.4834,
      "step": 9252
    },
    {
      "epoch": 0.9760548523206751,
      "grad_norm": 0.4169006645679474,
      "learning_rate": 2.055515113590538e-05,
      "loss": 1.4814,
      "step": 9253
    },
    {
      "epoch": 0.9761603375527426,
      "grad_norm": 0.4058118760585785,
      "learning_rate": 2.0526641699922274e-05,
      "loss": 1.4778,
      "step": 9254
    },
    {
      "epoch": 0.9762658227848101,
      "grad_norm": 0.440136581659317,
      "learning_rate": 2.0498171805752038e-05,
      "loss": 1.4669,
      "step": 9255
    },
    {
      "epoch": 0.9763713080168777,
      "grad_norm": 0.3998717665672302,
      "learning_rate": 2.0469741398551272e-05,
      "loss": 1.4589,
      "step": 9256
    },
    {
      "epoch": 0.9764767932489451,
      "grad_norm": 0.41716086864471436,
      "learning_rate": 2.0441350423552625e-05,
      "loss": 1.5096,
      "step": 9257
    },
    {
      "epoch": 0.9765822784810126,
      "grad_norm": 0.43359634280204773,
      "learning_rate": 2.0412998826064695e-05,
      "loss": 1.4654,
      "step": 9258
    },
    {
      "epoch": 0.9766877637130802,
      "grad_norm": 0.4217475354671478,
      "learning_rate": 2.0384686551471954e-05,
      "loss": 1.4759,
      "step": 9259
    },
    {
      "epoch": 0.9767932489451476,
      "grad_norm": 0.41475754976272583,
      "learning_rate": 2.0356413545234603e-05,
      "loss": 1.4628,
      "step": 9260
    },
    {
      "epoch": 0.9768987341772152,
      "grad_norm": 0.4203028380870819,
      "learning_rate": 2.0328179752888504e-05,
      "loss": 1.4866,
      "step": 9261
    },
    {
      "epoch": 0.9770042194092827,
      "grad_norm": 0.41151440143585205,
      "learning_rate": 2.029998512004507e-05,
      "loss": 1.4797,
      "step": 9262
    },
    {
      "epoch": 0.9771097046413502,
      "grad_norm": 0.4004734456539154,
      "learning_rate": 2.0271829592391113e-05,
      "loss": 1.4902,
      "step": 9263
    },
    {
      "epoch": 0.9772151898734177,
      "grad_norm": 0.4614029824733734,
      "learning_rate": 2.0243713115688823e-05,
      "loss": 1.5083,
      "step": 9264
    },
    {
      "epoch": 0.9773206751054853,
      "grad_norm": 0.43761423230171204,
      "learning_rate": 2.021563563577556e-05,
      "loss": 1.4815,
      "step": 9265
    },
    {
      "epoch": 0.9774261603375527,
      "grad_norm": 0.426094114780426,
      "learning_rate": 2.0187597098563862e-05,
      "loss": 1.5057,
      "step": 9266
    },
    {
      "epoch": 0.9775316455696202,
      "grad_norm": 0.3993452489376068,
      "learning_rate": 2.0159597450041257e-05,
      "loss": 1.4761,
      "step": 9267
    },
    {
      "epoch": 0.9776371308016878,
      "grad_norm": 0.43500107526779175,
      "learning_rate": 2.0131636636270178e-05,
      "loss": 1.477,
      "step": 9268
    },
    {
      "epoch": 0.9777426160337552,
      "grad_norm": 0.45615869760513306,
      "learning_rate": 2.0103714603387898e-05,
      "loss": 1.4863,
      "step": 9269
    },
    {
      "epoch": 0.9778481012658228,
      "grad_norm": 0.4383704662322998,
      "learning_rate": 2.0075831297606357e-05,
      "loss": 1.4893,
      "step": 9270
    },
    {
      "epoch": 0.9779535864978903,
      "grad_norm": 0.41047871112823486,
      "learning_rate": 2.004798666521213e-05,
      "loss": 1.4418,
      "step": 9271
    },
    {
      "epoch": 0.9780590717299578,
      "grad_norm": 0.41966596245765686,
      "learning_rate": 2.0020180652566292e-05,
      "loss": 1.4769,
      "step": 9272
    },
    {
      "epoch": 0.9781645569620253,
      "grad_norm": 0.4164266884326935,
      "learning_rate": 1.999241320610428e-05,
      "loss": 1.4869,
      "step": 9273
    },
    {
      "epoch": 0.9782700421940929,
      "grad_norm": 0.3983272314071655,
      "learning_rate": 1.996468427233586e-05,
      "loss": 1.4991,
      "step": 9274
    },
    {
      "epoch": 0.9783755274261603,
      "grad_norm": 0.45602473616600037,
      "learning_rate": 1.9936993797844958e-05,
      "loss": 1.4895,
      "step": 9275
    },
    {
      "epoch": 0.9784810126582278,
      "grad_norm": 0.4188278019428253,
      "learning_rate": 1.9909341729289613e-05,
      "loss": 1.4588,
      "step": 9276
    },
    {
      "epoch": 0.9785864978902954,
      "grad_norm": 0.3984006643295288,
      "learning_rate": 1.9881728013401843e-05,
      "loss": 1.465,
      "step": 9277
    },
    {
      "epoch": 0.9786919831223628,
      "grad_norm": 0.41702425479888916,
      "learning_rate": 1.9854152596987523e-05,
      "loss": 1.4988,
      "step": 9278
    },
    {
      "epoch": 0.9787974683544304,
      "grad_norm": 0.40520593523979187,
      "learning_rate": 1.9826615426926342e-05,
      "loss": 1.4731,
      "step": 9279
    },
    {
      "epoch": 0.9789029535864979,
      "grad_norm": 0.4267204701900482,
      "learning_rate": 1.9799116450171627e-05,
      "loss": 1.436,
      "step": 9280
    },
    {
      "epoch": 0.9790084388185654,
      "grad_norm": 0.4072263836860657,
