{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 7180,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001392757660167131,
      "grad_norm": 2.8908393383026123,
      "learning_rate": 4.861111111111111e-07,
      "loss": 9.5583,
      "step": 1
    },
    {
      "epoch": 0.002785515320334262,
      "grad_norm": 2.8815290927886963,
      "learning_rate": 9.722222222222222e-07,
      "loss": 9.6058,
      "step": 2
    },
    {
      "epoch": 0.004178272980501393,
      "grad_norm": 2.6509928703308105,
      "learning_rate": 1.458333333333333e-06,
      "loss": 9.549,
      "step": 3
    },
    {
      "epoch": 0.005571030640668524,
      "grad_norm": 2.6149637699127197,
      "learning_rate": 1.9444444444444444e-06,
      "loss": 9.5588,
      "step": 4
    },
    {
      "epoch": 0.006963788300835654,
      "grad_norm": 3.3958027362823486,
      "learning_rate": 2.4305555555555552e-06,
      "loss": 9.584,
      "step": 5
    },
    {
      "epoch": 0.008356545961002786,
      "grad_norm": 2.5388824939727783,
      "learning_rate": 2.916666666666666e-06,
      "loss": 9.5191,
      "step": 6
    },
    {
      "epoch": 0.009749303621169917,
      "grad_norm": 2.521509885787964,
      "learning_rate": 3.4027777777777774e-06,
      "loss": 9.5599,
      "step": 7
    },
    {
      "epoch": 0.011142061281337047,
      "grad_norm": 2.5034923553466797,
      "learning_rate": 3.888888888888889e-06,
      "loss": 9.5376,
      "step": 8
    },
    {
      "epoch": 0.012534818941504178,
      "grad_norm": 2.9331307411193848,
      "learning_rate": 4.375e-06,
      "loss": 9.5846,
      "step": 9
    },
    {
      "epoch": 0.013927576601671309,
      "grad_norm": 2.613980531692505,
      "learning_rate": 4.8611111111111105e-06,
      "loss": 9.5551,
      "step": 10
    },
    {
      "epoch": 0.01532033426183844,
      "grad_norm": 2.579277276992798,
      "learning_rate": 5.347222222222222e-06,
      "loss": 9.5235,
      "step": 11
    },
    {
      "epoch": 0.016713091922005572,
      "grad_norm": 2.4165875911712646,
      "learning_rate": 5.833333333333332e-06,
      "loss": 9.4958,
      "step": 12
    },
    {
      "epoch": 0.018105849582172703,
      "grad_norm": 2.5124220848083496,
      "learning_rate": 6.319444444444444e-06,
      "loss": 9.5346,
      "step": 13
    },
    {
      "epoch": 0.019498607242339833,
      "grad_norm": 2.4491915702819824,
      "learning_rate": 6.805555555555555e-06,
      "loss": 9.5037,
      "step": 14
    },
    {
      "epoch": 0.020891364902506964,
      "grad_norm": 2.270697593688965,
      "learning_rate": 7.291666666666667e-06,
      "loss": 9.5149,
      "step": 15
    },
    {
      "epoch": 0.022284122562674095,
      "grad_norm": 2.4208738803863525,
      "learning_rate": 7.777777777777777e-06,
      "loss": 9.5268,
      "step": 16
    },
    {
      "epoch": 0.023676880222841225,
      "grad_norm": 2.4248204231262207,
      "learning_rate": 8.263888888888888e-06,
      "loss": 9.5157,
      "step": 17
    },
    {
      "epoch": 0.025069637883008356,
      "grad_norm": 2.4280309677124023,
      "learning_rate": 8.75e-06,
      "loss": 9.4933,
      "step": 18
    },
    {
      "epoch": 0.026462395543175487,
      "grad_norm": 3.640717029571533,
      "learning_rate": 9.23611111111111e-06,
      "loss": 9.5245,
      "step": 19
    },
    {
      "epoch": 0.027855153203342618,
      "grad_norm": 2.6451287269592285,
      "learning_rate": 9.722222222222221e-06,
      "loss": 9.5873,
      "step": 20
    },
    {
      "epoch": 0.02924791086350975,
      "grad_norm": 2.3361599445343018,
      "learning_rate": 1.0208333333333334e-05,
      "loss": 9.5214,
      "step": 21
    },
    {
      "epoch": 0.03064066852367688,
      "grad_norm": 2.440614700317383,
      "learning_rate": 1.0694444444444444e-05,
      "loss": 9.5212,
      "step": 22
    },
    {
      "epoch": 0.03203342618384401,
      "grad_norm": 2.296144962310791,
      "learning_rate": 1.1180555555555554e-05,
      "loss": 9.4204,
      "step": 23
    },
    {
      "epoch": 0.033426183844011144,
      "grad_norm": 2.450535774230957,
      "learning_rate": 1.1666666666666665e-05,
      "loss": 9.5338,
      "step": 24
    },
    {
      "epoch": 0.034818941504178275,
      "grad_norm": 2.552171230316162,
      "learning_rate": 1.2152777777777777e-05,
      "loss": 9.5031,
      "step": 25
    },
    {
      "epoch": 0.036211699164345405,
      "grad_norm": 2.9361586570739746,
      "learning_rate": 1.2638888888888888e-05,
      "loss": 9.4962,
      "step": 26
    },
    {
      "epoch": 0.037604456824512536,
      "grad_norm": 2.195554733276367,
      "learning_rate": 1.3124999999999999e-05,
      "loss": 9.4691,
      "step": 27
    },
    {
      "epoch": 0.03899721448467967,
      "grad_norm": 2.3282525539398193,
      "learning_rate": 1.361111111111111e-05,
      "loss": 9.4448,
      "step": 28
    },
    {
      "epoch": 0.0403899721448468,
      "grad_norm": 2.380357503890991,
      "learning_rate": 1.409722222222222e-05,
      "loss": 9.4604,
      "step": 29
    },
    {
      "epoch": 0.04178272980501393,
      "grad_norm": 2.1941680908203125,
      "learning_rate": 1.4583333333333333e-05,
      "loss": 9.5046,
      "step": 30
    },
    {
      "epoch": 0.04317548746518106,
      "grad_norm": 2.200751543045044,
      "learning_rate": 1.5069444444444444e-05,
      "loss": 9.4559,
      "step": 31
    },
    {
      "epoch": 0.04456824512534819,
      "grad_norm": 2.4295594692230225,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 9.5226,
      "step": 32
    },
    {
      "epoch": 0.04596100278551532,
      "grad_norm": 2.3435890674591064,
      "learning_rate": 1.6041666666666666e-05,
      "loss": 9.4528,
      "step": 33
    },
    {
      "epoch": 0.04735376044568245,
      "grad_norm": 3.0860228538513184,
      "learning_rate": 1.6527777777777777e-05,
      "loss": 9.4985,
      "step": 34
    },
    {
      "epoch": 0.04874651810584958,
      "grad_norm": 2.2005224227905273,
      "learning_rate": 1.7013888888888888e-05,
      "loss": 9.4564,
      "step": 35
    },
    {
      "epoch": 0.05013927576601671,
      "grad_norm": 2.600155830383301,
      "learning_rate": 1.75e-05,
      "loss": 9.4709,
      "step": 36
    },
    {
      "epoch": 0.05153203342618384,
      "grad_norm": 2.273688793182373,
      "learning_rate": 1.798611111111111e-05,
      "loss": 9.5097,
      "step": 37
    },
    {
      "epoch": 0.052924791086350974,
      "grad_norm": 2.3479037284851074,
      "learning_rate": 1.847222222222222e-05,
      "loss": 9.4957,
      "step": 38
    },
    {
      "epoch": 0.054317548746518104,
      "grad_norm": 2.629591464996338,
      "learning_rate": 1.895833333333333e-05,
      "loss": 9.5332,
      "step": 39
    },
    {
      "epoch": 0.055710306406685235,
      "grad_norm": 2.1469650268554688,
      "learning_rate": 1.9444444444444442e-05,
      "loss": 9.4229,
      "step": 40
    },
    {
      "epoch": 0.057103064066852366,
      "grad_norm": 2.278468132019043,
      "learning_rate": 1.9930555555555553e-05,
      "loss": 9.4944,
      "step": 41
    },
    {
      "epoch": 0.0584958217270195,
      "grad_norm": 1.952836275100708,
      "learning_rate": 2.0416666666666667e-05,
      "loss": 9.4444,
      "step": 42
    },
    {
      "epoch": 0.05988857938718663,
      "grad_norm": 2.0497686862945557,
      "learning_rate": 2.0902777777777775e-05,
      "loss": 9.4685,
      "step": 43
    },
    {
      "epoch": 0.06128133704735376,
      "grad_norm": 2.3993453979492188,
      "learning_rate": 2.138888888888889e-05,
      "loss": 9.4273,
      "step": 44
    },
    {
      "epoch": 0.06267409470752089,
      "grad_norm": 2.091829299926758,
      "learning_rate": 2.1874999999999996e-05,
      "loss": 9.4461,
      "step": 45
    },
    {
      "epoch": 0.06406685236768803,
      "grad_norm": 2.043539047241211,
      "learning_rate": 2.2361111111111107e-05,
      "loss": 9.4678,
      "step": 46
    },
    {
      "epoch": 0.06545961002785515,
      "grad_norm": 2.401531219482422,
      "learning_rate": 2.284722222222222e-05,
      "loss": 9.4719,
      "step": 47
    },
    {
      "epoch": 0.06685236768802229,
      "grad_norm": 1.9853050708770752,
      "learning_rate": 2.333333333333333e-05,
      "loss": 9.4406,
      "step": 48
    },
    {
      "epoch": 0.06824512534818941,
      "grad_norm": 1.8597462177276611,
      "learning_rate": 2.3819444444444443e-05,
      "loss": 9.4266,
      "step": 49
    },
    {
      "epoch": 0.06963788300835655,
      "grad_norm": 2.1002442836761475,
      "learning_rate": 2.4305555555555554e-05,
      "loss": 9.4377,
      "step": 50
    },
    {
      "epoch": 0.07103064066852367,
      "grad_norm": 2.0721702575683594,
      "learning_rate": 2.4791666666666665e-05,
      "loss": 9.378,
      "step": 51
    },
    {
      "epoch": 0.07242339832869081,
      "grad_norm": 2.1750292778015137,
      "learning_rate": 2.5277777777777776e-05,
      "loss": 9.4031,
      "step": 52
    },
    {
      "epoch": 0.07381615598885793,
      "grad_norm": 2.0567100048065186,
      "learning_rate": 2.5763888888888887e-05,
      "loss": 9.4175,
      "step": 53
    },
    {
      "epoch": 0.07520891364902507,
      "grad_norm": 1.9709560871124268,
      "learning_rate": 2.6249999999999998e-05,
      "loss": 9.4306,
      "step": 54
    },
    {
      "epoch": 0.0766016713091922,
      "grad_norm": 2.089951992034912,
      "learning_rate": 2.673611111111111e-05,
      "loss": 9.3999,
      "step": 55
    },
    {
      "epoch": 0.07799442896935933,
      "grad_norm": 2.057718515396118,
      "learning_rate": 2.722222222222222e-05,
      "loss": 9.4385,
      "step": 56
    },
    {
      "epoch": 0.07938718662952646,
      "grad_norm": 2.2145371437072754,
      "learning_rate": 2.770833333333333e-05,
      "loss": 9.4618,
      "step": 57
    },
    {
      "epoch": 0.0807799442896936,
      "grad_norm": 2.0873260498046875,
      "learning_rate": 2.819444444444444e-05,
      "loss": 9.4036,
      "step": 58
    },
    {
      "epoch": 0.08217270194986072,
      "grad_norm": 2.0868418216705322,
      "learning_rate": 2.8680555555555552e-05,
      "loss": 9.4577,
      "step": 59
    },
    {
      "epoch": 0.08356545961002786,
      "grad_norm": 2.0237319469451904,
      "learning_rate": 2.9166666666666666e-05,
      "loss": 9.4307,
      "step": 60
    },
    {
      "epoch": 0.08495821727019498,
      "grad_norm": 1.981904149055481,
      "learning_rate": 2.9652777777777774e-05,
      "loss": 9.3878,
      "step": 61
    },
    {
      "epoch": 0.08635097493036212,
      "grad_norm": 1.8592807054519653,
      "learning_rate": 3.0138888888888888e-05,
      "loss": 9.4487,
      "step": 62
    },
    {
      "epoch": 0.08774373259052924,
      "grad_norm": 2.1793065071105957,
      "learning_rate": 3.0625e-05,
      "loss": 9.4536,
      "step": 63
    },
    {
      "epoch": 0.08913649025069638,
      "grad_norm": 1.99411141872406,
      "learning_rate": 3.111111111111111e-05,
      "loss": 9.4251,
      "step": 64
    },
    {
      "epoch": 0.0905292479108635,
      "grad_norm": 2.7959253787994385,
      "learning_rate": 3.159722222222222e-05,
      "loss": 9.4595,
      "step": 65
    },
    {
      "epoch": 0.09192200557103064,
      "grad_norm": 1.9733268022537231,
      "learning_rate": 3.208333333333333e-05,
      "loss": 9.448,
      "step": 66
    },
    {
      "epoch": 0.09331476323119778,
      "grad_norm": 1.9736051559448242,
      "learning_rate": 3.256944444444444e-05,
      "loss": 9.421,
      "step": 67
    },
    {
      "epoch": 0.0947075208913649,
      "grad_norm": 2.1202316284179688,
      "learning_rate": 3.3055555555555553e-05,
      "loss": 9.4151,
      "step": 68
    },
    {
      "epoch": 0.09610027855153204,
      "grad_norm": 1.9673542976379395,
      "learning_rate": 3.3541666666666664e-05,
      "loss": 9.433,
      "step": 69
    },
    {
      "epoch": 0.09749303621169916,
      "grad_norm": 2.3342862129211426,
      "learning_rate": 3.4027777777777775e-05,
      "loss": 9.4349,
      "step": 70
    },
    {
      "epoch": 0.0988857938718663,
      "grad_norm": 2.252122402191162,
      "learning_rate": 3.4513888888888886e-05,
      "loss": 9.3299,
      "step": 71
    },
    {
      "epoch": 0.10027855153203342,
      "grad_norm": 1.9738399982452393,
      "learning_rate": 3.5e-05,
      "loss": 9.3686,
      "step": 72
    },
    {
      "epoch": 0.10167130919220056,
      "grad_norm": 1.9239985942840576,
      "learning_rate": 3.548611111111111e-05,
      "loss": 9.3583,
      "step": 73
    },
    {
      "epoch": 0.10306406685236769,
      "grad_norm": 2.0035247802734375,
      "learning_rate": 3.597222222222222e-05,
      "loss": 9.3615,
      "step": 74
    },
    {
      "epoch": 0.10445682451253482,
      "grad_norm": 2.1391425132751465,
      "learning_rate": 3.645833333333333e-05,
      "loss": 9.3704,
      "step": 75
    },
    {
      "epoch": 0.10584958217270195,
      "grad_norm": 2.5549750328063965,
      "learning_rate": 3.694444444444444e-05,
      "loss": 9.3467,
      "step": 76
    },
    {
      "epoch": 0.10724233983286909,
      "grad_norm": 3.201791763305664,
      "learning_rate": 3.743055555555555e-05,
      "loss": 9.3759,
      "step": 77
    },
    {
      "epoch": 0.10863509749303621,
      "grad_norm": 2.117368698120117,
      "learning_rate": 3.791666666666666e-05,
      "loss": 9.4053,
      "step": 78
    },
    {
      "epoch": 0.11002785515320335,
      "grad_norm": 2.5113062858581543,
      "learning_rate": 3.840277777777778e-05,
      "loss": 9.3376,
      "step": 79
    },
    {
      "epoch": 0.11142061281337047,
      "grad_norm": 2.192519187927246,
      "learning_rate": 3.8888888888888884e-05,
      "loss": 9.417,
      "step": 80
    },
    {
      "epoch": 0.11281337047353761,
      "grad_norm": 2.4800620079040527,
      "learning_rate": 3.9374999999999995e-05,
      "loss": 9.3578,
      "step": 81
    },
    {
      "epoch": 0.11420612813370473,
      "grad_norm": 2.34975004196167,
      "learning_rate": 3.9861111111111106e-05,
      "loss": 9.3705,
      "step": 82
    },
    {
      "epoch": 0.11559888579387187,
      "grad_norm": 2.3390395641326904,
      "learning_rate": 4.034722222222222e-05,
      "loss": 9.336,
      "step": 83
    },
    {
      "epoch": 0.116991643454039,
      "grad_norm": 2.483818531036377,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 9.3954,
      "step": 84
    },
    {
      "epoch": 0.11838440111420613,
      "grad_norm": 2.2996625900268555,
      "learning_rate": 4.131944444444444e-05,
      "loss": 9.3094,
      "step": 85
    },
    {
      "epoch": 0.11977715877437325,
      "grad_norm": 2.236759901046753,
      "learning_rate": 4.180555555555555e-05,
      "loss": 9.3359,
      "step": 86
    },
    {
      "epoch": 0.12116991643454039,
      "grad_norm": 2.3694021701812744,
      "learning_rate": 4.229166666666666e-05,
      "loss": 9.3774,
      "step": 87
    },
    {
      "epoch": 0.12256267409470752,
      "grad_norm": 2.4825735092163086,
      "learning_rate": 4.277777777777778e-05,
      "loss": 9.2735,
      "step": 88
    },
    {
      "epoch": 0.12395543175487465,
      "grad_norm": 2.1388025283813477,
      "learning_rate": 4.326388888888889e-05,
      "loss": 9.3976,
      "step": 89
    },
    {
      "epoch": 0.12534818941504178,
      "grad_norm": 2.3984382152557373,
      "learning_rate": 4.374999999999999e-05,
      "loss": 9.3027,
      "step": 90
    },
    {
      "epoch": 0.12674094707520892,
      "grad_norm": 2.3104631900787354,
      "learning_rate": 4.4236111111111104e-05,
      "loss": 9.2651,
      "step": 91
    },
    {
      "epoch": 0.12813370473537605,
      "grad_norm": 2.2130093574523926,
      "learning_rate": 4.4722222222222215e-05,
      "loss": 9.2949,
      "step": 92
    },
    {
      "epoch": 0.12952646239554316,
      "grad_norm": 2.3408825397491455,
      "learning_rate": 4.520833333333333e-05,
      "loss": 9.4334,
      "step": 93
    },
    {
      "epoch": 0.1309192200557103,
      "grad_norm": 2.701294183731079,
      "learning_rate": 4.569444444444444e-05,
      "loss": 9.2697,
      "step": 94
    },
    {
      "epoch": 0.13231197771587744,
      "grad_norm": 2.449645757675171,
      "learning_rate": 4.6180555555555554e-05,
      "loss": 9.2557,
      "step": 95
    },
    {
      "epoch": 0.13370473537604458,
      "grad_norm": 2.396810531616211,
      "learning_rate": 4.666666666666666e-05,
      "loss": 9.2513,
      "step": 96
    },
    {
      "epoch": 0.13509749303621169,
      "grad_norm": 2.6208698749542236,
      "learning_rate": 4.7152777777777776e-05,
      "loss": 9.3046,
      "step": 97
    },
    {
      "epoch": 0.13649025069637882,
      "grad_norm": 2.6074302196502686,
      "learning_rate": 4.7638888888888887e-05,
      "loss": 9.34,
      "step": 98
    },
    {
      "epoch": 0.13788300835654596,
      "grad_norm": 2.6555991172790527,
      "learning_rate": 4.8125e-05,
      "loss": 9.3138,
      "step": 99
    },
    {
      "epoch": 0.1392757660167131,
      "grad_norm": 2.6097066402435303,
      "learning_rate": 4.861111111111111e-05,
      "loss": 9.3208,
      "step": 100
    },
    {
      "epoch": 0.14066852367688024,
      "grad_norm": 2.5649614334106445,
      "learning_rate": 4.909722222222221e-05,
      "loss": 9.2206,
      "step": 101
    },
    {
      "epoch": 0.14206128133704735,
      "grad_norm": 2.6494956016540527,
      "learning_rate": 4.958333333333333e-05,
      "loss": 9.3438,
      "step": 102
    },
    {
      "epoch": 0.14345403899721448,
      "grad_norm": 2.6111457347869873,
      "learning_rate": 5.006944444444444e-05,
      "loss": 9.2774,
      "step": 103
    },
    {
      "epoch": 0.14484679665738162,
      "grad_norm": 2.6649367809295654,
      "learning_rate": 5.055555555555555e-05,
      "loss": 9.2373,
      "step": 104
    },
    {
      "epoch": 0.14623955431754876,
      "grad_norm": 2.546994924545288,
      "learning_rate": 5.104166666666666e-05,
      "loss": 9.2823,
      "step": 105
    },
    {
      "epoch": 0.14763231197771587,
      "grad_norm": 2.5940351486206055,
      "learning_rate": 5.1527777777777774e-05,
      "loss": 9.2889,
      "step": 106
    },
    {
      "epoch": 0.149025069637883,
      "grad_norm": 2.6945221424102783,
      "learning_rate": 5.2013888888888885e-05,
      "loss": 9.1161,
      "step": 107
    },
    {
      "epoch": 0.15041782729805014,
      "grad_norm": 3.0894527435302734,
      "learning_rate": 5.2499999999999995e-05,
      "loss": 9.1788,
      "step": 108
    },
    {
      "epoch": 0.15181058495821728,
      "grad_norm": 2.867466926574707,
      "learning_rate": 5.2986111111111106e-05,
      "loss": 9.1579,
      "step": 109
    },
    {
      "epoch": 0.1532033426183844,
      "grad_norm": 2.731473445892334,
      "learning_rate": 5.347222222222222e-05,
      "loss": 9.264,
      "step": 110
    },
    {
      "epoch": 0.15459610027855153,
      "grad_norm": 2.9084222316741943,
      "learning_rate": 5.395833333333333e-05,
      "loss": 9.3098,
      "step": 111
    },
    {
      "epoch": 0.15598885793871867,
      "grad_norm": 2.7787702083587646,
      "learning_rate": 5.444444444444444e-05,
      "loss": 9.3831,
      "step": 112
    },
    {
      "epoch": 0.1573816155988858,
      "grad_norm": 3.490419387817383,
      "learning_rate": 5.493055555555555e-05,
      "loss": 9.2544,
      "step": 113
    },
    {
      "epoch": 0.15877437325905291,
      "grad_norm": 3.0516364574432373,
      "learning_rate": 5.541666666666666e-05,
      "loss": 9.2826,
      "step": 114
    },
    {
      "epoch": 0.16016713091922005,
      "grad_norm": 3.0757088661193848,
      "learning_rate": 5.590277777777778e-05,
      "loss": 9.221,
      "step": 115
    },
    {
      "epoch": 0.1615598885793872,
      "grad_norm": 3.052006244659424,
      "learning_rate": 5.638888888888888e-05,
      "loss": 9.0494,
      "step": 116
    },
    {
      "epoch": 0.16295264623955433,
      "grad_norm": 2.6261179447174072,
      "learning_rate": 5.687499999999999e-05,
      "loss": 9.2244,
      "step": 117
    },
    {
      "epoch": 0.16434540389972144,
      "grad_norm": 2.775188684463501,
      "learning_rate": 5.7361111111111104e-05,
      "loss": 9.2703,
      "step": 118
    },
    {
      "epoch": 0.16573816155988857,
      "grad_norm": 2.6947615146636963,
      "learning_rate": 5.7847222222222215e-05,
      "loss": 9.1994,
      "step": 119
    },
    {
      "epoch": 0.1671309192200557,
      "grad_norm": 3.2321982383728027,
      "learning_rate": 5.833333333333333e-05,
      "loss": 9.0563,
      "step": 120
    },
    {
      "epoch": 0.16852367688022285,
      "grad_norm": 2.8258464336395264,
      "learning_rate": 5.881944444444444e-05,
      "loss": 9.2179,
      "step": 121
    },
    {
      "epoch": 0.16991643454038996,
      "grad_norm": 3.2115259170532227,
      "learning_rate": 5.930555555555555e-05,
      "loss": 9.195,
      "step": 122
    },
    {
      "epoch": 0.1713091922005571,
      "grad_norm": 2.8792707920074463,
      "learning_rate": 5.979166666666666e-05,
      "loss": 9.02,
      "step": 123
    },
    {
      "epoch": 0.17270194986072424,
      "grad_norm": 2.866339683532715,
      "learning_rate": 6.0277777777777776e-05,
      "loss": 9.1544,
      "step": 124
    },
    {
      "epoch": 0.17409470752089137,
      "grad_norm": 3.1229407787323,
      "learning_rate": 6.076388888888889e-05,
      "loss": 9.1192,
      "step": 125
    },
    {
      "epoch": 0.17548746518105848,
      "grad_norm": 3.0198867321014404,
      "learning_rate": 6.125e-05,
      "loss": 9.0572,
      "step": 126
    },
    {
      "epoch": 0.17688022284122562,
      "grad_norm": 3.060091018676758,
      "learning_rate": 6.173611111111111e-05,
      "loss": 9.1694,
      "step": 127
    },
    {
      "epoch": 0.17827298050139276,
      "grad_norm": 3.0378992557525635,
      "learning_rate": 6.222222222222222e-05,
      "loss": 9.0026,
      "step": 128
    },
    {
      "epoch": 0.1796657381615599,
      "grad_norm": 3.2583515644073486,
      "learning_rate": 6.270833333333333e-05,
      "loss": 9.0214,
      "step": 129
    },
    {
      "epoch": 0.181058495821727,
      "grad_norm": 3.5013654232025146,
      "learning_rate": 6.319444444444444e-05,
      "loss": 9.1353,
      "step": 130
    },
    {
      "epoch": 0.18245125348189414,
      "grad_norm": 3.2136590480804443,
      "learning_rate": 6.368055555555555e-05,
      "loss": 9.0555,
      "step": 131
    },
    {
      "epoch": 0.18384401114206128,
      "grad_norm": 3.704627513885498,
      "learning_rate": 6.416666666666666e-05,
      "loss": 9.0568,
      "step": 132
    },
    {
      "epoch": 0.18523676880222842,
      "grad_norm": 3.6550254821777344,
      "learning_rate": 6.465277777777777e-05,
      "loss": 9.0778,
      "step": 133
    },
    {
      "epoch": 0.18662952646239556,
      "grad_norm": 3.029344320297241,
      "learning_rate": 6.513888888888889e-05,
      "loss": 9.0669,
      "step": 134
    },
    {
      "epoch": 0.18802228412256267,
      "grad_norm": 3.116588592529297,
      "learning_rate": 6.5625e-05,
      "loss": 9.0169,
      "step": 135
    },
    {
      "epoch": 0.1894150417827298,
      "grad_norm": 2.9819633960723877,
      "learning_rate": 6.611111111111111e-05,
      "loss": 9.1648,
      "step": 136
    },
    {
      "epoch": 0.19080779944289694,
      "grad_norm": 3.1296842098236084,
      "learning_rate": 6.659722222222222e-05,
      "loss": 9.0234,
      "step": 137
    },
    {
      "epoch": 0.19220055710306408,
      "grad_norm": 3.7833495140075684,
      "learning_rate": 6.708333333333333e-05,
      "loss": 9.1396,
      "step": 138
    },
    {
      "epoch": 0.1935933147632312,
      "grad_norm": 3.1753551959991455,
      "learning_rate": 6.756944444444444e-05,
      "loss": 9.0495,
      "step": 139
    },
    {
      "epoch": 0.19498607242339833,
      "grad_norm": 3.087599515914917,
      "learning_rate": 6.805555555555555e-05,
      "loss": 8.8913,
      "step": 140
    },
    {
      "epoch": 0.19637883008356546,
      "grad_norm": 3.865417957305908,
      "learning_rate": 6.854166666666666e-05,
      "loss": 9.2012,
      "step": 141
    },
    {
      "epoch": 0.1977715877437326,
      "grad_norm": 3.1938564777374268,
      "learning_rate": 6.902777777777777e-05,
      "loss": 9.1955,
      "step": 142
    },
    {
      "epoch": 0.1991643454038997,
      "grad_norm": 3.7016489505767822,
      "learning_rate": 6.951388888888888e-05,
      "loss": 8.9795,
      "step": 143
    },
    {
      "epoch": 0.20055710306406685,
      "grad_norm": 3.168267250061035,
      "learning_rate": 7e-05,
      "loss": 9.0605,
      "step": 144
    },
    {
      "epoch": 0.201949860724234,
      "grad_norm": 3.670396566390991,
      "learning_rate": 6.99900511654349e-05,
      "loss": 9.1897,
      "step": 145
    },
    {
      "epoch": 0.20334261838440112,
      "grad_norm": 3.8943371772766113,
      "learning_rate": 6.998010233086982e-05,
      "loss": 8.989,
      "step": 146
    },
    {
      "epoch": 0.20473537604456823,
      "grad_norm": 3.5359532833099365,
      "learning_rate": 6.99701534963047e-05,
      "loss": 8.9466,
      "step": 147
    },
    {
      "epoch": 0.20612813370473537,
      "grad_norm": 3.0849764347076416,
      "learning_rate": 6.996020466173962e-05,
      "loss": 9.0618,
      "step": 148
    },
    {
      "epoch": 0.2075208913649025,
      "grad_norm": 3.2177481651306152,
      "learning_rate": 6.995025582717453e-05,
      "loss": 9.0375,
      "step": 149
    },
    {
      "epoch": 0.20891364902506965,
      "grad_norm": 3.689551830291748,
      "learning_rate": 6.994030699260943e-05,
      "loss": 8.9582,
      "step": 150
    },
    {
      "epoch": 0.21030640668523676,
      "grad_norm": 3.431204319000244,
      "learning_rate": 6.993035815804434e-05,
      "loss": 9.0407,
      "step": 151
    },
    {
      "epoch": 0.2116991643454039,
      "grad_norm": 3.074554443359375,
      "learning_rate": 6.992040932347924e-05,
      "loss": 8.869,
      "step": 152
    },
    {
      "epoch": 0.21309192200557103,
      "grad_norm": 3.3490922451019287,
      "learning_rate": 6.991046048891414e-05,
      "loss": 9.0683,
      "step": 153
    },
    {
      "epoch": 0.21448467966573817,
      "grad_norm": 3.005607843399048,
      "learning_rate": 6.990051165434906e-05,
      "loss": 9.1739,
      "step": 154
    },
    {
      "epoch": 0.21587743732590528,
      "grad_norm": 3.483210563659668,
      "learning_rate": 6.989056281978397e-05,
      "loss": 9.0874,
      "step": 155
    },
    {
      "epoch": 0.21727019498607242,
      "grad_norm": 3.241920232772827,
      "learning_rate": 6.988061398521887e-05,
      "loss": 9.0501,
      "step": 156
    },
    {
      "epoch": 0.21866295264623956,
      "grad_norm": 3.062758207321167,
      "learning_rate": 6.987066515065377e-05,
      "loss": 9.1537,
      "step": 157
    },
    {
      "epoch": 0.2200557103064067,
      "grad_norm": 3.220777988433838,
      "learning_rate": 6.986071631608868e-05,
      "loss": 8.8843,
      "step": 158
    },
    {
      "epoch": 0.2214484679665738,
      "grad_norm": 3.283903121948242,
      "learning_rate": 6.985076748152358e-05,
      "loss": 8.9421,
      "step": 159
    },
    {
      "epoch": 0.22284122562674094,
      "grad_norm": 3.5407607555389404,
      "learning_rate": 6.98408186469585e-05,
      "loss": 9.0808,
      "step": 160
    },
    {
      "epoch": 0.22423398328690808,
      "grad_norm": 3.2570924758911133,
      "learning_rate": 6.98308698123934e-05,
      "loss": 8.792,
      "step": 161
    },
    {
      "epoch": 0.22562674094707522,
      "grad_norm": 3.170731544494629,
      "learning_rate": 6.982092097782831e-05,
      "loss": 9.017,
      "step": 162
    },
    {
      "epoch": 0.22701949860724233,
      "grad_norm": 3.8973240852355957,
      "learning_rate": 6.981097214326321e-05,
      "loss": 8.9749,
      "step": 163
    },
    {
      "epoch": 0.22841225626740946,
      "grad_norm": 4.280517101287842,
      "learning_rate": 6.980102330869812e-05,
      "loss": 8.9395,
      "step": 164
    },
    {
      "epoch": 0.2298050139275766,
      "grad_norm": 3.2216036319732666,
      "learning_rate": 6.979107447413302e-05,
      "loss": 8.9109,
      "step": 165
    },
    {
      "epoch": 0.23119777158774374,
      "grad_norm": 3.2336678504943848,
      "learning_rate": 6.978112563956792e-05,
      "loss": 9.0659,
      "step": 166
    },
    {
      "epoch": 0.23259052924791088,
      "grad_norm": 3.7737042903900146,
      "learning_rate": 6.977117680500283e-05,
      "loss": 8.9384,
      "step": 167
    },
    {
      "epoch": 0.233983286908078,
      "grad_norm": 3.4278862476348877,
      "learning_rate": 6.976122797043775e-05,
      "loss": 8.9576,
      "step": 168
    },
    {
      "epoch": 0.23537604456824512,
      "grad_norm": 3.465973377227783,
      "learning_rate": 6.975127913587265e-05,
      "loss": 8.9898,
      "step": 169
    },
    {
      "epoch": 0.23676880222841226,
      "grad_norm": 5.112400531768799,
      "learning_rate": 6.974133030130755e-05,
      "loss": 8.7912,
      "step": 170
    },
    {
      "epoch": 0.2381615598885794,
      "grad_norm": 4.257119178771973,
      "learning_rate": 6.973138146674246e-05,
      "loss": 8.899,
      "step": 171
    },
    {
      "epoch": 0.2395543175487465,
      "grad_norm": 4.034311771392822,
      "learning_rate": 6.972143263217736e-05,
      "loss": 8.8859,
      "step": 172
    },
    {
      "epoch": 0.24094707520891365,
      "grad_norm": 3.9154863357543945,
      "learning_rate": 6.971148379761228e-05,
      "loss": 9.074,
      "step": 173
    },
    {
      "epoch": 0.24233983286908078,
      "grad_norm": 4.049985885620117,
      "learning_rate": 6.970153496304718e-05,
      "loss": 8.9757,
      "step": 174
    },
    {
      "epoch": 0.24373259052924792,
      "grad_norm": 3.8165013790130615,
      "learning_rate": 6.969158612848209e-05,
      "loss": 8.9547,
      "step": 175
    },
    {
      "epoch": 0.24512534818941503,
      "grad_norm": 4.293961048126221,
      "learning_rate": 6.968163729391699e-05,
      "loss": 8.8375,
      "step": 176
    },
    {
      "epoch": 0.24651810584958217,
      "grad_norm": 5.501155376434326,
      "learning_rate": 6.96716884593519e-05,
      "loss": 8.9973,
      "step": 177
    },
    {
      "epoch": 0.2479108635097493,
      "grad_norm": 3.8478055000305176,
      "learning_rate": 6.96617396247868e-05,
      "loss": 8.8513,
      "step": 178
    },
    {
      "epoch": 0.24930362116991645,
      "grad_norm": 3.494331121444702,
      "learning_rate": 6.965179079022172e-05,
      "loss": 9.106,
      "step": 179
    },
    {
      "epoch": 0.25069637883008355,
      "grad_norm": 3.3035213947296143,
      "learning_rate": 6.964184195565662e-05,
      "loss": 8.8163,
      "step": 180
    },
    {
      "epoch": 0.2520891364902507,
      "grad_norm": 3.44191575050354,
      "learning_rate": 6.963189312109153e-05,
      "loss": 9.0062,
      "step": 181
    },
    {
      "epoch": 0.25348189415041783,
      "grad_norm": 4.08702278137207,
      "learning_rate": 6.962194428652643e-05,
      "loss": 8.8288,
      "step": 182
    },
    {
      "epoch": 0.25487465181058494,
      "grad_norm": 3.422994375228882,
      "learning_rate": 6.961199545196133e-05,
      "loss": 8.7787,
      "step": 183
    },
    {
      "epoch": 0.2562674094707521,
      "grad_norm": 3.4032821655273438,
      "learning_rate": 6.960204661739624e-05,
      "loss": 8.876,
      "step": 184
    },
    {
      "epoch": 0.2576601671309192,
      "grad_norm": 3.357682228088379,
      "learning_rate": 6.959209778283116e-05,
      "loss": 9.0388,
      "step": 185
    },
    {
      "epoch": 0.2590529247910863,
      "grad_norm": 3.4004476070404053,
      "learning_rate": 6.958214894826605e-05,
      "loss": 9.0445,
      "step": 186
    },
    {
      "epoch": 0.2604456824512535,
      "grad_norm": 3.507704734802246,
      "learning_rate": 6.957220011370096e-05,
      "loss": 8.7071,
      "step": 187
    },
    {
      "epoch": 0.2618384401114206,
      "grad_norm": 3.7336301803588867,
      "learning_rate": 6.956225127913587e-05,
      "loss": 9.0218,
      "step": 188
    },
    {
      "epoch": 0.26323119777158777,
      "grad_norm": 3.0426900386810303,
      "learning_rate": 6.955230244457077e-05,
      "loss": 8.9227,
      "step": 189
    },
    {
      "epoch": 0.2646239554317549,
      "grad_norm": 3.8145570755004883,
      "learning_rate": 6.954235361000568e-05,
      "loss": 9.0258,
      "step": 190
    },
    {
      "epoch": 0.266016713091922,
      "grad_norm": 3.6831650733947754,
      "learning_rate": 6.953240477544058e-05,
      "loss": 9.0218,
      "step": 191
    },
    {
      "epoch": 0.26740947075208915,
      "grad_norm": 3.8541195392608643,
      "learning_rate": 6.952245594087548e-05,
      "loss": 8.9664,
      "step": 192
    },
    {
      "epoch": 0.26880222841225626,
      "grad_norm": 4.092911720275879,
      "learning_rate": 6.95125071063104e-05,
      "loss": 8.8958,
      "step": 193
    },
    {
      "epoch": 0.27019498607242337,
      "grad_norm": 3.52112078666687,
      "learning_rate": 6.95025582717453e-05,
      "loss": 8.7399,
      "step": 194
    },
    {
      "epoch": 0.27158774373259054,
      "grad_norm": 3.9229259490966797,
      "learning_rate": 6.949260943718021e-05,
      "loss": 8.503,
      "step": 195
    },
    {
      "epoch": 0.27298050139275765,
      "grad_norm": 4.027508735656738,
      "learning_rate": 6.948266060261511e-05,
      "loss": 8.8499,
      "step": 196
    },
    {
      "epoch": 0.2743732590529248,
      "grad_norm": 3.772533655166626,
      "learning_rate": 6.947271176805002e-05,
      "loss": 8.8647,
      "step": 197
    },
    {
      "epoch": 0.2757660167130919,
      "grad_norm": 3.626054048538208,
      "learning_rate": 6.946276293348492e-05,
      "loss": 8.7417,
      "step": 198
    },
    {
      "epoch": 0.27715877437325903,
      "grad_norm": 3.7706661224365234,
      "learning_rate": 6.945281409891984e-05,
      "loss": 8.675,
      "step": 199
    },
    {
      "epoch": 0.2785515320334262,
      "grad_norm": 3.55424165725708,
      "learning_rate": 6.944286526435474e-05,
      "loss": 8.7761,
      "step": 200
    },
    {
      "epoch": 0.2799442896935933,
      "grad_norm": 4.005819320678711,
      "learning_rate": 6.943291642978965e-05,
      "loss": 8.7846,
      "step": 201
    },
    {
      "epoch": 0.28133704735376047,
      "grad_norm": 3.900658369064331,
      "learning_rate": 6.942296759522455e-05,
      "loss": 8.7952,
      "step": 202
    },
    {
      "epoch": 0.2827298050139276,
      "grad_norm": 3.804995059967041,
      "learning_rate": 6.941301876065946e-05,
      "loss": 8.7566,
      "step": 203
    },
    {
      "epoch": 0.2841225626740947,
      "grad_norm": 4.131643295288086,
      "learning_rate": 6.940306992609437e-05,
      "loss": 8.8913,
      "step": 204
    },
    {
      "epoch": 0.28551532033426186,
      "grad_norm": 3.9271464347839355,
      "learning_rate": 6.939312109152928e-05,
      "loss": 8.69,
      "step": 205
    },
    {
      "epoch": 0.28690807799442897,
      "grad_norm": 5.086146354675293,
      "learning_rate": 6.938317225696418e-05,
      "loss": 8.7908,
      "step": 206
    },
    {
      "epoch": 0.2883008356545961,
      "grad_norm": 4.286420822143555,
      "learning_rate": 6.937322342239909e-05,
      "loss": 8.4524,
      "step": 207
    },
    {
      "epoch": 0.28969359331476324,
      "grad_norm": 3.5584073066711426,
      "learning_rate": 6.936327458783399e-05,
      "loss": 8.9401,
      "step": 208
    },
    {
      "epoch": 0.29108635097493035,
      "grad_norm": 3.8784470558166504,
      "learning_rate": 6.93533257532689e-05,