      "learning_rate": 1.9771655613750312e-05,
      "loss": 1.4747,
      "step": 9281
    },
    {
      "epoch": 0.9791139240506329,
      "grad_norm": 0.40923407673835754,
      "learning_rate": 1.9744232864762798e-05,
      "loss": 1.4754,
      "step": 9282
    },
    {
      "epoch": 0.9792194092827005,
      "grad_norm": 0.406464159488678,
      "learning_rate": 1.971684815038283e-05,
      "loss": 1.4756,
      "step": 9283
    },
    {
      "epoch": 0.9793248945147679,
      "grad_norm": 0.41171348094940186,
      "learning_rate": 1.9689501417857458e-05,
      "loss": 1.4873,
      "step": 9284
    },
    {
      "epoch": 0.9794303797468354,
      "grad_norm": 0.4129951000213623,
      "learning_rate": 1.9662192614506883e-05,
      "loss": 1.4979,
      "step": 9285
    },
    {
      "epoch": 0.979535864978903,
      "grad_norm": 0.4049890637397766,
      "learning_rate": 1.9634921687724354e-05,
      "loss": 1.4657,
      "step": 9286
    },
    {
      "epoch": 0.9796413502109704,
      "grad_norm": 0.423679381608963,
      "learning_rate": 1.960768858497612e-05,
      "loss": 1.4898,
      "step": 9287
    },
    {
      "epoch": 0.979746835443038,
      "grad_norm": 0.41526612639427185,
      "learning_rate": 1.9580493253801253e-05,
      "loss": 1.4776,
      "step": 9288
    },
    {
      "epoch": 0.9798523206751055,
      "grad_norm": 0.4138219356536865,
      "learning_rate": 1.9553335641811623e-05,
      "loss": 1.4729,
      "step": 9289
    },
    {
      "epoch": 0.979957805907173,
      "grad_norm": 0.4148155748844147,
      "learning_rate": 1.952621569669175e-05,
      "loss": 1.4926,
      "step": 9290
    },
    {
      "epoch": 0.9800632911392405,
      "grad_norm": 0.3987088203430176,
      "learning_rate": 1.9499133366198684e-05,
      "loss": 1.4524,
      "step": 9291
    },
    {
      "epoch": 0.9801687763713081,
      "grad_norm": 0.42366352677345276,
      "learning_rate": 1.947208859816199e-05,
      "loss": 1.4562,
      "step": 9292
    },
    {
      "epoch": 0.9802742616033755,
      "grad_norm": 0.4108063578605652,
      "learning_rate": 1.9445081340483534e-05,
      "loss": 1.4961,
      "step": 9293
    },
    {
      "epoch": 0.980379746835443,
      "grad_norm": 0.4230404794216156,
      "learning_rate": 1.9418111541137484e-05,
      "loss": 1.502,
      "step": 9294
    },
    {
      "epoch": 0.9804852320675106,
      "grad_norm": 0.4407208263874054,
      "learning_rate": 1.939117914817016e-05,
      "loss": 1.4918,
      "step": 9295
    },
    {
      "epoch": 0.980590717299578,
      "grad_norm": 0.4105750620365143,
      "learning_rate": 1.936428410969991e-05,
      "loss": 1.4586,
      "step": 9296
    },
    {
      "epoch": 0.9806962025316456,
      "grad_norm": 0.42926931381225586,
      "learning_rate": 1.933742637391708e-05,
      "loss": 1.4716,
      "step": 9297
    },
    {
      "epoch": 0.9808016877637131,
      "grad_norm": 0.42408859729766846,
      "learning_rate": 1.9310605889083842e-05,
      "loss": 1.5031,
      "step": 9298
    },
    {
      "epoch": 0.9809071729957806,
      "grad_norm": 0.4241902232170105,
      "learning_rate": 1.9283822603534143e-05,
      "loss": 1.4835,
      "step": 9299
    },
    {
      "epoch": 0.9810126582278481,
      "grad_norm": 0.40558022260665894,
      "learning_rate": 1.9257076465673605e-05,
      "loss": 1.4956,
      "step": 9300
    },
    {
      "epoch": 0.9811181434599157,
      "grad_norm": 0.4021380543708801,
      "learning_rate": 1.923036742397937e-05,
      "loss": 1.4691,
      "step": 9301
    },
    {
      "epoch": 0.9812236286919831,
      "grad_norm": 0.41294413805007935,
      "learning_rate": 1.9203695427000086e-05,
      "loss": 1.4787,
      "step": 9302
    },
    {
      "epoch": 0.9813291139240506,
      "grad_norm": 0.41563257575035095,
      "learning_rate": 1.9177060423355717e-05,
      "loss": 1.4989,
      "step": 9303
    },
    {
      "epoch": 0.9814345991561182,
      "grad_norm": 0.41813787817955017,
      "learning_rate": 1.9150462361737524e-05,
      "loss": 1.4605,
      "step": 9304
    },
    {
      "epoch": 0.9815400843881856,
      "grad_norm": 0.44080349802970886,
      "learning_rate": 1.912390119090793e-05,
      "loss": 1.4871,
      "step": 9305
    },
    {
      "epoch": 0.9816455696202532,
      "grad_norm": 0.4605334401130676,
      "learning_rate": 1.909737685970039e-05,
      "loss": 1.487,
      "step": 9306
    },
    {
      "epoch": 0.9817510548523207,
      "grad_norm": 0.40913528203964233,
      "learning_rate": 1.9070889317019377e-05,
      "loss": 1.5157,
      "step": 9307
    },
    {
      "epoch": 0.9818565400843882,
      "grad_norm": 0.4434133768081665,
      "learning_rate": 1.904443851184018e-05,
      "loss": 1.494,
      "step": 9308
    },
    {
      "epoch": 0.9819620253164557,
      "grad_norm": 0.41206395626068115,
      "learning_rate": 1.90180243932089e-05,
      "loss": 1.4345,
      "step": 9309
    },
    {