      "loss": 8.9656,
      "step": 209
    },
    {
      "epoch": 0.2924791086350975,
      "grad_norm": 3.98789644241333,
      "learning_rate": 6.93433769187038e-05,
      "loss": 8.825,
      "step": 210
    },
    {
      "epoch": 0.2938718662952646,
      "grad_norm": 4.665347576141357,
      "learning_rate": 6.93334280841387e-05,
      "loss": 8.8567,
      "step": 211
    },
    {
      "epoch": 0.29526462395543174,
      "grad_norm": 4.211238861083984,
      "learning_rate": 6.932347924957362e-05,
      "loss": 8.5163,
      "step": 212
    },
    {
      "epoch": 0.2966573816155989,
      "grad_norm": 4.115207195281982,
      "learning_rate": 6.931353041500852e-05,
      "loss": 8.8728,
      "step": 213
    },
    {
      "epoch": 0.298050139275766,
      "grad_norm": 4.007952690124512,
      "learning_rate": 6.930358158044343e-05,
      "loss": 8.8066,
      "step": 214
    },
    {
      "epoch": 0.2994428969359331,
      "grad_norm": 4.000673770904541,
      "learning_rate": 6.929363274587833e-05,
      "loss": 8.8258,
      "step": 215
    },
    {
      "epoch": 0.3008356545961003,
      "grad_norm": 3.824923038482666,
      "learning_rate": 6.928368391131324e-05,
      "loss": 8.6051,
      "step": 216
    },
    {
      "epoch": 0.3022284122562674,
      "grad_norm": 4.339374542236328,
      "learning_rate": 6.927373507674814e-05,
      "loss": 8.8224,
      "step": 217
    },
    {
      "epoch": 0.30362116991643456,
      "grad_norm": 4.101476669311523,
      "learning_rate": 6.926378624218306e-05,
      "loss": 8.4742,
      "step": 218
    },
    {
      "epoch": 0.3050139275766017,
      "grad_norm": 3.2088582515716553,
      "learning_rate": 6.925383740761796e-05,
      "loss": 9.0615,
      "step": 219
    },
    {
      "epoch": 0.3064066852367688,
      "grad_norm": 3.446979522705078,
      "learning_rate": 6.924388857305287e-05,
      "loss": 8.8705,
      "step": 220
    },
    {
      "epoch": 0.30779944289693595,
      "grad_norm": 3.8296539783477783,
      "learning_rate": 6.923393973848777e-05,
      "loss": 8.6779,
      "step": 221
    },
    {
      "epoch": 0.30919220055710306,
      "grad_norm": 3.905626058578491,
      "learning_rate": 6.922399090392267e-05,
      "loss": 8.5484,
      "step": 222
    },
    {
      "epoch": 0.31058495821727017,
      "grad_norm": 3.4928548336029053,
      "learning_rate": 6.921404206935758e-05,
      "loss": 8.7129,
      "step": 223
    },
    {
      "epoch": 0.31197771587743733,
      "grad_norm": 3.5804247856140137,
      "learning_rate": 6.92040932347925e-05,
      "loss": 8.571,
      "step": 224
    },
    {
      "epoch": 0.31337047353760444,
      "grad_norm": 4.050597667694092,
      "learning_rate": 6.919414440022739e-05,
      "loss": 8.6514,
      "step": 225
    },
    {
      "epoch": 0.3147632311977716,
      "grad_norm": 4.444650650024414,
      "learning_rate": 6.91841955656623e-05,
      "loss": 8.5167,
      "step": 226
    },
    {
      "epoch": 0.3161559888579387,
      "grad_norm": 4.613363265991211,
      "learning_rate": 6.917424673109721e-05,
      "loss": 8.7485,
      "step": 227
    },
    {
      "epoch": 0.31754874651810583,
      "grad_norm": 3.857363700866699,
      "learning_rate": 6.916429789653211e-05,
      "loss": 8.4424,
      "step": 228
    },
    {
      "epoch": 0.318941504178273,
      "grad_norm": 3.773397922515869,
      "learning_rate": 6.915434906196703e-05,
      "loss": 8.3823,
      "step": 229
    },
    {
      "epoch": 0.3203342618384401,
      "grad_norm": 3.7881274223327637,
      "learning_rate": 6.914440022740192e-05,
      "loss": 8.5923,
      "step": 230
    },
    {
      "epoch": 0.32172701949860727,
      "grad_norm": 3.938821792602539,
      "learning_rate": 6.913445139283684e-05,
      "loss": 8.6306,
      "step": 231
    },
    {
      "epoch": 0.3231197771587744,
      "grad_norm": 4.051789283752441,
      "learning_rate": 6.912450255827174e-05,
      "loss": 8.7096,
      "step": 232
    },
    {
      "epoch": 0.3245125348189415,
      "grad_norm": 3.930919647216797,
      "learning_rate": 6.911455372370665e-05,
      "loss": 8.8334,
      "step": 233
    },
    {
      "epoch": 0.32590529247910865,
      "grad_norm": 5.816878318786621,
      "learning_rate": 6.910460488914155e-05,
      "loss": 8.6405,
      "step": 234
    },
    {
      "epoch": 0.32729805013927576,
      "grad_norm": 4.822601318359375,
      "learning_rate": 6.909465605457645e-05,
      "loss": 8.7494,
      "step": 235
    },
    {
      "epoch": 0.3286908077994429,
      "grad_norm": 4.636408805847168,
      "learning_rate": 6.908470722001136e-05,
      "loss": 8.7079,
      "step": 236
    },
    {
      "epoch": 0.33008356545961004,
      "grad_norm": 3.6052465438842773,
      "learning_rate": 6.907475838544628e-05,
      "loss": 8.3915,
      "step": 237
    },
    {
      "epoch": 0.33147632311977715,
      "grad_norm": 3.939838409423828,
      "learning_rate": 6.906480955088118e-05,
      "loss": 8.7672,
      "step": 238
    },
    {
      "epoch": 0.3328690807799443,
      "grad_norm": 3.9982073307037354,
      "learning_rate": 6.905486071631608e-05,
      "loss": 8.3698,
      "step": 239
    },
    {
      "epoch": 0.3342618384401114,
      "grad_norm": 3.5854127407073975,
      "learning_rate": 6.904491188175099e-05,
      "loss": 8.6274,
      "step": 240
    },
    {
      "epoch": 0.33565459610027853,
      "grad_norm": 3.827282428741455,
      "learning_rate": 6.903496304718589e-05,
      "loss": 8.5594,
      "step": 241
    },
    {
      "epoch": 0.3370473537604457,
      "grad_norm": 3.9238662719726562,
      "learning_rate": 6.90250142126208e-05,
      "loss": 8.5622,
      "step": 242
    },
    {
      "epoch": 0.3384401114206128,
      "grad_norm": 4.059924602508545,
      "learning_rate": 6.901506537805571e-05,
      "loss": 8.631,
      "step": 243
    },
    {
      "epoch": 0.3398328690807799,
      "grad_norm": 3.882917642593384,
      "learning_rate": 6.900511654349062e-05,
      "loss": 8.5415,
      "step": 244
    },
    {
      "epoch": 0.3412256267409471,
      "grad_norm": 3.731807231903076,
      "learning_rate": 6.899516770892552e-05,
      "loss": 8.7059,
      "step": 245
    },
    {
      "epoch": 0.3426183844011142,
      "grad_norm": 4.963943958282471,
      "learning_rate": 6.898521887436043e-05,
      "loss": 8.2728,
      "step": 246
    },
    {
      "epoch": 0.34401114206128136,
      "grad_norm": 3.780099391937256,
      "learning_rate": 6.897527003979533e-05,
      "loss": 8.6473,
      "step": 247
    },
    {
      "epoch": 0.34540389972144847,
      "grad_norm": 3.502495527267456,
      "learning_rate": 6.896532120523023e-05,
      "loss": 8.64,
      "step": 248
    },
    {
      "epoch": 0.3467966573816156,
      "grad_norm": 4.078880310058594,
      "learning_rate": 6.895537237066515e-05,
      "loss": 8.6649,
      "step": 249
    },
    {
      "epoch": 0.34818941504178275,
      "grad_norm": 3.965594530105591,
      "learning_rate": 6.894542353610004e-05,
      "loss": 8.4793,
      "step": 250
    },
    {
      "epoch": 0.34958217270194986,
      "grad_norm": 3.7059476375579834,
      "learning_rate": 6.893547470153496e-05,
      "loss": 8.6176,
      "step": 251
    },
    {
      "epoch": 0.35097493036211697,
      "grad_norm": 3.8155102729797363,
      "learning_rate": 6.892552586696986e-05,
      "loss": 8.5968,
      "step": 252
    },
    {
      "epoch": 0.35236768802228413,
      "grad_norm": 3.9428980350494385,
      "learning_rate": 6.891557703240477e-05,
      "loss": 8.7197,
      "step": 253
    },
    {
      "epoch": 0.35376044568245124,
      "grad_norm": 3.9024670124053955,
      "learning_rate": 6.890562819783967e-05,
      "loss": 8.33,
      "step": 254
    },
    {
      "epoch": 0.3551532033426184,
      "grad_norm": 3.6394572257995605,
      "learning_rate": 6.889567936327458e-05,
      "loss": 8.4934,
      "step": 255
    },
    {
      "epoch": 0.3565459610027855,
      "grad_norm": 3.7295875549316406,
      "learning_rate": 6.888573052870948e-05,
      "loss": 8.1607,
      "step": 256
    },
    {
      "epoch": 0.3579387186629526,
      "grad_norm": 3.547926187515259,
      "learning_rate": 6.88757816941444e-05,
      "loss": 8.5152,
      "step": 257
    },
    {
      "epoch": 0.3593314763231198,
      "grad_norm": 4.115841865539551,
      "learning_rate": 6.88658328595793e-05,
      "loss": 8.3158,
      "step": 258
    },
    {
      "epoch": 0.3607242339832869,
      "grad_norm": 3.5963521003723145,
      "learning_rate": 6.88558840250142e-05,
      "loss": 8.0662,
      "step": 259
    },
    {
      "epoch": 0.362116991643454,
      "grad_norm": 3.4882774353027344,
      "learning_rate": 6.884593519044911e-05,
      "loss": 8.2915,
      "step": 260
    },
    {
      "epoch": 0.3635097493036212,
      "grad_norm": 4.000843524932861,
      "learning_rate": 6.883598635588401e-05,
      "loss": 8.459,
      "step": 261
    },
    {
      "epoch": 0.3649025069637883,
      "grad_norm": 3.437972068786621,
      "learning_rate": 6.882603752131893e-05,
      "loss": 8.6511,
      "step": 262
    },
    {
      "epoch": 0.36629526462395545,
      "grad_norm": 3.727342367172241,
      "learning_rate": 6.881608868675384e-05,
      "loss": 8.6024,
      "step": 263
    },
    {
      "epoch": 0.36768802228412256,
      "grad_norm": 3.84175968170166,
      "learning_rate": 6.880613985218874e-05,
      "loss": 8.5866,
      "step": 264
    },
    {
      "epoch": 0.36908077994428967,
      "grad_norm": 4.956315517425537,
      "learning_rate": 6.879619101762364e-05,
      "loss": 7.9471,
      "step": 265
    },
    {
      "epoch": 0.37047353760445684,
      "grad_norm": 4.100083351135254,
      "learning_rate": 6.878624218305855e-05,
      "loss": 8.5005,
      "step": 266
    },
    {
      "epoch": 0.37186629526462395,
      "grad_norm": 4.244174003601074,
      "learning_rate": 6.877629334849345e-05,
      "loss": 8.5043,
      "step": 267
    },
    {
      "epoch": 0.3732590529247911,
      "grad_norm": 4.182929039001465,
      "learning_rate": 6.876634451392837e-05,
      "loss": 8.6353,
      "step": 268
    },
    {
      "epoch": 0.3746518105849582,
      "grad_norm": 4.116519927978516,
      "learning_rate": 6.875639567936326e-05,
      "loss": 8.2668,
      "step": 269
    },
    {
      "epoch": 0.37604456824512533,
      "grad_norm": 4.627055644989014,
      "learning_rate": 6.874644684479818e-05,
      "loss": 8.3179,
      "step": 270
    },
    {
      "epoch": 0.3774373259052925,
      "grad_norm": 4.048768520355225,
      "learning_rate": 6.873649801023308e-05,
      "loss": 8.2831,
      "step": 271
    },
    {
      "epoch": 0.3788300835654596,
      "grad_norm": 4.0451340675354,
      "learning_rate": 6.872654917566799e-05,
      "loss": 8.4852,
      "step": 272
    },
    {
      "epoch": 0.3802228412256267,
      "grad_norm": 3.7257699966430664,
      "learning_rate": 6.871660034110289e-05,
      "loss": 8.0703,
      "step": 273
    },
    {
      "epoch": 0.3816155988857939,
      "grad_norm": 4.093203544616699,
      "learning_rate": 6.87066515065378e-05,
      "loss": 8.304,
      "step": 274
    },
    {
      "epoch": 0.383008356545961,
      "grad_norm": 3.7395145893096924,
      "learning_rate": 6.86967026719727e-05,
      "loss": 8.3956,
      "step": 275
    },
    {
      "epoch": 0.38440111420612816,
      "grad_norm": 4.134032726287842,
      "learning_rate": 6.868675383740762e-05,
      "loss": 8.1094,
      "step": 276
    },
    {
      "epoch": 0.38579387186629527,
      "grad_norm": 4.843389511108398,
      "learning_rate": 6.867680500284252e-05,
      "loss": 8.3974,
      "step": 277
    },
    {
      "epoch": 0.3871866295264624,
      "grad_norm": 3.7470033168792725,
      "learning_rate": 6.866685616827742e-05,
      "loss": 8.3527,
      "step": 278
    },
    {
      "epoch": 0.38857938718662954,
      "grad_norm": 3.5585482120513916,
      "learning_rate": 6.865690733371233e-05,
      "loss": 8.2012,
      "step": 279
    },
    {
      "epoch": 0.38997214484679665,
      "grad_norm": 4.271474361419678,
      "learning_rate": 6.864695849914723e-05,
      "loss": 8.6701,
      "step": 280
    },
    {
      "epoch": 0.39136490250696376,
      "grad_norm": 4.250993728637695,
      "learning_rate": 6.863700966458214e-05,
      "loss": 8.3565,
      "step": 281
    },
    {
      "epoch": 0.39275766016713093,
      "grad_norm": 3.6277823448181152,
      "learning_rate": 6.862706083001705e-05,
      "loss": 8.2638,
      "step": 282
    },
    {
      "epoch": 0.39415041782729804,
      "grad_norm": 4.192377090454102,
      "learning_rate": 6.861711199545196e-05,
      "loss": 8.1367,
      "step": 283
    },
    {
      "epoch": 0.3955431754874652,
      "grad_norm": 3.8814239501953125,
      "learning_rate": 6.860716316088686e-05,
      "loss": 8.2968,
      "step": 284
    },
    {
      "epoch": 0.3969359331476323,
      "grad_norm": 3.861527442932129,
      "learning_rate": 6.859721432632177e-05,
      "loss": 8.2238,
      "step": 285
    },
    {
      "epoch": 0.3983286908077994,
      "grad_norm": 3.7376210689544678,
      "learning_rate": 6.858726549175667e-05,
      "loss": 8.2223,
      "step": 286
    },
    {
      "epoch": 0.3997214484679666,
      "grad_norm": 4.001213550567627,
      "learning_rate": 6.857731665719159e-05,
      "loss": 8.3963,
      "step": 287
    },
    {
      "epoch": 0.4011142061281337,
      "grad_norm": 3.7330715656280518,
      "learning_rate": 6.856736782262649e-05,
      "loss": 8.2626,
      "step": 288
    },
    {
      "epoch": 0.4025069637883008,
      "grad_norm": 4.302996635437012,
      "learning_rate": 6.85574189880614e-05,
      "loss": 8.3355,
      "step": 289
    },
    {
      "epoch": 0.403899721448468,
      "grad_norm": 3.770094156265259,
      "learning_rate": 6.85474701534963e-05,
      "loss": 8.3885,
      "step": 290
    },
    {
      "epoch": 0.4052924791086351,
      "grad_norm": 4.166201591491699,
      "learning_rate": 6.85375213189312e-05,
      "loss": 8.2349,
      "step": 291
    },
    {
      "epoch": 0.40668523676880225,
      "grad_norm": 3.9205539226531982,
      "learning_rate": 6.852757248436611e-05,
      "loss": 8.3096,
      "step": 292
    },
    {
      "epoch": 0.40807799442896936,
      "grad_norm": 3.5723092555999756,
      "learning_rate": 6.851762364980103e-05,
      "loss": 8.5081,
      "step": 293
    },
    {
      "epoch": 0.40947075208913647,
      "grad_norm": 4.560377597808838,
      "learning_rate": 6.850767481523592e-05,
      "loss": 8.2322,
      "step": 294
    },
    {
      "epoch": 0.41086350974930363,
      "grad_norm": 4.03375244140625,
      "learning_rate": 6.849772598067083e-05,
      "loss": 8.2581,
      "step": 295
    },
    {
      "epoch": 0.41225626740947074,
      "grad_norm": 4.069296836853027,
      "learning_rate": 6.848777714610574e-05,
      "loss": 8.1669,
      "step": 296
    },
    {
      "epoch": 0.4136490250696379,
      "grad_norm": 4.926348686218262,
      "learning_rate": 6.847782831154064e-05,
      "loss": 8.1485,
      "step": 297
    },
    {
      "epoch": 0.415041782729805,
      "grad_norm": 4.371025562286377,
      "learning_rate": 6.846787947697555e-05,
      "loss": 7.9984,
      "step": 298
    },
    {
      "epoch": 0.41643454038997213,
      "grad_norm": 4.272550582885742,
      "learning_rate": 6.845793064241045e-05,
      "loss": 8.0296,
      "step": 299
    },
    {
      "epoch": 0.4178272980501393,
      "grad_norm": 4.450153350830078,
      "learning_rate": 6.844798180784535e-05,
      "loss": 8.2459,
      "step": 300
    },
    {
      "epoch": 0.4192200557103064,
      "grad_norm": 4.329676151275635,
      "learning_rate": 6.843803297328027e-05,
      "loss": 8.247,
      "step": 301
    },
    {
      "epoch": 0.4206128133704735,
      "grad_norm": 4.585980415344238,
      "learning_rate": 6.842808413871518e-05,
      "loss": 8.3604,
      "step": 302
    },
    {
      "epoch": 0.4220055710306407,
      "grad_norm": 3.8479998111724854,
      "learning_rate": 6.841813530415008e-05,
      "loss": 8.4521,
      "step": 303
    },
    {
      "epoch": 0.4233983286908078,
      "grad_norm": 4.2120256423950195,
      "learning_rate": 6.840818646958498e-05,
      "loss": 8.1643,
      "step": 304
    },
    {
      "epoch": 0.42479108635097496,
      "grad_norm": 4.131726264953613,
      "learning_rate": 6.839823763501989e-05,
      "loss": 8.3119,
      "step": 305
    },
    {
      "epoch": 0.42618384401114207,
      "grad_norm": 3.7550110816955566,
      "learning_rate": 6.838828880045479e-05,
      "loss": 7.8546,
      "step": 306
    },
    {
      "epoch": 0.4275766016713092,
      "grad_norm": 4.501293182373047,
      "learning_rate": 6.837833996588971e-05,
      "loss": 8.387,
      "step": 307
    },
    {
      "epoch": 0.42896935933147634,
      "grad_norm": 3.984358310699463,
      "learning_rate": 6.83683911313246e-05,
      "loss": 8.3124,
      "step": 308
    },
    {
      "epoch": 0.43036211699164345,
      "grad_norm": 4.523743152618408,
      "learning_rate": 6.835844229675952e-05,
      "loss": 8.1569,
      "step": 309
    },
    {
      "epoch": 0.43175487465181056,
      "grad_norm": 3.7779624462127686,
      "learning_rate": 6.834849346219442e-05,
      "loss": 8.1798,
      "step": 310
    },
    {
      "epoch": 0.4331476323119777,
      "grad_norm": 4.647845268249512,
      "learning_rate": 6.833854462762933e-05,
      "loss": 7.9793,
      "step": 311
    },
    {
      "epoch": 0.43454038997214484,
      "grad_norm": 3.5851359367370605,
      "learning_rate": 6.832859579306424e-05,
      "loss": 8.1629,
      "step": 312
    },
    {
      "epoch": 0.435933147632312,
      "grad_norm": 4.379482269287109,
      "learning_rate": 6.831864695849913e-05,
      "loss": 8.2956,
      "step": 313
    },
    {
      "epoch": 0.4373259052924791,
      "grad_norm": 4.4390645027160645,
      "learning_rate": 6.830869812393405e-05,
      "loss": 7.9394,
      "step": 314
    },
    {
      "epoch": 0.4387186629526462,
      "grad_norm": 4.138993740081787,
      "learning_rate": 6.829874928936896e-05,
      "loss": 8.4355,
      "step": 315
    },
    {
      "epoch": 0.4401114206128134,
      "grad_norm": 4.070654392242432,
      "learning_rate": 6.828880045480386e-05,
      "loss": 8.2139,
      "step": 316
    },
    {
      "epoch": 0.4415041782729805,
      "grad_norm": 4.666057109832764,
      "learning_rate": 6.827885162023876e-05,
      "loss": 8.4277,
      "step": 317
    },
    {
      "epoch": 0.4428969359331476,
      "grad_norm": 4.2710185050964355,
      "learning_rate": 6.826890278567367e-05,
      "loss": 8.2777,
      "step": 318
    },
    {
      "epoch": 0.44428969359331477,
      "grad_norm": 4.994112014770508,
      "learning_rate": 6.825895395110857e-05,
      "loss": 8.1703,
      "step": 319
    },
    {
      "epoch": 0.4456824512534819,
      "grad_norm": 3.9322173595428467,
      "learning_rate": 6.824900511654349e-05,
      "loss": 8.0619,
      "step": 320
    },
    {
      "epoch": 0.44707520891364905,
      "grad_norm": 4.775113105773926,
      "learning_rate": 6.82390562819784e-05,
      "loss": 8.3888,
      "step": 321
    },
    {
      "epoch": 0.44846796657381616,
      "grad_norm": 4.775972366333008,
      "learning_rate": 6.82291074474133e-05,
      "loss": 8.2665,
      "step": 322
    },
    {
      "epoch": 0.44986072423398327,
      "grad_norm": 4.2928290367126465,
      "learning_rate": 6.82191586128482e-05,
      "loss": 8.2028,
      "step": 323
    },
    {
      "epoch": 0.45125348189415043,
      "grad_norm": 3.857879877090454,
      "learning_rate": 6.820920977828311e-05,
      "loss": 8.2462,
      "step": 324
    },
    {
      "epoch": 0.45264623955431754,
      "grad_norm": 3.9455676078796387,
      "learning_rate": 6.819926094371801e-05,
      "loss": 8.2978,
      "step": 325
    },
    {
      "epoch": 0.45403899721448465,
      "grad_norm": 4.97908878326416,
      "learning_rate": 6.818931210915293e-05,
      "loss": 8.0867,
      "step": 326
    },
    {
      "epoch": 0.4554317548746518,
      "grad_norm": 4.033157825469971,
      "learning_rate": 6.817936327458783e-05,
      "loss": 8.4074,
      "step": 327
    },
    {
      "epoch": 0.4568245125348189,
      "grad_norm": 4.895842552185059,
      "learning_rate": 6.816941444002274e-05,
      "loss": 8.3586,
      "step": 328
    },
    {
      "epoch": 0.4582172701949861,
      "grad_norm": 7.515446186065674,
      "learning_rate": 6.815946560545764e-05,
      "loss": 7.9998,
      "step": 329
    },
    {
      "epoch": 0.4596100278551532,
      "grad_norm": 4.775269031524658,
      "learning_rate": 6.814951677089254e-05,
      "loss": 8.3389,
      "step": 330
    },
    {
      "epoch": 0.4610027855153203,
      "grad_norm": 4.500304222106934,
      "learning_rate": 6.813956793632745e-05,
      "loss": 8.1957,
      "step": 331
    },
    {
      "epoch": 0.4623955431754875,
      "grad_norm": 3.8328864574432373,
      "learning_rate": 6.812961910176237e-05,
      "loss": 8.4248,
      "step": 332
    },
    {
      "epoch": 0.4637883008356546,
      "grad_norm": 4.228978157043457,
      "learning_rate": 6.811967026719726e-05,
      "loss": 7.9624,
      "step": 333
    },
    {
      "epoch": 0.46518105849582175,
      "grad_norm": 3.9632623195648193,
      "learning_rate": 6.810972143263217e-05,
      "loss": 7.9722,
      "step": 334
    },
    {
      "epoch": 0.46657381615598886,
      "grad_norm": 4.006036758422852,
      "learning_rate": 6.809977259806708e-05,
      "loss": 8.0736,
      "step": 335
    },
    {
      "epoch": 0.467966573816156,
      "grad_norm": 3.991527557373047,
      "learning_rate": 6.808982376350198e-05,
      "loss": 8.3156,
      "step": 336
    },
    {
      "epoch": 0.46935933147632314,
      "grad_norm": 3.9454505443573,
      "learning_rate": 6.80798749289369e-05,
      "loss": 8.2869,
      "step": 337
    },
    {
      "epoch": 0.47075208913649025,
      "grad_norm": 4.333049774169922,
      "learning_rate": 6.806992609437179e-05,
      "loss": 8.0926,
      "step": 338
    },
    {
      "epoch": 0.47214484679665736,
      "grad_norm": 4.2421112060546875,
      "learning_rate": 6.80599772598067e-05,
      "loss": 8.3053,
      "step": 339
    },
    {
      "epoch": 0.4735376044568245,
      "grad_norm": 4.790825366973877,
      "learning_rate": 6.805002842524161e-05,
      "loss": 7.8001,
      "step": 340
    },
    {
      "epoch": 0.47493036211699163,
      "grad_norm": 4.551433086395264,
      "learning_rate": 6.804007959067652e-05,
      "loss": 8.0157,
      "step": 341
    },
    {
      "epoch": 0.4763231197771588,
      "grad_norm": 4.332050800323486,
      "learning_rate": 6.803013075611142e-05,
      "loss": 8.0316,
      "step": 342
    },
    {
      "epoch": 0.4777158774373259,
      "grad_norm": 4.675700664520264,
      "learning_rate": 6.802018192154632e-05,
      "loss": 8.0046,
      "step": 343
    },
    {
      "epoch": 0.479108635097493,
      "grad_norm": 4.037833213806152,
      "learning_rate": 6.801023308698123e-05,
      "loss": 8.1756,
      "step": 344
    },
    {
      "epoch": 0.4805013927576602,
      "grad_norm": 4.044549942016602,
      "learning_rate": 6.800028425241615e-05,
      "loss": 7.8956,
      "step": 345
    },
    {
      "epoch": 0.4818941504178273,
      "grad_norm": 4.56648588180542,
      "learning_rate": 6.799033541785105e-05,
      "loss": 7.8569,
      "step": 346
    },
    {
      "epoch": 0.4832869080779944,
      "grad_norm": 4.27932071685791,
      "learning_rate": 6.798038658328595e-05,
      "loss": 7.983,
      "step": 347
    },
    {
      "epoch": 0.48467966573816157,
      "grad_norm": 4.309334754943848,
      "learning_rate": 6.797043774872086e-05,
      "loss": 8.0926,
      "step": 348
    },
    {
      "epoch": 0.4860724233983287,
      "grad_norm": 4.4279961585998535,
      "learning_rate": 6.796048891415576e-05,
      "loss": 8.1277,
      "step": 349
    },
    {
      "epoch": 0.48746518105849584,
      "grad_norm": 4.202175140380859,
      "learning_rate": 6.795054007959067e-05,
      "loss": 8.2918,
      "step": 350
    },
    {
      "epoch": 0.48885793871866295,
      "grad_norm": 4.153756618499756,
      "learning_rate": 6.794059124502558e-05,
      "loss": 8.0533,
      "step": 351
    },
    {
      "epoch": 0.49025069637883006,
      "grad_norm": 3.8882815837860107,
      "learning_rate": 6.793064241046048e-05,
      "loss": 8.2729,
      "step": 352
    },
    {
      "epoch": 0.49164345403899723,
      "grad_norm": 4.373082160949707,
      "learning_rate": 6.792069357589539e-05,
      "loss": 8.2598,
      "step": 353
    },
    {
      "epoch": 0.49303621169916434,
      "grad_norm": 4.185801982879639,
      "learning_rate": 6.79107447413303e-05,
      "loss": 8.1271,
      "step": 354
    },
    {
      "epoch": 0.49442896935933145,
      "grad_norm": 4.643874645233154,
      "learning_rate": 6.79007959067652e-05,
      "loss": 8.1628,
      "step": 355
    },
    {
      "epoch": 0.4958217270194986,
      "grad_norm": 4.2254414558410645,
      "learning_rate": 6.78908470722001e-05,
      "loss": 8.0874,
      "step": 356
    },
    {
      "epoch": 0.4972144846796657,
      "grad_norm": 3.6543078422546387,
      "learning_rate": 6.788089823763501e-05,
      "loss": 8.12,
      "step": 357
    },
    {
      "epoch": 0.4986072423398329,
      "grad_norm": 4.454078197479248,
      "learning_rate": 6.787094940306991e-05,
      "loss": 8.1818,
      "step": 358
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.394839286804199,
      "learning_rate": 6.786100056850483e-05,
      "loss": 7.8266,
      "step": 359
    },
    {
      "epoch": 0.5013927576601671,
      "grad_norm": 4.51784610748291,
      "learning_rate": 6.785105173393973e-05,
      "loss": 8.377,
      "step": 360
    },
    {
      "epoch": 0.5027855153203342,
      "grad_norm": 3.762207269668579,
      "learning_rate": 6.784110289937464e-05,
      "loss": 8.0772,
      "step": 361
    },
    {
      "epoch": 0.5041782729805014,
      "grad_norm": 4.488733291625977,
      "learning_rate": 6.783115406480954e-05,
      "loss": 7.9061,
      "step": 362
    },
    {
      "epoch": 0.5055710306406686,
      "grad_norm": 3.9269416332244873,
      "learning_rate": 6.782120523024445e-05,
      "loss": 8.1671,
      "step": 363
    },
    {
      "epoch": 0.5069637883008357,
      "grad_norm": 4.34641170501709,
      "learning_rate": 6.781125639567935e-05,
      "loss": 8.2714,
      "step": 364
    },
    {
      "epoch": 0.5083565459610028,
      "grad_norm": 5.105216979980469,
      "learning_rate": 6.780130756111427e-05,
      "loss": 8.1929,
      "step": 365
    },
    {
      "epoch": 0.5097493036211699,
      "grad_norm": 4.120882987976074,
      "learning_rate": 6.779135872654917e-05,
      "loss": 8.2241,
      "step": 366
    },
    {
      "epoch": 0.5111420612813371,
      "grad_norm": 4.087029933929443,
      "learning_rate": 6.778140989198408e-05,
      "loss": 7.9492,
      "step": 367
    },
    {
      "epoch": 0.5125348189415042,
      "grad_norm": 3.694951057434082,
      "learning_rate": 6.777146105741898e-05,
      "loss": 7.9803,
      "step": 368
    },
    {
      "epoch": 0.5139275766016713,
      "grad_norm": 4.075655460357666,
      "learning_rate": 6.776151222285389e-05,
      "loss": 7.917,
      "step": 369
    },
    {
      "epoch": 0.5153203342618384,
      "grad_norm": 4.9306254386901855,
      "learning_rate": 6.77515633882888e-05,
      "loss": 7.8565,
      "step": 370
    },
    {
      "epoch": 0.5167130919220055,
      "grad_norm": 4.179455280303955,
      "learning_rate": 6.77416145537237e-05,
      "loss": 7.8994,
      "step": 371
    },
    {
      "epoch": 0.5181058495821727,
      "grad_norm": 4.032535552978516,
      "learning_rate": 6.773166571915861e-05,
      "loss": 8.0329,
      "step": 372
    },
    {
      "epoch": 0.5194986072423399,
      "grad_norm": 3.875140905380249,
      "learning_rate": 6.772171688459351e-05,
      "loss": 7.4966,
      "step": 373
    },
    {
      "epoch": 0.520891364902507,
      "grad_norm": 4.436954498291016,
      "learning_rate": 6.771176805002842e-05,
      "loss": 8.0252,
      "step": 374
    },
    {
      "epoch": 0.5222841225626741,
      "grad_norm": 3.833369493484497,
      "learning_rate": 6.770181921546332e-05,
      "loss": 8.0937,
      "step": 375
    },
    {
      "epoch": 0.5236768802228412,
      "grad_norm": 3.9323227405548096,
      "learning_rate": 6.769187038089824e-05,
      "loss": 7.9147,
      "step": 376
    },
    {
      "epoch": 0.5250696378830083,
      "grad_norm": 4.252800941467285,
      "learning_rate": 6.768192154633313e-05,
      "loss": 7.9065,
      "step": 377
    },
    {
      "epoch": 0.5264623955431755,
      "grad_norm": 4.6514482498168945,
      "learning_rate": 6.767197271176805e-05,
      "loss": 7.7514,
      "step": 378
    },
    {
      "epoch": 0.5278551532033426,
      "grad_norm": 4.116487503051758,
      "learning_rate": 6.766202387720295e-05,
      "loss": 8.1345,
      "step": 379
    },
    {
      "epoch": 0.5292479108635098,
      "grad_norm": 4.300816059112549,
      "learning_rate": 6.765207504263786e-05,
      "loss": 7.8827,
      "step": 380
    },
    {
      "epoch": 0.5306406685236769,
      "grad_norm": 3.882556676864624,
      "learning_rate": 6.764212620807276e-05,
      "loss": 7.9054,
      "step": 381
    },
    {
      "epoch": 0.532033426183844,
      "grad_norm": 4.058185577392578,
      "learning_rate": 6.763217737350767e-05,
      "loss": 7.7686,
      "step": 382
    },
    {
      "epoch": 0.5334261838440112,
      "grad_norm": 3.925845146179199,
      "learning_rate": 6.762222853894257e-05,
      "loss": 7.9461,
      "step": 383
    },
    {
      "epoch": 0.5348189415041783,
      "grad_norm": 4.54960823059082,
      "learning_rate": 6.761227970437749e-05,
      "loss": 7.9756,
      "step": 384
    },
    {
      "epoch": 0.5362116991643454,
      "grad_norm": 4.5480427742004395,
      "learning_rate": 6.760233086981239e-05,
      "loss": 7.9223,
      "step": 385
    },
    {
      "epoch": 0.5376044568245125,
      "grad_norm": 3.820448875427246,
      "learning_rate": 6.75923820352473e-05,
      "loss": 7.9005,
      "step": 386
    },
    {
      "epoch": 0.5389972144846796,
      "grad_norm": 4.343704700469971,
      "learning_rate": 6.75824332006822e-05,
      "loss": 7.5845,
      "step": 387
    },
    {
      "epoch": 0.5403899721448467,
      "grad_norm": 4.2667036056518555,
      "learning_rate": 6.75724843661171e-05,
      "loss": 7.914,
      "step": 388
    },
    {
      "epoch": 0.541782729805014,
      "grad_norm": 3.9239776134490967,
      "learning_rate": 6.756253553155201e-05,
      "loss": 8.2154,
      "step": 389
    },
    {
      "epoch": 0.5431754874651811,
      "grad_norm": 3.943493604660034,
      "learning_rate": 6.755258669698692e-05,
      "loss": 7.4672,
      "step": 390
    },
    {
      "epoch": 0.5445682451253482,
      "grad_norm": 4.2096266746521,
      "learning_rate": 6.754263786242182e-05,
      "loss": 7.7497,
      "step": 391
    },
    {
      "epoch": 0.5459610027855153,
      "grad_norm": 3.7932655811309814,
      "learning_rate": 6.753268902785673e-05,
      "loss": 8.2188,
      "step": 392
    },
    {
      "epoch": 0.5473537604456824,
      "grad_norm": 4.857273578643799,
      "learning_rate": 6.752274019329164e-05,
      "loss": 7.8588,
      "step": 393
    },
    {
      "epoch": 0.5487465181058496,
      "grad_norm": 4.0276007652282715,
      "learning_rate": 6.751279135872654e-05,
      "loss": 7.9237,
      "step": 394
    },
    {
      "epoch": 0.5501392757660167,
      "grad_norm": 4.130008697509766,
      "learning_rate": 6.750284252416146e-05,
      "loss": 7.5528,
      "step": 395
    },
    {
      "epoch": 0.5515320334261838,
      "grad_norm": 4.636471271514893,
      "learning_rate": 6.749289368959635e-05,
      "loss": 7.684,
      "step": 396
    },
    {
      "epoch": 0.552924791086351,
      "grad_norm": 4.785216331481934,
      "learning_rate": 6.748294485503125e-05,
      "loss": 8.0515,
      "step": 397
    },
    {
      "epoch": 0.5543175487465181,
      "grad_norm": 4.659693241119385,
      "learning_rate": 6.747299602046617e-05,
      "loss": 7.8735,
      "step": 398
    },
    {
      "epoch": 0.5557103064066853,
      "grad_norm": 4.613758563995361,
      "learning_rate": 6.746304718590108e-05,
      "loss": 7.964,
      "step": 399
    },
    {
      "epoch": 0.5571030640668524,
      "grad_norm": 4.028876781463623,
      "learning_rate": 6.745309835133598e-05,
      "loss": 7.5535,
      "step": 400
    },
    {
      "epoch": 0.5584958217270195,
      "grad_norm": 6.227209091186523,
      "learning_rate": 6.744314951677088e-05,
      "loss": 7.8115,
      "step": 401
    },
    {
      "epoch": 0.5598885793871866,
      "grad_norm": 5.188333511352539,
      "learning_rate": 6.743320068220579e-05,
      "loss": 7.6271,
      "step": 402
    },
    {
      "epoch": 0.5612813370473537,
      "grad_norm": 5.935398101806641,
      "learning_rate": 6.74232518476407e-05,
      "loss": 8.2987,
      "step": 403
    },
    {
      "epoch": 0.5626740947075209,
      "grad_norm": 4.565037727355957,
      "learning_rate": 6.741330301307561e-05,
      "loss": 7.4819,
      "step": 404
    },
    {
      "epoch": 0.564066852367688,
      "grad_norm": 4.986330032348633,
      "learning_rate": 6.740335417851051e-05,
      "loss": 7.6441,
      "step": 405
    },
    {
      "epoch": 0.5654596100278552,
      "grad_norm": 4.388195991516113,
      "learning_rate": 6.739340534394542e-05,
      "loss": 7.6117,
      "step": 406
    },
    {
      "epoch": 0.5668523676880223,
      "grad_norm": 4.697558403015137,
      "learning_rate": 6.738345650938032e-05,
      "loss": 7.7427,
      "step": 407
    },
    {
      "epoch": 0.5682451253481894,
      "grad_norm": 4.241237640380859,
      "learning_rate": 6.737350767481523e-05,
      "loss": 7.9022,
      "step": 408
    },
    {
      "epoch": 0.5696378830083565,
      "grad_norm": 4.273233413696289,
      "learning_rate": 6.736355884025014e-05,
      "loss": 7.637,
      "step": 409
    },
    {
      "epoch": 0.5710306406685237,
      "grad_norm": 4.361372947692871,
      "learning_rate": 6.735361000568505e-05,
      "loss": 7.5757,
      "step": 410
    },
    {
      "epoch": 0.5724233983286908,
      "grad_norm": 4.265485763549805,
      "learning_rate": 6.734366117111995e-05,
      "loss": 7.6315,
      "step": 411
    },
    {
      "epoch": 0.5738161559888579,
      "grad_norm": 4.280587673187256,
      "learning_rate": 6.733371233655486e-05,
      "loss": 7.6017,
      "step": 412
    },
    {
      "epoch": 0.575208913649025,
      "grad_norm": 5.131948471069336,
      "learning_rate": 6.732376350198976e-05,
      "loss": 7.973,
      "step": 413
    },
    {
      "epoch": 0.5766016713091922,
      "grad_norm": 4.602144241333008,
      "learning_rate": 6.731381466742466e-05,
      "loss": 7.8972,
      "step": 414
    },
    {
      "epoch": 0.5779944289693594,
      "grad_norm": 4.569875240325928,
      "learning_rate": 6.730386583285958e-05,
      "loss": 7.9031,
      "step": 415
    },
    {
      "epoch": 0.5793871866295265,
      "grad_norm": 4.378690719604492,
      "learning_rate": 6.729391699829447e-05,
      "loss": 7.7355,
      "step": 416
    },
    {
      "epoch": 0.5807799442896936,
      "grad_norm": 5.248675346374512,
      "learning_rate": 6.728396816372939e-05,
      "loss": 7.4518,
      "step": 417
    },
    {
      "epoch": 0.5821727019498607,
      "grad_norm": 5.819039344787598,
      "learning_rate": 6.727401932916429e-05,
      "loss": 7.7974,
      "step": 418
    },
    {