      "epoch": 0.9820675105485233,
      "grad_norm": 0.4508298635482788,
      "learning_rate": 1.899164691024229e-05,
      "loss": 1.4925,
      "step": 9310
    },
    {
      "epoch": 0.9821729957805907,
      "grad_norm": 0.43016839027404785,
      "learning_rate": 1.8965306012127665e-05,
      "loss": 1.464,
      "step": 9311
    },
    {
      "epoch": 0.9822784810126582,
      "grad_norm": 0.3965884745121002,
      "learning_rate": 1.8939001648122847e-05,
      "loss": 1.4872,
      "step": 9312
    },
    {
      "epoch": 0.9823839662447258,
      "grad_norm": 0.41789162158966064,
      "learning_rate": 1.8912733767556005e-05,
      "loss": 1.4708,
      "step": 9313
    },
    {
      "epoch": 0.9824894514767932,
      "grad_norm": 0.4155394732952118,
      "learning_rate": 1.8886502319825606e-05,
      "loss": 1.5208,
      "step": 9314
    },
    {
      "epoch": 0.9825949367088608,
      "grad_norm": 0.40481072664260864,
      "learning_rate": 1.8860307254400307e-05,
      "loss": 1.4554,
      "step": 9315
    },
    {
      "epoch": 0.9827004219409282,
      "grad_norm": 0.40905100107192993,
      "learning_rate": 1.883414852081882e-05,
      "loss": 1.5075,
      "step": 9316
    },
    {
      "epoch": 0.9828059071729958,
      "grad_norm": 0.42430728673934937,
      "learning_rate": 1.8808026068689887e-05,
      "loss": 1.4601,
      "step": 9317
    },
    {
      "epoch": 0.9829113924050633,
      "grad_norm": 0.4392472207546234,
      "learning_rate": 1.87819398476921e-05,
      "loss": 1.4612,
      "step": 9318
    },
    {
      "epoch": 0.9830168776371307,
      "grad_norm": 0.4241482615470886,
      "learning_rate": 1.8755889807573868e-05,
      "loss": 1.4666,
      "step": 9319
    },
    {
      "epoch": 0.9831223628691983,
      "grad_norm": 0.4253840446472168,
      "learning_rate": 1.872987589815331e-05,
      "loss": 1.4576,
      "step": 9320
    },
    {
      "epoch": 0.9832278481012658,
      "grad_norm": 0.3936906158924103,
      "learning_rate": 1.870389806931811e-05,
      "loss": 1.5007,
      "step": 9321
    },
    {
      "epoch": 0.9833333333333333,
      "grad_norm": 0.45343899726867676,
      "learning_rate": 1.8677956271025497e-05,
      "loss": 1.4946,
      "step": 9322
    },
    {
      "epoch": 0.9834388185654008,
      "grad_norm": 0.39700406789779663,
      "learning_rate": 1.865205045330207e-05,
      "loss": 1.546,
      "step": 9323
    },
    {
      "epoch": 0.9835443037974684,
      "grad_norm": 0.4086047112941742,
      "learning_rate": 1.8626180566243758e-05,
      "loss": 1.4917,
      "step": 9324
    },
    {
      "epoch": 0.9836497890295358,
      "grad_norm": 0.4209315776824951,
      "learning_rate": 1.8600346560015723e-05,
      "loss": 1.4773,
      "step": 9325
    },
    {
      "epoch": 0.9837552742616034,
      "grad_norm": 0.44885584712028503,
      "learning_rate": 1.8574548384852206e-05,
      "loss": 1.4954,
      "step": 9326
    },
    {
      "epoch": 0.9838607594936709,
      "grad_norm": 0.4245237708091736,
      "learning_rate": 1.8548785991056514e-05,
      "loss": 1.4984,
      "step": 9327
    },
    {
      "epoch": 0.9839662447257383,
      "grad_norm": 0.42421185970306396,
      "learning_rate": 1.8523059329000848e-05,
      "loss": 1.4559,
      "step": 9328
    },
    {
      "epoch": 0.9840717299578059,
      "grad_norm": 0.4430307149887085,
      "learning_rate": 1.8497368349126255e-05,
      "loss": 1.4642,
      "step": 9329
    },
    {
      "epoch": 0.9841772151898734,
      "grad_norm": 0.4064572751522064,
      "learning_rate": 1.8471713001942538e-05,
      "loss": 1.4581,
      "step": 9330
    },
    {
      "epoch": 0.9842827004219409,
      "grad_norm": 0.42986389994621277,
      "learning_rate": 1.84460932380281e-05,
      "loss": 1.4979,
      "step": 9331
    },
    {
      "epoch": 0.9843881856540084,
      "grad_norm": 0.4210165739059448,
      "learning_rate": 1.842050900802993e-05,
      "loss": 1.465,
      "step": 9332
    },
    {
      "epoch": 0.984493670886076,
      "grad_norm": 0.4063452184200287,
      "learning_rate": 1.8394960262663446e-05,
      "loss": 1.5021,
      "step": 9333
    },
    {
      "epoch": 0.9845991561181434,
      "grad_norm": 0.41390594840049744,
      "learning_rate": 1.8369446952712427e-05,
      "loss": 1.5027,
      "step": 9334
    },
    {
      "epoch": 0.984704641350211,
      "grad_norm": 0.4068928062915802,
      "learning_rate": 1.834396902902892e-05,
      "loss": 1.4561,
      "step": 9335
    },
    {
      "epoch": 0.9848101265822785,
      "grad_norm": 0.42157015204429626,
      "learning_rate": 1.8318526442533124e-05,
      "loss": 1.4861,
      "step": 9336
    },
    {
      "epoch": 0.984915611814346,
      "grad_norm": 0.3931799829006195,
      "learning_rate": 1.8293119144213324e-05,
      "loss": 1.4875,
      "step": 9337
    },
    {
      "epoch": 0.9850210970464135,
      "grad_norm": 0.39976540207862854,
      "learning_rate": 1.826774708512579e-05,