      "epoch": 0.5835654596100278,
      "grad_norm": 4.727663040161133,
      "learning_rate": 6.72640704945992e-05,
      "loss": 7.3703,
      "step": 419
    },
    {
      "epoch": 0.584958217270195,
      "grad_norm": 3.7929654121398926,
      "learning_rate": 6.725412166003411e-05,
      "loss": 7.989,
      "step": 420
    },
    {
      "epoch": 0.5863509749303621,
      "grad_norm": 4.322558403015137,
      "learning_rate": 6.7244172825469e-05,
      "loss": 7.877,
      "step": 421
    },
    {
      "epoch": 0.5877437325905293,
      "grad_norm": 4.5659613609313965,
      "learning_rate": 6.723422399090391e-05,
      "loss": 7.8418,
      "step": 422
    },
    {
      "epoch": 0.5891364902506964,
      "grad_norm": 4.2391133308410645,
      "learning_rate": 6.722427515633883e-05,
      "loss": 7.6219,
      "step": 423
    },
    {
      "epoch": 0.5905292479108635,
      "grad_norm": 4.069179058074951,
      "learning_rate": 6.721432632177373e-05,
      "loss": 7.9688,
      "step": 424
    },
    {
      "epoch": 0.5919220055710307,
      "grad_norm": 4.541570663452148,
      "learning_rate": 6.720437748720864e-05,
      "loss": 7.9766,
      "step": 425
    },
    {
      "epoch": 0.5933147632311978,
      "grad_norm": 4.306227207183838,
      "learning_rate": 6.719442865264354e-05,
      "loss": 7.7377,
      "step": 426
    },
    {
      "epoch": 0.5947075208913649,
      "grad_norm": 4.158151626586914,
      "learning_rate": 6.718447981807844e-05,
      "loss": 7.5197,
      "step": 427
    },
    {
      "epoch": 0.596100278551532,
      "grad_norm": 4.7426228523254395,
      "learning_rate": 6.717453098351336e-05,
      "loss": 7.9801,
      "step": 428
    },
    {
      "epoch": 0.5974930362116991,
      "grad_norm": 4.626999378204346,
      "learning_rate": 6.716458214894827e-05,
      "loss": 7.5905,
      "step": 429
    },
    {
      "epoch": 0.5988857938718662,
      "grad_norm": 5.324058532714844,
      "learning_rate": 6.715463331438317e-05,
      "loss": 8.0235,
      "step": 430
    },
    {
      "epoch": 0.6002785515320335,
      "grad_norm": 4.126290798187256,
      "learning_rate": 6.714468447981807e-05,
      "loss": 7.8866,
      "step": 431
    },
    {
      "epoch": 0.6016713091922006,
      "grad_norm": 3.7322535514831543,
      "learning_rate": 6.713473564525298e-05,
      "loss": 7.4555,
      "step": 432
    },
    {
      "epoch": 0.6030640668523677,
      "grad_norm": 4.0403876304626465,
      "learning_rate": 6.712478681068788e-05,
      "loss": 7.7761,
      "step": 433
    },
    {
      "epoch": 0.6044568245125348,
      "grad_norm": 4.520780563354492,
      "learning_rate": 6.71148379761228e-05,
      "loss": 7.9322,
      "step": 434
    },
    {
      "epoch": 0.6058495821727019,
      "grad_norm": 3.943120002746582,
      "learning_rate": 6.710488914155769e-05,
      "loss": 7.4862,
      "step": 435
    },
    {
      "epoch": 0.6072423398328691,
      "grad_norm": 5.254693984985352,
      "learning_rate": 6.709494030699261e-05,
      "loss": 7.375,
      "step": 436
    },
    {
      "epoch": 0.6086350974930362,
      "grad_norm": 3.7695791721343994,
      "learning_rate": 6.708499147242751e-05,
      "loss": 7.7116,
      "step": 437
    },
    {
      "epoch": 0.6100278551532033,
      "grad_norm": 4.416331768035889,
      "learning_rate": 6.707504263786242e-05,
      "loss": 7.9933,
      "step": 438
    },
    {
      "epoch": 0.6114206128133705,
      "grad_norm": 3.907776117324829,
      "learning_rate": 6.706509380329732e-05,
      "loss": 7.9828,
      "step": 439
    },
    {
      "epoch": 0.6128133704735376,
      "grad_norm": 4.00132942199707,
      "learning_rate": 6.705514496873222e-05,
      "loss": 7.4383,
      "step": 440
    },
    {
      "epoch": 0.6142061281337048,
      "grad_norm": 3.9430370330810547,
      "learning_rate": 6.704519613416713e-05,
      "loss": 7.5872,
      "step": 441
    },
    {
      "epoch": 0.6155988857938719,
      "grad_norm": 4.282450199127197,
      "learning_rate": 6.703524729960205e-05,
      "loss": 7.2941,
      "step": 442
    },
    {
      "epoch": 0.616991643454039,
      "grad_norm": 4.128746509552002,
      "learning_rate": 6.702529846503695e-05,
      "loss": 7.5871,
      "step": 443
    },
    {
      "epoch": 0.6183844011142061,
      "grad_norm": 4.742000102996826,
      "learning_rate": 6.701534963047185e-05,
      "loss": 7.7283,
      "step": 444
    },
    {
      "epoch": 0.6197771587743732,
      "grad_norm": 4.397759437561035,
      "learning_rate": 6.700540079590676e-05,
      "loss": 7.4465,
      "step": 445
    },
    {
      "epoch": 0.6211699164345403,
      "grad_norm": 5.393130302429199,
      "learning_rate": 6.699545196134166e-05,
      "loss": 8.0633,
      "step": 446
    },
    {
      "epoch": 0.6225626740947076,
      "grad_norm": 4.592435836791992,
      "learning_rate": 6.698550312677657e-05,
      "loss": 7.5154,
      "step": 447
    },
    {
      "epoch": 0.6239554317548747,
      "grad_norm": 6.1243109703063965,
      "learning_rate": 6.697555429221148e-05,
      "loss": 7.3979,
      "step": 448
    },
    {
      "epoch": 0.6253481894150418,
      "grad_norm": 5.653916835784912,
      "learning_rate": 6.696560545764639e-05,
      "loss": 7.3102,
      "step": 449
    },
    {
      "epoch": 0.6267409470752089,
      "grad_norm": 4.128157138824463,
      "learning_rate": 6.695565662308129e-05,
      "loss": 7.6995,
      "step": 450
    },
    {
      "epoch": 0.628133704735376,
      "grad_norm": 4.498085975646973,
      "learning_rate": 6.69457077885162e-05,
      "loss": 7.3869,
      "step": 451
    },
    {
      "epoch": 0.6295264623955432,
      "grad_norm": 4.387475490570068,
      "learning_rate": 6.69357589539511e-05,
      "loss": 7.3913,
      "step": 452
    },
    {
      "epoch": 0.6309192200557103,
      "grad_norm": 4.345146179199219,
      "learning_rate": 6.692581011938602e-05,
      "loss": 7.7166,
      "step": 453
    },
    {
      "epoch": 0.6323119777158774,
      "grad_norm": 4.356617450714111,
      "learning_rate": 6.691586128482092e-05,
      "loss": 7.2329,
      "step": 454
    },
    {
      "epoch": 0.6337047353760445,
      "grad_norm": 4.563919544219971,
      "learning_rate": 6.690591245025583e-05,
      "loss": 7.4578,
      "step": 455
    },
    {
      "epoch": 0.6350974930362117,
      "grad_norm": 3.964571952819824,
      "learning_rate": 6.689596361569073e-05,
      "loss": 7.3362,
      "step": 456
    },
    {
      "epoch": 0.6364902506963789,
      "grad_norm": 3.775155782699585,
      "learning_rate": 6.688601478112563e-05,
      "loss": 7.1363,
      "step": 457
    },
    {
      "epoch": 0.637883008356546,
      "grad_norm": 4.498113632202148,
      "learning_rate": 6.687606594656054e-05,
      "loss": 7.3597,
      "step": 458
    },
    {
      "epoch": 0.6392757660167131,
      "grad_norm": 4.065338134765625,
      "learning_rate": 6.686611711199546e-05,
      "loss": 7.8226,
      "step": 459
    },
    {
      "epoch": 0.6406685236768802,
      "grad_norm": 4.917151927947998,
      "learning_rate": 6.685616827743035e-05,
      "loss": 7.6501,
      "step": 460
    },
    {
      "epoch": 0.6420612813370473,
      "grad_norm": 4.308130741119385,
      "learning_rate": 6.684621944286526e-05,
      "loss": 7.6445,
      "step": 461
    },
    {
      "epoch": 0.6434540389972145,
      "grad_norm": 3.901914596557617,
      "learning_rate": 6.683627060830017e-05,
      "loss": 7.8137,
      "step": 462
    },
    {
      "epoch": 0.6448467966573816,
      "grad_norm": 3.91216778755188,
      "learning_rate": 6.682632177373507e-05,
      "loss": 7.7947,
      "step": 463
    },
    {
      "epoch": 0.6462395543175488,
      "grad_norm": 4.771979331970215,
      "learning_rate": 6.681637293916998e-05,
      "loss": 7.6896,
      "step": 464
    },
    {
      "epoch": 0.6476323119777159,
      "grad_norm": 5.148728847503662,
      "learning_rate": 6.680642410460488e-05,
      "loss": 7.4895,
      "step": 465
    },
    {
      "epoch": 0.649025069637883,
      "grad_norm": 4.330319404602051,
      "learning_rate": 6.679647527003978e-05,
      "loss": 7.6376,
      "step": 466
    },
    {
      "epoch": 0.6504178272980501,
      "grad_norm": 4.768857002258301,
      "learning_rate": 6.67865264354747e-05,
      "loss": 7.7503,
      "step": 467
    },
    {
      "epoch": 0.6518105849582173,
      "grad_norm": 4.0677900314331055,
      "learning_rate": 6.67765776009096e-05,
      "loss": 7.5107,
      "step": 468
    },
    {
      "epoch": 0.6532033426183844,
      "grad_norm": 5.170895576477051,
      "learning_rate": 6.676662876634451e-05,
      "loss": 7.6665,
      "step": 469
    },
    {
      "epoch": 0.6545961002785515,
      "grad_norm": 6.822413444519043,
      "learning_rate": 6.675667993177941e-05,
      "loss": 7.3319,
      "step": 470
    },
    {
      "epoch": 0.6559888579387186,
      "grad_norm": 4.619007587432861,
      "learning_rate": 6.674673109721432e-05,
      "loss": 7.4256,
      "step": 471
    },
    {
      "epoch": 0.6573816155988857,
      "grad_norm": 4.023612022399902,
      "learning_rate": 6.673678226264922e-05,
      "loss": 7.5926,
      "step": 472
    },
    {
      "epoch": 0.658774373259053,
      "grad_norm": 4.313607215881348,
      "learning_rate": 6.672683342808414e-05,
      "loss": 7.7134,
      "step": 473
    },
    {
      "epoch": 0.6601671309192201,
      "grad_norm": 4.842207908630371,
      "learning_rate": 6.671688459351903e-05,
      "loss": 7.4796,
      "step": 474
    },
    {
      "epoch": 0.6615598885793872,
      "grad_norm": 8.598404884338379,
      "learning_rate": 6.670693575895395e-05,
      "loss": 7.584,
      "step": 475
    },
    {
      "epoch": 0.6629526462395543,
      "grad_norm": 4.583897113800049,
      "learning_rate": 6.669698692438885e-05,
      "loss": 7.6584,
      "step": 476
    },
    {
      "epoch": 0.6643454038997214,
      "grad_norm": 5.060597896575928,
      "learning_rate": 6.668703808982376e-05,
      "loss": 7.2289,
      "step": 477
    },
    {
      "epoch": 0.6657381615598886,
      "grad_norm": 4.662160873413086,
      "learning_rate": 6.667708925525867e-05,
      "loss": 7.1151,
      "step": 478
    },
    {
      "epoch": 0.6671309192200557,
      "grad_norm": 4.092073917388916,
      "learning_rate": 6.666714042069356e-05,
      "loss": 7.0661,
      "step": 479
    },
    {
      "epoch": 0.6685236768802229,
      "grad_norm": 4.688446998596191,
      "learning_rate": 6.665719158612847e-05,
      "loss": 7.8714,
      "step": 480
    },
    {
      "epoch": 0.66991643454039,
      "grad_norm": 6.650211334228516,
      "learning_rate": 6.664724275156339e-05,
      "loss": 7.4456,
      "step": 481
    },
    {
      "epoch": 0.6713091922005571,
      "grad_norm": 4.948154926300049,
      "learning_rate": 6.663729391699829e-05,
      "loss": 7.3168,
      "step": 482
    },
    {
      "epoch": 0.6727019498607242,
      "grad_norm": 4.459090232849121,
      "learning_rate": 6.66273450824332e-05,
      "loss": 7.7445,
      "step": 483
    },
    {
      "epoch": 0.6740947075208914,
      "grad_norm": 4.993417263031006,
      "learning_rate": 6.66173962478681e-05,
      "loss": 7.5161,
      "step": 484
    },
    {
      "epoch": 0.6754874651810585,
      "grad_norm": 4.263315200805664,
      "learning_rate": 6.6607447413303e-05,
      "loss": 7.7636,
      "step": 485
    },
    {
      "epoch": 0.6768802228412256,
      "grad_norm": 3.9328417778015137,
      "learning_rate": 6.659749857873792e-05,
      "loss": 7.7585,
      "step": 486
    },
    {
      "epoch": 0.6782729805013927,
      "grad_norm": 4.226478576660156,
      "learning_rate": 6.658754974417282e-05,
      "loss": 7.7491,
      "step": 487
    },
    {
      "epoch": 0.6796657381615598,
      "grad_norm": 4.163766860961914,
      "learning_rate": 6.657760090960773e-05,
      "loss": 7.2713,
      "step": 488
    },
    {
      "epoch": 0.6810584958217271,
      "grad_norm": 4.728254318237305,
      "learning_rate": 6.656765207504263e-05,
      "loss": 7.5866,
      "step": 489
    },
    {
      "epoch": 0.6824512534818942,
      "grad_norm": 4.2728657722473145,
      "learning_rate": 6.655770324047754e-05,
      "loss": 7.4582,
      "step": 490
    },
    {
      "epoch": 0.6838440111420613,
      "grad_norm": 4.443821430206299,
      "learning_rate": 6.654775440591244e-05,
      "loss": 7.4313,
      "step": 491
    },
    {
      "epoch": 0.6852367688022284,
      "grad_norm": 4.36879301071167,
      "learning_rate": 6.653780557134736e-05,
      "loss": 7.3954,
      "step": 492
    },
    {
      "epoch": 0.6866295264623955,
      "grad_norm": 3.9841506481170654,
      "learning_rate": 6.652785673678226e-05,
      "loss": 7.3566,
      "step": 493
    },
    {
      "epoch": 0.6880222841225627,
      "grad_norm": 4.223026752471924,
      "learning_rate": 6.651790790221717e-05,
      "loss": 7.4146,
      "step": 494
    },
    {
      "epoch": 0.6894150417827298,
      "grad_norm": 4.6790008544921875,
      "learning_rate": 6.650795906765207e-05,
      "loss": 7.3331,
      "step": 495
    },
    {
      "epoch": 0.6908077994428969,
      "grad_norm": 4.291107654571533,
      "learning_rate": 6.649801023308697e-05,
      "loss": 7.2829,
      "step": 496
    },
    {
      "epoch": 0.692200557103064,
      "grad_norm": 4.316932678222656,
      "learning_rate": 6.648806139852188e-05,
      "loss": 7.41,
      "step": 497
    },
    {
      "epoch": 0.6935933147632312,
      "grad_norm": 4.292764663696289,
      "learning_rate": 6.64781125639568e-05,
      "loss": 7.2584,
      "step": 498
    },
    {
      "epoch": 0.6949860724233984,
      "grad_norm": 4.277196407318115,
      "learning_rate": 6.646816372939169e-05,
      "loss": 7.3955,
      "step": 499
    },
    {
      "epoch": 0.6963788300835655,
      "grad_norm": 4.318631649017334,
      "learning_rate": 6.64582148948266e-05,
      "loss": 7.3698,
      "step": 500
    },
    {
      "epoch": 0.6977715877437326,
      "grad_norm": 4.1690263748168945,
      "learning_rate": 6.644826606026151e-05,
      "loss": 7.2454,
      "step": 501
    },
    {
      "epoch": 0.6991643454038997,
      "grad_norm": 3.916719675064087,
      "learning_rate": 6.643831722569641e-05,
      "loss": 7.1476,
      "step": 502
    },
    {
      "epoch": 0.7005571030640668,
      "grad_norm": 3.9907374382019043,
      "learning_rate": 6.642836839113132e-05,
      "loss": 7.4952,
      "step": 503
    },
    {
      "epoch": 0.7019498607242339,
      "grad_norm": 4.071834564208984,
      "learning_rate": 6.641841955656622e-05,
      "loss": 7.492,
      "step": 504
    },
    {
      "epoch": 0.7033426183844012,
      "grad_norm": 4.0513458251953125,
      "learning_rate": 6.640847072200112e-05,
      "loss": 7.4342,
      "step": 505
    },
    {
      "epoch": 0.7047353760445683,
      "grad_norm": 4.521594524383545,
      "learning_rate": 6.639852188743604e-05,
      "loss": 7.3739,
      "step": 506
    },
    {
      "epoch": 0.7061281337047354,
      "grad_norm": 4.986645221710205,
      "learning_rate": 6.638857305287095e-05,
      "loss": 7.55,
      "step": 507
    },
    {
      "epoch": 0.7075208913649025,
      "grad_norm": 4.119948387145996,
      "learning_rate": 6.637862421830585e-05,
      "loss": 7.7317,
      "step": 508
    },
    {
      "epoch": 0.7089136490250696,
      "grad_norm": 5.314901351928711,
      "learning_rate": 6.636867538374075e-05,
      "loss": 7.7665,
      "step": 509
    },
    {
      "epoch": 0.7103064066852368,
      "grad_norm": 4.315718650817871,
      "learning_rate": 6.635872654917566e-05,
      "loss": 7.1784,
      "step": 510
    },
    {
      "epoch": 0.7116991643454039,
      "grad_norm": 3.93410325050354,
      "learning_rate": 6.634877771461058e-05,
      "loss": 7.4012,
      "step": 511
    },
    {
      "epoch": 0.713091922005571,
      "grad_norm": 4.619295597076416,
      "learning_rate": 6.633882888004548e-05,
      "loss": 7.178,
      "step": 512
    },
    {
      "epoch": 0.7144846796657381,
      "grad_norm": 4.923532485961914,
      "learning_rate": 6.632888004548038e-05,
      "loss": 7.4176,
      "step": 513
    },
    {
      "epoch": 0.7158774373259053,
      "grad_norm": 5.159674167633057,
      "learning_rate": 6.631893121091529e-05,
      "loss": 7.2846,
      "step": 514
    },
    {
      "epoch": 0.7172701949860725,
      "grad_norm": 3.867479085922241,
      "learning_rate": 6.630898237635019e-05,
      "loss": 7.142,
      "step": 515
    },
    {
      "epoch": 0.7186629526462396,
      "grad_norm": 4.792901039123535,
      "learning_rate": 6.62990335417851e-05,
      "loss": 7.4178,
      "step": 516
    },
    {
      "epoch": 0.7200557103064067,
      "grad_norm": 4.785666465759277,
      "learning_rate": 6.628908470722001e-05,
      "loss": 7.1373,
      "step": 517
    },
    {
      "epoch": 0.7214484679665738,
      "grad_norm": 5.257824420928955,
      "learning_rate": 6.62791358726549e-05,
      "loss": 7.3563,
      "step": 518
    },
    {
      "epoch": 0.7228412256267409,
      "grad_norm": 4.598160266876221,
      "learning_rate": 6.626918703808982e-05,
      "loss": 7.2522,
      "step": 519
    },
    {
      "epoch": 0.724233983286908,
      "grad_norm": 4.3474040031433105,
      "learning_rate": 6.625923820352473e-05,
      "loss": 7.7297,
      "step": 520
    },
    {
      "epoch": 0.7256267409470752,
      "grad_norm": 4.432811737060547,
      "learning_rate": 6.624928936895963e-05,
      "loss": 7.3761,
      "step": 521
    },
    {
      "epoch": 0.7270194986072424,
      "grad_norm": 3.7676665782928467,
      "learning_rate": 6.623934053439453e-05,
      "loss": 7.2723,
      "step": 522
    },
    {
      "epoch": 0.7284122562674095,
      "grad_norm": 4.212183952331543,
      "learning_rate": 6.622939169982944e-05,
      "loss": 7.1899,
      "step": 523
    },
    {
      "epoch": 0.7298050139275766,
      "grad_norm": 3.817636013031006,
      "learning_rate": 6.621944286526434e-05,
      "loss": 7.0919,
      "step": 524
    },
    {
      "epoch": 0.7311977715877437,
      "grad_norm": 3.760101079940796,
      "learning_rate": 6.620949403069926e-05,
      "loss": 7.0633,
      "step": 525
    },
    {
      "epoch": 0.7325905292479109,
      "grad_norm": 4.175741672515869,
      "learning_rate": 6.619954519613416e-05,
      "loss": 7.053,
      "step": 526
    },
    {
      "epoch": 0.733983286908078,
      "grad_norm": 4.742794036865234,
      "learning_rate": 6.618959636156907e-05,
      "loss": 7.4681,
      "step": 527
    },
    {
      "epoch": 0.7353760445682451,
      "grad_norm": 5.325410842895508,
      "learning_rate": 6.617964752700397e-05,
      "loss": 7.5734,
      "step": 528
    },
    {
      "epoch": 0.7367688022284122,
      "grad_norm": 4.095676422119141,
      "learning_rate": 6.616969869243888e-05,
      "loss": 7.5398,
      "step": 529
    },
    {
      "epoch": 0.7381615598885793,
      "grad_norm": 5.706197738647461,
      "learning_rate": 6.615974985787378e-05,
      "loss": 7.4122,
      "step": 530
    },
    {
      "epoch": 0.7395543175487466,
      "grad_norm": 4.12428617477417,
      "learning_rate": 6.61498010233087e-05,
      "loss": 7.1195,
      "step": 531
    },
    {
      "epoch": 0.7409470752089137,
      "grad_norm": 4.2953691482543945,
      "learning_rate": 6.61398521887436e-05,
      "loss": 6.9206,
      "step": 532
    },
    {
      "epoch": 0.7423398328690808,
      "grad_norm": 4.770367622375488,
      "learning_rate": 6.61299033541785e-05,
      "loss": 7.0371,
      "step": 533
    },
    {
      "epoch": 0.7437325905292479,
      "grad_norm": 4.273123264312744,
      "learning_rate": 6.611995451961341e-05,
      "loss": 6.5632,
      "step": 534
    },
    {
      "epoch": 0.745125348189415,
      "grad_norm": 5.233716011047363,
      "learning_rate": 6.611000568504831e-05,
      "loss": 7.3385,
      "step": 535
    },
    {
      "epoch": 0.7465181058495822,
      "grad_norm": 4.813663959503174,
      "learning_rate": 6.610005685048323e-05,
      "loss": 7.2633,
      "step": 536
    },
    {
      "epoch": 0.7479108635097493,
      "grad_norm": 4.405749797821045,
      "learning_rate": 6.609010801591814e-05,
      "loss": 7.4617,
      "step": 537
    },
    {
      "epoch": 0.7493036211699164,
      "grad_norm": 4.777441024780273,
      "learning_rate": 6.608015918135304e-05,
      "loss": 7.5012,
      "step": 538
    },
    {
      "epoch": 0.7506963788300836,
      "grad_norm": 3.9238295555114746,
      "learning_rate": 6.607021034678794e-05,
      "loss": 7.3552,
      "step": 539
    },
    {
      "epoch": 0.7520891364902507,
      "grad_norm": 5.207787990570068,
      "learning_rate": 6.606026151222285e-05,
      "loss": 6.8543,
      "step": 540
    },
    {
      "epoch": 0.7534818941504178,
      "grad_norm": 4.217895984649658,
      "learning_rate": 6.605031267765775e-05,
      "loss": 7.2798,
      "step": 541
    },
    {
      "epoch": 0.754874651810585,
      "grad_norm": 4.337287425994873,
      "learning_rate": 6.604036384309267e-05,
      "loss": 7.2047,
      "step": 542
    },
    {
      "epoch": 0.7562674094707521,
      "grad_norm": 5.529717922210693,
      "learning_rate": 6.603041500852756e-05,
      "loss": 7.3142,
      "step": 543
    },
    {
      "epoch": 0.7576601671309192,
      "grad_norm": 4.267019271850586,
      "learning_rate": 6.602046617396248e-05,
      "loss": 6.9866,
      "step": 544
    },
    {
      "epoch": 0.7590529247910863,
      "grad_norm": 4.3455352783203125,
      "learning_rate": 6.601051733939738e-05,
      "loss": 6.9549,
      "step": 545
    },
    {
      "epoch": 0.7604456824512534,
      "grad_norm": 4.68610143661499,
      "learning_rate": 6.600056850483229e-05,
      "loss": 7.5779,
      "step": 546
    },
    {
      "epoch": 0.7618384401114207,
      "grad_norm": 5.094757080078125,
      "learning_rate": 6.599061967026719e-05,
      "loss": 7.4815,
      "step": 547
    },
    {
      "epoch": 0.7632311977715878,
      "grad_norm": 4.233945846557617,
      "learning_rate": 6.59806708357021e-05,
      "loss": 7.3032,
      "step": 548
    },
    {
      "epoch": 0.7646239554317549,
      "grad_norm": 4.713217258453369,
      "learning_rate": 6.5970722001137e-05,
      "loss": 7.3833,
      "step": 549
    },
    {
      "epoch": 0.766016713091922,
      "grad_norm": 6.269781112670898,
      "learning_rate": 6.596077316657192e-05,
      "loss": 7.212,
      "step": 550
    },
    {
      "epoch": 0.7674094707520891,
      "grad_norm": 4.588368892669678,
      "learning_rate": 6.595082433200682e-05,
      "loss": 7.448,
      "step": 551
    },
    {
      "epoch": 0.7688022284122563,
      "grad_norm": 4.187366008758545,
      "learning_rate": 6.594087549744172e-05,
      "loss": 7.0892,
      "step": 552
    },
    {
      "epoch": 0.7701949860724234,
      "grad_norm": 4.566118240356445,
      "learning_rate": 6.593092666287663e-05,
      "loss": 6.8282,
      "step": 553
    },
    {
      "epoch": 0.7715877437325905,
      "grad_norm": 5.366352081298828,
      "learning_rate": 6.592097782831153e-05,
      "loss": 7.4461,
      "step": 554
    },
    {
      "epoch": 0.7729805013927576,
      "grad_norm": 4.582744598388672,
      "learning_rate": 6.591102899374644e-05,
      "loss": 7.0199,
      "step": 555
    },
    {
      "epoch": 0.7743732590529248,
      "grad_norm": 3.6933884620666504,
      "learning_rate": 6.590108015918135e-05,
      "loss": 6.9732,
      "step": 556
    },
    {
      "epoch": 0.775766016713092,
      "grad_norm": 4.238746166229248,
      "learning_rate": 6.589113132461624e-05,
      "loss": 7.621,
      "step": 557
    },
    {
      "epoch": 0.7771587743732591,
      "grad_norm": 5.870996475219727,
      "learning_rate": 6.588118249005116e-05,
      "loss": 7.1813,
      "step": 558
    },
    {
      "epoch": 0.7785515320334262,
      "grad_norm": 4.53729248046875,
      "learning_rate": 6.587123365548607e-05,
      "loss": 7.2066,
      "step": 559
    },
    {
      "epoch": 0.7799442896935933,
      "grad_norm": 5.991088390350342,
      "learning_rate": 6.586128482092097e-05,
      "loss": 7.1529,
      "step": 560
    },
    {
      "epoch": 0.7813370473537604,
      "grad_norm": 7.669244289398193,
      "learning_rate": 6.585133598635589e-05,
      "loss": 7.3021,
      "step": 561
    },
    {
      "epoch": 0.7827298050139275,
      "grad_norm": 7.90692663192749,
      "learning_rate": 6.584138715179078e-05,
      "loss": 7.4128,
      "step": 562
    },
    {
      "epoch": 0.7841225626740947,
      "grad_norm": 5.141284942626953,
      "learning_rate": 6.583143831722568e-05,
      "loss": 7.0841,
      "step": 563
    },
    {
      "epoch": 0.7855153203342619,
      "grad_norm": 5.19080924987793,
      "learning_rate": 6.58214894826606e-05,
      "loss": 7.2239,
      "step": 564
    },
    {
      "epoch": 0.786908077994429,
      "grad_norm": 4.016646862030029,
      "learning_rate": 6.58115406480955e-05,
      "loss": 6.8987,
      "step": 565
    },
    {
      "epoch": 0.7883008356545961,
      "grad_norm": 4.256837368011475,
      "learning_rate": 6.580159181353041e-05,
      "loss": 6.9851,
      "step": 566
    },
    {
      "epoch": 0.7896935933147632,
      "grad_norm": 4.62431001663208,
      "learning_rate": 6.579164297896531e-05,
      "loss": 6.9119,
      "step": 567
    },
    {
      "epoch": 0.7910863509749304,
      "grad_norm": 4.101048946380615,
      "learning_rate": 6.578169414440022e-05,
      "loss": 7.0689,
      "step": 568
    },
    {
      "epoch": 0.7924791086350975,
      "grad_norm": 4.175568103790283,
      "learning_rate": 6.577174530983513e-05,
      "loss": 7.0084,
      "step": 569
    },
    {
      "epoch": 0.7938718662952646,
      "grad_norm": 4.61623477935791,
      "learning_rate": 6.576179647527004e-05,
      "loss": 7.0022,
      "step": 570
    },
    {
      "epoch": 0.7952646239554317,
      "grad_norm": 4.684569358825684,
      "learning_rate": 6.575184764070494e-05,
      "loss": 7.2584,
      "step": 571
    },
    {
      "epoch": 0.7966573816155988,
      "grad_norm": 5.298076629638672,
      "learning_rate": 6.574189880613985e-05,
      "loss": 7.5078,
      "step": 572
    },
    {
      "epoch": 0.7980501392757661,
      "grad_norm": 4.708342552185059,
      "learning_rate": 6.573194997157475e-05,
      "loss": 7.0561,
      "step": 573
    },
    {
      "epoch": 0.7994428969359332,
      "grad_norm": 4.700664520263672,
      "learning_rate": 6.572200113700965e-05,
      "loss": 6.9512,
      "step": 574
    },
    {
      "epoch": 0.8008356545961003,
      "grad_norm": 4.937999248504639,
      "learning_rate": 6.571205230244457e-05,
      "loss": 7.3912,
      "step": 575
    },
    {
      "epoch": 0.8022284122562674,
      "grad_norm": 4.863896369934082,
      "learning_rate": 6.570210346787948e-05,
      "loss": 7.213,
      "step": 576
    },
    {
      "epoch": 0.8036211699164345,
      "grad_norm": 4.428140640258789,
      "learning_rate": 6.569215463331438e-05,
      "loss": 7.0807,
      "step": 577
    },
    {
      "epoch": 0.8050139275766016,
      "grad_norm": 4.518229961395264,
      "learning_rate": 6.568220579874928e-05,
      "loss": 7.2289,
      "step": 578
    },
    {
      "epoch": 0.8064066852367688,
      "grad_norm": 4.143688201904297,
      "learning_rate": 6.567225696418419e-05,
      "loss": 7.0841,
      "step": 579
    },
    {
      "epoch": 0.807799442896936,
      "grad_norm": 4.04298734664917,
      "learning_rate": 6.566230812961909e-05,
      "loss": 6.8355,
      "step": 580
    },
    {
      "epoch": 0.8091922005571031,
      "grad_norm": 3.7745213508605957,
      "learning_rate": 6.565235929505401e-05,
      "loss": 6.7909,
      "step": 581
    },
    {
      "epoch": 0.8105849582172702,
      "grad_norm": 4.9903483390808105,
      "learning_rate": 6.56424104604889e-05,
      "loss": 7.1245,
      "step": 582
    },
    {
      "epoch": 0.8119777158774373,
      "grad_norm": 4.211728096008301,
      "learning_rate": 6.563246162592382e-05,
      "loss": 7.1137,
      "step": 583
    },
    {
      "epoch": 0.8133704735376045,
      "grad_norm": 3.8617141246795654,
      "learning_rate": 6.562251279135872e-05,
      "loss": 7.6446,
      "step": 584
    },
    {
      "epoch": 0.8147632311977716,
      "grad_norm": 3.9979238510131836,
      "learning_rate": 6.561256395679363e-05,
      "loss": 7.1116,
      "step": 585
    },
    {
      "epoch": 0.8161559888579387,
      "grad_norm": 4.649385929107666,
      "learning_rate": 6.560261512222853e-05,
      "loss": 7.062,
      "step": 586
    },
    {
      "epoch": 0.8175487465181058,
      "grad_norm": 4.784141540527344,
      "learning_rate": 6.559266628766343e-05,
      "loss": 6.5499,
      "step": 587
    },
    {
      "epoch": 0.8189415041782729,
      "grad_norm": 6.651003360748291,
      "learning_rate": 6.558271745309834e-05,
      "loss": 7.5502,
      "step": 588
    },
    {
      "epoch": 0.8203342618384402,
      "grad_norm": 4.67574405670166,
      "learning_rate": 6.557276861853326e-05,
      "loss": 6.9083,
      "step": 589
    },
    {
      "epoch": 0.8217270194986073,
      "grad_norm": 5.133337497711182,
      "learning_rate": 6.556281978396816e-05,
      "loss": 7.2497,
      "step": 590
    },
    {
      "epoch": 0.8231197771587744,
      "grad_norm": 4.505591869354248,
      "learning_rate": 6.555287094940306e-05,
      "loss": 7.2742,
      "step": 591
    },
    {
      "epoch": 0.8245125348189415,
      "grad_norm": 4.713627338409424,
      "learning_rate": 6.554292211483797e-05,
      "loss": 7.1844,
      "step": 592
    },
    {
      "epoch": 0.8259052924791086,
      "grad_norm": 4.773281097412109,
      "learning_rate": 6.553297328027287e-05,
      "loss": 6.9555,
      "step": 593
    },
    {
      "epoch": 0.8272980501392758,
      "grad_norm": 4.636007308959961,
      "learning_rate": 6.552302444570779e-05,
      "loss": 6.8225,
      "step": 594
    },
    {
      "epoch": 0.8286908077994429,
      "grad_norm": 4.04464054107666,
      "learning_rate": 6.55130756111427e-05,
      "loss": 7.1568,
      "step": 595
    },
    {
      "epoch": 0.83008356545961,
      "grad_norm": 4.1389994621276855,
      "learning_rate": 6.55031267765776e-05,
      "loss": 7.1188,
      "step": 596
    },
    {
      "epoch": 0.8314763231197771,
      "grad_norm": 4.6370086669921875,
      "learning_rate": 6.54931779420125e-05,
      "loss": 6.9095,
      "step": 597
    },
    {
      "epoch": 0.8328690807799443,
      "grad_norm": 5.183930397033691,
      "learning_rate": 6.54832291074474e-05,
      "loss": 7.2433,
      "step": 598
    },
    {
      "epoch": 0.8342618384401114,
      "grad_norm": 5.050951957702637,
      "learning_rate": 6.547328027288231e-05,
      "loss": 7.2824,
      "step": 599
    },
    {
      "epoch": 0.8356545961002786,
      "grad_norm": 4.163578987121582,
      "learning_rate": 6.546333143831723e-05,
      "loss": 6.8322,
      "step": 600
    },
    {
      "epoch": 0.8370473537604457,
      "grad_norm": 4.691806316375732,
      "learning_rate": 6.545338260375212e-05,
      "loss": 7.0893,
      "step": 601
    },
    {
      "epoch": 0.8384401114206128,
      "grad_norm": 4.006926536560059,
      "learning_rate": 6.544343376918704e-05,
      "loss": 6.8651,
      "step": 602
    },
    {
      "epoch": 0.8398328690807799,
      "grad_norm": 4.914746284484863,
      "learning_rate": 6.543348493462194e-05,
      "loss": 6.9708,
      "step": 603
    },
    {
      "epoch": 0.841225626740947,
      "grad_norm": 4.245934009552002,
      "learning_rate": 6.542353610005684e-05,
      "loss": 6.7655,
      "step": 604
    },
    {
      "epoch": 0.8426183844011143,
      "grad_norm": 4.1350860595703125,
      "learning_rate": 6.541358726549175e-05,
      "loss": 7.2147,
      "step": 605
    },
    {
      "epoch": 0.8440111420612814,
      "grad_norm": 4.082108974456787,
      "learning_rate": 6.540363843092665e-05,
      "loss": 7.0206,
      "step": 606
    },
    {
      "epoch": 0.8454038997214485,
      "grad_norm": 4.45560884475708,
      "learning_rate": 6.539368959636156e-05,
      "loss": 6.652,
      "step": 607
    },
    {
      "epoch": 0.8467966573816156,
      "grad_norm": 4.454686641693115,
      "learning_rate": 6.538374076179647e-05,
      "loss": 7.4298,
      "step": 608
    },
    {
      "epoch": 0.8481894150417827,
      "grad_norm": 4.466418743133545,
      "learning_rate": 6.537379192723138e-05,
      "loss": 7.5666,
      "step": 609
    },
    {
      "epoch": 0.8495821727019499,
      "grad_norm": 3.795778512954712,
      "learning_rate": 6.536384309266628e-05,
      "loss": 7.0145,
      "step": 610
    },
    {
      "epoch": 0.850974930362117,
      "grad_norm": 4.251140117645264,
      "learning_rate": 6.535389425810119e-05,
      "loss": 7.2689,
      "step": 611
    },
    {
      "epoch": 0.8523676880222841,
      "grad_norm": 4.9372406005859375,
      "learning_rate": 6.534394542353609e-05,
      "loss": 7.038,
      "step": 612
    },
    {
      "epoch": 0.8537604456824512,
      "grad_norm": 4.656927108764648,
      "learning_rate": 6.5333996588971e-05,
      "loss": 7.2663,
      "step": 613
    },
    {
      "epoch": 0.8551532033426184,
      "grad_norm": 4.058319091796875,
      "learning_rate": 6.532404775440591e-05,
      "loss": 7.1084,
      "step": 614
    },
    {
      "epoch": 0.8565459610027855,
      "grad_norm": 4.106003761291504,
      "learning_rate": 6.531409891984082e-05,
      "loss": 6.6016,
      "step": 615
    },
    {
      "epoch": 0.8579387186629527,
      "grad_norm": 6.479761600494385,
      "learning_rate": 6.530415008527572e-05,
      "loss": 7.2559,
      "step": 616
    },
    {
      "epoch": 0.8593314763231198,
      "grad_norm": 4.016816139221191,
      "learning_rate": 6.529420125071062e-05,
      "loss": 7.1392,
      "step": 617
    },
    {
      "epoch": 0.8607242339832869,
      "grad_norm": 4.016937732696533,
      "learning_rate": 6.528425241614553e-05,
      "loss": 6.538,
      "step": 618
    },
    {
      "epoch": 0.862116991643454,
      "grad_norm": 4.666384696960449,
      "learning_rate": 6.527430358158045e-05,
      "loss": 6.7963,
      "step": 619
    },
    {
      "epoch": 0.8635097493036211,
      "grad_norm": 4.206920623779297,
      "learning_rate": 6.526435474701535e-05,
      "loss": 7.2645,
      "step": 620
    },
    {
      "epoch": 0.8649025069637883,
      "grad_norm": 4.449158191680908,
      "learning_rate": 6.525440591245024e-05,
      "loss": 7.0839,
      "step": 621
    },
    {
      "epoch": 0.8662952646239555,
      "grad_norm": 6.0619964599609375,
      "learning_rate": 6.524445707788516e-05,
      "loss": 7.0886,
      "step": 622
    },
    {
      "epoch": 0.8676880222841226,
      "grad_norm": 4.405318737030029,
      "learning_rate": 6.523450824332006e-05,
      "loss": 6.8355,
      "step": 623
    },
    {
      "epoch": 0.8690807799442897,
      "grad_norm": 4.437068462371826,
      "learning_rate": 6.522455940875497e-05,
      "loss": 7.0541,
      "step": 624
    },
    {
      "epoch": 0.8704735376044568,
      "grad_norm": 4.335897922515869,
      "learning_rate": 6.521461057418988e-05,
      "loss": 7.2348,
      "step": 625
    },
    {
      "epoch": 0.871866295264624,
      "grad_norm": 4.303770065307617,
      "learning_rate": 6.520466173962477e-05,
      "loss": 7.127,
      "step": 626
    },
    {
      "epoch": 0.8732590529247911,
      "grad_norm": 4.896326065063477,
      "learning_rate": 6.519471290505969e-05,
      "loss": 7.2584,
      "step": 627
    },
    {
      "epoch": 0.8746518105849582,
      "grad_norm": 4.51694917678833,
      "learning_rate": 6.51847640704946e-05,
      "loss": 6.6806,
      "step": 628
    },
    {
      "epoch": 0.8760445682451253,
      "grad_norm": 4.327085018157959,
      "learning_rate": 6.51748152359295e-05,
      "loss": 7.2585,
      "step": 629
    },
    {
      "epoch": 0.8774373259052924,
      "grad_norm": 9.302055358886719,
      "learning_rate": 6.51648664013644e-05,
      "loss": 6.5966,
      "step": 630
    },
    {
      "epoch": 0.8788300835654597,
      "grad_norm": 4.5956597328186035,
      "learning_rate": 6.515491756679931e-05,
      "loss": 7.3094,
      "step": 631
    },
    {
      "epoch": 0.8802228412256268,
      "grad_norm": 4.286032199859619,
      "learning_rate": 6.514496873223421e-05,
      "loss": 6.7288,
      "step": 632
    },
    {
      "epoch": 0.8816155988857939,
      "grad_norm": 4.732513427734375,
      "learning_rate": 6.513501989766913e-05,
      "loss": 7.3134,
      "step": 633
    },
    {
      "epoch": 0.883008356545961,
      "grad_norm": 5.090188026428223,
      "learning_rate": 6.512507106310403e-05,
      "loss": 6.529,
      "step": 634
    },
    {
      "epoch": 0.8844011142061281,
      "grad_norm": 4.097725868225098,
      "learning_rate": 6.511512222853894e-05,
      "loss": 6.4284,
      "step": 635
    },
    {
      "epoch": 0.8857938718662952,
      "grad_norm": 4.353823184967041,
      "learning_rate": 6.510517339397384e-05,
      "loss": 7.0736,
      "step": 636
    },
    {
      "epoch": 0.8871866295264624,
      "grad_norm": 5.056899547576904,
      "learning_rate": 6.509522455940875e-05,
      "loss": 7.3754,
      "step": 637
    },
    {
      "epoch": 0.8885793871866295,
      "grad_norm": 4.356399059295654,
      "learning_rate": 6.508527572484365e-05,
      "loss": 6.7254,
      "step": 638
    },
    {
      "epoch": 0.8899721448467967,
      "grad_norm": 4.472954273223877,
      "learning_rate": 6.507532689027857e-05,
      "loss": 7.1479,
      "step": 639
    },
    {
      "epoch": 0.8913649025069638,
      "grad_norm": 4.049185752868652,
      "learning_rate": 6.506537805571346e-05,
      "loss": 7.0321,
      "step": 640
    },
    {
      "epoch": 0.8927576601671309,
      "grad_norm": 4.421054363250732,
      "learning_rate": 6.505542922114838e-05,
      "loss": 6.8861,
      "step": 641
    },
    {
      "epoch": 0.8941504178272981,
      "grad_norm": 4.027243137359619,
      "learning_rate": 6.504548038658328e-05,
      "loss": 7.047,
      "step": 642
    },
    {
      "epoch": 0.8955431754874652,
      "grad_norm": 4.731100559234619,
      "learning_rate": 6.503553155201818e-05,
      "loss": 6.6766,
      "step": 643
    },
    {
      "epoch": 0.8969359331476323,
      "grad_norm": 4.997005939483643,
      "learning_rate": 6.50255827174531e-05,
      "loss": 6.964,
      "step": 644
    },
    {
      "epoch": 0.8983286908077994,
      "grad_norm": 4.219837188720703,
      "learning_rate": 6.501563388288799e-05,
      "loss": 7.0504,
      "step": 645
    },
    {
      "epoch": 0.8997214484679665,
      "grad_norm": 4.365358829498291,
      "learning_rate": 6.50056850483229e-05,
      "loss": 6.7434,
      "step": 646
    },
    {
      "epoch": 0.9011142061281338,
      "grad_norm": 4.429034233093262,
      "learning_rate": 6.499573621375781e-05,
      "loss": 6.408,
      "step": 647
    },
    {
      "epoch": 0.9025069637883009,
      "grad_norm": 4.476465702056885,
      "learning_rate": 6.498578737919272e-05,
      "loss": 6.9395,
      "step": 648
    },
    {
      "epoch": 0.903899721448468,
      "grad_norm": 3.8384079933166504,
      "learning_rate": 6.497583854462762e-05,
      "loss": 6.5423,
      "step": 649
    },
    {
      "epoch": 0.9052924791086351,
      "grad_norm": 4.713837623596191,
      "learning_rate": 6.496588971006253e-05,
      "loss": 7.1157,
      "step": 650
    },
    {
      "epoch": 0.9066852367688022,
      "grad_norm": 4.122503280639648,
      "learning_rate": 6.495594087549743e-05,
      "loss": 6.6999,
      "step": 651
    },
    {
      "epoch": 0.9080779944289693,
      "grad_norm": 5.028951644897461,
      "learning_rate": 6.494599204093235e-05,
      "loss": 6.7512,
      "step": 652
    },
    {
      "epoch": 0.9094707520891365,
      "grad_norm": 4.512173175811768,
      "learning_rate": 6.493604320636725e-05,
      "loss": 6.8538,
      "step": 653
    },
    {
      "epoch": 0.9108635097493036,
      "grad_norm": 4.629344463348389,
      "learning_rate": 6.492609437180216e-05,
      "loss": 6.8566,
      "step": 654
    },
    {
      "epoch": 0.9122562674094707,
      "grad_norm": 4.417125701904297,
      "learning_rate": 6.491614553723706e-05,
      "loss": 7.0721,
      "step": 655
    },
    {
      "epoch": 0.9136490250696379,
      "grad_norm": 4.363222122192383,
      "learning_rate": 6.490619670267196e-05,
      "loss": 6.8486,
      "step": 656
    },
    {
      "epoch": 0.915041782729805,
      "grad_norm": 4.881872177124023,
      "learning_rate": 6.489624786810687e-05,
      "loss": 6.7772,
      "step": 657
    },
    {
      "epoch": 0.9164345403899722,
      "grad_norm": 6.637619495391846,
      "learning_rate": 6.488629903354179e-05,
      "loss": 7.021,
      "step": 658
    },
    {
      "epoch": 0.9178272980501393,
      "grad_norm": 4.867405891418457,
      "learning_rate": 6.487635019897669e-05,
      "loss": 7.0793,
      "step": 659
    },
    {
      "epoch": 0.9192200557103064,
      "grad_norm": 6.487813472747803,
      "learning_rate": 6.48664013644116e-05,
      "loss": 6.1775,
      "step": 660
    },
    {
      "epoch": 0.9206128133704735,
      "grad_norm": 3.9346566200256348,
      "learning_rate": 6.48564525298465e-05,
      "loss": 6.9936,
      "step": 661
    },
    {
      "epoch": 0.9220055710306406,
      "grad_norm": 4.957814693450928,
      "learning_rate": 6.48465036952814e-05,
      "loss": 6.7964,
      "step": 662
    },
    {
      "epoch": 0.9233983286908078,
      "grad_norm": 4.183300495147705,
      "learning_rate": 6.48365548607163e-05,
      "loss": 6.7053,
      "step": 663
    },
    {
      "epoch": 0.924791086350975,
      "grad_norm": 4.829319000244141,
      "learning_rate": 6.482660602615122e-05,
      "loss": 6.8848,
      "step": 664
    },
    {
      "epoch": 0.9261838440111421,
      "grad_norm": 4.310740947723389,
      "learning_rate": 6.481665719158611e-05,
      "loss": 6.8218,
      "step": 665
    },
    {
      "epoch": 0.9275766016713092,
      "grad_norm": 4.705667972564697,
      "learning_rate": 6.480670835702103e-05,
      "loss": 6.938,
      "step": 666
    },
    {
      "epoch": 0.9289693593314763,
      "grad_norm": 5.496898651123047,
      "learning_rate": 6.479675952245594e-05,
      "loss": 6.8801,
      "step": 667
    },
    {
      "epoch": 0.9303621169916435,
      "grad_norm": 4.279628753662109,
      "learning_rate": 6.478681068789084e-05,
      "loss": 6.7455,
      "step": 668
    },
    {
      "epoch": 0.9317548746518106,
      "grad_norm": 5.3207244873046875,
      "learning_rate": 6.477686185332574e-05,
      "loss": 6.8616,
      "step": 669
    },
    {
      "epoch": 0.9331476323119777,
      "grad_norm": 4.7314934730529785,
      "learning_rate": 6.476691301876065e-05,
      "loss": 6.5074,
      "step": 670
    },
    {
      "epoch": 0.9345403899721448,
      "grad_norm": 6.029335021972656,
      "learning_rate": 6.475696418419555e-05,
      "loss": 6.8448,
      "step": 671
    },
    {
      "epoch": 0.935933147632312,
      "grad_norm": 4.07269287109375,
      "learning_rate": 6.474701534963047e-05,
      "loss": 6.6289,
      "step": 672
    },
    {
      "epoch": 0.9373259052924791,
      "grad_norm": 4.37167501449585,
      "learning_rate": 6.473706651506537e-05,
      "loss": 6.3312,
      "step": 673
    },
    {
      "epoch": 0.9387186629526463,
      "grad_norm": 4.4684367179870605,
      "learning_rate": 6.472711768050028e-05,
      "loss": 6.6532,
      "step": 674
    },
    {
      "epoch": 0.9401114206128134,
      "grad_norm": 4.46663761138916,
      "learning_rate": 6.471716884593518e-05,
      "loss": 6.7704,
      "step": 675
    },
    {
      "epoch": 0.9415041782729805,
      "grad_norm": 4.736171722412109,
      "learning_rate": 6.470722001137009e-05,
      "loss": 6.7296,
      "step": 676
    },
    {
      "epoch": 0.9428969359331476,
      "grad_norm": 4.486076831817627,
      "learning_rate": 6.4697271176805e-05,
      "loss": 6.6435,
      "step": 677
    },
    {
      "epoch": 0.9442896935933147,
      "grad_norm": 4.807506561279297,
      "learning_rate": 6.468732234223991e-05,
      "loss": 6.6223,
      "step": 678
    },
    {
      "epoch": 0.9456824512534819,
      "grad_norm": 6.329336166381836,
      "learning_rate": 6.467737350767481e-05,
      "loss": 6.631,
      "step": 679
    },
    {
      "epoch": 0.947075208913649,
      "grad_norm": 5.18367338180542,
      "learning_rate": 6.466742467310972e-05,
      "loss": 6.7117,
      "step": 680
    },
    {
      "epoch": 0.9484679665738162,
      "grad_norm": 5.845064640045166,
      "learning_rate": 6.465747583854462e-05,
      "loss": 7.154,
      "step": 681
    },
    {
      "epoch": 0.9498607242339833,
      "grad_norm": 4.247320175170898,
      "learning_rate": 6.464752700397952e-05,
      "loss": 6.5966,
      "step": 682
    },
    {
      "epoch": 0.9512534818941504,
      "grad_norm": 4.217019557952881,
      "learning_rate": 6.463757816941444e-05,
      "loss": 6.8241,
      "step": 683
    },
    {
      "epoch": 0.9526462395543176,
      "grad_norm": 4.208700180053711,
      "learning_rate": 6.462762933484933e-05,
      "loss": 7.0147,
      "step": 684
    },
    {
      "epoch": 0.9540389972144847,
      "grad_norm": 3.9506630897521973,
      "learning_rate": 6.461768050028425e-05,
      "loss": 6.1799,
      "step": 685
    },
    {
      "epoch": 0.9554317548746518,
      "grad_norm": 4.633251190185547,
      "learning_rate": 6.460773166571915e-05,
      "loss": 6.586,
      "step": 686
    },
    {
      "epoch": 0.9568245125348189,
      "grad_norm": 4.501231670379639,
      "learning_rate": 6.459778283115406e-05,
      "loss": 6.69,
      "step": 687
    },
    {
      "epoch": 0.958217270194986,
      "grad_norm": 4.209427356719971,
      "learning_rate": 6.458783399658896e-05,
      "loss": 6.65,
      "step": 688
    },
    {
      "epoch": 0.9596100278551533,
      "grad_norm": 3.9768435955047607,
      "learning_rate": 6.457788516202387e-05,
      "loss": 6.829,
      "step": 689
    },
    {
      "epoch": 0.9610027855153204,
      "grad_norm": 4.2762532234191895,
      "learning_rate": 6.456793632745877e-05,
      "loss": 6.7769,
      "step": 690
    },
    {
      "epoch": 0.9623955431754875,
      "grad_norm": 4.354592800140381,
      "learning_rate": 6.455798749289369e-05,
      "loss": 6.8221,
      "step": 691
    },
    {
      "epoch": 0.9637883008356546,
      "grad_norm": 4.69321346282959,
      "learning_rate": 6.454803865832859e-05,
      "loss": 6.6603,
      "step": 692
    },
    {
      "epoch": 0.9651810584958217,
      "grad_norm": 4.885592937469482,
      "learning_rate": 6.45380898237635e-05,
      "loss": 6.8437,
      "step": 693
    },
    {
      "epoch": 0.9665738161559888,
      "grad_norm": 4.524178981781006,
      "learning_rate": 6.45281409891984e-05,
      "loss": 6.8486,
      "step": 694
    },
    {
      "epoch": 0.967966573816156,
      "grad_norm": 5.346595287322998,
      "learning_rate": 6.45181921546333e-05,
      "loss": 6.6868,
      "step": 695
    },
    {
      "epoch": 0.9693593314763231,
      "grad_norm": 5.525434970855713,
      "learning_rate": 6.450824332006821e-05,
      "loss": 6.5876,
      "step": 696
    },
    {
      "epoch": 0.9707520891364902,
      "grad_norm": 5.755575180053711,
      "learning_rate": 6.449829448550313e-05,
      "loss": 6.8912,
      "step": 697
    },
    {
      "epoch": 0.9721448467966574,
      "grad_norm": 4.54886531829834,
      "learning_rate": 6.448834565093803e-05,
      "loss": 6.8111,
      "step": 698
    },
    {
      "epoch": 0.9735376044568245,
      "grad_norm": 4.352001667022705,
      "learning_rate": 6.447839681637293e-05,
      "loss": 6.6811,
      "step": 699
    },
    {
      "epoch": 0.9749303621169917,
      "grad_norm": 4.668796062469482,
      "learning_rate": 6.446844798180784e-05,
      "loss": 6.5804,
      "step": 700
    },
    {
      "epoch": 0.9763231197771588,
      "grad_norm": 4.026281356811523,
      "learning_rate": 6.445849914724274e-05,
      "loss": 6.8115,
      "step": 701
    },
    {
      "epoch": 0.9777158774373259,
      "grad_norm": 4.048024654388428,
      "learning_rate": 6.444855031267766e-05,
      "loss": 7.0243,
      "step": 702
    },
    {
      "epoch": 0.979108635097493,
      "grad_norm": 4.501350402832031,
      "learning_rate": 6.443860147811256e-05,
      "loss": 6.6124,
      "step": 703
    },
    {
      "epoch": 0.9805013927576601,
      "grad_norm": 3.951732635498047,
      "learning_rate": 6.442865264354745e-05,
      "loss": 6.655,
      "step": 704
    },
    {
      "epoch": 0.9818941504178273,
      "grad_norm": 4.734480857849121,
      "learning_rate": 6.441870380898237e-05,
      "loss": 6.8814,
      "step": 705
    },
    {
      "epoch": 0.9832869080779945,
      "grad_norm": 5.22509765625,
      "learning_rate": 6.440875497441728e-05,
      "loss": 6.7328,
      "step": 706
    },
    {
      "epoch": 0.9846796657381616,
      "grad_norm": 4.631306171417236,
      "learning_rate": 6.439880613985218e-05,
      "loss": 6.925,
      "step": 707
    },
    {
      "epoch": 0.9860724233983287,
      "grad_norm": 4.773249626159668,
      "learning_rate": 6.43888573052871e-05,
      "loss": 6.3109,
      "step": 708
    },
    {
      "epoch": 0.9874651810584958,
      "grad_norm": 6.017326354980469,
      "learning_rate": 6.437890847072199e-05,
      "loss": 6.6903,
      "step": 709
    },
    {
      "epoch": 0.9888579387186629,
      "grad_norm": 4.226189136505127,
      "learning_rate": 6.43689596361569e-05,
      "loss": 6.7982,
      "step": 710
    },
    {
      "epoch": 0.9902506963788301,
      "grad_norm": 5.146072864532471,
      "learning_rate": 6.435901080159181e-05,
      "loss": 6.6298,
      "step": 711
    },
    {
      "epoch": 0.9916434540389972,
      "grad_norm": 4.559305191040039,
      "learning_rate": 6.434906196702671e-05,
      "loss": 6.7671,
      "step": 712
    },
    {
      "epoch": 0.9930362116991643,
      "grad_norm": 4.815158367156982,
      "learning_rate": 6.433911313246162e-05,
      "loss": 7.0803,
      "step": 713
    },
    {
      "epoch": 0.9944289693593314,
      "grad_norm": 4.042872905731201,
      "learning_rate": 6.432916429789652e-05,
      "loss": 6.8644,
      "step": 714
    },
    {
      "epoch": 0.9958217270194986,
      "grad_norm": 5.020940780639648,
      "learning_rate": 6.431921546333143e-05,
      "loss": 6.006,
      "step": 715
    },
    {
      "epoch": 0.9972144846796658,
      "grad_norm": 4.264742374420166,
      "learning_rate": 6.430926662876634e-05,
      "loss": 6.5417,
      "step": 716
    },
    {
      "epoch": 0.9986072423398329,
      "grad_norm": 4.906822681427002,
      "learning_rate": 6.429931779420125e-05,
      "loss": 6.5299,
      "step": 717
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.636160373687744,
      "learning_rate": 6.428936895963615e-05,
      "loss": 7.2965,
      "step": 718
    },
    {
      "epoch": 1.001392757660167,
      "grad_norm": 4.3554911613464355,
      "learning_rate": 6.427942012507106e-05,
      "loss": 6.2307,
      "step": 719
    },
    {
      "epoch": 1.0027855153203342,
      "grad_norm": 4.056817054748535,
      "learning_rate": 6.426947129050596e-05,
      "loss": 6.0645,
      "step": 720
    },
    {
      "epoch": 1.0041782729805013,
      "grad_norm": 4.727801322937012,
      "learning_rate": 6.425952245594086e-05,
      "loss": 6.7937,
      "step": 721
    },
    {
      "epoch": 1.0055710306406684,
      "grad_norm": 4.018832683563232,
      "learning_rate": 6.424957362137578e-05,
      "loss": 6.2987,
      "step": 722
    },
    {
      "epoch": 1.0069637883008355,
      "grad_norm": 4.134768962860107,
      "learning_rate": 6.423962478681069e-05,
      "loss": 6.3883,
      "step": 723
    },
    {
      "epoch": 1.0083565459610029,
      "grad_norm": 4.1882710456848145,
      "learning_rate": 6.422967595224559e-05,
      "loss": 6.0631,
      "step": 724
    },
    {
      "epoch": 1.00974930362117,
      "grad_norm": 4.130897045135498,
      "learning_rate": 6.42197271176805e-05,
      "loss": 6.6044,
      "step": 725
    },
    {
      "epoch": 1.011142061281337,
      "grad_norm": 4.134621620178223,
      "learning_rate": 6.42097782831154e-05,
      "loss": 6.4306,
      "step": 726
    },
    {
      "epoch": 1.0125348189415042,
      "grad_norm": 4.082250595092773,
      "learning_rate": 6.41998294485503e-05,
      "loss": 6.7674,
      "step": 727
    },
    {
      "epoch": 1.0139275766016713,
      "grad_norm": 4.204432487487793,
      "learning_rate": 6.418988061398521e-05,
      "loss": 5.7767,
      "step": 728
    },
    {
      "epoch": 1.0153203342618384,
      "grad_norm": 3.8960046768188477,
      "learning_rate": 6.417993177942011e-05,
      "loss": 6.7049,
      "step": 729
    },
    {
      "epoch": 1.0167130919220055,
      "grad_norm": 4.618951320648193,
      "learning_rate": 6.416998294485503e-05,
      "loss": 6.3384,
      "step": 730
    },
    {
      "epoch": 1.0181058495821727,
      "grad_norm": 4.198172092437744,
      "learning_rate": 6.416003411028993e-05,
      "loss": 6.2329,
      "step": 731
    },
    {
      "epoch": 1.0194986072423398,
      "grad_norm": 4.364832401275635,
      "learning_rate": 6.415008527572484e-05,
      "loss": 6.6683,
      "step": 732
    },
    {
      "epoch": 1.0208913649025069,
      "grad_norm": 4.310018062591553,
      "learning_rate": 6.414013644115974e-05,
      "loss": 6.4213,
      "step": 733
    },
    {
      "epoch": 1.0222841225626742,
      "grad_norm": 4.494109153747559,
      "learning_rate": 6.413018760659464e-05,
      "loss": 6.8315,
      "step": 734
    },
    {
      "epoch": 1.0236768802228413,
      "grad_norm": 4.4797892570495605,
      "learning_rate": 6.412023877202956e-05,
      "loss": 6.0234,
      "step": 735
    },
    {
      "epoch": 1.0250696378830084,
      "grad_norm": 4.1381964683532715,
      "learning_rate": 6.411028993746447e-05,
      "loss": 6.4072,
      "step": 736
    },
    {
      "epoch": 1.0264623955431755,
      "grad_norm": 3.576007604598999,
      "learning_rate": 6.410034110289937e-05,
      "loss": 5.8613,
      "step": 737
    },
    {
      "epoch": 1.0278551532033426,
      "grad_norm": 4.08755350112915,
      "learning_rate": 6.409039226833427e-05,
      "loss": 5.5571,
      "step": 738
    },
    {
      "epoch": 1.0292479108635098,
      "grad_norm": 5.901718616485596,
      "learning_rate": 6.408044343376918e-05,
      "loss": 6.4184,
      "step": 739
    },
    {
      "epoch": 1.0306406685236769,
      "grad_norm": 3.864762306213379,
      "learning_rate": 6.407049459920408e-05,
      "loss": 5.9471,
      "step": 740
    },
    {
      "epoch": 1.032033426183844,
      "grad_norm": 4.546471118927002,
      "learning_rate": 6.4060545764639e-05,
      "loss": 6.2533,
      "step": 741
    },
    {
      "epoch": 1.033426183844011,
      "grad_norm": 3.783642530441284,
      "learning_rate": 6.40505969300739e-05,
      "loss": 6.6279,
      "step": 742
    },
    {
      "epoch": 1.0348189415041782,
      "grad_norm": 4.507882118225098,
      "learning_rate": 6.404064809550881e-05,
      "loss": 6.4094,
      "step": 743
    },
    {
      "epoch": 1.0362116991643453,
      "grad_norm": 4.0628790855407715,
      "learning_rate": 6.403069926094371e-05,
      "loss": 6.7273,
      "step": 744
    },
    {
      "epoch": 1.0376044568245126,
      "grad_norm": 4.369492053985596,
      "learning_rate": 6.402075042637862e-05,
      "loss": 6.3731,
      "step": 745
    },
    {
      "epoch": 1.0389972144846797,
      "grad_norm": 3.927105188369751,
      "learning_rate": 6.401080159181352e-05,
      "loss": 6.0893,
      "step": 746
    },
    {
      "epoch": 1.0403899721448469,
      "grad_norm": 4.238670825958252,
      "learning_rate": 6.400085275724844e-05,
      "loss": 6.2224,
      "step": 747
    },
    {
      "epoch": 1.041782729805014,
      "grad_norm": 4.1663103103637695,
      "learning_rate": 6.399090392268333e-05,
      "loss": 6.3216,
      "step": 748
    },
    {
      "epoch": 1.043175487465181,
      "grad_norm": 5.088364601135254,
      "learning_rate": 6.398095508811825e-05,
      "loss": 5.7618,
      "step": 749
    },
    {
      "epoch": 1.0445682451253482,
      "grad_norm": 4.116178035736084,
      "learning_rate": 6.397100625355315e-05,
      "loss": 6.4555,
      "step": 750
    },
    {
      "epoch": 1.0459610027855153,
      "grad_norm": 3.701273202896118,
      "learning_rate": 6.396105741898805e-05,
      "loss": 6.2116,
      "step": 751
    },
    {
      "epoch": 1.0473537604456824,
      "grad_norm": 4.711391925811768,
      "learning_rate": 6.395110858442296e-05,
      "loss": 6.5953,
      "step": 752
    },
    {
      "epoch": 1.0487465181058495,
      "grad_norm": 4.637911319732666,
      "learning_rate": 6.394115974985786e-05,
      "loss": 6.6138,
      "step": 753
    },
    {
      "epoch": 1.0501392757660166,
      "grad_norm": 4.483766555786133,
      "learning_rate": 6.393121091529277e-05,
      "loss": 6.5945,
      "step": 754
    },
    {
      "epoch": 1.051532033426184,
      "grad_norm": 3.9736287593841553,
      "learning_rate": 6.392126208072768e-05,
      "loss": 6.7025,
      "step": 755
    },
    {
      "epoch": 1.052924791086351,
      "grad_norm": 4.410802841186523,
      "learning_rate": 6.391131324616259e-05,
      "loss": 6.4296,
      "step": 756
    },
    {
      "epoch": 1.0543175487465182,
      "grad_norm": 4.418691158294678,
      "learning_rate": 6.390136441159749e-05,
      "loss": 6.5703,
      "step": 757
    },
    {
      "epoch": 1.0557103064066853,
      "grad_norm": 4.40813684463501,
      "learning_rate": 6.38914155770324e-05,
      "loss": 6.2693,
      "step": 758
    },
    {
      "epoch": 1.0571030640668524,
      "grad_norm": 3.891906499862671,
      "learning_rate": 6.38814667424673e-05,
      "loss": 6.3551,
      "step": 759
    },
    {
      "epoch": 1.0584958217270195,
      "grad_norm": 4.650675296783447,
      "learning_rate": 6.387151790790222e-05,
      "loss": 6.7067,
      "step": 760
    },
    {
      "epoch": 1.0598885793871866,
      "grad_norm": 4.566289901733398,
      "learning_rate": 6.386156907333712e-05,
      "loss": 6.4354,
      "step": 761
    },
    {
      "epoch": 1.0612813370473537,
      "grad_norm": 4.452826023101807,
      "learning_rate": 6.385162023877203e-05,
      "loss": 6.0533,
      "step": 762
    },
    {
      "epoch": 1.0626740947075208,
      "grad_norm": 3.898571491241455,
      "learning_rate": 6.384167140420693e-05,
      "loss": 6.054,
      "step": 763
    },
    {
      "epoch": 1.064066852367688,
      "grad_norm": 3.97408127784729,
      "learning_rate": 6.383172256964183e-05,
      "loss": 6.556,
      "step": 764
    },
    {
      "epoch": 1.065459610027855,
      "grad_norm": 4.9180755615234375,
      "learning_rate": 6.382177373507674e-05,
      "loss": 5.8876,
      "step": 765
    },
    {
      "epoch": 1.0668523676880224,
      "grad_norm": 4.586277484893799,
      "learning_rate": 6.381182490051166e-05,
      "loss": 6.3013,
      "step": 766
    },
    {
      "epoch": 1.0682451253481895,
      "grad_norm": 3.9467272758483887,
      "learning_rate": 6.380187606594655e-05,
      "loss": 6.6091,
      "step": 767
    },
    {
      "epoch": 1.0696378830083566,
      "grad_norm": 3.837667226791382,
      "learning_rate": 6.379192723138146e-05,
      "loss": 6.3902,
      "step": 768
    },
    {
      "epoch": 1.0710306406685237,
      "grad_norm": 4.07935905456543,
      "learning_rate": 6.378197839681637e-05,
      "loss": 6.0804,
      "step": 769
    },
    {
      "epoch": 1.0724233983286908,
      "grad_norm": 4.16187858581543,
      "learning_rate": 6.377202956225127e-05,
      "loss": 6.5067,
      "step": 770
    },
    {
      "epoch": 1.073816155988858,
      "grad_norm": 3.817171335220337,
      "learning_rate": 6.376208072768618e-05,
      "loss": 6.6519,
      "step": 771
    },
    {
      "epoch": 1.075208913649025,
      "grad_norm": 3.7888450622558594,
      "learning_rate": 6.375213189312108e-05,
      "loss": 6.3623,
      "step": 772
    },
    {
      "epoch": 1.0766016713091922,
      "grad_norm": 6.532416343688965,
      "learning_rate": 6.374218305855599e-05,
      "loss": 6.1087,
      "step": 773
    },
    {
      "epoch": 1.0779944289693593,
      "grad_norm": 3.9244799613952637,
      "learning_rate": 6.37322342239909e-05,
      "loss": 6.3221,
      "step": 774
    },
    {
      "epoch": 1.0793871866295264,
      "grad_norm": 3.6228652000427246,
      "learning_rate": 6.372228538942581e-05,
      "loss": 6.0867,
      "step": 775
    },
    {
      "epoch": 1.0807799442896937,
      "grad_norm": 4.465415954589844,
      "learning_rate": 6.371233655486071e-05,
      "loss": 6.3228,
      "step": 776
    },
    {
      "epoch": 1.0821727019498608,
      "grad_norm": 4.021289348602295,
      "learning_rate": 6.370238772029562e-05,
      "loss": 6.1475,
      "step": 777
    },
    {
      "epoch": 1.083565459610028,
      "grad_norm": 4.0907392501831055,
      "learning_rate": 6.369243888573052e-05,
      "loss": 5.8969,
      "step": 778
    },
    {
      "epoch": 1.084958217270195,
      "grad_norm": 3.709568500518799,
      "learning_rate": 6.368249005116542e-05,
      "loss": 5.8814,
      "step": 779
    },
    {
      "epoch": 1.0863509749303621,
      "grad_norm": 4.08675479888916,
      "learning_rate": 6.367254121660034e-05,
      "loss": 6.4905,
      "step": 780
    },
    {
      "epoch": 1.0877437325905293,
      "grad_norm": 4.593238830566406,
      "learning_rate": 6.366259238203524e-05,
      "loss": 6.1173,
      "step": 781
    },
    {
      "epoch": 1.0891364902506964,
      "grad_norm": 4.014584064483643,
      "learning_rate": 6.365264354747015e-05,
      "loss": 5.8148,
      "step": 782
    },
    {
      "epoch": 1.0905292479108635,
      "grad_norm": 4.625162601470947,
      "learning_rate": 6.364269471290505e-05,
      "loss": 6.3614,
      "step": 783
    },
    {
      "epoch": 1.0919220055710306,
      "grad_norm": 4.606537342071533,
      "learning_rate": 6.363274587833996e-05,
      "loss": 6.1496,
      "step": 784
    },
    {
      "epoch": 1.0933147632311977,
      "grad_norm": 4.191118240356445,
      "learning_rate": 6.362279704377487e-05,
      "loss": 6.619,
      "step": 785
    },
    {
      "epoch": 1.0947075208913648,
      "grad_norm": 3.9036128520965576,
      "learning_rate": 6.361284820920978e-05,
      "loss": 6.5763,
      "step": 786
    },
    {
      "epoch": 1.0961002785515321,
      "grad_norm": 4.702339172363281,
      "learning_rate": 6.360289937464467e-05,
      "loss": 6.4284,
      "step": 787
    },
    {
      "epoch": 1.0974930362116992,
      "grad_norm": 4.698242664337158,
      "learning_rate": 6.359295054007959e-05,
      "loss": 5.9575,
      "step": 788
    },
    {
      "epoch": 1.0988857938718664,
      "grad_norm": 4.833227157592773,
      "learning_rate": 6.358300170551449e-05,
      "loss": 6.1474,
      "step": 789
    },
    {
      "epoch": 1.1002785515320335,
      "grad_norm": 4.603786945343018,
      "learning_rate": 6.35730528709494e-05,
      "loss": 6.1101,
      "step": 790
    },
    {
      "epoch": 1.1016713091922006,
      "grad_norm": 4.136576175689697,
      "learning_rate": 6.356310403638431e-05,
      "loss": 6.2686,
      "step": 791
    },
    {
      "epoch": 1.1030640668523677,
      "grad_norm": 4.749037742614746,
      "learning_rate": 6.35531552018192e-05,
      "loss": 6.1223,
      "step": 792
    },
    {
      "epoch": 1.1044568245125348,
      "grad_norm": 3.8633317947387695,
      "learning_rate": 6.354320636725412e-05,
      "loss": 6.1911,
      "step": 793
    },
    {
      "epoch": 1.105849582172702,
      "grad_norm": 3.7554922103881836,
      "learning_rate": 6.353325753268902e-05,
      "loss": 6.0206,
      "step": 794
    },
    {
      "epoch": 1.107242339832869,
      "grad_norm": 4.667807579040527,
      "learning_rate": 6.352330869812393e-05,
      "loss": 5.9013,
      "step": 795
    },
    {
      "epoch": 1.1086350974930361,
      "grad_norm": 3.96821665763855,
      "learning_rate": 6.351335986355883e-05,
      "loss": 5.9689,
      "step": 796
    },
    {
      "epoch": 1.1100278551532035,
      "grad_norm": 3.9820969104766846,
      "learning_rate": 6.350341102899374e-05,
      "loss": 5.959,
      "step": 797
    },
    {
      "epoch": 1.1114206128133706,
      "grad_norm": 4.256950378417969,
      "learning_rate": 6.349346219442864e-05,
      "loss": 6.0054,
      "step": 798
    },
    {
      "epoch": 1.1128133704735377,
      "grad_norm": 5.839420318603516,
      "learning_rate": 6.348351335986356e-05,
      "loss": 5.7755,
      "step": 799
    },
    {
      "epoch": 1.1142061281337048,
      "grad_norm": 3.9156510829925537,
      "learning_rate": 6.347356452529846e-05,
      "loss": 6.0378,
      "step": 800
    },
    {
      "epoch": 1.115598885793872,
      "grad_norm": 3.987285614013672,
      "learning_rate": 6.346361569073337e-05,
      "loss": 6.0337,
      "step": 801
    },
    {
      "epoch": 1.116991643454039,
      "grad_norm": 4.3043293952941895,
      "learning_rate": 6.345366685616827e-05,
      "loss": 5.9843,
      "step": 802
    },
    {
      "epoch": 1.1183844011142061,
      "grad_norm": 4.419588088989258,
      "learning_rate": 6.344371802160318e-05,
      "loss": 6.3288,
      "step": 803
    },
    {
      "epoch": 1.1197771587743732,
      "grad_norm": 4.007911205291748,
      "learning_rate": 6.343376918703808e-05,
      "loss": 6.5837,
      "step": 804
    },
    {
      "epoch": 1.1211699164345403,
      "grad_norm": 4.451642036437988,
      "learning_rate": 6.3423820352473e-05,
      "loss": 5.979,
      "step": 805
    },
    {
      "epoch": 1.1225626740947074,
      "grad_norm": 3.8725087642669678,
      "learning_rate": 6.34138715179079e-05,
      "loss": 6.2178,
      "step": 806
    },
    {
      "epoch": 1.1239554317548746,
      "grad_norm": 4.166767120361328,
      "learning_rate": 6.34039226833428e-05,
      "loss": 6.4615,
      "step": 807
    },
    {
      "epoch": 1.1253481894150417,
      "grad_norm": 5.3962721824646,
      "learning_rate": 6.339397384877771e-05,
      "loss": 5.6746,
      "step": 808
    },
    {
      "epoch": 1.126740947075209,
      "grad_norm": 4.789846420288086,
      "learning_rate": 6.338402501421261e-05,
      "loss": 6.1639,
      "step": 809
    },
    {
      "epoch": 1.128133704735376,
      "grad_norm": 4.104311466217041,
      "learning_rate": 6.337407617964752e-05,
      "loss": 6.1073,
      "step": 810
    },
    {
      "epoch": 1.1295264623955432,
      "grad_norm": 3.9826443195343018,
      "learning_rate": 6.336412734508242e-05,
      "loss": 5.5452,
      "step": 811
    },
    {
      "epoch": 1.1309192200557103,
      "grad_norm": 4.094364643096924,
      "learning_rate": 6.335417851051733e-05,
      "loss": 6.1048,
      "step": 812
    },
    {
      "epoch": 1.1323119777158774,
      "grad_norm": 4.958089828491211,
      "learning_rate": 6.334422967595224e-05,
      "loss": 5.8935,
      "step": 813
    },
    {
      "epoch": 1.1337047353760445,
      "grad_norm": 4.460774898529053,
      "learning_rate": 6.333428084138715e-05,
      "loss": 6.2401,
      "step": 814
    },
    {
      "epoch": 1.1350974930362117,
      "grad_norm": 4.0596513748168945,
      "learning_rate": 6.332433200682205e-05,
      "loss": 6.0652,
      "step": 815
    },
    {
      "epoch": 1.1364902506963788,
      "grad_norm": 3.9710872173309326,
      "learning_rate": 6.331438317225696e-05,
      "loss": 5.9651,
      "step": 816
    },
    {
      "epoch": 1.1378830083565459,
      "grad_norm": 4.852348804473877,
      "learning_rate": 6.330443433769186e-05,
      "loss": 6.3512,
      "step": 817
    },
    {
      "epoch": 1.1392757660167132,
      "grad_norm": 4.2245683670043945,
      "learning_rate": 6.329448550312678e-05,
      "loss": 5.7914,
      "step": 818
    },
    {
      "epoch": 1.1406685236768803,
      "grad_norm": 3.978698492050171,
      "learning_rate": 6.328453666856168e-05,
      "loss": 5.9029,
      "step": 819
    },
    {
      "epoch": 1.1420612813370474,
      "grad_norm": 5.231683731079102,
      "learning_rate": 6.327458783399659e-05,
      "loss": 5.9183,
      "step": 820
    },
    {
      "epoch": 1.1434540389972145,
      "grad_norm": 4.089069366455078,
      "learning_rate": 6.326463899943149e-05,
      "loss": 6.038,
      "step": 821
    },
    {
      "epoch": 1.1448467966573816,
      "grad_norm": 5.573705196380615,
      "learning_rate": 6.32546901648664e-05,
      "loss": 6.1843,
      "step": 822
    },
    {
      "epoch": 1.1462395543175488,
      "grad_norm": 3.8310587406158447,
      "learning_rate": 6.32447413303013e-05,
      "loss": 5.6359,
      "step": 823
    },
    {
      "epoch": 1.1476323119777159,
      "grad_norm": 4.118086814880371,
      "learning_rate": 6.323479249573621e-05,
      "loss": 5.8516,
      "step": 824
    },
    {
      "epoch": 1.149025069637883,
      "grad_norm": 4.978765964508057,
      "learning_rate": 6.322484366117112e-05,
      "loss": 6.0637,
      "step": 825
    },
    {
      "epoch": 1.15041782729805,
      "grad_norm": 3.9830269813537598,
      "learning_rate": 6.321489482660602e-05,
      "loss": 6.0359,
      "step": 826
    },
    {
      "epoch": 1.1518105849582172,
      "grad_norm": 4.337949752807617,
      "learning_rate": 6.320494599204093e-05,
      "loss": 5.6737,
      "step": 827
    },
    {
      "epoch": 1.1532033426183843,
      "grad_norm": 4.050119876861572,
      "learning_rate": 6.319499715747583e-05,
      "loss": 5.9892,
      "step": 828
    },
    {
      "epoch": 1.1545961002785514,
      "grad_norm": 4.22210693359375,
      "learning_rate": 6.318504832291074e-05,
      "loss": 6.4008,
      "step": 829
    },
    {
      "epoch": 1.1559888579387188,
      "grad_norm": 3.744811534881592,
      "learning_rate": 6.317509948834565e-05,
      "loss": 5.7396,
      "step": 830
    },
    {
      "epoch": 1.1573816155988859,
      "grad_norm": 4.1426262855529785,
      "learning_rate": 6.316515065378054e-05,
      "loss": 6.0505,
      "step": 831
    },
    {
      "epoch": 1.158774373259053,
      "grad_norm": 4.062531471252441,
      "learning_rate": 6.315520181921546e-05,
      "loss": 6.2989,
      "step": 832
    },
    {
      "epoch": 1.16016713091922,
      "grad_norm": 4.024596214294434,
      "learning_rate": 6.314525298465037e-05,
      "loss": 5.9928,
      "step": 833
    },
    {
      "epoch": 1.1615598885793872,
      "grad_norm": 4.976029872894287,
      "learning_rate": 6.313530415008527e-05,
      "loss": 6.171,
      "step": 834
    },
    {
      "epoch": 1.1629526462395543,
      "grad_norm": 4.145542144775391,
      "learning_rate": 6.312535531552017e-05,
      "loss": 6.0952,
      "step": 835
    },
    {
      "epoch": 1.1643454038997214,
      "grad_norm": 4.032221794128418,
      "learning_rate": 6.311540648095508e-05,
      "loss": 6.1791,
      "step": 836
    },
    {
      "epoch": 1.1657381615598885,
      "grad_norm": 4.050756454467773,
      "learning_rate": 6.310545764638998e-05,
      "loss": 6.4111,
      "step": 837
    },
    {
      "epoch": 1.1671309192200556,
      "grad_norm": 5.001420021057129,
      "learning_rate": 6.30955088118249e-05,
      "loss": 5.6097,
      "step": 838
    },
    {
      "epoch": 1.168523676880223,