      "loss": 1.4696,
      "step": 9338
    },
    {
      "epoch": 0.985126582278481,
      "grad_norm": 0.40549996495246887,
      "learning_rate": 1.824241021639465e-05,
      "loss": 1.4769,
      "step": 9339
    },
    {
      "epoch": 0.9852320675105485,
      "grad_norm": 0.39983344078063965,
      "learning_rate": 1.8217108489211845e-05,
      "loss": 1.5113,
      "step": 9340
    },
    {
      "epoch": 0.985337552742616,
      "grad_norm": 0.44870525598526,
      "learning_rate": 1.8191841854836994e-05,
      "loss": 1.508,
      "step": 9341
    },
    {
      "epoch": 0.9854430379746836,
      "grad_norm": 0.4412899315357208,
      "learning_rate": 1.8166610264597328e-05,
      "loss": 1.5076,
      "step": 9342
    },
    {
      "epoch": 0.985548523206751,
      "grad_norm": 0.42450806498527527,
      "learning_rate": 1.8141413669887598e-05,
      "loss": 1.5215,
      "step": 9343
    },
    {
      "epoch": 0.9856540084388186,
      "grad_norm": 0.4241693913936615,
      "learning_rate": 1.8116252022169935e-05,
      "loss": 1.4754,
      "step": 9344
    },
    {
      "epoch": 0.9857594936708861,
      "grad_norm": 0.4270302355289459,
      "learning_rate": 1.809112527297383e-05,
      "loss": 1.4897,
      "step": 9345
    },
    {
      "epoch": 0.9858649789029535,
      "grad_norm": 0.3991768956184387,
      "learning_rate": 1.8066033373895962e-05,
      "loss": 1.5017,
      "step": 9346
    },
    {
      "epoch": 0.9859704641350211,
      "grad_norm": 0.4303628206253052,
      "learning_rate": 1.804097627660017e-05,
      "loss": 1.5028,
      "step": 9347
    },
    {
      "epoch": 0.9860759493670886,
      "grad_norm": 0.40191367268562317,
      "learning_rate": 1.8015953932817347e-05,
      "loss": 1.5174,
      "step": 9348
    },
    {
      "epoch": 0.9861814345991561,
      "grad_norm": 0.414073646068573,
      "learning_rate": 1.799096629434529e-05,
      "loss": 1.4858,
      "step": 9349
    },
    {
      "epoch": 0.9862869198312236,
      "grad_norm": 0.3921477496623993,
      "learning_rate": 1.7966013313048696e-05,
      "loss": 1.4696,
      "step": 9350
    },
    {
      "epoch": 0.9863924050632912,
      "grad_norm": 0.4376528561115265,
      "learning_rate": 1.794109494085898e-05,
      "loss": 1.5231,
      "step": 9351
    },
    {
      "epoch": 0.9864978902953586,
      "grad_norm": 0.4133533537387848,
      "learning_rate": 1.7916211129774273e-05,
      "loss": 1.4892,
      "step": 9352
    },
    {
      "epoch": 0.9866033755274262,
      "grad_norm": 0.47707170248031616,
      "learning_rate": 1.7891361831859263e-05,
      "loss": 1.5059,
      "step": 9353
    },
    {
      "epoch": 0.9867088607594937,
      "grad_norm": 0.4317123293876648,
      "learning_rate": 1.78665469992451e-05,
      "loss": 1.4695,
      "step": 9354
    },
    {
      "epoch": 0.9868143459915611,
      "grad_norm": 0.4074588418006897,
      "learning_rate": 1.7841766584129377e-05,
      "loss": 1.4988,
      "step": 9355
    },
    {
      "epoch": 0.9869198312236287,
      "grad_norm": 0.43971219658851624,
      "learning_rate": 1.7817020538775933e-05,
      "loss": 1.479,
      "step": 9356
    },
    {
      "epoch": 0.9870253164556962,
      "grad_norm": 0.4198453724384308,
      "learning_rate": 1.779230881551485e-05,
      "loss": 1.4675,
      "step": 9357
    },
    {
      "epoch": 0.9871308016877637,
      "grad_norm": 0.4048851728439331,
      "learning_rate": 1.7767631366742332e-05,
      "loss": 1.4631,
      "step": 9358
    },
    {
      "epoch": 0.9872362869198312,
      "grad_norm": 0.4280070662498474,
      "learning_rate": 1.7742988144920578e-05,
      "loss": 1.5103,
      "step": 9359
    },
    {
      "epoch": 0.9873417721518988,
      "grad_norm": 0.40196070075035095,
      "learning_rate": 1.7718379102577752e-05,
      "loss": 1.4855,
      "step": 9360
    },
    {
      "epoch": 0.9874472573839662,
      "grad_norm": 0.4261288046836853,
      "learning_rate": 1.7693804192307827e-05,
      "loss": 1.4829,
      "step": 9361
    },
    {
      "epoch": 0.9875527426160338,
      "grad_norm": 0.40299469232559204,
      "learning_rate": 1.7669263366770554e-05,
      "loss": 1.4925,
      "step": 9362
    },
    {
      "epoch": 0.9876582278481013,
      "grad_norm": 0.43445515632629395,
      "learning_rate": 1.7644756578691348e-05,
      "loss": 1.4827,
      "step": 9363
    },
    {
      "epoch": 0.9877637130801687,
      "grad_norm": 0.43816348910331726,
      "learning_rate": 1.7620283780861163e-05,
      "loss": 1.4904,
      "step": 9364
    },
    {
      "epoch": 0.9878691983122363,
      "grad_norm": 0.4279087781906128,
      "learning_rate": 1.759584492613646e-05,
      "loss": 1.4878,
      "step": 9365
    },
    {
      "epoch": 0.9879746835443038,
      "grad_norm": 0.4332057237625122,
      "learning_rate": 1.757143996743906e-05,
      "loss": 1.5004,
      "step": 9366
    },
    {
      "epoch": 0.9880801687763713,
      "grad_norm": 0.4087739884853363,
      "learning_rate": 1.75470688577561e-05,
      "loss": 1.4455,
      "step": 9367
    },
    {
      "epoch": 0.9881856540084388,
      "grad_norm": 0.4181956648826599,
      "learning_rate": 1.7522731550139926e-05,
      "loss": 1.4949,
      "step": 9368
    },
    {
      "epoch": 0.9882911392405064,
      "grad_norm": 0.42697709798812866,
      "learning_rate": 1.7498427997707978e-05,
      "loss": 1.4721,
      "step": 9369
    },
    {
      "epoch": 0.9883966244725738,
      "grad_norm": 0.4501078724861145,
      "learning_rate": 1.7474158153642745e-05,
      "loss": 1.5217,
      "step": 9370
    },
    {
      "epoch": 0.9885021097046414,
      "grad_norm": 0.4283146560192108,
      "learning_rate": 1.744992197119162e-05,
      "loss": 1.4699,
      "step": 9371
    },
    {
      "epoch": 0.9886075949367089,
      "grad_norm": 0.395073801279068,
      "learning_rate": 1.7425719403666873e-05,
      "loss": 1.4714,
      "step": 9372
    },
    {
      "epoch": 0.9887130801687763,
      "grad_norm": 0.40190500020980835,
      "learning_rate": 1.7401550404445523e-05,
      "loss": 1.4857,
      "step": 9373
    },
    {
      "epoch": 0.9888185654008439,
      "grad_norm": 0.41541337966918945,
      "learning_rate": 1.737741492696922e-05,
      "loss": 1.4331,
      "step": 9374
    },
    {
      "epoch": 0.9889240506329114,
      "grad_norm": 0.40836068987846375,
      "learning_rate": 1.735331292474423e-05,
      "loss": 1.5041,
      "step": 9375
    },
    {
      "epoch": 0.9890295358649789,
      "grad_norm": 0.3903765082359314,
      "learning_rate": 1.73292443513413e-05,
      "loss": 1.4428,
      "step": 9376
    },
    {
      "epoch": 0.9891350210970464,
      "grad_norm": 0.41661468148231506,
      "learning_rate": 1.730520916039554e-05,
      "loss": 1.4502,
      "step": 9377
    },
    {
      "epoch": 0.989240506329114,
      "grad_norm": 0.4188127815723419,
      "learning_rate": 1.728120730560641e-05,
      "loss": 1.5105,
      "step": 9378
    },
    {
      "epoch": 0.9893459915611814,
      "grad_norm": 0.4256414771080017,
      "learning_rate": 1.7257238740737548e-05,
      "loss": 1.4926,
      "step": 9379
    },
    {
      "epoch": 0.989451476793249,
      "grad_norm": 0.40655890107154846,
      "learning_rate": 1.7233303419616745e-05,
      "loss": 1.4738,
      "step": 9380
    },
    {
      "epoch": 0.9895569620253165,
      "grad_norm": 0.4350295662879944,
      "learning_rate": 1.720940129613584e-05,
      "loss": 1.4408,
      "step": 9381
    },
    {
      "epoch": 0.989662447257384,
      "grad_norm": 0.4115702509880066,
      "learning_rate": 1.718553232425059e-05,
      "loss": 1.4874,
      "step": 9382
    },
    {
      "epoch": 0.9897679324894515,
      "grad_norm": 0.43595340847969055,
      "learning_rate": 1.7161696457980646e-05,
      "loss": 1.4917,
      "step": 9383
    },
    {
      "epoch": 0.9898734177215189,
      "grad_norm": 0.42772090435028076,
      "learning_rate": 1.7137893651409406e-05,
      "loss": 1.4736,
      "step": 9384
    },
    {
      "epoch": 0.9899789029535865,
      "grad_norm": 0.4015375077724457,
      "learning_rate": 1.7114123858683976e-05,
      "loss": 1.4666,
      "step": 9385
    },
    {
      "epoch": 0.990084388185654,
      "grad_norm": 0.4261699616909027,
      "learning_rate": 1.7090387034015054e-05,
      "loss": 1.4738,
      "step": 9386
    },
    {
      "epoch": 0.9901898734177215,
      "grad_norm": 0.42017146944999695,
      "learning_rate": 1.7066683131676825e-05,
      "loss": 1.4937,
      "step": 9387
    },
    {
      "epoch": 0.990295358649789,
      "grad_norm": 0.4136383831501007,
      "learning_rate": 1.704301210600693e-05,
      "loss": 1.4649,
      "step": 9388
    },
    {
      "epoch": 0.9904008438818566,
      "grad_norm": 0.42785435914993286,
      "learning_rate": 1.7019373911406307e-05,
      "loss": 1.4825,
      "step": 9389
    },
    {
      "epoch": 0.990506329113924,
      "grad_norm": 0.4480273127555847,
      "learning_rate": 1.699576850233916e-05,
      "loss": 1.4794,
      "step": 9390
    },
    {
      "epoch": 0.9906118143459915,
      "grad_norm": 0.4351575970649719,
      "learning_rate": 1.697219583333286e-05,
      "loss": 1.4851,
      "step": 9391
    },
    {
      "epoch": 0.9907172995780591,
      "grad_norm": 0.3972213864326477,
      "learning_rate": 1.694865585897781e-05,
      "loss": 1.4876,
      "step": 9392
    },
    {
      "epoch": 0.9908227848101265,
      "grad_norm": 0.4147304594516754,
      "learning_rate": 1.6925148533927435e-05,
      "loss": 1.4884,
      "step": 9393
    },
    {
      "epoch": 0.9909282700421941,
      "grad_norm": 0.4648360013961792,
      "learning_rate": 1.690167381289802e-05,
      "loss": 1.4689,
      "step": 9394
    },
    {
      "epoch": 0.9910337552742616,
      "grad_norm": 0.4486113488674164,
      "learning_rate": 1.6878231650668686e-05,
      "loss": 1.5359,
      "step": 9395