      "grad_norm": 6.472048282623291,
      "learning_rate": 6.30855599772598e-05,
      "loss": 6.3765,
      "step": 839
    },
    {
      "epoch": 1.16991643454039,
      "grad_norm": 4.391728401184082,
      "learning_rate": 6.307561114269471e-05,
      "loss": 6.5625,
      "step": 840
    },
    {
      "epoch": 1.1713091922005572,
      "grad_norm": 4.546194553375244,
      "learning_rate": 6.306566230812961e-05,
      "loss": 5.5986,
      "step": 841
    },
    {
      "epoch": 1.1727019498607243,
      "grad_norm": 5.060868740081787,
      "learning_rate": 6.305571347356452e-05,
      "loss": 5.7148,
      "step": 842
    },
    {
      "epoch": 1.1740947075208914,
      "grad_norm": 4.985201358795166,
      "learning_rate": 6.304576463899943e-05,
      "loss": 5.9249,
      "step": 843
    },
    {
      "epoch": 1.1754874651810585,
      "grad_norm": 4.410891532897949,
      "learning_rate": 6.303581580443434e-05,
      "loss": 6.3042,
      "step": 844
    },
    {
      "epoch": 1.1768802228412256,
      "grad_norm": 3.9722976684570312,
      "learning_rate": 6.302586696986924e-05,
      "loss": 6.1316,
      "step": 845
    },
    {
      "epoch": 1.1782729805013927,
      "grad_norm": 4.459285259246826,
      "learning_rate": 6.301591813530415e-05,
      "loss": 6.1813,
      "step": 846
    },
    {
      "epoch": 1.1796657381615598,
      "grad_norm": 4.7457075119018555,
      "learning_rate": 6.300596930073905e-05,
      "loss": 5.5983,
      "step": 847
    },
    {
      "epoch": 1.181058495821727,
      "grad_norm": 4.224425792694092,
      "learning_rate": 6.299602046617395e-05,
      "loss": 5.9078,
      "step": 848
    },
    {
      "epoch": 1.182451253481894,
      "grad_norm": 4.264475345611572,
      "learning_rate": 6.298607163160887e-05,
      "loss": 5.9993,
      "step": 849
    },
    {
      "epoch": 1.1838440111420612,
      "grad_norm": 4.388229846954346,
      "learning_rate": 6.297612279704378e-05,
      "loss": 5.8776,
      "step": 850
    },
    {
      "epoch": 1.1852367688022285,
      "grad_norm": 4.292633056640625,
      "learning_rate": 6.296617396247868e-05,
      "loss": 5.742,
      "step": 851
    },
    {
      "epoch": 1.1866295264623956,
      "grad_norm": 3.895598888397217,
      "learning_rate": 6.295622512791358e-05,
      "loss": 5.6649,
      "step": 852
    },
    {
      "epoch": 1.1880222841225627,
      "grad_norm": 4.261716842651367,
      "learning_rate": 6.294627629334849e-05,
      "loss": 5.9047,
      "step": 853
    },
    {
      "epoch": 1.1894150417827298,
      "grad_norm": 4.003418445587158,
      "learning_rate": 6.293632745878339e-05,
      "loss": 5.6362,
      "step": 854
    },
    {
      "epoch": 1.190807799442897,
      "grad_norm": 4.029558181762695,
      "learning_rate": 6.29263786242183e-05,
      "loss": 5.9837,
      "step": 855
    },
    {
      "epoch": 1.192200557103064,
      "grad_norm": 4.504885673522949,
      "learning_rate": 6.29164297896532e-05,
      "loss": 6.1724,
      "step": 856
    },
    {
      "epoch": 1.1935933147632312,
      "grad_norm": 4.499607563018799,
      "learning_rate": 6.290648095508812e-05,
      "loss": 5.9879,
      "step": 857
    },
    {
      "epoch": 1.1949860724233983,
      "grad_norm": 4.0793585777282715,
      "learning_rate": 6.289653212052302e-05,
      "loss": 5.8677,
      "step": 858
    },
    {
      "epoch": 1.1963788300835654,
      "grad_norm": 4.758726596832275,
      "learning_rate": 6.288658328595793e-05,
      "loss": 5.6511,
      "step": 859
    },
    {
      "epoch": 1.1977715877437327,
      "grad_norm": 6.490820407867432,
      "learning_rate": 6.287663445139283e-05,
      "loss": 6.3513,
      "step": 860
    },
    {
      "epoch": 1.1991643454038998,
      "grad_norm": 4.483609199523926,
      "learning_rate": 6.286668561682773e-05,
      "loss": 5.9408,
      "step": 861
    },
    {
      "epoch": 1.200557103064067,
      "grad_norm": 3.9243340492248535,
      "learning_rate": 6.285673678226264e-05,
      "loss": 5.7365,
      "step": 862
    },
    {
      "epoch": 1.201949860724234,
      "grad_norm": 5.145392417907715,
      "learning_rate": 6.284678794769756e-05,
      "loss": 6.0241,
      "step": 863
    },
    {
      "epoch": 1.2033426183844012,
      "grad_norm": 5.115814208984375,
      "learning_rate": 6.283683911313246e-05,
      "loss": 6.0067,
      "step": 864
    },
    {
      "epoch": 1.2047353760445683,
      "grad_norm": 4.754467964172363,
      "learning_rate": 6.282689027856736e-05,
      "loss": 6.2052,
      "step": 865
    },
    {
      "epoch": 1.2061281337047354,
      "grad_norm": 4.245777606964111,
      "learning_rate": 6.281694144400227e-05,
      "loss": 6.0602,
      "step": 866
    },
    {
      "epoch": 1.2075208913649025,
      "grad_norm": 4.805239677429199,
      "learning_rate": 6.280699260943717e-05,
      "loss": 6.0325,
      "step": 867
    },
    {
      "epoch": 1.2089136490250696,
      "grad_norm": 4.5950236320495605,
      "learning_rate": 6.279704377487208e-05,
      "loss": 5.8024,
      "step": 868
    },
    {
      "epoch": 1.2103064066852367,
      "grad_norm": 4.148370265960693,
      "learning_rate": 6.278709494030699e-05,
      "loss": 5.753,
      "step": 869
    },
    {
      "epoch": 1.2116991643454038,
      "grad_norm": 4.147583484649658,
      "learning_rate": 6.277714610574188e-05,
      "loss": 6.1212,
      "step": 870
    },
    {
      "epoch": 1.213091922005571,
      "grad_norm": 4.20698356628418,
      "learning_rate": 6.27671972711768e-05,
      "loss": 6.0757,
      "step": 871
    },
    {
      "epoch": 1.2144846796657383,
      "grad_norm": 4.499080181121826,
      "learning_rate": 6.27572484366117e-05,
      "loss": 6.1316,
      "step": 872
    },
    {
      "epoch": 1.2158774373259054,
      "grad_norm": 4.6047868728637695,
      "learning_rate": 6.274729960204661e-05,
      "loss": 5.973,
      "step": 873
    },
    {
      "epoch": 1.2172701949860725,
      "grad_norm": 5.152363300323486,
      "learning_rate": 6.273735076748153e-05,
      "loss": 5.8796,
      "step": 874
    },
    {
      "epoch": 1.2186629526462396,
      "grad_norm": 5.067062854766846,
      "learning_rate": 6.272740193291642e-05,
      "loss": 6.1757,
      "step": 875
    },
    {
      "epoch": 1.2200557103064067,
      "grad_norm": 3.904432535171509,
      "learning_rate": 6.271745309835134e-05,
      "loss": 6.1742,
      "step": 876
    },
    {
      "epoch": 1.2214484679665738,
      "grad_norm": 4.014632701873779,
      "learning_rate": 6.270750426378624e-05,
      "loss": 5.6793,
      "step": 877
    },
    {
      "epoch": 1.222841225626741,
      "grad_norm": 3.9907188415527344,
      "learning_rate": 6.269755542922114e-05,
      "loss": 6.129,
      "step": 878
    },
    {
      "epoch": 1.224233983286908,
      "grad_norm": 4.22652530670166,
      "learning_rate": 6.268760659465605e-05,
      "loss": 5.8635,
      "step": 879
    },
    {
      "epoch": 1.2256267409470751,
      "grad_norm": 4.706326007843018,
      "learning_rate": 6.267765776009095e-05,
      "loss": 6.4357,
      "step": 880
    },
    {
      "epoch": 1.2270194986072422,
      "grad_norm": 4.264747619628906,
      "learning_rate": 6.266770892552586e-05,
      "loss": 5.7687,
      "step": 881
    },
    {
      "epoch": 1.2284122562674096,
      "grad_norm": 4.919692516326904,
      "learning_rate": 6.265776009096077e-05,
      "loss": 5.8591,
      "step": 882
    },
    {
      "epoch": 1.2298050139275767,
      "grad_norm": 4.682892799377441,
      "learning_rate": 6.264781125639568e-05,
      "loss": 5.3311,
      "step": 883
    },
    {
      "epoch": 1.2311977715877438,
      "grad_norm": 4.29823112487793,
      "learning_rate": 6.263786242183058e-05,
      "loss": 5.2996,
      "step": 884
    },
    {
      "epoch": 1.232590529247911,
      "grad_norm": 4.313159465789795,
      "learning_rate": 6.262791358726549e-05,
      "loss": 5.9182,
      "step": 885
    },
    {
      "epoch": 1.233983286908078,
      "grad_norm": 4.632452487945557,
      "learning_rate": 6.261796475270039e-05,
      "loss": 5.4301,
      "step": 886
    },
    {
      "epoch": 1.2353760445682451,
      "grad_norm": 4.984564781188965,
      "learning_rate": 6.26080159181353e-05,
      "loss": 6.1695,
      "step": 887
    },
    {
      "epoch": 1.2367688022284122,
      "grad_norm": 4.197230339050293,
      "learning_rate": 6.259806708357021e-05,
      "loss": 5.8395,
      "step": 888
    },
    {
      "epoch": 1.2381615598885793,
      "grad_norm": 4.51923942565918,
      "learning_rate": 6.258811824900512e-05,
      "loss": 5.9988,
      "step": 889
    },
    {
      "epoch": 1.2395543175487465,
      "grad_norm": 4.738932132720947,
      "learning_rate": 6.257816941444002e-05,
      "loss": 5.5031,
      "step": 890
    },
    {
      "epoch": 1.2409470752089136,
      "grad_norm": 4.4604082107543945,
      "learning_rate": 6.256822057987492e-05,
      "loss": 5.5689,
      "step": 891
    },
    {
      "epoch": 1.2423398328690807,
      "grad_norm": 4.486937522888184,
      "learning_rate": 6.255827174530983e-05,
      "loss": 5.9191,
      "step": 892
    },
    {
      "epoch": 1.243732590529248,
      "grad_norm": 5.033269882202148,
      "learning_rate": 6.254832291074473e-05,
      "loss": 6.1984,
      "step": 893
    },
    {
      "epoch": 1.2451253481894151,
      "grad_norm": 3.9928624629974365,
      "learning_rate": 6.253837407617965e-05,
      "loss": 5.8658,
      "step": 894
    },
    {
      "epoch": 1.2465181058495822,
      "grad_norm": 3.888197183609009,
      "learning_rate": 6.252842524161454e-05,
      "loss": 5.5798,
      "step": 895
    },
    {
      "epoch": 1.2479108635097493,
      "grad_norm": 4.839468002319336,
      "learning_rate": 6.251847640704946e-05,
      "loss": 5.523,
      "step": 896
    },
    {
      "epoch": 1.2493036211699164,
      "grad_norm": 4.236422538757324,
      "learning_rate": 6.250852757248436e-05,
      "loss": 5.6453,
      "step": 897
    },
    {
      "epoch": 1.2506963788300836,
      "grad_norm": 4.6251397132873535,
      "learning_rate": 6.249857873791927e-05,
      "loss": 5.9148,
      "step": 898
    },
    {
      "epoch": 1.2520891364902507,
      "grad_norm": 5.287298202514648,
      "learning_rate": 6.248862990335417e-05,
      "loss": 5.96,
      "step": 899
    },
    {
      "epoch": 1.2534818941504178,
      "grad_norm": 4.414423942565918,
      "learning_rate": 6.247868106878907e-05,
      "loss": 5.7607,
      "step": 900
    },
    {
      "epoch": 1.2548746518105849,
      "grad_norm": 4.138334274291992,
      "learning_rate": 6.246873223422399e-05,
      "loss": 5.8021,
      "step": 901
    },
    {
      "epoch": 1.2562674094707522,
      "grad_norm": 4.3649749755859375,
      "learning_rate": 6.24587833996589e-05,
      "loss": 6.0395,
      "step": 902
    },
    {
      "epoch": 1.2576601671309193,
      "grad_norm": 3.8956754207611084,
      "learning_rate": 6.24488345650938e-05,
      "loss": 5.4605,
      "step": 903
    },
    {
      "epoch": 1.2590529247910864,
      "grad_norm": 4.663502216339111,
      "learning_rate": 6.24388857305287e-05,
      "loss": 6.1287,
      "step": 904
    },
    {
      "epoch": 1.2604456824512535,
      "grad_norm": 3.9989981651306152,
      "learning_rate": 6.242893689596361e-05,
      "loss": 5.9307,
      "step": 905
    },
    {
      "epoch": 1.2618384401114207,
      "grad_norm": 4.225207328796387,
      "learning_rate": 6.241898806139851e-05,
      "loss": 6.0563,
      "step": 906
    },
    {
      "epoch": 1.2632311977715878,
      "grad_norm": 4.495071887969971,
      "learning_rate": 6.240903922683343e-05,
      "loss": 5.7611,
      "step": 907
    },
    {
      "epoch": 1.2646239554317549,
      "grad_norm": 4.2864508628845215,
      "learning_rate": 6.239909039226833e-05,
      "loss": 5.8417,
      "step": 908
    },
    {
      "epoch": 1.266016713091922,
      "grad_norm": 4.327315330505371,
      "learning_rate": 6.238914155770324e-05,
      "loss": 5.904,
      "step": 909
    },
    {
      "epoch": 1.267409470752089,
      "grad_norm": 4.177762031555176,
      "learning_rate": 6.237919272313814e-05,
      "loss": 5.8393,
      "step": 910
    },
    {
      "epoch": 1.2688022284122562,
      "grad_norm": 4.884469509124756,
      "learning_rate": 6.236924388857305e-05,
      "loss": 6.0954,
      "step": 911
    },
    {
      "epoch": 1.2701949860724233,
      "grad_norm": 4.177820205688477,
      "learning_rate": 6.235929505400795e-05,
      "loss": 5.5047,
      "step": 912
    },
    {
      "epoch": 1.2715877437325904,
      "grad_norm": 4.735227584838867,
      "learning_rate": 6.234934621944287e-05,
      "loss": 5.7605,
      "step": 913
    },
    {
      "epoch": 1.2729805013927575,
      "grad_norm": 4.475788593292236,
      "learning_rate": 6.233939738487776e-05,
      "loss": 5.4821,
      "step": 914
    },
    {
      "epoch": 1.2743732590529249,
      "grad_norm": 4.623432159423828,
      "learning_rate": 6.232944855031268e-05,
      "loss": 6.0002,
      "step": 915
    },
    {
      "epoch": 1.275766016713092,
      "grad_norm": 4.811022758483887,
      "learning_rate": 6.231949971574758e-05,
      "loss": 5.6893,
      "step": 916
    },
    {
      "epoch": 1.277158774373259,
      "grad_norm": 4.068156719207764,
      "learning_rate": 6.230955088118248e-05,
      "loss": 6.2387,
      "step": 917
    },
    {
      "epoch": 1.2785515320334262,
      "grad_norm": 3.8421871662139893,
      "learning_rate": 6.229960204661739e-05,
      "loss": 5.6591,
      "step": 918
    },
    {
      "epoch": 1.2799442896935933,
      "grad_norm": 3.815600633621216,
      "learning_rate": 6.228965321205229e-05,
      "loss": 5.6912,
      "step": 919
    },
    {
      "epoch": 1.2813370473537604,
      "grad_norm": 4.52863883972168,
      "learning_rate": 6.22797043774872e-05,
      "loss": 6.0405,
      "step": 920
    },
    {
      "epoch": 1.2827298050139275,
      "grad_norm": 4.008754730224609,
      "learning_rate": 6.226975554292211e-05,
      "loss": 6.003,
      "step": 921
    },
    {
      "epoch": 1.2841225626740946,
      "grad_norm": 4.417988300323486,
      "learning_rate": 6.225980670835702e-05,
      "loss": 5.611,
      "step": 922
    },
    {
      "epoch": 1.285515320334262,
      "grad_norm": 3.9807190895080566,
      "learning_rate": 6.224985787379192e-05,
      "loss": 5.4935,
      "step": 923
    },
    {
      "epoch": 1.286908077994429,
      "grad_norm": 4.328763484954834,
      "learning_rate": 6.223990903922683e-05,
      "loss": 6.0448,
      "step": 924
    },
    {
      "epoch": 1.2883008356545962,
      "grad_norm": 5.232528209686279,
      "learning_rate": 6.222996020466173e-05,
      "loss": 5.8381,
      "step": 925
    },
    {
      "epoch": 1.2896935933147633,
      "grad_norm": 4.714798450469971,
      "learning_rate": 6.222001137009665e-05,
      "loss": 6.0027,
      "step": 926
    },
    {
      "epoch": 1.2910863509749304,
      "grad_norm": 4.252414226531982,
      "learning_rate": 6.221006253553155e-05,
      "loss": 6.1091,
      "step": 927
    },
    {
      "epoch": 1.2924791086350975,
      "grad_norm": 4.840965747833252,
      "learning_rate": 6.220011370096646e-05,
      "loss": 5.7015,
      "step": 928
    },
    {
      "epoch": 1.2938718662952646,
      "grad_norm": 4.268723487854004,
      "learning_rate": 6.219016486640136e-05,
      "loss": 5.6105,
      "step": 929
    },
    {
      "epoch": 1.2952646239554317,
      "grad_norm": 4.211379528045654,
      "learning_rate": 6.218021603183626e-05,
      "loss": 5.5058,
      "step": 930
    },
    {
      "epoch": 1.2966573816155988,
      "grad_norm": 4.496066093444824,
      "learning_rate": 6.217026719727117e-05,
      "loss": 5.664,
      "step": 931
    },
    {
      "epoch": 1.298050139275766,
      "grad_norm": 4.24421501159668,
      "learning_rate": 6.216031836270609e-05,
      "loss": 5.8674,
      "step": 932
    },
    {
      "epoch": 1.299442896935933,
      "grad_norm": 4.360576152801514,
      "learning_rate": 6.215036952814099e-05,
      "loss": 5.7603,
      "step": 933
    },
    {
      "epoch": 1.3008356545961002,
      "grad_norm": 4.054847717285156,
      "learning_rate": 6.21404206935759e-05,
      "loss": 5.8031,
      "step": 934
    },
    {
      "epoch": 1.3022284122562673,
      "grad_norm": 5.198960304260254,
      "learning_rate": 6.21304718590108e-05,
      "loss": 5.6797,
      "step": 935
    },
    {
      "epoch": 1.3036211699164346,
      "grad_norm": 4.035637378692627,
      "learning_rate": 6.21205230244457e-05,
      "loss": 5.6923,
      "step": 936
    },
    {
      "epoch": 1.3050139275766017,
      "grad_norm": 4.253226280212402,
      "learning_rate": 6.21105741898806e-05,
      "loss": 5.7791,
      "step": 937
    },
    {
      "epoch": 1.3064066852367688,
      "grad_norm": 4.880700588226318,
      "learning_rate": 6.210062535531551e-05,
      "loss": 5.876,
      "step": 938
    },
    {
      "epoch": 1.307799442896936,
      "grad_norm": 5.53300142288208,
      "learning_rate": 6.209067652075041e-05,
      "loss": 6.1308,
      "step": 939
    },
    {
      "epoch": 1.309192200557103,
      "grad_norm": 4.972252368927002,
      "learning_rate": 6.208072768618533e-05,
      "loss": 5.8277,
      "step": 940
    },
    {
      "epoch": 1.3105849582172702,
      "grad_norm": 5.015955448150635,
      "learning_rate": 6.207077885162024e-05,
      "loss": 5.9938,
      "step": 941
    },
    {
      "epoch": 1.3119777158774373,
      "grad_norm": 5.864644527435303,
      "learning_rate": 6.206083001705514e-05,
      "loss": 5.6609,
      "step": 942
    },
    {
      "epoch": 1.3133704735376044,
      "grad_norm": 7.571382522583008,
      "learning_rate": 6.205088118249004e-05,
      "loss": 5.5346,
      "step": 943
    },
    {
      "epoch": 1.3147632311977717,
      "grad_norm": 4.031362533569336,
      "learning_rate": 6.204093234792495e-05,
      "loss": 5.9596,
      "step": 944
    },
    {
      "epoch": 1.3161559888579388,
      "grad_norm": 4.2481608390808105,
      "learning_rate": 6.203098351335985e-05,
      "loss": 5.9164,
      "step": 945
    },
    {
      "epoch": 1.317548746518106,
      "grad_norm": 5.277892589569092,
      "learning_rate": 6.202103467879477e-05,
      "loss": 5.7101,
      "step": 946
    },
    {
      "epoch": 1.318941504178273,
      "grad_norm": 4.687514781951904,
      "learning_rate": 6.201108584422967e-05,
      "loss": 5.9427,
      "step": 947
    },
    {
      "epoch": 1.3203342618384402,
      "grad_norm": 3.968960762023926,
      "learning_rate": 6.200113700966458e-05,
      "loss": 5.7682,
      "step": 948
    },
    {
      "epoch": 1.3217270194986073,
      "grad_norm": 4.346205711364746,
      "learning_rate": 6.199118817509948e-05,
      "loss": 5.6468,
      "step": 949
    },
    {
      "epoch": 1.3231197771587744,
      "grad_norm": 4.177730560302734,
      "learning_rate": 6.198123934053439e-05,
      "loss": 5.9089,
      "step": 950
    },
    {
      "epoch": 1.3245125348189415,
      "grad_norm": 4.508565902709961,
      "learning_rate": 6.197129050596929e-05,
      "loss": 5.9606,
      "step": 951
    },
    {
      "epoch": 1.3259052924791086,
      "grad_norm": 5.372434139251709,
      "learning_rate": 6.196134167140421e-05,
      "loss": 5.928,
      "step": 952
    },
    {
      "epoch": 1.3272980501392757,
      "grad_norm": 3.947396755218506,
      "learning_rate": 6.19513928368391e-05,
      "loss": 5.9353,
      "step": 953
    },
    {
      "epoch": 1.3286908077994428,
      "grad_norm": 5.454845428466797,
      "learning_rate": 6.194144400227402e-05,
      "loss": 6.1017,
      "step": 954
    },
    {
      "epoch": 1.33008356545961,
      "grad_norm": 4.1838908195495605,
      "learning_rate": 6.193149516770892e-05,
      "loss": 5.1224,
      "step": 955
    },
    {
      "epoch": 1.331476323119777,
      "grad_norm": 4.65695333480835,
      "learning_rate": 6.192154633314382e-05,
      "loss": 6.1432,
      "step": 956
    },
    {
      "epoch": 1.3328690807799444,
      "grad_norm": 4.286771774291992,
      "learning_rate": 6.191159749857874e-05,
      "loss": 5.5884,
      "step": 957
    },
    {
      "epoch": 1.3342618384401115,
      "grad_norm": 4.350987911224365,
      "learning_rate": 6.190164866401363e-05,
      "loss": 6.1831,
      "step": 958
    },
    {
      "epoch": 1.3356545961002786,
      "grad_norm": 4.22835636138916,
      "learning_rate": 6.189169982944855e-05,
      "loss": 5.999,
      "step": 959
    },
    {
      "epoch": 1.3370473537604457,
      "grad_norm": 5.089576244354248,
      "learning_rate": 6.188175099488345e-05,
      "loss": 5.7303,
      "step": 960
    },
    {
      "epoch": 1.3384401114206128,
      "grad_norm": 4.221466541290283,
      "learning_rate": 6.187180216031836e-05,
      "loss": 5.778,
      "step": 961
    },
    {
      "epoch": 1.33983286908078,
      "grad_norm": 5.235956192016602,
      "learning_rate": 6.186185332575326e-05,
      "loss": 5.3759,
      "step": 962
    },
    {
      "epoch": 1.341225626740947,
      "grad_norm": 4.110906600952148,
      "learning_rate": 6.185190449118817e-05,
      "loss": 6.1872,
      "step": 963
    },
    {
      "epoch": 1.3426183844011141,
      "grad_norm": 6.1621994972229,
      "learning_rate": 6.184195565662307e-05,
      "loss": 5.6299,
      "step": 964
    },
    {
      "epoch": 1.3440111420612815,
      "grad_norm": 4.876462459564209,
      "learning_rate": 6.183200682205799e-05,
      "loss": 5.7035,
      "step": 965
    },
    {
      "epoch": 1.3454038997214486,
      "grad_norm": 4.035791873931885,
      "learning_rate": 6.182205798749289e-05,
      "loss": 5.9675,
      "step": 966
    },
    {
      "epoch": 1.3467966573816157,
      "grad_norm": 5.427221298217773,
      "learning_rate": 6.18121091529278e-05,
      "loss": 5.9598,
      "step": 967
    },
    {
      "epoch": 1.3481894150417828,
      "grad_norm": 4.459470748901367,
      "learning_rate": 6.18021603183627e-05,
      "loss": 6.2836,
      "step": 968
    },
    {
      "epoch": 1.34958217270195,
      "grad_norm": 4.781061172485352,
      "learning_rate": 6.17922114837976e-05,
      "loss": 5.4981,
      "step": 969
    },
    {
      "epoch": 1.350974930362117,
      "grad_norm": 4.126856327056885,
      "learning_rate": 6.178226264923251e-05,
      "loss": 5.426,
      "step": 970
    },
    {
      "epoch": 1.3523676880222841,
      "grad_norm": 5.340152740478516,
      "learning_rate": 6.177231381466743e-05,
      "loss": 5.7245,
      "step": 971
    },
    {
      "epoch": 1.3537604456824512,
      "grad_norm": 3.787040948867798,
      "learning_rate": 6.176236498010233e-05,
      "loss": 5.8324,
      "step": 972
    },
    {
      "epoch": 1.3551532033426184,
      "grad_norm": 4.234292984008789,
      "learning_rate": 6.175241614553723e-05,
      "loss": 5.6917,
      "step": 973
    },
    {
      "epoch": 1.3565459610027855,
      "grad_norm": 4.556074142456055,
      "learning_rate": 6.174246731097214e-05,
      "loss": 5.7874,
      "step": 974
    },
    {
      "epoch": 1.3579387186629526,
      "grad_norm": 5.05952262878418,
      "learning_rate": 6.173251847640704e-05,
      "loss": 5.3797,
      "step": 975
    },
    {
      "epoch": 1.3593314763231197,
      "grad_norm": 4.6962504386901855,
      "learning_rate": 6.172256964184195e-05,
      "loss": 6.0405,
      "step": 976
    },
    {
      "epoch": 1.3607242339832868,
      "grad_norm": 4.279208183288574,
      "learning_rate": 6.171262080727686e-05,
      "loss": 5.4048,
      "step": 977
    },
    {
      "epoch": 1.362116991643454,
      "grad_norm": 3.862438440322876,
      "learning_rate": 6.170267197271175e-05,
      "loss": 5.5516,
      "step": 978
    },
    {
      "epoch": 1.3635097493036212,
      "grad_norm": 4.292626857757568,
      "learning_rate": 6.169272313814667e-05,
      "loss": 5.7087,
      "step": 979
    },
    {
      "epoch": 1.3649025069637883,
      "grad_norm": 4.5584588050842285,
      "learning_rate": 6.168277430358158e-05,
      "loss": 5.8243,
      "step": 980
    },
    {
      "epoch": 1.3662952646239555,
      "grad_norm": 5.150217533111572,
      "learning_rate": 6.167282546901648e-05,
      "loss": 5.679,
      "step": 981
    },
    {
      "epoch": 1.3676880222841226,
      "grad_norm": 4.363905906677246,
      "learning_rate": 6.166287663445138e-05,
      "loss": 5.2874,
      "step": 982
    },
    {
      "epoch": 1.3690807799442897,
      "grad_norm": 4.296676158905029,
      "learning_rate": 6.165292779988629e-05,
      "loss": 5.6455,
      "step": 983
    },
    {
      "epoch": 1.3704735376044568,
      "grad_norm": 3.8754825592041016,
      "learning_rate": 6.16429789653212e-05,
      "loss": 5.6513,
      "step": 984
    },
    {
      "epoch": 1.371866295264624,
      "grad_norm": 3.8887650966644287,
      "learning_rate": 6.163303013075611e-05,
      "loss": 5.0899,
      "step": 985
    },
    {
      "epoch": 1.3732590529247912,
      "grad_norm": 3.969674825668335,
      "learning_rate": 6.162308129619101e-05,
      "loss": 5.2191,
      "step": 986
    },
    {
      "epoch": 1.3746518105849583,
      "grad_norm": 4.9046783447265625,
      "learning_rate": 6.161313246162592e-05,
      "loss": 5.2244,
      "step": 987
    },
    {
      "epoch": 1.3760445682451254,
      "grad_norm": 4.269351005554199,
      "learning_rate": 6.160318362706082e-05,
      "loss": 5.659,
      "step": 988
    },
    {
      "epoch": 1.3774373259052926,
      "grad_norm": 4.419259071350098,
      "learning_rate": 6.159323479249573e-05,
      "loss": 5.7227,
      "step": 989
    },
    {
      "epoch": 1.3788300835654597,
      "grad_norm": 3.671412944793701,
      "learning_rate": 6.158328595793064e-05,
      "loss": 5.4442,
      "step": 990
    },
    {
      "epoch": 1.3802228412256268,
      "grad_norm": 11.605569839477539,
      "learning_rate": 6.157333712336555e-05,
      "loss": 5.3918,
      "step": 991
    },
    {
      "epoch": 1.3816155988857939,
      "grad_norm": 4.40183687210083,
      "learning_rate": 6.156338828880045e-05,
      "loss": 5.4837,
      "step": 992
    },
    {
      "epoch": 1.383008356545961,
      "grad_norm": 6.407291889190674,
      "learning_rate": 6.155343945423536e-05,
      "loss": 5.7197,
      "step": 993
    },
    {
      "epoch": 1.384401114206128,
      "grad_norm": 4.451000213623047,
      "learning_rate": 6.154349061967026e-05,
      "loss": 6.1491,
      "step": 994
    },
    {
      "epoch": 1.3857938718662952,
      "grad_norm": 4.515896797180176,
      "learning_rate": 6.153354178510516e-05,
      "loss": 5.5752,
      "step": 995
    },
    {
      "epoch": 1.3871866295264623,
      "grad_norm": 5.924260139465332,
      "learning_rate": 6.152359295054008e-05,
      "loss": 5.4646,
      "step": 996
    },
    {
      "epoch": 1.3885793871866294,
      "grad_norm": 4.985195159912109,
      "learning_rate": 6.151364411597497e-05,
      "loss": 5.9752,
      "step": 997
    },
    {
      "epoch": 1.3899721448467965,
      "grad_norm": 4.6726298332214355,
      "learning_rate": 6.150369528140989e-05,
      "loss": 5.7832,
      "step": 998
    },
    {
      "epoch": 1.3913649025069637,
      "grad_norm": 4.758754730224609,
      "learning_rate": 6.14937464468448e-05,
      "loss": 6.0052,
      "step": 999
    },
    {
      "epoch": 1.392757660167131,
      "grad_norm": 4.372791767120361,
      "learning_rate": 6.14837976122797e-05,
      "loss": 5.267,
      "step": 1000
    },
    {
      "epoch": 1.394150417827298,
      "grad_norm": 4.5017828941345215,
      "learning_rate": 6.14738487777146e-05,
      "loss": 5.7571,
      "step": 1001
    },
    {
      "epoch": 1.3955431754874652,
      "grad_norm": 4.74721622467041,
      "learning_rate": 6.14638999431495e-05,
      "loss": 5.2723,
      "step": 1002
    },
    {
      "epoch": 1.3969359331476323,
      "grad_norm": 4.530693054199219,
      "learning_rate": 6.145395110858441e-05,
      "loss": 5.2973,
      "step": 1003
    },
    {
      "epoch": 1.3983286908077994,
      "grad_norm": 3.9919230937957764,
      "learning_rate": 6.144400227401933e-05,
      "loss": 5.5479,
      "step": 1004
    },
    {
      "epoch": 1.3997214484679665,
      "grad_norm": 4.236709117889404,
      "learning_rate": 6.143405343945423e-05,
      "loss": 5.4198,
      "step": 1005
    },
    {
      "epoch": 1.4011142061281336,
      "grad_norm": 4.950026988983154,
      "learning_rate": 6.142410460488914e-05,
      "loss": 5.3283,
      "step": 1006
    },
    {
      "epoch": 1.4025069637883008,
      "grad_norm": 4.193325996398926,
      "learning_rate": 6.141415577032404e-05,
      "loss": 5.4861,
      "step": 1007
    },
    {
      "epoch": 1.403899721448468,
      "grad_norm": 3.9612650871276855,
      "learning_rate": 6.140420693575894e-05,
      "loss": 5.3246,
      "step": 1008
    },
    {
      "epoch": 1.4052924791086352,
      "grad_norm": 6.121573448181152,
      "learning_rate": 6.139425810119385e-05,
      "loss": 5.6648,
      "step": 1009
    },
    {
      "epoch": 1.4066852367688023,
      "grad_norm": 4.664121150970459,
      "learning_rate": 6.138430926662877e-05,
      "loss": 5.4626,
      "step": 1010
    },
    {
      "epoch": 1.4080779944289694,
      "grad_norm": 5.541447639465332,
      "learning_rate": 6.137436043206367e-05,
      "loss": 6.0914,
      "step": 1011
    },
    {
      "epoch": 1.4094707520891365,
      "grad_norm": 5.062973976135254,
      "learning_rate": 6.136441159749857e-05,
      "loss": 5.7945,
      "step": 1012
    },
    {
      "epoch": 1.4108635097493036,
      "grad_norm": 4.804659843444824,
      "learning_rate": 6.135446276293348e-05,
      "loss": 5.6537,
      "step": 1013
    },
    {
      "epoch": 1.4122562674094707,
      "grad_norm": 5.041073322296143,
      "learning_rate": 6.134451392836838e-05,
      "loss": 5.7487,
      "step": 1014
    },
    {
      "epoch": 1.4136490250696379,
      "grad_norm": 4.3413310050964355,
      "learning_rate": 6.13345650938033e-05,
      "loss": 5.4096,
      "step": 1015
    },
    {
      "epoch": 1.415041782729805,
      "grad_norm": 5.038837432861328,
      "learning_rate": 6.13246162592382e-05,
      "loss": 5.3533,
      "step": 1016
    },
    {
      "epoch": 1.416434540389972,
      "grad_norm": 4.292027950286865,
      "learning_rate": 6.131466742467311e-05,
      "loss": 5.0028,
      "step": 1017
    },
    {
      "epoch": 1.4178272980501392,
      "grad_norm": 5.166059494018555,
      "learning_rate": 6.130471859010801e-05,
      "loss": 5.7436,
      "step": 1018
    },
    {
      "epoch": 1.4192200557103063,
      "grad_norm": 4.512571811676025,
      "learning_rate": 6.129476975554292e-05,
      "loss": 5.8252,
      "step": 1019
    },
    {
      "epoch": 1.4206128133704734,
      "grad_norm": 5.226598739624023,
      "learning_rate": 6.128482092097782e-05,
      "loss": 5.2616,
      "step": 1020
    },
    {
      "epoch": 1.4220055710306407,
      "grad_norm": 4.709335803985596,
      "learning_rate": 6.127487208641274e-05,
      "loss": 5.5318,
      "step": 1021
    },
    {
      "epoch": 1.4233983286908078,
      "grad_norm": 4.278756618499756,
      "learning_rate": 6.126492325184763e-05,
      "loss": 5.1678,
      "step": 1022
    },
    {
      "epoch": 1.424791086350975,
      "grad_norm": 3.8250346183776855,
      "learning_rate": 6.125497441728255e-05,
      "loss": 5.5314,
      "step": 1023
    },
    {
      "epoch": 1.426183844011142,
      "grad_norm": 4.649449825286865,
      "learning_rate": 6.124502558271745e-05,
      "loss": 4.8598,
      "step": 1024
    },
    {
      "epoch": 1.4275766016713092,
      "grad_norm": 4.135913372039795,
      "learning_rate": 6.123507674815235e-05,
      "loss": 5.3614,
      "step": 1025
    },
    {
      "epoch": 1.4289693593314763,
      "grad_norm": 5.673917293548584,
      "learning_rate": 6.122512791358726e-05,
      "loss": 5.6734,
      "step": 1026
    },
    {
      "epoch": 1.4303621169916434,
      "grad_norm": 4.483156681060791,
      "learning_rate": 6.121517907902216e-05,
      "loss": 5.7856,
      "step": 1027
    },
    {
      "epoch": 1.4317548746518105,
      "grad_norm": 7.711784362792969,
      "learning_rate": 6.120523024445707e-05,
      "loss": 5.2445,
      "step": 1028
    },
    {
      "epoch": 1.4331476323119778,
      "grad_norm": 4.991926670074463,
      "learning_rate": 6.119528140989198e-05,
      "loss": 5.5703,
      "step": 1029
    },
    {
      "epoch": 1.434540389972145,
      "grad_norm": 4.263853073120117,
      "learning_rate": 6.118533257532689e-05,
      "loss": 5.4425,
      "step": 1030
    },
    {
      "epoch": 1.435933147632312,
      "grad_norm": 4.021599292755127,
      "learning_rate": 6.117538374076179e-05,
      "loss": 5.3901,
      "step": 1031
    },
    {
      "epoch": 1.4373259052924792,
      "grad_norm": 4.648274898529053,
      "learning_rate": 6.11654349061967e-05,
      "loss": 5.3981,
      "step": 1032
    },
    {
      "epoch": 1.4387186629526463,
      "grad_norm": 4.237360000610352,
      "learning_rate": 6.11554860716316e-05,
      "loss": 6.0682,
      "step": 1033
    },
    {
      "epoch": 1.4401114206128134,
      "grad_norm": 4.454067707061768,
      "learning_rate": 6.11455372370665e-05,
      "loss": 5.6093,
      "step": 1034
    },
    {
      "epoch": 1.4415041782729805,
      "grad_norm": 6.402667999267578,
      "learning_rate": 6.113558840250142e-05,
      "loss": 5.6968,
      "step": 1035
    },
    {
      "epoch": 1.4428969359331476,
      "grad_norm": 4.66150426864624,
      "learning_rate": 6.112563956793631e-05,
      "loss": 6.0225,
      "step": 1036
    },
    {
      "epoch": 1.4442896935933147,
      "grad_norm": 4.917180061340332,
      "learning_rate": 6.111569073337123e-05,
      "loss": 5.4053,
      "step": 1037
    },
    {
      "epoch": 1.4456824512534818,
      "grad_norm": 4.644591808319092,
      "learning_rate": 6.110574189880613e-05,
      "loss": 5.4277,
      "step": 1038
    },
    {
      "epoch": 1.447075208913649,
      "grad_norm": 6.097512722015381,
      "learning_rate": 6.109579306424104e-05,
      "loss": 5.1661,
      "step": 1039
    },
    {
      "epoch": 1.448467966573816,
      "grad_norm": 4.1244025230407715,
      "learning_rate": 6.108584422967596e-05,
      "loss": 5.6467,
      "step": 1040
    },
    {
      "epoch": 1.4498607242339832,
      "grad_norm": 4.131136417388916,
      "learning_rate": 6.107589539511085e-05,
      "loss": 5.8465,
      "step": 1041
    },
    {
      "epoch": 1.4512534818941505,
      "grad_norm": 5.074693202972412,
      "learning_rate": 6.106594656054576e-05,
      "loss": 5.3554,
      "step": 1042
    },
    {
      "epoch": 1.4526462395543176,
      "grad_norm": 6.52849006652832,
      "learning_rate": 6.105599772598067e-05,
      "loss": 5.3501,
      "step": 1043
    },
    {
      "epoch": 1.4540389972144847,
      "grad_norm": 5.787081241607666,
      "learning_rate": 6.104604889141557e-05,
      "loss": 5.4217,
      "step": 1044
    },
    {
      "epoch": 1.4554317548746518,
      "grad_norm": 4.454585552215576,
      "learning_rate": 6.103610005685048e-05,
      "loss": 5.2065,
      "step": 1045
    },
    {
      "epoch": 1.456824512534819,
      "grad_norm": 5.2344279289245605,
      "learning_rate": 6.102615122228538e-05,
      "loss": 5.732,
      "step": 1046
    },
    {
      "epoch": 1.458217270194986,
      "grad_norm": 4.843843460083008,
      "learning_rate": 6.101620238772029e-05,
      "loss": 5.6284,
      "step": 1047
    },
    {
      "epoch": 1.4596100278551531,
      "grad_norm": 4.481762409210205,
      "learning_rate": 6.1006253553155195e-05,
      "loss": 5.4837,
      "step": 1048
    },
    {