    },
    {
      "epoch": 0.9911392405063291,
      "grad_norm": 0.41870036721229553,
      "learning_rate": 1.6854822002081265e-05,
      "loss": 1.4622,
      "step": 9396
    },
    {
      "epoch": 0.9912447257383966,
      "grad_norm": 0.40832987427711487,
      "learning_rate": 1.6831444822040207e-05,
      "loss": 1.4846,
      "step": 9397
    },
    {
      "epoch": 0.9913502109704642,
      "grad_norm": 0.39077070355415344,
      "learning_rate": 1.6808100065512536e-05,
      "loss": 1.4909,
      "step": 9398
    },
    {
      "epoch": 0.9914556962025316,
      "grad_norm": 0.41672998666763306,
      "learning_rate": 1.67847876875277e-05,
      "loss": 1.5076,
      "step": 9399
    },
    {
      "epoch": 0.9915611814345991,
      "grad_norm": 0.42821717262268066,
      "learning_rate": 1.6761507643177553e-05,
      "loss": 1.4917,
      "step": 9400
    },
    {
      "epoch": 0.9916666666666667,
      "grad_norm": 0.42270588874816895,
      "learning_rate": 1.673825988761623e-05,
      "loss": 1.4375,
      "step": 9401
    },
    {
      "epoch": 0.9917721518987341,
      "grad_norm": 0.43380117416381836,
      "learning_rate": 1.671504437606004e-05,
      "loss": 1.4832,
      "step": 9402
    },
    {
      "epoch": 0.9918776371308017,
      "grad_norm": 0.48586305975914,
      "learning_rate": 1.6691861063787444e-05,
      "loss": 1.4342,
      "step": 9403
    },
    {
      "epoch": 0.9919831223628692,
      "grad_norm": 0.4231918156147003,
      "learning_rate": 1.666870990613889e-05,
      "loss": 1.4704,
      "step": 9404
    },
    {
      "epoch": 0.9920886075949367,
      "grad_norm": 0.4204140305519104,
      "learning_rate": 1.6645590858516798e-05,
      "loss": 1.4609,
      "step": 9405
    },
    {
      "epoch": 0.9921940928270042,
      "grad_norm": 0.4043950140476227,
      "learning_rate": 1.662250387638544e-05,
      "loss": 1.4548,
      "step": 9406
    },
    {
      "epoch": 0.9922995780590718,
      "grad_norm": 0.40539121627807617,
      "learning_rate": 1.6599448915270843e-05,
      "loss": 1.4638,
      "step": 9407
    },
    {
      "epoch": 0.9924050632911392,
      "grad_norm": 0.4334234893321991,
      "learning_rate": 1.657642593076074e-05,
      "loss": 1.4846,
      "step": 9408
    },
    {
      "epoch": 0.9925105485232067,
      "grad_norm": 0.40530937910079956,
      "learning_rate": 1.655343487850443e-05,
      "loss": 1.4879,
      "step": 9409
    },
    {
      "epoch": 0.9926160337552743,
      "grad_norm": 0.4055024981498718,
      "learning_rate": 1.6530475714212752e-05,
      "loss": 1.4713,
      "step": 9410
    },
    {
      "epoch": 0.9927215189873417,
      "grad_norm": 0.42495599389076233,
      "learning_rate": 1.6507548393657978e-05,
      "loss": 1.5232,
      "step": 9411
    },
    {
      "epoch": 0.9928270042194093,
      "grad_norm": 0.4067240357398987,
      "learning_rate": 1.6484652872673692e-05,
      "loss": 1.4391,
      "step": 9412
    },
    {
      "epoch": 0.9929324894514768,
      "grad_norm": 0.4060860574245453,
      "learning_rate": 1.6461789107154772e-05,
      "loss": 1.4521,
      "step": 9413
    },
    {
      "epoch": 0.9930379746835443,
      "grad_norm": 0.4327999949455261,
      "learning_rate": 1.6438957053057234e-05,
      "loss": 1.5112,
      "step": 9414
    },
    {
      "epoch": 0.9931434599156118,
      "grad_norm": 0.47486430406570435,
      "learning_rate": 1.6416156666398208e-05,
      "loss": 1.5041,
      "step": 9415
    },
    {
      "epoch": 0.9932489451476794,
      "grad_norm": 0.41256389021873474,
      "learning_rate": 1.6393387903255822e-05,
      "loss": 1.4746,
      "step": 9416
    },
    {
      "epoch": 0.9933544303797468,
      "grad_norm": 0.4513871669769287,
      "learning_rate": 1.63706507197691e-05,
      "loss": 1.5179,
      "step": 9417
    },
    {
      "epoch": 0.9934599156118143,
      "grad_norm": 0.4124189615249634,
      "learning_rate": 1.634794507213793e-05,
      "loss": 1.4735,
      "step": 9418
    },
    {
      "epoch": 0.9935654008438819,
      "grad_norm": 0.4506292939186096,
      "learning_rate": 1.6325270916622947e-05,
      "loss": 1.506,
      "step": 9419
    },
    {
      "epoch": 0.9936708860759493,
      "grad_norm": 0.41866230964660645,
      "learning_rate": 1.6302628209545423e-05,
      "loss": 1.4476,
      "step": 9420
    },
    {
      "epoch": 0.9937763713080169,
      "grad_norm": 0.4115118384361267,
      "learning_rate": 1.6280016907287243e-05,
      "loss": 1.5209,
      "step": 9421
    },
    {
      "epoch": 0.9938818565400844,
      "grad_norm": 0.4035124182701111,
      "learning_rate": 1.6257436966290764e-05,
      "loss": 1.5037,
      "step": 9422
    },
    {
      "epoch": 0.9939873417721519,
      "grad_norm": 0.42556723952293396,
      "learning_rate": 1.623488834305878e-05,
      "loss": 1.4688,
      "step": 9423
    },
    {
      "epoch": 0.9940928270042194,
      "grad_norm": 0.40388932824134827,