      "epoch": 1.4610027855153203,
      "grad_norm": 4.521838188171387,
      "learning_rate": 6.0996304718590106e-05,
      "loss": 5.5173,
      "step": 1049
    },
    {
      "epoch": 1.4623955431754876,
      "grad_norm": 4.370110511779785,
      "learning_rate": 6.098635588402501e-05,
      "loss": 5.377,
      "step": 1050
    },
    {
      "epoch": 1.4637883008356547,
      "grad_norm": 4.479923725128174,
      "learning_rate": 6.0976407049459914e-05,
      "loss": 5.6361,
      "step": 1051
    },
    {
      "epoch": 1.4651810584958218,
      "grad_norm": 4.193077087402344,
      "learning_rate": 6.096645821489482e-05,
      "loss": 5.5678,
      "step": 1052
    },
    {
      "epoch": 1.466573816155989,
      "grad_norm": 4.2606201171875,
      "learning_rate": 6.095650938032973e-05,
      "loss": 5.7236,
      "step": 1053
    },
    {
      "epoch": 1.467966573816156,
      "grad_norm": 4.565093994140625,
      "learning_rate": 6.094656054576463e-05,
      "loss": 5.4948,
      "step": 1054
    },
    {
      "epoch": 1.4693593314763231,
      "grad_norm": 4.295390605926514,
      "learning_rate": 6.0936611711199544e-05,
      "loss": 5.2635,
      "step": 1055
    },
    {
      "epoch": 1.4707520891364902,
      "grad_norm": 4.787730693817139,
      "learning_rate": 6.092666287663444e-05,
      "loss": 5.4875,
      "step": 1056
    },
    {
      "epoch": 1.4721448467966574,
      "grad_norm": 5.01603364944458,
      "learning_rate": 6.091671404206935e-05,
      "loss": 5.441,
      "step": 1057
    },
    {
      "epoch": 1.4735376044568245,
      "grad_norm": 7.9381513595581055,
      "learning_rate": 6.0906765207504256e-05,
      "loss": 5.5475,
      "step": 1058
    },
    {
      "epoch": 1.4749303621169916,
      "grad_norm": 4.661684036254883,
      "learning_rate": 6.089681637293917e-05,
      "loss": 5.6567,
      "step": 1059
    },
    {
      "epoch": 1.4763231197771587,
      "grad_norm": 4.058624744415283,
      "learning_rate": 6.088686753837407e-05,
      "loss": 5.0811,
      "step": 1060
    },
    {
      "epoch": 1.4777158774373258,
      "grad_norm": 4.226449012756348,
      "learning_rate": 6.0876918703808975e-05,
      "loss": 5.6178,
      "step": 1061
    },
    {
      "epoch": 1.479108635097493,
      "grad_norm": 4.778773784637451,
      "learning_rate": 6.086696986924388e-05,
      "loss": 5.9535,
      "step": 1062
    },
    {
      "epoch": 1.4805013927576602,
      "grad_norm": 5.2507195472717285,
      "learning_rate": 6.085702103467879e-05,
      "loss": 5.4184,
      "step": 1063
    },
    {
      "epoch": 1.4818941504178273,
      "grad_norm": 4.835323333740234,
      "learning_rate": 6.08470722001137e-05,
      "loss": 5.4343,
      "step": 1064
    },
    {
      "epoch": 1.4832869080779945,
      "grad_norm": 4.113168716430664,
      "learning_rate": 6.0837123365548605e-05,
      "loss": 5.5131,
      "step": 1065
    },
    {
      "epoch": 1.4846796657381616,
      "grad_norm": 5.054234504699707,
      "learning_rate": 6.08271745309835e-05,
      "loss": 5.1919,
      "step": 1066
    },
    {
      "epoch": 1.4860724233983287,
      "grad_norm": 4.585954666137695,
      "learning_rate": 6.081722569641841e-05,
      "loss": 5.4392,
      "step": 1067
    },
    {
      "epoch": 1.4874651810584958,
      "grad_norm": 4.478633880615234,
      "learning_rate": 6.0807276861853324e-05,
      "loss": 5.1403,
      "step": 1068
    },
    {
      "epoch": 1.488857938718663,
      "grad_norm": 4.157210350036621,
      "learning_rate": 6.079732802728823e-05,
      "loss": 5.3341,
      "step": 1069
    },
    {
      "epoch": 1.49025069637883,
      "grad_norm": 4.241940021514893,
      "learning_rate": 6.078737919272313e-05,
      "loss": 5.5515,
      "step": 1070
    },
    {
      "epoch": 1.4916434540389973,
      "grad_norm": 4.752518177032471,
      "learning_rate": 6.0777430358158036e-05,
      "loss": 5.4118,
      "step": 1071
    },
    {
      "epoch": 1.4930362116991645,
      "grad_norm": 4.162990093231201,
      "learning_rate": 6.076748152359295e-05,
      "loss": 5.3923,
      "step": 1072
    },
    {
      "epoch": 1.4944289693593316,
      "grad_norm": 6.27323579788208,
      "learning_rate": 6.075753268902785e-05,
      "loss": 5.3445,
      "step": 1073
    },
    {
      "epoch": 1.4958217270194987,
      "grad_norm": 3.8739569187164307,
      "learning_rate": 6.074758385446276e-05,
      "loss": 5.7046,
      "step": 1074
    },
    {
      "epoch": 1.4972144846796658,
      "grad_norm": 4.954774856567383,
      "learning_rate": 6.073763501989766e-05,
      "loss": 5.2156,
      "step": 1075
    },
    {
      "epoch": 1.498607242339833,
      "grad_norm": 4.059577941894531,
      "learning_rate": 6.072768618533257e-05,
      "loss": 5.1942,
      "step": 1076
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.890732288360596,
      "learning_rate": 6.0717737350767474e-05,
      "loss": 5.3576,
      "step": 1077
    },
    {
      "epoch": 1.501392757660167,
      "grad_norm": 4.2035298347473145,
      "learning_rate": 6.0707788516202385e-05,
      "loss": 5.3458,
      "step": 1078
    },
    {
      "epoch": 1.5027855153203342,
      "grad_norm": 4.541333198547363,
      "learning_rate": 6.069783968163729e-05,
      "loss": 5.855,
      "step": 1079
    },
    {
      "epoch": 1.5041782729805013,
      "grad_norm": 4.331189155578613,
      "learning_rate": 6.0687890847072193e-05,
      "loss": 5.4807,
      "step": 1080
    },
    {
      "epoch": 1.5055710306406684,
      "grad_norm": 4.317488193511963,
      "learning_rate": 6.06779420125071e-05,
      "loss": 5.7551,
      "step": 1081
    },
    {
      "epoch": 1.5069637883008355,
      "grad_norm": 4.339413166046143,
      "learning_rate": 6.066799317794201e-05,
      "loss": 5.0802,
      "step": 1082
    },
    {
      "epoch": 1.5083565459610027,
      "grad_norm": 4.928534984588623,
      "learning_rate": 6.065804434337691e-05,
      "loss": 5.745,
      "step": 1083
    },
    {
      "epoch": 1.5097493036211698,
      "grad_norm": 3.766230583190918,
      "learning_rate": 6.064809550881182e-05,
      "loss": 5.3793,
      "step": 1084
    },
    {
      "epoch": 1.511142061281337,
      "grad_norm": 4.159375190734863,
      "learning_rate": 6.063814667424672e-05,
      "loss": 5.3454,
      "step": 1085
    },
    {
      "epoch": 1.5125348189415042,
      "grad_norm": 4.078312873840332,
      "learning_rate": 6.062819783968163e-05,
      "loss": 4.8024,
      "step": 1086
    },
    {
      "epoch": 1.5139275766016713,
      "grad_norm": 4.715867042541504,
      "learning_rate": 6.0618249005116536e-05,
      "loss": 5.3176,
      "step": 1087
    },
    {
      "epoch": 1.5153203342618384,
      "grad_norm": 6.341579914093018,
      "learning_rate": 6.0608300170551446e-05,
      "loss": 5.4676,
      "step": 1088
    },
    {
      "epoch": 1.5167130919220055,
      "grad_norm": 4.221826553344727,
      "learning_rate": 6.059835133598635e-05,
      "loss": 5.2045,
      "step": 1089
    },
    {
      "epoch": 1.5181058495821727,
      "grad_norm": 4.108441352844238,
      "learning_rate": 6.0588402501421255e-05,
      "loss": 4.8501,
      "step": 1090
    },
    {
      "epoch": 1.51949860724234,
      "grad_norm": 4.748363018035889,
      "learning_rate": 6.057845366685616e-05,
      "loss": 5.6348,
      "step": 1091
    },
    {
      "epoch": 1.520891364902507,
      "grad_norm": 5.1843132972717285,
      "learning_rate": 6.056850483229107e-05,
      "loss": 5.2767,
      "step": 1092
    },
    {
      "epoch": 1.5222841225626742,
      "grad_norm": 7.4928436279296875,
      "learning_rate": 6.055855599772598e-05,
      "loss": 5.3334,
      "step": 1093
    },
    {
      "epoch": 1.5236768802228413,
      "grad_norm": 4.244089126586914,
      "learning_rate": 6.0548607163160884e-05,
      "loss": 5.6978,
      "step": 1094
    },
    {
      "epoch": 1.5250696378830084,
      "grad_norm": 5.482670783996582,
      "learning_rate": 6.053865832859578e-05,
      "loss": 5.0617,
      "step": 1095
    },
    {
      "epoch": 1.5264623955431755,
      "grad_norm": 4.926762104034424,
      "learning_rate": 6.052870949403069e-05,
      "loss": 5.0647,
      "step": 1096
    },
    {
      "epoch": 1.5278551532033426,
      "grad_norm": 4.932678699493408,
      "learning_rate": 6.05187606594656e-05,
      "loss": 5.175,
      "step": 1097
    },
    {
      "epoch": 1.5292479108635098,
      "grad_norm": 4.782625198364258,
      "learning_rate": 6.050881182490051e-05,
      "loss": 5.2746,
      "step": 1098
    },
    {
      "epoch": 1.5306406685236769,
      "grad_norm": 5.648792743682861,
      "learning_rate": 6.049886299033542e-05,
      "loss": 6.1578,
      "step": 1099
    },
    {
      "epoch": 1.532033426183844,
      "grad_norm": 4.749377727508545,
      "learning_rate": 6.0488914155770316e-05,
      "loss": 5.2959,
      "step": 1100
    },
    {
      "epoch": 1.533426183844011,
      "grad_norm": 4.50879430770874,
      "learning_rate": 6.0478965321205226e-05,
      "loss": 5.4286,
      "step": 1101
    },
    {
      "epoch": 1.5348189415041782,
      "grad_norm": 5.779273986816406,
      "learning_rate": 6.046901648664013e-05,
      "loss": 5.2275,
      "step": 1102
    },
    {
      "epoch": 1.5362116991643453,
      "grad_norm": 5.6755690574646,
      "learning_rate": 6.045906765207504e-05,
      "loss": 5.3256,
      "step": 1103
    },
    {
      "epoch": 1.5376044568245124,
      "grad_norm": 4.357021331787109,
      "learning_rate": 6.0449118817509945e-05,
      "loss": 5.2706,
      "step": 1104
    },
    {
      "epoch": 1.5389972144846795,
      "grad_norm": 4.6982502937316895,
      "learning_rate": 6.043916998294485e-05,
      "loss": 5.273,
      "step": 1105
    },
    {
      "epoch": 1.5403899721448466,
      "grad_norm": 4.713165283203125,
      "learning_rate": 6.0429221148379754e-05,
      "loss": 5.6245,
      "step": 1106
    },
    {
      "epoch": 1.541782729805014,
      "grad_norm": 4.421494960784912,
      "learning_rate": 6.0419272313814664e-05,
      "loss": 5.2232,
      "step": 1107
    },
    {
      "epoch": 1.543175487465181,
      "grad_norm": 4.202319622039795,
      "learning_rate": 6.040932347924957e-05,
      "loss": 5.5722,
      "step": 1108
    },
    {
      "epoch": 1.5445682451253482,
      "grad_norm": 4.8798089027404785,
      "learning_rate": 6.039937464468448e-05,
      "loss": 5.61,
      "step": 1109
    },
    {
      "epoch": 1.5459610027855153,
      "grad_norm": 3.93410587310791,
      "learning_rate": 6.038942581011938e-05,
      "loss": 5.9666,
      "step": 1110
    },
    {
      "epoch": 1.5473537604456824,
      "grad_norm": 4.741140842437744,
      "learning_rate": 6.037947697555429e-05,
      "loss": 5.4437,
      "step": 1111
    },
    {
      "epoch": 1.5487465181058497,
      "grad_norm": 4.0675177574157715,
      "learning_rate": 6.036952814098919e-05,
      "loss": 5.6989,
      "step": 1112
    },
    {
      "epoch": 1.5501392757660168,
      "grad_norm": 4.973630428314209,
      "learning_rate": 6.03595793064241e-05,
      "loss": 5.7861,
      "step": 1113
    },
    {
      "epoch": 1.551532033426184,
      "grad_norm": 3.842400550842285,
      "learning_rate": 6.0349630471859e-05,
      "loss": 5.1912,
      "step": 1114
    },
    {
      "epoch": 1.552924791086351,
      "grad_norm": 5.069960594177246,
      "learning_rate": 6.033968163729391e-05,
      "loss": 5.3714,
      "step": 1115
    },
    {
      "epoch": 1.5543175487465182,
      "grad_norm": 4.760850429534912,
      "learning_rate": 6.0329732802728815e-05,
      "loss": 5.0273,
      "step": 1116
    },
    {
      "epoch": 1.5557103064066853,
      "grad_norm": 4.261945724487305,
      "learning_rate": 6.0319783968163726e-05,
      "loss": 4.9191,
      "step": 1117
    },
    {
      "epoch": 1.5571030640668524,
      "grad_norm": 4.528093338012695,
      "learning_rate": 6.030983513359863e-05,
      "loss": 4.8012,
      "step": 1118
    },
    {
      "epoch": 1.5584958217270195,
      "grad_norm": 3.9830808639526367,
      "learning_rate": 6.0299886299033534e-05,
      "loss": 5.2467,
      "step": 1119
    },
    {
      "epoch": 1.5598885793871866,
      "grad_norm": 4.702970504760742,
      "learning_rate": 6.028993746446844e-05,
      "loss": 5.3754,
      "step": 1120
    },
    {
      "epoch": 1.5612813370473537,
      "grad_norm": 5.310110569000244,
      "learning_rate": 6.027998862990335e-05,
      "loss": 5.2087,
      "step": 1121
    },
    {
      "epoch": 1.5626740947075208,
      "grad_norm": 4.7408127784729,
      "learning_rate": 6.027003979533826e-05,
      "loss": 5.0399,
      "step": 1122
    },
    {
      "epoch": 1.564066852367688,
      "grad_norm": 5.498900890350342,
      "learning_rate": 6.0260090960773164e-05,
      "loss": 5.0449,
      "step": 1123
    },
    {
      "epoch": 1.565459610027855,
      "grad_norm": 5.8566484451293945,
      "learning_rate": 6.025014212620806e-05,
      "loss": 5.5611,
      "step": 1124
    },
    {
      "epoch": 1.5668523676880222,
      "grad_norm": 4.719356060028076,
      "learning_rate": 6.024019329164297e-05,
      "loss": 5.1937,
      "step": 1125
    },
    {
      "epoch": 1.5682451253481893,
      "grad_norm": 5.191523551940918,
      "learning_rate": 6.023024445707788e-05,
      "loss": 5.4316,
      "step": 1126
    },
    {
      "epoch": 1.5696378830083564,
      "grad_norm": 4.1674652099609375,
      "learning_rate": 6.022029562251279e-05,
      "loss": 5.3627,
      "step": 1127
    },
    {
      "epoch": 1.5710306406685237,
      "grad_norm": 4.102329254150391,
      "learning_rate": 6.02103467879477e-05,
      "loss": 5.3458,
      "step": 1128
    },
    {
      "epoch": 1.5724233983286908,
      "grad_norm": 5.134708881378174,
      "learning_rate": 6.0200397953382595e-05,
      "loss": 5.4587,
      "step": 1129
    },
    {
      "epoch": 1.573816155988858,
      "grad_norm": 4.33356237411499,
      "learning_rate": 6.0190449118817506e-05,
      "loss": 5.099,
      "step": 1130
    },
    {
      "epoch": 1.575208913649025,
      "grad_norm": 5.126018047332764,
      "learning_rate": 6.018050028425241e-05,
      "loss": 5.8141,
      "step": 1131
    },
    {
      "epoch": 1.5766016713091922,
      "grad_norm": 4.2986321449279785,
      "learning_rate": 6.017055144968732e-05,
      "loss": 5.7301,
      "step": 1132
    },
    {
      "epoch": 1.5779944289693595,
      "grad_norm": 5.645534992218018,
      "learning_rate": 6.0160602615122225e-05,
      "loss": 5.5876,
      "step": 1133
    },
    {
      "epoch": 1.5793871866295266,
      "grad_norm": 4.3288373947143555,
      "learning_rate": 6.015065378055713e-05,
      "loss": 4.9602,
      "step": 1134
    },
    {
      "epoch": 1.5807799442896937,
      "grad_norm": 4.655366897583008,
      "learning_rate": 6.014070494599203e-05,
      "loss": 5.3268,
      "step": 1135
    },
    {
      "epoch": 1.5821727019498608,
      "grad_norm": 4.137612819671631,
      "learning_rate": 6.0130756111426944e-05,
      "loss": 5.232,
      "step": 1136
    },
    {
      "epoch": 1.583565459610028,
      "grad_norm": 4.0097150802612305,
      "learning_rate": 6.012080727686185e-05,
      "loss": 4.9411,
      "step": 1137
    },
    {
      "epoch": 1.584958217270195,
      "grad_norm": 4.4006242752075195,
      "learning_rate": 6.011085844229676e-05,
      "loss": 5.2862,
      "step": 1138
    },
    {
      "epoch": 1.5863509749303621,
      "grad_norm": 6.664628982543945,
      "learning_rate": 6.0100909607731656e-05,
      "loss": 5.4074,
      "step": 1139
    },
    {
      "epoch": 1.5877437325905293,
      "grad_norm": 5.05947732925415,
      "learning_rate": 6.009096077316657e-05,
      "loss": 4.9289,
      "step": 1140
    },
    {
      "epoch": 1.5891364902506964,
      "grad_norm": 4.076658248901367,
      "learning_rate": 6.008101193860147e-05,
      "loss": 5.2478,
      "step": 1141
    },
    {
      "epoch": 1.5905292479108635,
      "grad_norm": 4.9287309646606445,
      "learning_rate": 6.007106310403638e-05,
      "loss": 5.638,
      "step": 1142
    },
    {
      "epoch": 1.5919220055710306,
      "grad_norm": 4.976055145263672,
      "learning_rate": 6.0061114269471286e-05,
      "loss": 4.9995,
      "step": 1143
    },
    {
      "epoch": 1.5933147632311977,
      "grad_norm": 4.467281818389893,
      "learning_rate": 6.005116543490619e-05,
      "loss": 5.0632,
      "step": 1144
    },
    {
      "epoch": 1.5947075208913648,
      "grad_norm": 10.33066463470459,
      "learning_rate": 6.0041216600341094e-05,
      "loss": 5.5778,
      "step": 1145
    },
    {
      "epoch": 1.596100278551532,
      "grad_norm": 4.165136814117432,
      "learning_rate": 6.0031267765776005e-05,
      "loss": 5.0624,
      "step": 1146
    },
    {
      "epoch": 1.597493036211699,
      "grad_norm": 4.316961765289307,
      "learning_rate": 6.0021318931210916e-05,
      "loss": 5.7191,
      "step": 1147
    },
    {
      "epoch": 1.5988857938718661,
      "grad_norm": 4.557807445526123,
      "learning_rate": 6.001137009664582e-05,
      "loss": 5.7798,
      "step": 1148
    },
    {
      "epoch": 1.6002785515320335,
      "grad_norm": 5.020558834075928,
      "learning_rate": 6.000142126208072e-05,
      "loss": 5.1111,
      "step": 1149
    },
    {
      "epoch": 1.6016713091922006,
      "grad_norm": 4.93497371673584,
      "learning_rate": 5.999147242751563e-05,
      "loss": 5.0666,
      "step": 1150
    },
    {
      "epoch": 1.6030640668523677,
      "grad_norm": 4.127545356750488,
      "learning_rate": 5.998152359295054e-05,
      "loss": 5.3712,
      "step": 1151
    },
    {
      "epoch": 1.6044568245125348,
      "grad_norm": 5.075907230377197,
      "learning_rate": 5.997157475838544e-05,
      "loss": 5.0848,
      "step": 1152
    },
    {
      "epoch": 1.605849582172702,
      "grad_norm": 4.087526321411133,
      "learning_rate": 5.996162592382034e-05,
      "loss": 4.9592,
      "step": 1153
    },
    {
      "epoch": 1.6072423398328692,
      "grad_norm": 4.246445178985596,
      "learning_rate": 5.995167708925525e-05,
      "loss": 5.1955,
      "step": 1154
    },
    {
      "epoch": 1.6086350974930363,
      "grad_norm": 5.3127617835998535,
      "learning_rate": 5.994172825469016e-05,
      "loss": 5.2628,
      "step": 1155
    },
    {
      "epoch": 1.6100278551532035,
      "grad_norm": 4.276273250579834,
      "learning_rate": 5.9931779420125066e-05,
      "loss": 5.3533,
      "step": 1156
    },
    {
      "epoch": 1.6114206128133706,
      "grad_norm": 4.265781402587891,
      "learning_rate": 5.992183058555998e-05,
      "loss": 5.4358,
      "step": 1157
    },
    {
      "epoch": 1.6128133704735377,
      "grad_norm": 5.589931011199951,
      "learning_rate": 5.9911881750994874e-05,
      "loss": 5.4369,
      "step": 1158
    },
    {
      "epoch": 1.6142061281337048,
      "grad_norm": 5.889350414276123,
      "learning_rate": 5.9901932916429785e-05,
      "loss": 5.1033,
      "step": 1159
    },
    {
      "epoch": 1.615598885793872,
      "grad_norm": 3.8457164764404297,
      "learning_rate": 5.989198408186469e-05,
      "loss": 5.4135,
      "step": 1160
    },
    {
      "epoch": 1.616991643454039,
      "grad_norm": 7.082851886749268,
      "learning_rate": 5.98820352472996e-05,
      "loss": 5.1141,
      "step": 1161
    },
    {
      "epoch": 1.6183844011142061,
      "grad_norm": 4.579845905303955,
      "learning_rate": 5.9872086412734504e-05,
      "loss": 5.0879,
      "step": 1162
    },
    {
      "epoch": 1.6197771587743732,
      "grad_norm": 4.529720306396484,
      "learning_rate": 5.986213757816941e-05,
      "loss": 4.983,
      "step": 1163
    },
    {
      "epoch": 1.6211699164345403,
      "grad_norm": 4.010012149810791,
      "learning_rate": 5.985218874360431e-05,
      "loss": 5.1261,
      "step": 1164
    },
    {
      "epoch": 1.6225626740947074,
      "grad_norm": 5.6930460929870605,
      "learning_rate": 5.984223990903922e-05,
      "loss": 5.7673,
      "step": 1165
    },
    {
      "epoch": 1.6239554317548746,
      "grad_norm": 4.764626502990723,
      "learning_rate": 5.983229107447413e-05,
      "loss": 5.0265,
      "step": 1166
    },
    {
      "epoch": 1.6253481894150417,
      "grad_norm": 4.306996822357178,
      "learning_rate": 5.982234223990904e-05,
      "loss": 5.1573,
      "step": 1167
    },
    {
      "epoch": 1.6267409470752088,
      "grad_norm": 3.859175682067871,
      "learning_rate": 5.9812393405343935e-05,
      "loss": 4.9783,
      "step": 1168
    },
    {
      "epoch": 1.6281337047353759,
      "grad_norm": 3.9782230854034424,
      "learning_rate": 5.9802444570778846e-05,
      "loss": 5.3576,
      "step": 1169
    },
    {
      "epoch": 1.6295264623955432,
      "grad_norm": 4.932841777801514,
      "learning_rate": 5.979249573621375e-05,
      "loss": 5.5647,
      "step": 1170
    },
    {
      "epoch": 1.6309192200557103,
      "grad_norm": 4.469640254974365,
      "learning_rate": 5.978254690164866e-05,
      "loss": 5.3512,
      "step": 1171
    },
    {
      "epoch": 1.6323119777158774,
      "grad_norm": 4.7257981300354,
      "learning_rate": 5.9772598067083565e-05,
      "loss": 5.4221,
      "step": 1172
    },
    {
      "epoch": 1.6337047353760445,
      "grad_norm": 4.525543212890625,
      "learning_rate": 5.976264923251847e-05,
      "loss": 5.2254,
      "step": 1173
    },
    {
      "epoch": 1.6350974930362117,
      "grad_norm": 4.767129421234131,
      "learning_rate": 5.975270039795337e-05,
      "loss": 4.7804,
      "step": 1174
    },
    {
      "epoch": 1.636490250696379,
      "grad_norm": 4.485895156860352,
      "learning_rate": 5.9742751563388284e-05,
      "loss": 5.3896,
      "step": 1175
    },
    {
      "epoch": 1.637883008356546,
      "grad_norm": 3.918351888656616,
      "learning_rate": 5.9732802728823195e-05,
      "loss": 5.066,
      "step": 1176
    },
    {
      "epoch": 1.6392757660167132,
      "grad_norm": 4.607816696166992,
      "learning_rate": 5.97228538942581e-05,
      "loss": 5.2288,
      "step": 1177
    },
    {
      "epoch": 1.6406685236768803,
      "grad_norm": 5.340575695037842,
      "learning_rate": 5.9712905059692996e-05,
      "loss": 5.02,
      "step": 1178
    },
    {
      "epoch": 1.6420612813370474,
      "grad_norm": 4.5595383644104,
      "learning_rate": 5.970295622512791e-05,
      "loss": 4.5346,
      "step": 1179
    },
    {
      "epoch": 1.6434540389972145,
      "grad_norm": 5.011353015899658,
      "learning_rate": 5.969300739056282e-05,
      "loss": 4.9287,
      "step": 1180
    },
    {
      "epoch": 1.6448467966573816,
      "grad_norm": 3.863245964050293,
      "learning_rate": 5.968305855599772e-05,
      "loss": 5.3482,
      "step": 1181
    },
    {
      "epoch": 1.6462395543175488,
      "grad_norm": 4.161905765533447,
      "learning_rate": 5.967310972143263e-05,
      "loss": 5.464,
      "step": 1182
    },
    {
      "epoch": 1.6476323119777159,
      "grad_norm": 4.196154594421387,
      "learning_rate": 5.966316088686753e-05,
      "loss": 5.4455,
      "step": 1183
    },
    {
      "epoch": 1.649025069637883,
      "grad_norm": 4.436659812927246,
      "learning_rate": 5.965321205230244e-05,
      "loss": 4.9179,
      "step": 1184
    },
    {
      "epoch": 1.65041782729805,
      "grad_norm": 3.823091506958008,
      "learning_rate": 5.9643263217737345e-05,
      "loss": 5.2561,
      "step": 1185
    },
    {
      "epoch": 1.6518105849582172,
      "grad_norm": 4.765748023986816,
      "learning_rate": 5.9633314383172256e-05,
      "loss": 5.0113,
      "step": 1186
    },
    {
      "epoch": 1.6532033426183843,
      "grad_norm": 4.5616865158081055,
      "learning_rate": 5.962336554860716e-05,
      "loss": 5.0939,
      "step": 1187
    },
    {
      "epoch": 1.6545961002785514,
      "grad_norm": 4.232777118682861,
      "learning_rate": 5.9613416714042064e-05,
      "loss": 5.4659,
      "step": 1188
    },
    {
      "epoch": 1.6559888579387185,
      "grad_norm": 5.522634983062744,
      "learning_rate": 5.960346787947697e-05,
      "loss": 4.767,
      "step": 1189
    },
    {
      "epoch": 1.6573816155988856,
      "grad_norm": 4.595826148986816,
      "learning_rate": 5.959351904491188e-05,
      "loss": 4.8471,
      "step": 1190
    },
    {
      "epoch": 1.658774373259053,
      "grad_norm": 4.422351360321045,
      "learning_rate": 5.958357021034678e-05,
      "loss": 4.5475,
      "step": 1191
    },
    {
      "epoch": 1.66016713091922,
      "grad_norm": 4.2058587074279785,
      "learning_rate": 5.9573621375781694e-05,
      "loss": 5.1124,
      "step": 1192
    },
    {
      "epoch": 1.6615598885793872,
      "grad_norm": 6.848339080810547,
      "learning_rate": 5.956367254121659e-05,
      "loss": 5.3895,
      "step": 1193
    },
    {
      "epoch": 1.6629526462395543,
      "grad_norm": 5.851674556732178,
      "learning_rate": 5.95537237066515e-05,
      "loss": 5.0473,
      "step": 1194
    },
    {
      "epoch": 1.6643454038997214,
      "grad_norm": 4.681227684020996,
      "learning_rate": 5.9543774872086406e-05,
      "loss": 5.1896,
      "step": 1195
    },
    {
      "epoch": 1.6657381615598887,
      "grad_norm": 8.910616874694824,
      "learning_rate": 5.953382603752132e-05,
      "loss": 5.2984,
      "step": 1196
    },
    {
      "epoch": 1.6671309192200559,
      "grad_norm": 4.738180637359619,
      "learning_rate": 5.9523877202956214e-05,
      "loss": 5.1951,
      "step": 1197
    },
    {
      "epoch": 1.668523676880223,
      "grad_norm": 4.421966552734375,
      "learning_rate": 5.9513928368391125e-05,
      "loss": 5.0105,
      "step": 1198
    },
    {
      "epoch": 1.66991643454039,
      "grad_norm": 4.483490943908691,
      "learning_rate": 5.950397953382603e-05,
      "loss": 4.8556,
      "step": 1199
    },
    {
      "epoch": 1.6713091922005572,
      "grad_norm": 4.8139729499816895,
      "learning_rate": 5.949403069926094e-05,
      "loss": 5.4025,
      "step": 1200
    },
    {
      "epoch": 1.6727019498607243,
      "grad_norm": 5.430883884429932,
      "learning_rate": 5.9484081864695844e-05,
      "loss": 5.1012,
      "step": 1201
    },
    {
      "epoch": 1.6740947075208914,
      "grad_norm": 4.058892726898193,
      "learning_rate": 5.947413303013075e-05,
      "loss": 4.9456,
      "step": 1202
    },
    {
      "epoch": 1.6754874651810585,
      "grad_norm": 4.60087251663208,
      "learning_rate": 5.946418419556565e-05,
      "loss": 5.1531,
      "step": 1203
    },
    {
      "epoch": 1.6768802228412256,
      "grad_norm": 5.3073272705078125,
      "learning_rate": 5.945423536100056e-05,
      "loss": 5.061,
      "step": 1204
    },
    {
      "epoch": 1.6782729805013927,
      "grad_norm": 5.039947509765625,
      "learning_rate": 5.9444286526435474e-05,
      "loss": 5.0029,
      "step": 1205
    },
    {
      "epoch": 1.6796657381615598,
      "grad_norm": 6.318732738494873,
      "learning_rate": 5.943433769187038e-05,
      "loss": 4.9684,
      "step": 1206
    },
    {
      "epoch": 1.681058495821727,
      "grad_norm": 4.097740173339844,
      "learning_rate": 5.9424388857305275e-05,
      "loss": 4.6838,
      "step": 1207
    },
    {
      "epoch": 1.682451253481894,
      "grad_norm": 3.9446823596954346,
      "learning_rate": 5.9414440022740186e-05,
      "loss": 4.6736,
      "step": 1208
    },
    {
      "epoch": 1.6838440111420612,
      "grad_norm": 4.035701751708984,
      "learning_rate": 5.94044911881751e-05,
      "loss": 5.2903,
      "step": 1209
    },
    {
      "epoch": 1.6852367688022283,
      "grad_norm": 5.2842583656311035,
      "learning_rate": 5.939454235361e-05,
      "loss": 5.4825,
      "step": 1210
    },
    {
      "epoch": 1.6866295264623954,
      "grad_norm": 4.32784366607666,
      "learning_rate": 5.938459351904491e-05,
      "loss": 5.0295,
      "step": 1211
    },
    {
      "epoch": 1.6880222841225627,
      "grad_norm": 4.509637355804443,
      "learning_rate": 5.937464468447981e-05,
      "loss": 5.1461,
      "step": 1212
    },
    {
      "epoch": 1.6894150417827298,
      "grad_norm": 4.043561935424805,
      "learning_rate": 5.936469584991472e-05,
      "loss": 4.7742,
      "step": 1213
    },
    {
      "epoch": 1.690807799442897,
      "grad_norm": 5.205489635467529,
      "learning_rate": 5.9354747015349624e-05,
      "loss": 4.6934,
      "step": 1214
    },
    {
      "epoch": 1.692200557103064,
      "grad_norm": 4.248347282409668,
      "learning_rate": 5.9344798180784535e-05,
      "loss": 5.4886,
      "step": 1215
    },
    {
      "epoch": 1.6935933147632312,
      "grad_norm": 4.5697197914123535,
      "learning_rate": 5.933484934621944e-05,
      "loss": 5.0912,
      "step": 1216
    },
    {
      "epoch": 1.6949860724233985,
      "grad_norm": 4.81688928604126,
      "learning_rate": 5.932490051165434e-05,
      "loss": 5.4905,
      "step": 1217
    },
    {
      "epoch": 1.6963788300835656,
      "grad_norm": 5.066283702850342,
      "learning_rate": 5.931495167708925e-05,
      "loss": 4.8654,
      "step": 1218
    },
    {
      "epoch": 1.6977715877437327,
      "grad_norm": 4.438613414764404,
      "learning_rate": 5.930500284252416e-05,
      "loss": 5.0706,
      "step": 1219
    },
    {
      "epoch": 1.6991643454038998,
      "grad_norm": 4.010351181030273,
      "learning_rate": 5.929505400795906e-05,
      "loss": 5.4198,
      "step": 1220
    },
    {
      "epoch": 1.700557103064067,
      "grad_norm": 3.8836967945098877,
      "learning_rate": 5.928510517339397e-05,
      "loss": 5.3495,
      "step": 1221
    },
    {
      "epoch": 1.701949860724234,
      "grad_norm": 9.083257675170898,
      "learning_rate": 5.927515633882887e-05,
      "loss": 5.0419,
      "step": 1222
    },
    {
      "epoch": 1.7033426183844012,
      "grad_norm": 4.5670623779296875,
      "learning_rate": 5.926520750426378e-05,
      "loss": 4.8311,
      "step": 1223
    },
    {
      "epoch": 1.7047353760445683,
      "grad_norm": 4.448812484741211,
      "learning_rate": 5.9255258669698685e-05,
      "loss": 4.8841,
      "step": 1224
    },
    {
      "epoch": 1.7061281337047354,
      "grad_norm": 4.513146877288818,
      "learning_rate": 5.9245309835133596e-05,
      "loss": 5.0136,
      "step": 1225
    },
    {
      "epoch": 1.7075208913649025,
      "grad_norm": 4.682188034057617,
      "learning_rate": 5.92353610005685e-05,
      "loss": 4.8929,
      "step": 1226
    },
    {
      "epoch": 1.7089136490250696,
      "grad_norm": 4.83218240737915,
      "learning_rate": 5.9225412166003404e-05,
      "loss": 5.3724,
      "step": 1227
    },
    {
      "epoch": 1.7103064066852367,
      "grad_norm": 4.96373987197876,
      "learning_rate": 5.921546333143831e-05,
      "loss": 5.282,
      "step": 1228
    },
    {
      "epoch": 1.7116991643454038,
      "grad_norm": 4.282477378845215,
      "learning_rate": 5.920551449687322e-05,
      "loss": 4.4825,
      "step": 1229
    },
    {
      "epoch": 1.713091922005571,
      "grad_norm": 5.492599010467529,
      "learning_rate": 5.919556566230812e-05,
      "loss": 5.0134,
      "step": 1230
    },
    {
      "epoch": 1.714484679665738,
      "grad_norm": 5.012602806091309,
      "learning_rate": 5.9185616827743034e-05,
      "loss": 5.3737,
      "step": 1231
    },
    {
      "epoch": 1.7158774373259051,
      "grad_norm": 7.142603874206543,
      "learning_rate": 5.917566799317793e-05,
      "loss": 5.2829,
      "step": 1232
    },
    {
      "epoch": 1.7172701949860725,
      "grad_norm": 5.519572734832764,
      "learning_rate": 5.916571915861284e-05,
      "loss": 5.2057,
      "step": 1233
    },
    {
      "epoch": 1.7186629526462396,
      "grad_norm": 4.653135776519775,
      "learning_rate": 5.915577032404775e-05,
      "loss": 4.9784,
      "step": 1234
    },
    {
      "epoch": 1.7200557103064067,
      "grad_norm": 5.516115188598633,
      "learning_rate": 5.914582148948266e-05,
      "loss": 5.2837,
      "step": 1235
    },
    {
      "epoch": 1.7214484679665738,
      "grad_norm": 4.786309242248535,
      "learning_rate": 5.913587265491757e-05,
      "loss": 5.5864,
      "step": 1236
    },
    {
      "epoch": 1.722841225626741,
      "grad_norm": 4.339987277984619,
      "learning_rate": 5.9125923820352465e-05,
      "loss": 4.6804,
      "step": 1237
    },
    {
      "epoch": 1.724233983286908,
      "grad_norm": 4.238324165344238,
      "learning_rate": 5.9115974985787376e-05,
      "loss": 5.0248,
      "step": 1238
    },
    {
      "epoch": 1.7256267409470754,
      "grad_norm": 4.193300247192383,
      "learning_rate": 5.910602615122228e-05,
      "loss": 4.8384,
      "step": 1239
    },
    {
      "epoch": 1.7270194986072425,
      "grad_norm": 4.547834873199463,
      "learning_rate": 5.909607731665719e-05,
      "loss": 5.26,
      "step": 1240
    },
    {
      "epoch": 1.7284122562674096,
      "grad_norm": 5.412426471710205,
      "learning_rate": 5.908612848209209e-05,
      "loss": 5.0807,
      "step": 1241
    },
    {
      "epoch": 1.7298050139275767,
      "grad_norm": 4.3400139808654785,
      "learning_rate": 5.9076179647527e-05,
      "loss": 4.8661,
      "step": 1242
    },
    {
      "epoch": 1.7311977715877438,
      "grad_norm": 4.30125093460083,
      "learning_rate": 5.90662308129619e-05,
      "loss": 5.465,
      "step": 1243
    },
    {
      "epoch": 1.732590529247911,
      "grad_norm": 7.581824779510498,
      "learning_rate": 5.9056281978396814e-05,
      "loss": 4.9617,
      "step": 1244
    },
    {
      "epoch": 1.733983286908078,
      "grad_norm": 4.4453020095825195,
      "learning_rate": 5.904633314383172e-05,
      "loss": 4.8422,
      "step": 1245
    },
    {
      "epoch": 1.7353760445682451,
      "grad_norm": 4.197065353393555,
      "learning_rate": 5.903638430926662e-05,
      "loss": 5.1462,
      "step": 1246
    },
    {
      "epoch": 1.7367688022284122,
      "grad_norm": 4.419856548309326,
      "learning_rate": 5.9026435474701526e-05,
      "loss": 5.35,
      "step": 1247
    },
    {
      "epoch": 1.7381615598885793,
      "grad_norm": 5.0929789543151855,
      "learning_rate": 5.901648664013644e-05,
      "loss": 4.7436,
      "step": 1248
    },
    {
      "epoch": 1.7395543175487465,
      "grad_norm": 5.040703296661377,
      "learning_rate": 5.900653780557134e-05,
      "loss": 5.0925,
      "step": 1249
    },
    {
      "epoch": 1.7409470752089136,
      "grad_norm": 3.77506160736084,
      "learning_rate": 5.899658897100625e-05,
      "loss": 4.6411,
      "step": 1250
    },
    {
      "epoch": 1.7423398328690807,
      "grad_norm": 4.22568416595459,
      "learning_rate": 5.898664013644115e-05,
      "loss": 5.2016,
      "step": 1251
    },
    {
      "epoch": 1.7437325905292478,
      "grad_norm": 4.241969585418701,
      "learning_rate": 5.897669130187606e-05,
      "loss": 5.059,
      "step": 1252
    },
    {
      "epoch": 1.745125348189415,
      "grad_norm": 4.156410217285156,
      "learning_rate": 5.8966742467310964e-05,
      "loss": 5.1263,
      "step": 1253
    },
    {
      "epoch": 1.7465181058495822,
      "grad_norm": 3.999236822128296,
      "learning_rate": 5.8956793632745875e-05,
      "loss": 4.89,
      "step": 1254
    },
    {
      "epoch": 1.7479108635097493,
      "grad_norm": 4.329075336456299,
      "learning_rate": 5.894684479818078e-05,
      "loss": 4.871,
      "step": 1255
    },
    {
      "epoch": 1.7493036211699164,
      "grad_norm": 5.040160179138184,
      "learning_rate": 5.893689596361568e-05,
      "loss": 4.9036,
      "step": 1256
    },
    {
      "epoch": 1.7506963788300836,
      "grad_norm": 4.441908836364746,
      "learning_rate": 5.892694712905059e-05,
      "loss": 5.5134,
      "step": 1257
    },
    {
      "epoch": 1.7520891364902507,
      "grad_norm": 6.066964149475098,
      "learning_rate": 5.89169982944855e-05,
      "loss": 5.083,
      "step": 1258
    },
    {
      "epoch": 1.7534818941504178,
      "grad_norm": 4.725863456726074,
      "learning_rate": 5.89070494599204e-05,
      "loss": 5.5413,
      "step": 1259
    },
    {
      "epoch": 1.754874651810585,
      "grad_norm": 4.986145496368408,
      "learning_rate": 5.889710062535531e-05,
      "loss": 5.0532,
      "step": 1260
    },
    {
      "epoch": 1.7562674094707522,
      "grad_norm": 4.244281768798828,
      "learning_rate": 5.888715179079021e-05,
      "loss": 5.3201,
      "step": 1261
    },
    {
      "epoch": 1.7576601671309193,
      "grad_norm": 4.933448791503906,
      "learning_rate": 5.887720295622512e-05,
      "loss": 5.2326,
      "step": 1262
    },
    {
      "epoch": 1.7590529247910864,
      "grad_norm": 4.831197738647461,
      "learning_rate": 5.886725412166003e-05,
      "loss": 5.0347,
      "step": 1263
    },
    {
      "epoch": 1.7604456824512535,
      "grad_norm": 6.52940034866333,
      "learning_rate": 5.8857305287094936e-05,
      "loss": 4.7476,
      "step": 1264
    },
    {
      "epoch": 1.7618384401114207,
      "grad_norm": 4.689281463623047,
      "learning_rate": 5.884735645252985e-05,
      "loss": 5.3087,
      "step": 1265
    },
    {
      "epoch": 1.7632311977715878,
      "grad_norm": 4.21406888961792,
      "learning_rate": 5.8837407617964744e-05,
      "loss": 5.3117,
      "step": 1266
    },
    {
      "epoch": 1.7646239554317549,
      "grad_norm": 4.888186931610107,
      "learning_rate": 5.8827458783399655e-05,
      "loss": 4.9891,
      "step": 1267
    },
    {
      "epoch": 1.766016713091922,
      "grad_norm": 6.084913730621338,
      "learning_rate": 5.881750994883456e-05,
      "loss": 4.9797,
      "step": 1268
    },
    {
      "epoch": 1.767409470752089,
      "grad_norm": 9.895624160766602,
      "learning_rate": 5.880756111426947e-05,
      "loss": 5.259,
      "step": 1269
    },
    {
      "epoch": 1.7688022284122562,
      "grad_norm": 4.0565571784973145,
      "learning_rate": 5.8797612279704374e-05,
      "loss": 4.9253,
      "step": 1270
    },
    {
      "epoch": 1.7701949860724233,
      "grad_norm": 5.655186653137207,
      "learning_rate": 5.878766344513928e-05,
      "loss": 4.9211,
      "step": 1271
    },
    {
      "epoch": 1.7715877437325904,
      "grad_norm": 3.942230463027954,
      "learning_rate": 5.877771461057418e-05,
      "loss": 5.1804,
      "step": 1272
    },
    {
      "epoch": 1.7729805013927575,
      "grad_norm": 5.159029483795166,
      "learning_rate": 5.876776577600909e-05,
      "loss": 5.0458,
      "step": 1273
    },
    {
      "epoch": 1.7743732590529246,
      "grad_norm": 5.114142894744873,
      "learning_rate": 5.8757816941444e-05,
      "loss": 4.7971,
      "step": 1274
    },
    {
      "epoch": 1.775766016713092,
      "grad_norm": 5.485936641693115,
      "learning_rate": 5.874786810687891e-05,
      "loss": 4.6306,
      "step": 1275
    },
    {
      "epoch": 1.777158774373259,
      "grad_norm": 6.276236057281494,
      "learning_rate": 5.8737919272313805e-05,
      "loss": 5.1194,
      "step": 1276
    },
    {
      "epoch": 1.7785515320334262,
      "grad_norm": 4.214354991912842,
      "learning_rate": 5.8727970437748716e-05,
      "loss": 4.6878,
      "step": 1277
    },
    {
      "epoch": 1.7799442896935933,
      "grad_norm": 6.638233184814453,
      "learning_rate": 5.871802160318362e-05,
      "loss": 4.8345,
      "step": 1278
    },
    {
      "epoch": 1.7813370473537604,
      "grad_norm": 4.968391418457031,
      "learning_rate": 5.870807276861853e-05,
      "loss": 4.3399,
      "step": 1279
    },
    {
      "epoch": 1.7827298050139275,
      "grad_norm": 4.857850074768066,
      "learning_rate": 5.8698123934053435e-05,
      "loss": 5.1188,
      "step": 1280
    },
    {
      "epoch": 1.7841225626740949,
      "grad_norm": 4.397397518157959,
      "learning_rate": 5.868817509948834e-05,
      "loss": 5.0782,
      "step": 1281
    },
    {
      "epoch": 1.785515320334262,
      "grad_norm": 4.862459659576416,
      "learning_rate": 5.8678226264923243e-05,
      "loss": 4.7992,
      "step": 1282
    },
    {
      "epoch": 1.786908077994429,
      "grad_norm": 4.83461856842041,
      "learning_rate": 5.8668277430358154e-05,
      "loss": 5.2456,
      "step": 1283
    },
    {
      "epoch": 1.7883008356545962,
      "grad_norm": 5.348004341125488,
      "learning_rate": 5.865832859579306e-05,
      "loss": 4.909,
      "step": 1284
    },
    {
      "epoch": 1.7896935933147633,
      "grad_norm": 3.98246169090271,
      "learning_rate": 5.864837976122796e-05,
      "loss": 4.9827,
      "step": 1285
    },
    {
      "epoch": 1.7910863509749304,
      "grad_norm": 4.485197067260742,
      "learning_rate": 5.8638430926662867e-05,
      "loss": 4.8303,
      "step": 1286
    },
    {
      "epoch": 1.7924791086350975,
      "grad_norm": 4.2006731033325195,
      "learning_rate": 5.862848209209778e-05,
      "loss": 4.3103,
      "step": 1287
    },
    {
      "epoch": 1.7938718662952646,
      "grad_norm": 5.053812503814697,
      "learning_rate": 5.861853325753269e-05,
      "loss": 5.1545,
      "step": 1288
    },
    {
      "epoch": 1.7952646239554317,
      "grad_norm": 5.753026962280273,
      "learning_rate": 5.860858442296759e-05,
      "loss": 5.0855,
      "step": 1289
    },
    {
      "epoch": 1.7966573816155988,
      "grad_norm": 4.728921413421631,
      "learning_rate": 5.859863558840249e-05,
      "loss": 4.6271,
      "step": 1290
    },
    {
      "epoch": 1.798050139275766,
      "grad_norm": 4.88857364654541,
      "learning_rate": 5.85886867538374e-05,
      "loss": 4.8991,
      "step": 1291
    },
    {
      "epoch": 1.799442896935933,
      "grad_norm": 4.64939022064209,
      "learning_rate": 5.857873791927231e-05,
      "loss": 4.7404,
      "step": 1292
    },
    {
      "epoch": 1.8008356545961002,
      "grad_norm": 4.134711265563965,
      "learning_rate": 5.8568789084707215e-05,
      "loss": 4.9327,
      "step": 1293
    },
    {
      "epoch": 1.8022284122562673,
      "grad_norm": 4.269225120544434,
      "learning_rate": 5.8558840250142126e-05,
      "loss": 5.1741,
      "step": 1294
    },
    {
      "epoch": 1.8036211699164344,
      "grad_norm": 4.530089378356934,
      "learning_rate": 5.8548891415577024e-05,
      "loss": 4.7819,
      "step": 1295
    },
    {
      "epoch": 1.8050139275766015,
      "grad_norm": 5.171576976776123,
      "learning_rate": 5.8538942581011934e-05,
      "loss": 5.1002,
      "step": 1296
    },
    {
      "epoch": 1.8064066852367688,
      "grad_norm": 4.574953079223633,
      "learning_rate": 5.852899374644684e-05,
      "loss": 5.2196,
      "step": 1297
    },
    {
      "epoch": 1.807799442896936,
      "grad_norm": 4.656442165374756,
      "learning_rate": 5.851904491188175e-05,
      "loss": 4.8261,
      "step": 1298
    },
    {
      "epoch": 1.809192200557103,
      "grad_norm": 5.161946773529053,
      "learning_rate": 5.8509096077316653e-05,
      "loss": 5.1474,
      "step": 1299
    },
    {
      "epoch": 1.8105849582172702,
      "grad_norm": 5.265983581542969,
      "learning_rate": 5.849914724275156e-05,
      "loss": 4.902,
      "step": 1300
    },
    {
      "epoch": 1.8119777158774373,
      "grad_norm": 4.652301788330078,
      "learning_rate": 5.848919840818646e-05,
      "loss": 4.722,
      "step": 1301
    },
    {
      "epoch": 1.8133704735376046,
      "grad_norm": 4.341517925262451,
      "learning_rate": 5.847924957362137e-05,
      "loss": 5.2609,
      "step": 1302
    },
    {
      "epoch": 1.8147632311977717,
      "grad_norm": 5.4265313148498535,
      "learning_rate": 5.8469300739056277e-05,
      "loss": 5.1165,
      "step": 1303
    },
    {
      "epoch": 1.8161559888579388,
      "grad_norm": 4.894313812255859,
      "learning_rate": 5.845935190449119e-05,
      "loss": 4.8027,
      "step": 1304
    },
    {
      "epoch": 1.817548746518106,
      "grad_norm": 4.270108222961426,
      "learning_rate": 5.8449403069926085e-05,
      "loss": 4.9183,
      "step": 1305
    },
    {
      "epoch": 1.818941504178273,
      "grad_norm": 6.259138107299805,
      "learning_rate": 5.8439454235360996e-05,
      "loss": 4.9659,
      "step": 1306
    },
    {
      "epoch": 1.8203342618384402,
      "grad_norm": 4.402987480163574,
      "learning_rate": 5.84295054007959e-05,
      "loss": 4.5167,
      "step": 1307
    },
    {
      "epoch": 1.8217270194986073,
      "grad_norm": 4.210943698883057,
      "learning_rate": 5.841955656623081e-05,
      "loss": 5.0627,
      "step": 1308
    },
    {
      "epoch": 1.8231197771587744,
      "grad_norm": 4.103539943695068,
      "learning_rate": 5.8409607731665715e-05,
      "loss": 5.1827,
      "step": 1309
    },
    {
      "epoch": 1.8245125348189415,
      "grad_norm": 6.2264299392700195,
      "learning_rate": 5.839965889710062e-05,
      "loss": 5.2231,
      "step": 1310
    },
    {
      "epoch": 1.8259052924791086,
      "grad_norm": 6.247641563415527,
      "learning_rate": 5.838971006253552e-05,
      "loss": 5.0428,
      "step": 1311
    },
    {
      "epoch": 1.8272980501392757,
      "grad_norm": 4.584635257720947,
      "learning_rate": 5.8379761227970434e-05,
      "loss": 4.9395,
      "step": 1312
    },
    {
      "epoch": 1.8286908077994428,
      "grad_norm": 3.6346242427825928,
      "learning_rate": 5.836981239340534e-05,
      "loss": 4.8325,
      "step": 1313
    },
    {
      "epoch": 1.83008356545961,
      "grad_norm": 4.009475231170654,
      "learning_rate": 5.835986355884025e-05,
      "loss": 4.6872,
      "step": 1314
    },
    {
      "epoch": 1.831476323119777,
      "grad_norm": 4.166362762451172,
      "learning_rate": 5.8349914724275146e-05,
      "loss": 5.1893,
      "step": 1315
    },
    {
      "epoch": 1.8328690807799441,
      "grad_norm": 5.342883586883545,
      "learning_rate": 5.8339965889710057e-05,
      "loss": 5.0689,
      "step": 1316
    },
    {
      "epoch": 1.8342618384401113,
      "grad_norm": 4.411332607269287,
      "learning_rate": 5.833001705514497e-05,
      "loss": 4.847,
      "step": 1317
    },
    {
      "epoch": 1.8356545961002786,
      "grad_norm": 5.047307968139648,
      "learning_rate": 5.832006822057987e-05,
      "loss": 5.1624,
      "step": 1318
    },
    {
      "epoch": 1.8370473537604457,
      "grad_norm": 5.890045166015625,
      "learning_rate": 5.831011938601478e-05,
      "loss": 4.8237,
      "step": 1319
    },
    {
      "epoch": 1.8384401114206128,
      "grad_norm": 5.271467208862305,
      "learning_rate": 5.830017055144968e-05,
      "loss": 5.0535,
      "step": 1320
    },
    {
      "epoch": 1.83983286908078,
      "grad_norm": 4.287512302398682,
      "learning_rate": 5.829022171688459e-05,
      "loss": 4.3351,
      "step": 1321
    },
    {
      "epoch": 1.841225626740947,
      "grad_norm": 4.520585536956787,
      "learning_rate": 5.8280272882319495e-05,
      "loss": 5.2462,
      "step": 1322
    },
    {
      "epoch": 1.8426183844011144,
      "grad_norm": 4.573328971862793,
      "learning_rate": 5.8270324047754405e-05,
      "loss": 4.6179,
      "step": 1323
    },
    {
      "epoch": 1.8440111420612815,
      "grad_norm": 4.49533748626709,
      "learning_rate": 5.82603752131893e-05,
      "loss": 5.1825,
      "step": 1324
    },
    {
      "epoch": 1.8454038997214486,
      "grad_norm": 4.156423568725586,
      "learning_rate": 5.8250426378624214e-05,
      "loss": 5.0544,
      "step": 1325
    },
    {
      "epoch": 1.8467966573816157,
      "grad_norm": 4.7064433097839355,
      "learning_rate": 5.824047754405912e-05,
      "loss": 5.0148,
      "step": 1326
    },
    {
      "epoch": 1.8481894150417828,
      "grad_norm": 4.827240467071533,
      "learning_rate": 5.823052870949403e-05,
      "loss": 4.6688,
      "step": 1327
    },
    {
      "epoch": 1.84958217270195,
      "grad_norm": 5.145040035247803,
      "learning_rate": 5.822057987492893e-05,
      "loss": 5.3015,
      "step": 1328
    },
    {
      "epoch": 1.850974930362117,
      "grad_norm": 7.89101505279541,
      "learning_rate": 5.821063104036384e-05,
      "loss": 4.8715,
      "step": 1329
    },
    {
      "epoch": 1.8523676880222841,
      "grad_norm": 4.7502923011779785,
      "learning_rate": 5.820068220579874e-05,
      "loss": 5.0511,
      "step": 1330
    },
    {
      "epoch": 1.8537604456824512,
      "grad_norm": 4.670762538909912,
      "learning_rate": 5.819073337123365e-05,
      "loss": 4.7304,
      "step": 1331
    },
    {
      "epoch": 1.8551532033426184,
      "grad_norm": 3.9312586784362793,
      "learning_rate": 5.8180784536668556e-05,
      "loss": 4.8302,
      "step": 1332
    },
    {
      "epoch": 1.8565459610027855,
      "grad_norm": 4.732985496520996,
      "learning_rate": 5.8170835702103467e-05,
      "loss": 4.81,
      "step": 1333
    },
    {
      "epoch": 1.8579387186629526,
      "grad_norm": 5.042375564575195,
      "learning_rate": 5.8160886867538364e-05,
      "loss": 4.7127,
      "step": 1334
    },
    {
      "epoch": 1.8593314763231197,
      "grad_norm": 4.160640239715576,
      "learning_rate": 5.8150938032973275e-05,
      "loss": 5.0566,
      "step": 1335
    },
    {
      "epoch": 1.8607242339832868,
      "grad_norm": 4.453600883483887,
      "learning_rate": 5.814098919840818e-05,
      "loss": 4.7741,
      "step": 1336
    },
    {
      "epoch": 1.862116991643454,
      "grad_norm": 4.191370964050293,
      "learning_rate": 5.813104036384309e-05,
      "loss": 4.9208,
      "step": 1337
    },
    {
      "epoch": 1.863509749303621,
      "grad_norm": 4.520490646362305,
      "learning_rate": 5.8121091529277994e-05,
      "loss": 4.9874,
      "step": 1338
    },
    {
      "epoch": 1.8649025069637883,
      "grad_norm": 5.264113426208496,
      "learning_rate": 5.81111426947129e-05,
      "loss": 4.6024,
      "step": 1339
    },
    {
      "epoch": 1.8662952646239555,
      "grad_norm": 4.082907199859619,
      "learning_rate": 5.81011938601478e-05,
      "loss": 5.0785,
      "step": 1340
    },
    {
      "epoch": 1.8676880222841226,
      "grad_norm": 3.9186766147613525,
      "learning_rate": 5.809124502558271e-05,
      "loss": 4.5587,
      "step": 1341
    },
    {
      "epoch": 1.8690807799442897,
      "grad_norm": 4.418737888336182,
      "learning_rate": 5.808129619101762e-05,
      "loss": 4.3297,
      "step": 1342
    },
    {
      "epoch": 1.8704735376044568,
      "grad_norm": 4.381232738494873,
      "learning_rate": 5.807134735645253e-05,
      "loss": 4.4724,
      "step": 1343
    },
    {
      "epoch": 1.8718662952646241,
      "grad_norm": 4.592318534851074,
      "learning_rate": 5.8061398521887425e-05,
      "loss": 4.6123,
      "step": 1344
    },
    {
      "epoch": 1.8732590529247912,
      "grad_norm": 6.614316463470459,
      "learning_rate": 5.8051449687322336e-05,
      "loss": 4.9827,
      "step": 1345
    },
    {
      "epoch": 1.8746518105849583,
      "grad_norm": 4.040852069854736,
      "learning_rate": 5.8041500852757247e-05,
      "loss": 4.7856,
      "step": 1346
    },
    {
      "epoch": 1.8760445682451254,
      "grad_norm": 4.913686275482178,
      "learning_rate": 5.803155201819215e-05,
      "loss": 5.0156,
      "step": 1347
    },
    {
      "epoch": 1.8774373259052926,
      "grad_norm": 5.432875633239746,
      "learning_rate": 5.802160318362706e-05,
      "loss": 4.8061,
      "step": 1348
    },
    {
      "epoch": 1.8788300835654597,
      "grad_norm": 4.370886325836182,
      "learning_rate": 5.801165434906196e-05,
      "loss": 4.8098,
      "step": 1349
    },
    {
      "epoch": 1.8802228412256268,
      "grad_norm": 4.216629505157471,
      "learning_rate": 5.800170551449687e-05,
      "loss": 4.3694,
      "step": 1350
    },
    {
      "epoch": 1.8816155988857939,
      "grad_norm": 4.49286413192749,
      "learning_rate": 5.7991756679931774e-05,
      "loss": 5.1319,
      "step": 1351
    },
    {
      "epoch": 1.883008356545961,
      "grad_norm": 4.426698684692383,
      "learning_rate": 5.7981807845366685e-05,
      "loss": 5.2308,
      "step": 1352
    },
    {
      "epoch": 1.884401114206128,
      "grad_norm": 4.446043491363525,
      "learning_rate": 5.797185901080159e-05,
      "loss": 4.736,
      "step": 1353
    },
    {
      "epoch": 1.8857938718662952,
      "grad_norm": 4.116539001464844,
      "learning_rate": 5.796191017623649e-05,
      "loss": 4.8607,
      "step": 1354
    },
    {
      "epoch": 1.8871866295264623,
      "grad_norm": 3.835434913635254,
      "learning_rate": 5.79519613416714e-05,
      "loss": 4.304,
      "step": 1355
    },
    {
      "epoch": 1.8885793871866294,
      "grad_norm": 4.829227447509766,
      "learning_rate": 5.794201250710631e-05,
      "loss": 4.9886,
      "step": 1356
    },
    {
      "epoch": 1.8899721448467965,
      "grad_norm": 6.427865982055664,
      "learning_rate": 5.793206367254121e-05,
      "loss": 4.2493,
      "step": 1357
    },
    {
      "epoch": 1.8913649025069637,
      "grad_norm": 7.588675498962402,
      "learning_rate": 5.792211483797612e-05,
      "loss": 5.2044,
      "step": 1358
    },
    {
      "epoch": 1.8927576601671308,
      "grad_norm": 5.770694255828857,
      "learning_rate": 5.791216600341102e-05,
      "loss": 4.5427,
      "step": 1359
    },
    {
      "epoch": 1.894150417827298,
      "grad_norm": 4.1365485191345215,
      "learning_rate": 5.790221716884593e-05,
      "loss": 5.0175,
      "step": 1360
    },
    {
      "epoch": 1.8955431754874652,
      "grad_norm": 3.9512853622436523,
      "learning_rate": 5.7892268334280835e-05,
      "loss": 5.1879,
      "step": 1361
    },
    {
      "epoch": 1.8969359331476323,
      "grad_norm": 4.1711530685424805,
      "learning_rate": 5.7882319499715746e-05,
      "loss": 4.5685,
      "step": 1362
    },
    {
      "epoch": 1.8983286908077994,
      "grad_norm": 4.351253986358643,
      "learning_rate": 5.787237066515065e-05,
      "loss": 4.6267,
      "step": 1363
    },
    {
      "epoch": 1.8997214484679665,
      "grad_norm": 4.42673397064209,
      "learning_rate": 5.7862421830585554e-05,
      "loss": 5.4112,
      "step": 1364
    },
    {
      "epoch": 1.9011142061281339,
      "grad_norm": 4.138251304626465,
      "learning_rate": 5.785247299602046e-05,
      "loss": 4.8453,
      "step": 1365
    },
    {
      "epoch": 1.902506963788301,
      "grad_norm": 4.287871360778809,
      "learning_rate": 5.784252416145537e-05,
      "loss": 4.9142,
      "step": 1366
    },
    {
      "epoch": 1.903899721448468,
      "grad_norm": 4.384206295013428,
      "learning_rate": 5.783257532689027e-05,
      "loss": 4.8998,
      "step": 1367
    },
    {
      "epoch": 1.9052924791086352,
      "grad_norm": 6.017545223236084,
      "learning_rate": 5.782262649232518e-05,
      "loss": 4.8551,
      "step": 1368
    },
    {
      "epoch": 1.9066852367688023,
      "grad_norm": 5.367592811584473,
      "learning_rate": 5.781267765776008e-05,
      "loss": 5.2526,
      "step": 1369
    },
    {
      "epoch": 1.9080779944289694,
      "grad_norm": 5.079102039337158,
      "learning_rate": 5.780272882319499e-05,
      "loss": 4.7068,
      "step": 1370
    },
    {
      "epoch": 1.9094707520891365,
      "grad_norm": 5.6808648109436035,
      "learning_rate": 5.7792779988629896e-05,
      "loss": 4.6206,
      "step": 1371
    },
    {
      "epoch": 1.9108635097493036,
      "grad_norm": 3.902747869491577,
      "learning_rate": 5.778283115406481e-05,
      "loss": 4.2311,
      "step": 1372
    },
    {
      "epoch": 1.9122562674094707,
      "grad_norm": 4.13630485534668,
      "learning_rate": 5.7772882319499704e-05,
      "loss": 4.94,
      "step": 1373
    },
    {
      "epoch": 1.9136490250696379,
      "grad_norm": 4.3250250816345215,
      "learning_rate": 5.7762933484934615e-05,
      "loss": 4.6258,
      "step": 1374
    },
    {
      "epoch": 1.915041782729805,
      "grad_norm": 4.0559186935424805,
      "learning_rate": 5.7752984650369526e-05,
      "loss": 5.2015,
      "step": 1375
    },
    {
      "epoch": 1.916434540389972,
      "grad_norm": 4.951042175292969,
      "learning_rate": 5.774303581580443e-05,
      "loss": 5.2368,
      "step": 1376
    },
    {
      "epoch": 1.9178272980501392,
      "grad_norm": 4.291134834289551,
      "learning_rate": 5.773308698123934e-05,
      "loss": 4.7236,
      "step": 1377
    },
    {
      "epoch": 1.9192200557103063,
      "grad_norm": 6.4045515060424805,
      "learning_rate": 5.772313814667424e-05,
      "loss": 5.3078,
      "step": 1378
    },
    {
      "epoch": 1.9206128133704734,
      "grad_norm": 4.630138874053955,
      "learning_rate": 5.771318931210915e-05,
      "loss": 4.6285,
      "step": 1379
    },
    {
      "epoch": 1.9220055710306405,
      "grad_norm": 5.62636661529541,
      "learning_rate": 5.770324047754405e-05,
      "loss": 4.3099,
      "step": 1380
    },
    {
      "epoch": 1.9233983286908078,
      "grad_norm": 4.544035911560059,
      "learning_rate": 5.7693291642978964e-05,
      "loss": 4.6767,
      "step": 1381
    },
    {
      "epoch": 1.924791086350975,
      "grad_norm": 4.146763324737549,
      "learning_rate": 5.768334280841387e-05,
      "loss": 4.5114,
      "step": 1382
    },
    {
      "epoch": 1.926183844011142,
      "grad_norm": 5.176914215087891,
      "learning_rate": 5.767339397384877e-05,
      "loss": 4.3382,
      "step": 1383
    },
    {
      "epoch": 1.9275766016713092,
      "grad_norm": 4.158557415008545,
      "learning_rate": 5.7663445139283676e-05,
      "loss": 5.1691,
      "step": 1384
    },
    {
      "epoch": 1.9289693593314763,
      "grad_norm": 5.481085777282715,
      "learning_rate": 5.765349630471859e-05,
      "loss": 4.883,
      "step": 1385
    },
    {
      "epoch": 1.9303621169916436,
      "grad_norm": 4.413912296295166,
      "learning_rate": 5.764354747015349e-05,
      "loss": 4.4083,
      "step": 1386
    },
    {
      "epoch": 1.9317548746518107,
      "grad_norm": 7.916518211364746,
      "learning_rate": 5.76335986355884e-05,
      "loss": 5.2687,
      "step": 1387
    },
    {
      "epoch": 1.9331476323119778,
      "grad_norm": 3.890836000442505,
      "learning_rate": 5.76236498010233e-05,
      "loss": 4.4788,
      "step": 1388
    },
    {
      "epoch": 1.934540389972145,
      "grad_norm": 4.351379871368408,
      "learning_rate": 5.761370096645821e-05,
      "loss": 4.386,
      "step": 1389
    },
    {
      "epoch": 1.935933147632312,
      "grad_norm": 6.305099010467529,
      "learning_rate": 5.7603752131893114e-05,
      "loss": 4.857,
      "step": 1390
    },
    {
      "epoch": 1.9373259052924792,
      "grad_norm": 5.599578857421875,
      "learning_rate": 5.7593803297328025e-05,
      "loss": 5.7696,
      "step": 1391
    },
    {
      "epoch": 1.9387186629526463,
      "grad_norm": 3.866544723510742,
      "learning_rate": 5.758385446276293e-05,
      "loss": 4.3325,
      "step": 1392
    },
    {
      "epoch": 1.9401114206128134,
      "grad_norm": 4.895143508911133,
      "learning_rate": 5.757390562819783e-05,
      "loss": 4.8292,
      "step": 1393
    },
    {
      "epoch": 1.9415041782729805,
      "grad_norm": 4.506255149841309,
      "learning_rate": 5.756395679363274e-05,
      "loss": 4.5273,
      "step": 1394
    },
    {
      "epoch": 1.9428969359331476,
      "grad_norm": 4.85635232925415,
      "learning_rate": 5.755400795906765e-05,
      "loss": 5.0543,
      "step": 1395
    },
    {
      "epoch": 1.9442896935933147,
      "grad_norm": 5.331640720367432,
      "learning_rate": 5.754405912450255e-05,
      "loss": 4.5892,
      "step": 1396
    },
    {
      "epoch": 1.9456824512534818,
      "grad_norm": 3.6229052543640137,
      "learning_rate": 5.753411028993746e-05,
      "loss": 4.6695,
      "step": 1397
    },
    {
      "epoch": 1.947075208913649,
      "grad_norm": 4.780959129333496,
      "learning_rate": 5.752416145537236e-05,
      "loss": 4.4121,
      "step": 1398
    },
    {
      "epoch": 1.948467966573816,
      "grad_norm": 4.7896809577941895,
      "learning_rate": 5.751421262080727e-05,
      "loss": 4.9175,
      "step": 1399
    },
    {
      "epoch": 1.9498607242339832,
      "grad_norm": 4.381043434143066,
      "learning_rate": 5.750426378624218e-05,
      "loss": 4.8115,
      "step": 1400
    },
    {
      "epoch": 1.9512534818941503,
      "grad_norm": 4.62701940536499,
      "learning_rate": 5.7494314951677086e-05,
      "loss": 4.9335,
      "step": 1401
    },
    {
      "epoch": 1.9526462395543176,
      "grad_norm": 4.220277309417725,
      "learning_rate": 5.7484366117112e-05,
      "loss": 4.5878,
      "step": 1402
    },
    {
      "epoch": 1.9540389972144847,
      "grad_norm": 5.747648239135742,
      "learning_rate": 5.7474417282546894e-05,
      "loss": 4.9775,
      "step": 1403
    },
    {
      "epoch": 1.9554317548746518,
      "grad_norm": 5.05411958694458,
      "learning_rate": 5.7464468447981805e-05,
      "loss": 4.7716,
      "step": 1404
    },
    {
      "epoch": 1.956824512534819,
      "grad_norm": 4.94806432723999,
      "learning_rate": 5.745451961341671e-05,
      "loss": 4.5957,
      "step": 1405
    },
    {
      "epoch": 1.958217270194986,
      "grad_norm": 4.678225994110107,
      "learning_rate": 5.744457077885162e-05,
      "loss": 5.2641,
      "step": 1406
    },
    {
      "epoch": 1.9596100278551534,
      "grad_norm": 3.895622491836548,
      "learning_rate": 5.7434621944286524e-05,
      "loss": 4.472,
      "step": 1407
    },
    {
      "epoch": 1.9610027855153205,
      "grad_norm": 4.450009346008301,
      "learning_rate": 5.742467310972143e-05,
      "loss": 4.5857,
      "step": 1408
    },
    {
      "epoch": 1.9623955431754876,
      "grad_norm": 4.481184005737305,
      "learning_rate": 5.741472427515633e-05,
      "loss": 4.3904,
      "step": 1409
    },
    {
      "epoch": 1.9637883008356547,
      "grad_norm": 7.812098979949951,
      "learning_rate": 5.740477544059124e-05,
      "loss": 4.6505,
      "step": 1410
    },
    {
      "epoch": 1.9651810584958218,
      "grad_norm": 5.122473239898682,
      "learning_rate": 5.739482660602615e-05,
      "loss": 5.3639,
      "step": 1411
    },
    {
      "epoch": 1.966573816155989,
      "grad_norm": 4.021027088165283,
      "learning_rate": 5.738487777146105e-05,
      "loss": 4.5642,
      "step": 1412
    },
    {
      "epoch": 1.967966573816156,
      "grad_norm": 4.1866536140441895,
      "learning_rate": 5.7374928936895955e-05,
      "loss": 4.5343,
      "step": 1413
    },
    {
      "epoch": 1.9693593314763231,
      "grad_norm": 5.105562686920166,
      "learning_rate": 5.7364980102330866e-05,
      "loss": 4.7158,
      "step": 1414
    },
    {
      "epoch": 1.9707520891364902,
      "grad_norm": 4.511186122894287,
      "learning_rate": 5.735503126776577e-05,
      "loss": 5.0681,
      "step": 1415
    },
    {
      "epoch": 1.9721448467966574,
      "grad_norm": 4.926334857940674,
      "learning_rate": 5.734508243320068e-05,
      "loss": 5.056,
      "step": 1416
    },
    {
      "epoch": 1.9735376044568245,
      "grad_norm": 5.372550010681152,
      "learning_rate": 5.733513359863558e-05,
      "loss": 4.9231,
      "step": 1417
    },
    {
      "epoch": 1.9749303621169916,
      "grad_norm": 4.97346830368042,
      "learning_rate": 5.732518476407049e-05,
      "loss": 4.8801,
      "step": 1418
    },
    {
      "epoch": 1.9763231197771587,
      "grad_norm": 4.553598403930664,
      "learning_rate": 5.731523592950539e-05,
      "loss": 4.871,
      "step": 1419
    },
    {
      "epoch": 1.9777158774373258,
      "grad_norm": 4.302711486816406,
      "learning_rate": 5.7305287094940304e-05,
      "loss": 4.4123,
      "step": 1420
    },
    {
      "epoch": 1.979108635097493,
      "grad_norm": 5.098697185516357,
      "learning_rate": 5.729533826037521e-05,
      "loss": 4.1993,
      "step": 1421
    },
    {
      "epoch": 1.98050139275766,
      "grad_norm": 5.240197658538818,
      "learning_rate": 5.728538942581011e-05,
      "loss": 4.6883,
      "step": 1422
    },
    {
      "epoch": 1.9818941504178273,
      "grad_norm": 5.173120498657227,
      "learning_rate": 5.7275440591245016e-05,
      "loss": 5.0508,
      "step": 1423
    },
    {
      "epoch": 1.9832869080779945,
      "grad_norm": 5.3469390869140625,
      "learning_rate": 5.726549175667993e-05,
      "loss": 5.003,
      "step": 1424
    },
    {
      "epoch": 1.9846796657381616,
      "grad_norm": 4.947526454925537,
      "learning_rate": 5.725554292211483e-05,
      "loss": 5.0937,
      "step": 1425
    },
    {
      "epoch": 1.9860724233983287,
      "grad_norm": 4.714720249176025,
      "learning_rate": 5.724559408754974e-05,
      "loss": 4.5825,
      "step": 1426
    },
    {
      "epoch": 1.9874651810584958,
      "grad_norm": 5.9887471199035645,
      "learning_rate": 5.723564525298464e-05,
      "loss": 4.2472,
      "step": 1427
    },
    {
      "epoch": 1.988857938718663,
      "grad_norm": 4.950433731079102,
      "learning_rate": 5.722569641841955e-05,
      "loss": 4.8435,
      "step": 1428
    },
    {
      "epoch": 1.9902506963788302,
      "grad_norm": 4.301307201385498,
      "learning_rate": 5.721574758385446e-05,
      "loss": 4.5511,
      "step": 1429
    },
    {
      "epoch": 1.9916434540389973,
      "grad_norm": 4.059776306152344,
      "learning_rate": 5.7205798749289365e-05,
      "loss": 4.4791,
      "step": 1430
    },
    {
      "epoch": 1.9930362116991645,
      "grad_norm": 6.272601127624512,
      "learning_rate": 5.7195849914724276e-05,
      "loss": 4.6268,
      "step": 1431
    },
    {
      "epoch": 1.9944289693593316,
      "grad_norm": 6.988149642944336,
      "learning_rate": 5.718590108015917e-05,
      "loss": 4.3559,
      "step": 1432
    },
    {
      "epoch": 1.9958217270194987,
      "grad_norm": 5.213869094848633,
      "learning_rate": 5.7175952245594084e-05,
      "loss": 4.905,
      "step": 1433
    },
    {
      "epoch": 1.9972144846796658,
      "grad_norm": 4.848118305206299,
      "learning_rate": 5.716600341102899e-05,
      "loss": 4.5364,
      "step": 1434
    },
    {
      "epoch": 1.998607242339833,
      "grad_norm": 5.0276899337768555,
      "learning_rate": 5.71560545764639e-05,
      "loss": 4.815,
      "step": 1435
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.235401153564453,
      "learning_rate": 5.71461057418988e-05,
      "loss": 4.4951,
      "step": 1436
    },
    {
      "epoch": 2.001392757660167,
      "grad_norm": 6.121840476989746,
      "learning_rate": 5.713615690733371e-05,
      "loss": 3.9839,
      "step": 1437
    },
    {
      "epoch": 2.002785515320334,
      "grad_norm": 4.8736748695373535,
      "learning_rate": 5.712620807276861e-05,
      "loss": 4.0053,
      "step": 1438
    },
    {
      "epoch": 2.0041782729805013,
      "grad_norm": 6.055631160736084,
      "learning_rate": 5.711625923820352e-05,
      "loss": 4.4934,
      "step": 1439
    },
    {
      "epoch": 2.0055710306406684,
      "grad_norm": 6.15172815322876,
      "learning_rate": 5.7106310403638426e-05,
      "loss": 4.4508,
      "step": 1440
    },
    {
      "epoch": 2.0069637883008355,
      "grad_norm": 3.874490976333618,
      "learning_rate": 5.709636156907334e-05,
      "loss": 4.2319,
      "step": 1441
    },
    {
      "epoch": 2.0083565459610027,
      "grad_norm": 4.357259273529053,
      "learning_rate": 5.7086412734508234e-05,
      "loss": 4.091,
      "step": 1442
    },
    {
      "epoch": 2.0097493036211698,
      "grad_norm": 4.5031328201293945,
      "learning_rate": 5.7076463899943145e-05,
      "loss": 4.3625,
      "step": 1443
    },
    {
      "epoch": 2.011142061281337,
      "grad_norm": 3.8418807983398438,
      "learning_rate": 5.706651506537805e-05,
      "loss": 3.8717,
      "step": 1444
    },
    {
      "epoch": 2.012534818941504,
      "grad_norm": 3.67148494720459,
      "learning_rate": 5.705656623081296e-05,
      "loss": 4.0714,
      "step": 1445
    },
    {
      "epoch": 2.013927576601671,
      "grad_norm": 4.8723320960998535,
      "learning_rate": 5.7046617396247864e-05,
      "loss": 4.1978,
      "step": 1446
    },
    {
      "epoch": 2.0153203342618387,
      "grad_norm": 4.041756629943848,
      "learning_rate": 5.703666856168277e-05,
      "loss": 4.7872,
      "step": 1447
    },
    {
      "epoch": 2.0167130919220058,
      "grad_norm": 4.530393600463867,
      "learning_rate": 5.702671972711767e-05,
      "loss": 4.1579,
      "step": 1448
    },
    {
      "epoch": 2.018105849582173,
      "grad_norm": 4.1396918296813965,
      "learning_rate": 5.701677089255258e-05,
      "loss": 4.4417,
      "step": 1449
    },
    {
      "epoch": 2.01949860724234,
      "grad_norm": 6.0312323570251465,
      "learning_rate": 5.700682205798749e-05,
      "loss": 4.3929,
      "step": 1450
    },
    {
      "epoch": 2.020891364902507,
      "grad_norm": 4.5816874504089355,
      "learning_rate": 5.69968732234224e-05,
      "loss": 4.5516,
      "step": 1451
    },
    {
      "epoch": 2.022284122562674,
      "grad_norm": 4.764605522155762,
      "learning_rate": 5.6986924388857295e-05,
      "loss": 4.2814,
      "step": 1452
    },
    {
      "epoch": 2.0236768802228413,
      "grad_norm": 4.251420021057129,
      "learning_rate": 5.6976975554292206e-05,
      "loss": 4.2084,
      "step": 1453
    },
    {
      "epoch": 2.0250696378830084,
      "grad_norm": 4.817270755767822,
      "learning_rate": 5.696702671972711e-05,
      "loss": 4.5213,
      "step": 1454
    },
    {
      "epoch": 2.0264623955431755,
      "grad_norm": 5.079899311065674,
      "learning_rate": 5.695707788516202e-05,
      "loss": 4.439,
      "step": 1455
    },
    {
      "epoch": 2.0278551532033426,
      "grad_norm": 5.405300617218018,
      "learning_rate": 5.694712905059692e-05,
      "loss": 3.9677,
      "step": 1456
    },
    {
      "epoch": 2.0292479108635098,
      "grad_norm": 3.9286108016967773,
      "learning_rate": 5.693718021603183e-05,
      "loss": 4.2411,
      "step": 1457
    },
    {
      "epoch": 2.030640668523677,
      "grad_norm": 3.5010762214660645,
      "learning_rate": 5.692723138146674e-05,
      "loss": 3.9355,
      "step": 1458
    },
    {
      "epoch": 2.032033426183844,
      "grad_norm": 4.042318820953369,
      "learning_rate": 5.6917282546901644e-05,
      "loss": 4.0139,
      "step": 1459
    },
    {
      "epoch": 2.033426183844011,
      "grad_norm": 4.314542293548584,
      "learning_rate": 5.6907333712336555e-05,
      "loss": 4.1083,
      "step": 1460
    },
    {
      "epoch": 2.034818941504178,
      "grad_norm": 4.028563976287842,
      "learning_rate": 5.689738487777145e-05,
      "loss": 4.2008,
      "step": 1461
    },
    {
      "epoch": 2.0362116991643453,
      "grad_norm": 5.131320476531982,
      "learning_rate": 5.688743604320636e-05,
      "loss": 4.6816,
      "step": 1462
    },
    {
      "epoch": 2.0376044568245124,
      "grad_norm": 4.556960105895996,
      "learning_rate": 5.687748720864127e-05,
      "loss": 4.4714,
      "step": 1463
    },
    {
      "epoch": 2.0389972144846795,
      "grad_norm": 4.0364251136779785,
      "learning_rate": 5.686753837407618e-05,
      "loss": 3.961,
      "step": 1464
    },
    {
      "epoch": 2.0403899721448466,
      "grad_norm": 3.756049871444702,
      "learning_rate": 5.685758953951108e-05,
      "loss": 4.0901,
      "step": 1465
    },
    {
      "epoch": 2.0417827298050137,
      "grad_norm": 3.9342846870422363,