      "learning_rate": 1.62123709941544e-05,
      "loss": 1.4537,
      "step": 9424
    },
    {
      "epoch": 0.994198312236287,
      "grad_norm": 0.40017464756965637,
      "learning_rate": 1.6189884876200976e-05,
      "loss": 1.4763,
      "step": 9425
    },
    {
      "epoch": 0.9943037974683544,
      "grad_norm": 0.4098859131336212,
      "learning_rate": 1.6167429945882038e-05,
      "loss": 1.514,
      "step": 9426
    },
    {
      "epoch": 0.994409282700422,
      "grad_norm": 0.4903082847595215,
      "learning_rate": 1.6145006159941168e-05,
      "loss": 1.5052,
      "step": 9427
    },
    {
      "epoch": 0.9945147679324895,
      "grad_norm": 0.448175847530365,
      "learning_rate": 1.6122613475181976e-05,
      "loss": 1.462,
      "step": 9428
    },
    {
      "epoch": 0.9946202531645569,
      "grad_norm": 0.4094811975955963,
      "learning_rate": 1.610025184846797e-05,
      "loss": 1.5111,
      "step": 9429
    },
    {
      "epoch": 0.9947257383966245,
      "grad_norm": 0.43230900168418884,
      "learning_rate": 1.6077921236722464e-05,
      "loss": 1.4393,
      "step": 9430
    },
    {
      "epoch": 0.994831223628692,
      "grad_norm": 0.4265371859073639,
      "learning_rate": 1.6055621596928567e-05,
      "loss": 1.4648,
      "step": 9431
    },
    {
      "epoch": 0.9949367088607595,
      "grad_norm": 0.4402812421321869,
      "learning_rate": 1.6033352886129e-05,
      "loss": 1.4972,
      "step": 9432
    },
    {
      "epoch": 0.995042194092827,
      "grad_norm": 0.38128983974456787,
      "learning_rate": 1.60111150614261e-05,
      "loss": 1.4803,
      "step": 9433
    },
    {
      "epoch": 0.9951476793248946,
      "grad_norm": 0.40581005811691284,
      "learning_rate": 1.5988908079981696e-05,
      "loss": 1.4758,
      "step": 9434
    },
    {
      "epoch": 0.995253164556962,
      "grad_norm": 0.42678025364875793,
      "learning_rate": 1.5966731899017015e-05,
      "loss": 1.478,
      "step": 9435
    },
    {
      "epoch": 0.9953586497890295,
      "grad_norm": 0.44256579875946045,
      "learning_rate": 1.5944586475812638e-05,
      "loss": 1.4658,
      "step": 9436
    },
    {
      "epoch": 0.9954641350210971,
      "grad_norm": 0.4607081115245819,
      "learning_rate": 1.592247176770838e-05,
      "loss": 1.4687,
      "step": 9437
    },
    {
      "epoch": 0.9955696202531645,
      "grad_norm": 0.43047481775283813,
      "learning_rate": 1.590038773210323e-05,
      "loss": 1.4775,
      "step": 9438
    },
    {
      "epoch": 0.9956751054852321,
      "grad_norm": 0.43549367785453796,
      "learning_rate": 1.587833432645528e-05,
      "loss": 1.5243,
      "step": 9439
    },
    {
      "epoch": 0.9957805907172996,
      "grad_norm": 0.4098023772239685,
      "learning_rate": 1.5856311508281594e-05,
      "loss": 1.4736,
      "step": 9440
    },
    {
      "epoch": 0.9958860759493671,
      "grad_norm": 0.41330060362815857,
      "learning_rate": 1.5834319235158193e-05,
      "loss": 1.4905,
      "step": 9441
    },
    {
      "epoch": 0.9959915611814346,
      "grad_norm": 0.4074678122997284,
      "learning_rate": 1.5812357464719904e-05,
      "loss": 1.5066,
      "step": 9442
    },
    {
      "epoch": 0.9960970464135022,
      "grad_norm": 0.4042770564556122,
      "learning_rate": 1.5790426154660347e-05,
      "loss": 1.4628,
      "step": 9443
    },
    {
      "epoch": 0.9962025316455696,
      "grad_norm": 0.43630751967430115,
      "learning_rate": 1.5768525262731804e-05,
      "loss": 1.4804,
      "step": 9444
    },
    {
      "epoch": 0.9963080168776371,
      "grad_norm": 0.40686318278312683,
      "learning_rate": 1.574665474674514e-05,
      "loss": 1.4822,
      "step": 9445
    },
    {
      "epoch": 0.9964135021097047,
      "grad_norm": 0.4143632650375366,
      "learning_rate": 1.5724814564569767e-05,
      "loss": 1.4769,
      "step": 9446
    },
    {
      "epoch": 0.9965189873417721,
      "grad_norm": 0.4255259335041046,
      "learning_rate": 1.57030046741335e-05,
      "loss": 1.4876,
      "step": 9447
    },
    {
      "epoch": 0.9966244725738397,
      "grad_norm": 0.38538506627082825,
      "learning_rate": 1.568122503342252e-05,
      "loss": 1.4923,
      "step": 9448
    },
    {
      "epoch": 0.9967299578059071,
      "grad_norm": 0.4252411723136902,
      "learning_rate": 1.5659475600481297e-05,
      "loss": 1.5136,
      "step": 9449
    },
    {
      "epoch": 0.9968354430379747,
      "grad_norm": 0.4337424337863922,
      "learning_rate": 1.5637756333412454e-05,
      "loss": 1.5092,
      "step": 9450
    },
    {
      "epoch": 0.9969409282700422,
      "grad_norm": 0.42196324467658997,
      "learning_rate": 1.5616067190376765e-05,
      "loss": 1.478,
      "step": 9451
    },
    {
      "epoch": 0.9970464135021097,
      "grad_norm": 0.4065301716327667,
      "learning_rate": 1.559440812959299e-05,
      "loss": 1.5299,
      "step": 9452
    },
    {
      "epoch": 0.9971518987341772,
      "grad_norm": 0.40432223677635193,
      "learning_rate": 1.5572779109337886e-05,
      "loss": 1.4636,
      "step": 9453
    },
    {
      "epoch": 0.9972573839662447,
      "grad_norm": 0.4223288893699646,
      "learning_rate": 1.555118008794605e-05,
      "loss": 1.4642,
      "step": 9454
    },
    {
      "epoch": 0.9973628691983122,
      "grad_norm": 0.4262867271900177,
      "learning_rate": 1.552961102380987e-05,
      "loss": 1.462,
      "step": 9455
    },
    {
      "epoch": 0.9974683544303797,
      "grad_norm": 0.41093143820762634,
      "learning_rate": 1.550807187537945e-05,
      "loss": 1.4688,
      "step": 9456
    },
    {
      "epoch": 0.9975738396624473,
      "grad_norm": 0.40972617268562317,
      "learning_rate": 1.5486562601162513e-05,
      "loss": 1.4575,
      "step": 9457
    },
    {
      "epoch": 0.9976793248945147,
      "grad_norm": 0.4389844536781311,
      "learning_rate": 1.5465083159724344e-05,
      "loss": 1.4596,
      "step": 9458
    },
    {
      "epoch": 0.9977848101265823,
      "grad_norm": 0.41310906410217285,
      "learning_rate": 1.544363350968769e-05,
      "loss": 1.4525,
      "step": 9459
    },
    {
      "epoch": 0.9978902953586498,
      "grad_norm": 0.4174353778362274,
      "learning_rate": 1.542221360973268e-05,
      "loss": 1.4883,
      "step": 9460
    },
    {
      "epoch": 0.9979957805907173,
      "grad_norm": 0.41183555126190186,
      "learning_rate": 1.5400823418596764e-05,
      "loss": 1.4778,
      "step": 9461
    },
    {
      "epoch": 0.9981012658227848,
      "grad_norm": 0.3995887339115143,
      "learning_rate": 1.537946289507462e-05,
      "loss": 1.4755,
      "step": 9462
    },
    {
      "epoch": 0.9982067510548523,
      "grad_norm": 0.45345938205718994,
      "learning_rate": 1.5358131998018067e-05,
      "loss": 1.4989,
      "step": 9463
    },
    {
      "epoch": 0.9983122362869198,
      "grad_norm": 0.4081273078918457,
      "learning_rate": 1.5336830686336012e-05,
      "loss": 1.499,
      "step": 9464
    },
    {
      "epoch": 0.9984177215189873,
      "grad_norm": 0.39568179845809937,
      "learning_rate": 1.5315558918994333e-05,
      "loss": 1.5205,
      "step": 9465
    },
    {
      "epoch": 0.9985232067510549,
      "grad_norm": 0.43256375193595886,
      "learning_rate": 1.5294316655015837e-05,
      "loss": 1.4661,
      "step": 9466
    },
    {
      "epoch": 0.9986286919831223,
      "grad_norm": 0.3998488783836365,
      "learning_rate": 1.527310385348017e-05,
      "loss": 1.474,
      "step": 9467
    },
    {
      "epoch": 0.9987341772151899,
      "grad_norm": 0.4185790419578552,
      "learning_rate": 1.5251920473523708e-05,
      "loss": 1.4758,
      "step": 9468
    },
    {
      "epoch": 0.9988396624472574,
      "grad_norm": 0.43349960446357727,
      "learning_rate": 1.523076647433954e-05,
      "loss": 1.4707,
      "step": 9469
    },
    {
      "epoch": 0.9989451476793249,
      "grad_norm": 0.4138621389865875,
      "learning_rate": 1.5209641815177312e-05,
      "loss": 1.4934,
      "step": 9470
    },
    {
      "epoch": 0.9990506329113924,
      "grad_norm": 0.4318915009498596,
      "learning_rate": 1.5188546455343223e-05,
      "loss": 1.4917,
      "step": 9471
    },
    {
      "epoch": 0.99915611814346,
      "grad_norm": 0.42510899901390076,
      "learning_rate": 1.5167480354199909e-05,
      "loss": 1.4478,
      "step": 9472
    },
    {
      "epoch": 0.9992616033755274,
      "grad_norm": 0.39650705456733704,
      "learning_rate": 1.5146443471166345e-05,
      "loss": 1.4837,
      "step": 9473
    },
    {
      "epoch": 0.9993670886075949,
      "grad_norm": 0.4121500551700592,
      "learning_rate": 1.5125435765717816e-05,
      "loss": 1.5017,
      "step": 9474
    },
    {
      "epoch": 0.9994725738396625,
      "grad_norm": 0.3991440236568451,
      "learning_rate": 1.5104457197385799e-05,
      "loss": 1.4771,
      "step": 9475
    },
    {
      "epoch": 0.9995780590717299,
      "grad_norm": 0.41974425315856934,
      "learning_rate": 1.508350772575791e-05,
      "loss": 1.4759,
      "step": 9476
    },
    {
      "epoch": 0.9996835443037975,
      "grad_norm": 0.4257811903953552,
      "learning_rate": 1.5062587310477816e-05,
      "loss": 1.469,
      "step": 9477
    },
    {
      "epoch": 0.999789029535865,
      "grad_norm": 0.3897549510002136,
      "learning_rate": 1.5041695911245136e-05,
      "loss": 1.5022,
      "step": 9478
    },
    {
      "epoch": 0.9998945147679325,
      "grad_norm": 0.4248228967189789,
      "learning_rate": 1.5020833487815421e-05,
      "loss": 1.4829,
      "step": 9479
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.152031421661377,
      "learning_rate": 1.5e-05,
      "loss": 1.4378,
      "step": 9480
    }
  ],
  "logging_steps": 1,
  "max_steps": 9480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6871337555631488e+16,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}