      "learning_rate": 5.6847640704945986e-05,
      "loss": 4.273,
      "step": 1466
    },
    {
      "epoch": 2.043175487465181,
      "grad_norm": 4.469848155975342,
      "learning_rate": 5.683769187038089e-05,
      "loss": 4.0942,
      "step": 1467
    },
    {
      "epoch": 2.0445682451253484,
      "grad_norm": 4.504777908325195,
      "learning_rate": 5.68277430358158e-05,
      "loss": 4.1311,
      "step": 1468
    },
    {
      "epoch": 2.0459610027855155,
      "grad_norm": 6.825126647949219,
      "learning_rate": 5.6817794201250705e-05,
      "loss": 4.2753,
      "step": 1469
    },
    {
      "epoch": 2.0473537604456826,
      "grad_norm": 4.323886394500732,
      "learning_rate": 5.6807845366685616e-05,
      "loss": 4.7228,
      "step": 1470
    },
    {
      "epoch": 2.0487465181058497,
      "grad_norm": 3.9217405319213867,
      "learning_rate": 5.6797896532120513e-05,
      "loss": 4.4236,
      "step": 1471
    },
    {
      "epoch": 2.050139275766017,
      "grad_norm": 3.924076795578003,
      "learning_rate": 5.6787947697555424e-05,
      "loss": 4.2186,
      "step": 1472
    },
    {
      "epoch": 2.051532033426184,
      "grad_norm": 4.097128868103027,
      "learning_rate": 5.677799886299033e-05,
      "loss": 4.4548,
      "step": 1473
    },
    {
      "epoch": 2.052924791086351,
      "grad_norm": 4.621747016906738,
      "learning_rate": 5.676805002842524e-05,
      "loss": 4.4286,
      "step": 1474
    },
    {
      "epoch": 2.054317548746518,
      "grad_norm": 4.6758856773376465,
      "learning_rate": 5.675810119386014e-05,
      "loss": 4.2658,
      "step": 1475
    },
    {
      "epoch": 2.0557103064066853,
      "grad_norm": 4.02954626083374,
      "learning_rate": 5.674815235929505e-05,
      "loss": 4.005,
      "step": 1476
    },
    {
      "epoch": 2.0571030640668524,
      "grad_norm": 3.7476682662963867,
      "learning_rate": 5.673820352472995e-05,
      "loss": 3.9305,
      "step": 1477
    },
    {
      "epoch": 2.0584958217270195,
      "grad_norm": 3.8278512954711914,
      "learning_rate": 5.672825469016486e-05,
      "loss": 4.4554,
      "step": 1478
    },
    {
      "epoch": 2.0598885793871866,
      "grad_norm": 4.185937881469727,
      "learning_rate": 5.6718305855599766e-05,
      "loss": 4.175,
      "step": 1479
    },
    {
      "epoch": 2.0612813370473537,
      "grad_norm": 3.753911018371582,
      "learning_rate": 5.670835702103468e-05,
      "loss": 4.3212,
      "step": 1480
    },
    {
      "epoch": 2.062674094707521,
      "grad_norm": 4.38871431350708,
      "learning_rate": 5.6698408186469575e-05,
      "loss": 4.4653,
      "step": 1481
    },
    {
      "epoch": 2.064066852367688,
      "grad_norm": 5.867580413818359,
      "learning_rate": 5.6688459351904485e-05,
      "loss": 4.1974,
      "step": 1482
    },
    {
      "epoch": 2.065459610027855,
      "grad_norm": 4.391541481018066,
      "learning_rate": 5.667851051733939e-05,
      "loss": 4.0392,
      "step": 1483
    },
    {
      "epoch": 2.066852367688022,
      "grad_norm": 3.861218214035034,
      "learning_rate": 5.66685616827743e-05,
      "loss": 4.1293,
      "step": 1484
    },
    {
      "epoch": 2.0682451253481893,
      "grad_norm": 4.242783069610596,
      "learning_rate": 5.665861284820921e-05,
      "loss": 4.0947,
      "step": 1485
    },
    {
      "epoch": 2.0696378830083564,
      "grad_norm": 4.565762519836426,
      "learning_rate": 5.664866401364411e-05,
      "loss": 4.0485,
      "step": 1486
    },
    {
      "epoch": 2.0710306406685235,
      "grad_norm": 5.0193963050842285,
      "learning_rate": 5.663871517907902e-05,
      "loss": 3.8263,
      "step": 1487
    },
    {
      "epoch": 2.0724233983286906,
      "grad_norm": 3.990995168685913,
      "learning_rate": 5.6628766344513923e-05,
      "loss": 4.1442,
      "step": 1488
    },
    {
      "epoch": 2.073816155988858,
      "grad_norm": 4.26665735244751,
      "learning_rate": 5.6618817509948834e-05,
      "loss": 4.2591,
      "step": 1489
    },
    {
      "epoch": 2.0752089136490253,
      "grad_norm": 5.239255428314209,
      "learning_rate": 5.660886867538374e-05,
      "loss": 4.3352,
      "step": 1490
    },
    {
      "epoch": 2.0766016713091924,
      "grad_norm": 4.049388885498047,
      "learning_rate": 5.659891984081864e-05,
      "loss": 3.8159,
      "step": 1491
    },
    {
      "epoch": 2.0779944289693595,
      "grad_norm": 4.513409614562988,
      "learning_rate": 5.6588971006253546e-05,
      "loss": 3.9364,
      "step": 1492
    },
    {
      "epoch": 2.0793871866295266,
      "grad_norm": 5.870997428894043,
      "learning_rate": 5.657902217168846e-05,
      "loss": 4.2862,
      "step": 1493
    },
    {
      "epoch": 2.0807799442896937,
      "grad_norm": 4.047566890716553,
      "learning_rate": 5.656907333712336e-05,
      "loss": 4.6385,
      "step": 1494
    },
    {
      "epoch": 2.082172701949861,
      "grad_norm": 4.456325531005859,
      "learning_rate": 5.655912450255827e-05,
      "loss": 4.2397,
      "step": 1495
    },
    {
      "epoch": 2.083565459610028,
      "grad_norm": 3.933844804763794,
      "learning_rate": 5.654917566799317e-05,
      "loss": 4.3366,
      "step": 1496
    },
    {
      "epoch": 2.084958217270195,
      "grad_norm": 4.979091644287109,
      "learning_rate": 5.653922683342808e-05,
      "loss": 4.1172,
      "step": 1497
    },
    {
      "epoch": 2.086350974930362,
      "grad_norm": 4.2216668128967285,
      "learning_rate": 5.6529277998862984e-05,
      "loss": 3.8411,
      "step": 1498
    },
    {
      "epoch": 2.0877437325905293,
      "grad_norm": 4.793976783752441,
      "learning_rate": 5.6519329164297895e-05,
      "loss": 4.4086,
      "step": 1499
    },
    {
      "epoch": 2.0891364902506964,
      "grad_norm": 4.201914310455322,
      "learning_rate": 5.650938032973279e-05,
      "loss": 4.7244,
      "step": 1500
    },
    {
      "epoch": 2.0905292479108635,
      "grad_norm": 4.169383525848389,
      "learning_rate": 5.6499431495167703e-05,
      "loss": 4.1203,
      "step": 1501
    },
    {
      "epoch": 2.0919220055710306,
      "grad_norm": 4.854778289794922,
      "learning_rate": 5.648948266060261e-05,
      "loss": 4.4488,
      "step": 1502
    },
    {
      "epoch": 2.0933147632311977,
      "grad_norm": 3.9533584117889404,
      "learning_rate": 5.647953382603752e-05,
      "loss": 4.0845,
      "step": 1503
    },
    {
      "epoch": 2.094707520891365,
      "grad_norm": 4.016109943389893,
      "learning_rate": 5.646958499147242e-05,
      "loss": 4.1466,
      "step": 1504
    },
    {
      "epoch": 2.096100278551532,
      "grad_norm": 4.391005516052246,
      "learning_rate": 5.6459636156907327e-05,
      "loss": 4.1644,
      "step": 1505
    },
    {
      "epoch": 2.097493036211699,
      "grad_norm": 3.59956693649292,
      "learning_rate": 5.644968732234223e-05,
      "loss": 3.8031,
      "step": 1506
    },
    {
      "epoch": 2.098885793871866,
      "grad_norm": 4.031566143035889,
      "learning_rate": 5.643973848777714e-05,
      "loss": 4.1626,
      "step": 1507
    },
    {
      "epoch": 2.1002785515320332,
      "grad_norm": 4.36628532409668,
      "learning_rate": 5.6429789653212046e-05,
      "loss": 4.2109,
      "step": 1508
    },
    {
      "epoch": 2.1016713091922004,
      "grad_norm": 4.576984882354736,
      "learning_rate": 5.6419840818646956e-05,
      "loss": 4.2118,
      "step": 1509
    },
    {
      "epoch": 2.103064066852368,
      "grad_norm": 4.1856794357299805,
      "learning_rate": 5.6409891984081854e-05,
      "loss": 4.9522,
      "step": 1510
    },
    {
      "epoch": 2.104456824512535,
      "grad_norm": 5.25437593460083,
      "learning_rate": 5.6399943149516765e-05,
      "loss": 4.12,
      "step": 1511
    },
    {
      "epoch": 2.105849582172702,
      "grad_norm": 4.273394584655762,
      "learning_rate": 5.638999431495167e-05,
      "loss": 4.6421,
      "step": 1512
    },
    {
      "epoch": 2.1072423398328692,
      "grad_norm": 4.079535007476807,
      "learning_rate": 5.638004548038658e-05,
      "loss": 3.9771,
      "step": 1513
    },
    {
      "epoch": 2.1086350974930363,
      "grad_norm": 4.427313804626465,
      "learning_rate": 5.637009664582149e-05,
      "loss": 4.2386,
      "step": 1514
    },
    {
      "epoch": 2.1100278551532035,
      "grad_norm": 4.569276809692383,
      "learning_rate": 5.636014781125639e-05,
      "loss": 4.9382,
      "step": 1515
    },
    {
      "epoch": 2.1114206128133706,
      "grad_norm": 4.124020099639893,
      "learning_rate": 5.63501989766913e-05,
      "loss": 4.2301,
      "step": 1516
    },
    {
      "epoch": 2.1128133704735377,
      "grad_norm": 4.050805568695068,
      "learning_rate": 5.63402501421262e-05,
      "loss": 3.7955,
      "step": 1517
    },
    {
      "epoch": 2.114206128133705,
      "grad_norm": 4.00832462310791,
      "learning_rate": 5.6330301307561113e-05,
      "loss": 4.3878,
      "step": 1518
    },
    {
      "epoch": 2.115598885793872,
      "grad_norm": 3.9795923233032227,
      "learning_rate": 5.632035247299602e-05,
      "loss": 4.6169,
      "step": 1519
    },
    {
      "epoch": 2.116991643454039,
      "grad_norm": 4.110733985900879,
      "learning_rate": 5.631040363843092e-05,
      "loss": 4.2281,
      "step": 1520
    },
    {
      "epoch": 2.118384401114206,
      "grad_norm": 3.6762120723724365,
      "learning_rate": 5.6300454803865826e-05,
      "loss": 3.7926,
      "step": 1521
    },
    {
      "epoch": 2.1197771587743732,
      "grad_norm": 5.0108256340026855,
      "learning_rate": 5.6290505969300736e-05,
      "loss": 3.8425,
      "step": 1522
    },
    {
      "epoch": 2.1211699164345403,
      "grad_norm": 4.064860820770264,
      "learning_rate": 5.628055713473564e-05,
      "loss": 4.1895,
      "step": 1523
    },
    {
      "epoch": 2.1225626740947074,
      "grad_norm": 4.2516703605651855,
      "learning_rate": 5.627060830017055e-05,
      "loss": 4.0237,
      "step": 1524
    },
    {
      "epoch": 2.1239554317548746,
      "grad_norm": 3.8588855266571045,
      "learning_rate": 5.626065946560545e-05,
      "loss": 3.8886,
      "step": 1525
    },
    {
      "epoch": 2.1253481894150417,
      "grad_norm": 4.1031575202941895,
      "learning_rate": 5.625071063104036e-05,
      "loss": 4.1133,
      "step": 1526
    },
    {
      "epoch": 2.1267409470752088,
      "grad_norm": 4.086325168609619,
      "learning_rate": 5.6240761796475264e-05,
      "loss": 3.9396,
      "step": 1527
    },
    {
      "epoch": 2.128133704735376,
      "grad_norm": 4.331142902374268,
      "learning_rate": 5.6230812961910174e-05,
      "loss": 4.292,
      "step": 1528
    },
    {
      "epoch": 2.129526462395543,
      "grad_norm": 4.0734381675720215,
      "learning_rate": 5.622086412734508e-05,
      "loss": 4.1097,
      "step": 1529
    },
    {
      "epoch": 2.13091922005571,
      "grad_norm": 4.513016700744629,
      "learning_rate": 5.621091529277998e-05,
      "loss": 4.2115,
      "step": 1530
    },
    {
      "epoch": 2.1323119777158777,
      "grad_norm": 4.545360088348389,
      "learning_rate": 5.620096645821489e-05,
      "loss": 4.2847,
      "step": 1531
    },
    {
      "epoch": 2.1337047353760448,
      "grad_norm": 4.601300239562988,
      "learning_rate": 5.61910176236498e-05,
      "loss": 4.2504,
      "step": 1532
    },
    {
      "epoch": 2.135097493036212,
      "grad_norm": 4.783657550811768,
      "learning_rate": 5.61810687890847e-05,
      "loss": 4.0739,
      "step": 1533
    },
    {
      "epoch": 2.136490250696379,
      "grad_norm": 4.459740161895752,
      "learning_rate": 5.617111995451961e-05,
      "loss": 4.1157,
      "step": 1534
    },
    {
      "epoch": 2.137883008356546,
      "grad_norm": 4.170529842376709,
      "learning_rate": 5.616117111995451e-05,
      "loss": 3.8235,
      "step": 1535
    },
    {
      "epoch": 2.139275766016713,
      "grad_norm": 4.170540809631348,
      "learning_rate": 5.615122228538942e-05,
      "loss": 4.309,
      "step": 1536
    },
    {
      "epoch": 2.1406685236768803,
      "grad_norm": 3.9952211380004883,
      "learning_rate": 5.6141273450824325e-05,
      "loss": 4.7077,
      "step": 1537
    },
    {
      "epoch": 2.1420612813370474,
      "grad_norm": 5.134542465209961,
      "learning_rate": 5.6131324616259236e-05,
      "loss": 3.7039,
      "step": 1538
    },
    {
      "epoch": 2.1434540389972145,
      "grad_norm": 4.411221981048584,
      "learning_rate": 5.612137578169413e-05,
      "loss": 4.0268,
      "step": 1539
    },
    {
      "epoch": 2.1448467966573816,
      "grad_norm": 4.216711044311523,
      "learning_rate": 5.6111426947129044e-05,
      "loss": 4.642,
      "step": 1540
    },
    {
      "epoch": 2.1462395543175488,
      "grad_norm": 4.027458190917969,
      "learning_rate": 5.6101478112563955e-05,
      "loss": 4.0452,
      "step": 1541
    },
    {
      "epoch": 2.147632311977716,
      "grad_norm": 5.340506076812744,
      "learning_rate": 5.609152927799886e-05,
      "loss": 4.4872,
      "step": 1542
    },
    {
      "epoch": 2.149025069637883,
      "grad_norm": 4.38348388671875,
      "learning_rate": 5.608158044343377e-05,
      "loss": 4.3302,
      "step": 1543
    },
    {
      "epoch": 2.15041782729805,
      "grad_norm": 4.876546382904053,
      "learning_rate": 5.607163160886867e-05,
      "loss": 3.5804,
      "step": 1544
    },
    {
      "epoch": 2.151810584958217,
      "grad_norm": 4.479104995727539,
      "learning_rate": 5.606168277430358e-05,
      "loss": 3.9673,
      "step": 1545
    },
    {
      "epoch": 2.1532033426183843,
      "grad_norm": 4.716986656188965,
      "learning_rate": 5.605173393973848e-05,
      "loss": 4.1712,
      "step": 1546
    },
    {
      "epoch": 2.1545961002785514,
      "grad_norm": 4.21389102935791,
      "learning_rate": 5.604178510517339e-05,
      "loss": 4.393,
      "step": 1547
    },
    {
      "epoch": 2.1559888579387185,
      "grad_norm": 5.340983867645264,
      "learning_rate": 5.60318362706083e-05,
      "loss": 4.8122,
      "step": 1548
    },
    {
      "epoch": 2.1573816155988856,
      "grad_norm": 4.190916538238525,
      "learning_rate": 5.60218874360432e-05,
      "loss": 4.5195,
      "step": 1549
    },
    {
      "epoch": 2.1587743732590527,
      "grad_norm": 4.221656799316406,
      "learning_rate": 5.6011938601478105e-05,
      "loss": 4.0933,
      "step": 1550
    },
    {
      "epoch": 2.16016713091922,
      "grad_norm": 3.9489657878875732,
      "learning_rate": 5.6001989766913016e-05,
      "loss": 3.8152,
      "step": 1551
    },
    {
      "epoch": 2.1615598885793874,
      "grad_norm": 4.408094882965088,
      "learning_rate": 5.599204093234792e-05,
      "loss": 3.9514,
      "step": 1552
    },
    {
      "epoch": 2.1629526462395545,
      "grad_norm": 4.304285049438477,
      "learning_rate": 5.598209209778283e-05,
      "loss": 4.5502,
      "step": 1553
    },
    {
      "epoch": 2.1643454038997216,
      "grad_norm": 4.349761962890625,
      "learning_rate": 5.597214326321773e-05,
      "loss": 3.9968,
      "step": 1554
    },
    {
      "epoch": 2.1657381615598887,
      "grad_norm": 4.542641639709473,
      "learning_rate": 5.596219442865264e-05,
      "loss": 3.821,
      "step": 1555
    },
    {
      "epoch": 2.167130919220056,
      "grad_norm": 3.8207194805145264,
      "learning_rate": 5.595224559408754e-05,
      "loss": 3.6482,
      "step": 1556
    },
    {
      "epoch": 2.168523676880223,
      "grad_norm": 4.042953014373779,
      "learning_rate": 5.5942296759522454e-05,
      "loss": 4.6605,
      "step": 1557
    },
    {
      "epoch": 2.16991643454039,
      "grad_norm": 4.262577533721924,
      "learning_rate": 5.593234792495736e-05,
      "loss": 4.587,
      "step": 1558
    },
    {
      "epoch": 2.171309192200557,
      "grad_norm": 4.657623291015625,
      "learning_rate": 5.592239909039226e-05,
      "loss": 4.4044,
      "step": 1559
    },
    {
      "epoch": 2.1727019498607243,
      "grad_norm": 4.761733531951904,
      "learning_rate": 5.5912450255827166e-05,
      "loss": 4.3288,
      "step": 1560
    },
    {
      "epoch": 2.1740947075208914,
      "grad_norm": 4.737750053405762,
      "learning_rate": 5.590250142126208e-05,
      "loss": 4.235,
      "step": 1561
    },
    {
      "epoch": 2.1754874651810585,
      "grad_norm": 4.017462253570557,
      "learning_rate": 5.589255258669698e-05,
      "loss": 4.2454,
      "step": 1562
    },
    {
      "epoch": 2.1768802228412256,
      "grad_norm": 4.1256585121154785,
      "learning_rate": 5.588260375213189e-05,
      "loss": 4.2497,
      "step": 1563
    },
    {
      "epoch": 2.1782729805013927,
      "grad_norm": 8.097579002380371,
      "learning_rate": 5.587265491756679e-05,
      "loss": 4.1784,
      "step": 1564
    },
    {
      "epoch": 2.17966573816156,
      "grad_norm": 4.777957916259766,
      "learning_rate": 5.58627060830017e-05,
      "loss": 4.0151,
      "step": 1565
    },
    {
      "epoch": 2.181058495821727,
      "grad_norm": 4.157647609710693,
      "learning_rate": 5.5852757248436604e-05,
      "loss": 3.9234,
      "step": 1566
    },
    {
      "epoch": 2.182451253481894,
      "grad_norm": 3.7936336994171143,
      "learning_rate": 5.5842808413871515e-05,
      "loss": 4.0243,
      "step": 1567
    },
    {
      "epoch": 2.183844011142061,
      "grad_norm": 5.389497756958008,
      "learning_rate": 5.5832859579306426e-05,
      "loss": 4.4381,
      "step": 1568
    },
    {
      "epoch": 2.1852367688022283,
      "grad_norm": 7.03663444519043,
      "learning_rate": 5.582291074474132e-05,
      "loss": 4.0193,
      "step": 1569
    },
    {
      "epoch": 2.1866295264623954,
      "grad_norm": 4.235686779022217,
      "learning_rate": 5.5812961910176234e-05,
      "loss": 4.0826,
      "step": 1570
    },
    {
      "epoch": 2.1880222841225625,
      "grad_norm": 4.34411096572876,
      "learning_rate": 5.580301307561114e-05,
      "loss": 4.3301,
      "step": 1571
    },
    {
      "epoch": 2.1894150417827296,
      "grad_norm": 4.225539684295654,
      "learning_rate": 5.579306424104605e-05,
      "loss": 4.5005,
      "step": 1572
    },
    {
      "epoch": 2.190807799442897,
      "grad_norm": 4.322992324829102,
      "learning_rate": 5.578311540648095e-05,
      "loss": 4.3125,
      "step": 1573
    },
    {
      "epoch": 2.1922005571030643,
      "grad_norm": 3.9500017166137695,
      "learning_rate": 5.577316657191586e-05,
      "loss": 3.9417,
      "step": 1574
    },
    {
      "epoch": 2.1935933147632314,
      "grad_norm": 3.9060707092285156,
      "learning_rate": 5.576321773735076e-05,
      "loss": 4.125,
      "step": 1575
    },
    {
      "epoch": 2.1949860724233985,
      "grad_norm": 4.2026286125183105,
      "learning_rate": 5.575326890278567e-05,
      "loss": 3.6776,
      "step": 1576
    },
    {
      "epoch": 2.1963788300835656,
      "grad_norm": 4.400326251983643,
      "learning_rate": 5.5743320068220576e-05,
      "loss": 4.4655,
      "step": 1577
    },
    {
      "epoch": 2.1977715877437327,
      "grad_norm": 5.644049644470215,
      "learning_rate": 5.573337123365549e-05,
      "loss": 4.0425,
      "step": 1578
    },
    {
      "epoch": 2.1991643454039,
      "grad_norm": 6.8138251304626465,
      "learning_rate": 5.5723422399090384e-05,
      "loss": 3.9512,
      "step": 1579
    },
    {
      "epoch": 2.200557103064067,
      "grad_norm": 5.015941619873047,
      "learning_rate": 5.5713473564525295e-05,
      "loss": 4.0271,
      "step": 1580
    },
    {
      "epoch": 2.201949860724234,
      "grad_norm": 4.783520698547363,
      "learning_rate": 5.57035247299602e-05,
      "loss": 4.2316,
      "step": 1581
    },
    {
      "epoch": 2.203342618384401,
      "grad_norm": 4.280081748962402,
      "learning_rate": 5.569357589539511e-05,
      "loss": 4.0603,
      "step": 1582
    },
    {
      "epoch": 2.2047353760445683,
      "grad_norm": 5.039388656616211,
      "learning_rate": 5.568362706083001e-05,
      "loss": 3.9209,
      "step": 1583
    },
    {
      "epoch": 2.2061281337047354,
      "grad_norm": 4.157800197601318,
      "learning_rate": 5.567367822626492e-05,
      "loss": 3.6996,
      "step": 1584
    },
    {
      "epoch": 2.2075208913649025,
      "grad_norm": 4.173427581787109,
      "learning_rate": 5.566372939169982e-05,
      "loss": 4.289,
      "step": 1585
    },
    {
      "epoch": 2.2089136490250696,
      "grad_norm": 4.1454267501831055,
      "learning_rate": 5.565378055713473e-05,
      "loss": 3.8151,
      "step": 1586
    },
    {
      "epoch": 2.2103064066852367,
      "grad_norm": 4.3231401443481445,
      "learning_rate": 5.564383172256964e-05,
      "loss": 4.2357,
      "step": 1587
    },
    {
      "epoch": 2.211699164345404,
      "grad_norm": 4.394160270690918,
      "learning_rate": 5.563388288800454e-05,
      "loss": 3.8161,
      "step": 1588
    },
    {
      "epoch": 2.213091922005571,
      "grad_norm": 5.923463344573975,
      "learning_rate": 5.5623934053439445e-05,
      "loss": 4.2231,
      "step": 1589
    },
    {
      "epoch": 2.214484679665738,
      "grad_norm": 4.116878032684326,
      "learning_rate": 5.5613985218874356e-05,
      "loss": 4.1096,
      "step": 1590
    },
    {
      "epoch": 2.215877437325905,
      "grad_norm": 4.993494033813477,
      "learning_rate": 5.560403638430926e-05,
      "loss": 4.1901,
      "step": 1591
    },
    {
      "epoch": 2.2172701949860723,
      "grad_norm": 4.756366729736328,
      "learning_rate": 5.559408754974417e-05,
      "loss": 4.1018,
      "step": 1592
    },
    {
      "epoch": 2.2186629526462394,
      "grad_norm": 4.143631935119629,
      "learning_rate": 5.558413871517907e-05,
      "loss": 4.0226,
      "step": 1593
    },
    {
      "epoch": 2.220055710306407,
      "grad_norm": 4.060478210449219,
      "learning_rate": 5.557418988061398e-05,
      "loss": 4.1714,
      "step": 1594
    },
    {
      "epoch": 2.2214484679665736,
      "grad_norm": 5.196919918060303,
      "learning_rate": 5.556424104604888e-05,
      "loss": 3.6259,
      "step": 1595
    },
    {
      "epoch": 2.222841225626741,
      "grad_norm": 4.83811616897583,
      "learning_rate": 5.5554292211483794e-05,
      "loss": 4.1439,
      "step": 1596
    },
    {
      "epoch": 2.2242339832869082,
      "grad_norm": 4.05018949508667,
      "learning_rate": 5.5544343376918705e-05,
      "loss": 3.5583,
      "step": 1597
    },
    {
      "epoch": 2.2256267409470754,
      "grad_norm": 4.4459638595581055,
      "learning_rate": 5.55343945423536e-05,
      "loss": 4.5046,
      "step": 1598
    },
    {
      "epoch": 2.2270194986072425,
      "grad_norm": 4.6933417320251465,
      "learning_rate": 5.552444570778851e-05,
      "loss": 4.2784,
      "step": 1599
    },
    {
      "epoch": 2.2284122562674096,
      "grad_norm": 4.105106830596924,
      "learning_rate": 5.551449687322342e-05,
      "loss": 4.1781,
      "step": 1600
    },
    {
      "epoch": 2.2298050139275767,
      "grad_norm": 4.288883209228516,
      "learning_rate": 5.550454803865833e-05,
      "loss": 4.2866,
      "step": 1601
    },
    {
      "epoch": 2.231197771587744,
      "grad_norm": 5.926295757293701,
      "learning_rate": 5.549459920409323e-05,
      "loss": 4.2175,
      "step": 1602
    },
    {
      "epoch": 2.232590529247911,
      "grad_norm": 3.661668539047241,
      "learning_rate": 5.5484650369528136e-05,
      "loss": 4.0713,
      "step": 1603
    },
    {
      "epoch": 2.233983286908078,
      "grad_norm": 3.785322904586792,
      "learning_rate": 5.547470153496304e-05,
      "loss": 4.1502,
      "step": 1604
    },
    {
      "epoch": 2.235376044568245,
      "grad_norm": 4.111382007598877,
      "learning_rate": 5.546475270039795e-05,
      "loss": 3.8881,
      "step": 1605
    },
    {
      "epoch": 2.2367688022284122,
      "grad_norm": 4.228845596313477,
      "learning_rate": 5.5454803865832855e-05,
      "loss": 3.7675,
      "step": 1606
    },
    {
      "epoch": 2.2381615598885793,
      "grad_norm": 3.797921895980835,
      "learning_rate": 5.5444855031267766e-05,
      "loss": 4.4189,
      "step": 1607
    },
    {
      "epoch": 2.2395543175487465,
      "grad_norm": 3.7675411701202393,
      "learning_rate": 5.543490619670266e-05,
      "loss": 3.5243,
      "step": 1608
    },
    {
      "epoch": 2.2409470752089136,
      "grad_norm": 4.016136646270752,
      "learning_rate": 5.5424957362137574e-05,
      "loss": 3.9185,
      "step": 1609
    },
    {
      "epoch": 2.2423398328690807,
      "grad_norm": 3.8654441833496094,
      "learning_rate": 5.541500852757248e-05,
      "loss": 3.8333,
      "step": 1610
    },
    {
      "epoch": 2.243732590529248,
      "grad_norm": 4.411679744720459,
      "learning_rate": 5.540505969300739e-05,
      "loss": 3.6448,
      "step": 1611
    },
    {
      "epoch": 2.245125348189415,
      "grad_norm": 4.223076820373535,
      "learning_rate": 5.539511085844229e-05,
      "loss": 4.1067,
      "step": 1612
    },
    {
      "epoch": 2.246518105849582,
      "grad_norm": 3.976116895675659,
      "learning_rate": 5.53851620238772e-05,
      "loss": 3.9666,
      "step": 1613
    },
    {
      "epoch": 2.247910863509749,
      "grad_norm": 4.159890174865723,
      "learning_rate": 5.53752131893121e-05,
      "loss": 4.2775,
      "step": 1614
    },
    {
      "epoch": 2.2493036211699167,
      "grad_norm": 4.389676570892334,
      "learning_rate": 5.536526435474701e-05,
      "loss": 4.1369,
      "step": 1615
    },
    {
      "epoch": 2.2506963788300833,
      "grad_norm": 5.163928985595703,
      "learning_rate": 5.5355315520181916e-05,
      "loss": 4.3889,
      "step": 1616
    },
    {
      "epoch": 2.252089136490251,
      "grad_norm": 4.48787784576416,
      "learning_rate": 5.534536668561683e-05,
      "loss": 4.2583,
      "step": 1617
    },
    {
      "epoch": 2.253481894150418,
      "grad_norm": 4.052666187286377,
      "learning_rate": 5.5335417851051724e-05,
      "loss": 4.0006,
      "step": 1618
    },
    {
      "epoch": 2.254874651810585,
      "grad_norm": 4.186406135559082,
      "learning_rate": 5.5325469016486635e-05,
      "loss": 4.1602,
      "step": 1619
    },
    {
      "epoch": 2.256267409470752,
      "grad_norm": 7.1013383865356445,
      "learning_rate": 5.531552018192154e-05,
      "loss": 4.0932,
      "step": 1620
    },
    {
      "epoch": 2.2576601671309193,
      "grad_norm": 3.8866419792175293,
      "learning_rate": 5.530557134735645e-05,
      "loss": 3.7892,
      "step": 1621
    },
    {
      "epoch": 2.2590529247910864,
      "grad_norm": 3.7705419063568115,
      "learning_rate": 5.529562251279136e-05,
      "loss": 3.7626,
      "step": 1622
    },
    {
      "epoch": 2.2604456824512535,
      "grad_norm": 3.958216428756714,
      "learning_rate": 5.528567367822626e-05,
      "loss": 4.2034,
      "step": 1623
    },
    {
      "epoch": 2.2618384401114207,
      "grad_norm": 4.2496771812438965,
      "learning_rate": 5.527572484366116e-05,
      "loss": 4.0623,
      "step": 1624
    },
    {
      "epoch": 2.2632311977715878,
      "grad_norm": 4.305548191070557,
      "learning_rate": 5.526577600909607e-05,
      "loss": 4.2394,
      "step": 1625
    },
    {
      "epoch": 2.264623955431755,
      "grad_norm": 4.462040901184082,
      "learning_rate": 5.5255827174530984e-05,
      "loss": 4.3093,
      "step": 1626
    },
    {
      "epoch": 2.266016713091922,
      "grad_norm": 4.265384197235107,
      "learning_rate": 5.524587833996588e-05,
      "loss": 3.8959,
      "step": 1627
    },
    {
      "epoch": 2.267409470752089,
      "grad_norm": 3.958977222442627,
      "learning_rate": 5.523592950540079e-05,
      "loss": 3.7723,
      "step": 1628
    },
    {
      "epoch": 2.268802228412256,
      "grad_norm": 4.284531116485596,
      "learning_rate": 5.5225980670835696e-05,
      "loss": 3.8134,
      "step": 1629
    },
    {
      "epoch": 2.2701949860724233,
      "grad_norm": 4.096482753753662,
      "learning_rate": 5.521603183627061e-05,
      "loss": 4.0498,
      "step": 1630
    },
    {
      "epoch": 2.2715877437325904,
      "grad_norm": 4.294747829437256,
      "learning_rate": 5.520608300170551e-05,
      "loss": 3.9642,
      "step": 1631
    },
    {
      "epoch": 2.2729805013927575,
      "grad_norm": 4.197109699249268,
      "learning_rate": 5.5196134167140415e-05,
      "loss": 4.0461,
      "step": 1632
    },
    {
      "epoch": 2.2743732590529246,
      "grad_norm": 3.998181104660034,
      "learning_rate": 5.518618533257532e-05,
      "loss": 4.4055,
      "step": 1633
    },
    {
      "epoch": 2.2757660167130918,
      "grad_norm": 4.521604061126709,
      "learning_rate": 5.517623649801023e-05,
      "loss": 3.6582,
      "step": 1634
    },
    {
      "epoch": 2.277158774373259,
      "grad_norm": 4.392470359802246,
      "learning_rate": 5.5166287663445134e-05,
      "loss": 4.3384,
      "step": 1635
    },
    {
      "epoch": 2.2785515320334264,
      "grad_norm": 4.117228984832764,
      "learning_rate": 5.5156338828880045e-05,
      "loss": 3.6421,
      "step": 1636
    },
    {
      "epoch": 2.279944289693593,
      "grad_norm": 4.706968784332275,
      "learning_rate": 5.514638999431494e-05,
      "loss": 4.5145,
      "step": 1637
    },
    {
      "epoch": 2.2813370473537606,
      "grad_norm": 4.281352519989014,
      "learning_rate": 5.513644115974985e-05,
      "loss": 3.9539,
      "step": 1638
    },
    {
      "epoch": 2.2827298050139277,
      "grad_norm": 4.370902061462402,
      "learning_rate": 5.512649232518476e-05,
      "loss": 3.755,
      "step": 1639
    },
    {
      "epoch": 2.284122562674095,
      "grad_norm": 4.418724060058594,
      "learning_rate": 5.511654349061967e-05,
      "loss": 3.9502,
      "step": 1640
    },
    {
      "epoch": 2.285515320334262,
      "grad_norm": 4.078126907348633,
      "learning_rate": 5.510659465605457e-05,
      "loss": 3.6778,
      "step": 1641
    },
    {
      "epoch": 2.286908077994429,
      "grad_norm": 4.293686389923096,
      "learning_rate": 5.5096645821489476e-05,
      "loss": 3.4956,
      "step": 1642
    },
    {
      "epoch": 2.288300835654596,
      "grad_norm": 5.51296854019165,
      "learning_rate": 5.508669698692438e-05,
      "loss": 4.2429,
      "step": 1643
    },
    {
      "epoch": 2.2896935933147633,
      "grad_norm": 4.143667221069336,
      "learning_rate": 5.507674815235929e-05,
      "loss": 4.142,
      "step": 1644
    },
    {
      "epoch": 2.2910863509749304,
      "grad_norm": 3.7358782291412354,
      "learning_rate": 5.5066799317794195e-05,
      "loss": 3.9213,
      "step": 1645
    },
    {
      "epoch": 2.2924791086350975,
      "grad_norm": 4.326070785522461,
      "learning_rate": 5.5056850483229106e-05,
      "loss": 3.5167,
      "step": 1646
    },
    {
      "epoch": 2.2938718662952646,
      "grad_norm": 4.527500629425049,
      "learning_rate": 5.5046901648664e-05,
      "loss": 3.7828,
      "step": 1647
    },
    {
      "epoch": 2.2952646239554317,
      "grad_norm": 4.36574649810791,
      "learning_rate": 5.5036952814098914e-05,
      "loss": 4.2175,
      "step": 1648
    },
    {
      "epoch": 2.296657381615599,
      "grad_norm": 3.914255142211914,
      "learning_rate": 5.502700397953382e-05,
      "loss": 3.7232,
      "step": 1649
    },
    {
      "epoch": 2.298050139275766,
      "grad_norm": 4.246000289916992,
      "learning_rate": 5.501705514496873e-05,
      "loss": 4.3355,
      "step": 1650
    },
    {
      "epoch": 2.299442896935933,
      "grad_norm": 5.617733955383301,
      "learning_rate": 5.500710631040364e-05,
      "loss": 3.876,
      "step": 1651
    },
    {
      "epoch": 2.3008356545961,
      "grad_norm": 3.8478708267211914,
      "learning_rate": 5.499715747583854e-05,
      "loss": 4.0928,
      "step": 1652
    },
    {
      "epoch": 2.3022284122562673,
      "grad_norm": 4.152899742126465,
      "learning_rate": 5.498720864127344e-05,
      "loss": 3.7558,
      "step": 1653
    },
    {
      "epoch": 2.3036211699164344,
      "grad_norm": 4.24927282333374,
      "learning_rate": 5.497725980670835e-05,
      "loss": 4.2481,
      "step": 1654
    },
    {
      "epoch": 2.3050139275766015,
      "grad_norm": 3.5137290954589844,
      "learning_rate": 5.496731097214326e-05,
      "loss": 3.8982,
      "step": 1655
    },
    {
      "epoch": 2.3064066852367686,
      "grad_norm": 3.882204294204712,
      "learning_rate": 5.495736213757817e-05,
      "loss": 3.545,
      "step": 1656
    },
    {
      "epoch": 2.307799442896936,
      "grad_norm": 4.548402786254883,
      "learning_rate": 5.494741330301307e-05,
      "loss": 4.0283,
      "step": 1657
    },
    {
      "epoch": 2.309192200557103,
      "grad_norm": 3.8759939670562744,
      "learning_rate": 5.4937464468447975e-05,
      "loss": 4.1759,
      "step": 1658
    },
    {
      "epoch": 2.3105849582172704,
      "grad_norm": 4.622664928436279,
      "learning_rate": 5.4927515633882886e-05,
      "loss": 4.1117,
      "step": 1659
    },
    {
      "epoch": 2.3119777158774375,
      "grad_norm": 4.528932571411133,
      "learning_rate": 5.491756679931779e-05,
      "loss": 3.9803,
      "step": 1660
    },
    {
      "epoch": 2.3133704735376046,
      "grad_norm": 3.8661184310913086,
      "learning_rate": 5.49076179647527e-05,
      "loss": 4.0703,
      "step": 1661
    },
    {
      "epoch": 2.3147632311977717,
      "grad_norm": 4.365589618682861,
      "learning_rate": 5.48976691301876e-05,
      "loss": 4.349,
      "step": 1662
    },
    {
      "epoch": 2.316155988857939,
      "grad_norm": 4.564374923706055,
      "learning_rate": 5.488772029562251e-05,
      "loss": 3.8839,
      "step": 1663
    },
    {
      "epoch": 2.317548746518106,
      "grad_norm": 3.963067054748535,
      "learning_rate": 5.487777146105741e-05,
      "loss": 3.8955,
      "step": 1664
    },
    {
      "epoch": 2.318941504178273,
      "grad_norm": 4.266085147857666,
      "learning_rate": 5.4867822626492324e-05,
      "loss": 3.9313,
      "step": 1665
    },
    {
      "epoch": 2.32033426183844,
      "grad_norm": 4.137028217315674,
      "learning_rate": 5.485787379192723e-05,
      "loss": 3.6683,
      "step": 1666
    },
    {
      "epoch": 2.3217270194986073,
      "grad_norm": 3.8003323078155518,
      "learning_rate": 5.484792495736213e-05,
      "loss": 4.1103,
      "step": 1667
    },
    {
      "epoch": 2.3231197771587744,
      "grad_norm": 4.106907844543457,
      "learning_rate": 5.4837976122797036e-05,
      "loss": 3.9397,
      "step": 1668
    },
    {
      "epoch": 2.3245125348189415,
      "grad_norm": 4.142886638641357,
      "learning_rate": 5.482802728823195e-05,
      "loss": 4.1045,
      "step": 1669
    },
    {
      "epoch": 2.3259052924791086,
      "grad_norm": 4.56312370300293,
      "learning_rate": 5.481807845366685e-05,
      "loss": 3.7163,
      "step": 1670
    },
    {
      "epoch": 2.3272980501392757,
      "grad_norm": 4.902716159820557,
      "learning_rate": 5.4808129619101755e-05,
      "loss": 3.5701,
      "step": 1671
    },
    {
      "epoch": 2.328690807799443,
      "grad_norm": 4.238404273986816,
      "learning_rate": 5.479818078453666e-05,
      "loss": 4.078,
      "step": 1672
    },
    {
      "epoch": 2.33008356545961,
      "grad_norm": 5.5242133140563965,
      "learning_rate": 5.478823194997157e-05,
      "loss": 3.8578,
      "step": 1673
    },
    {
      "epoch": 2.331476323119777,
      "grad_norm": 5.297257900238037,
      "learning_rate": 5.4778283115406474e-05,
      "loss": 3.6976,
      "step": 1674
    },
    {
      "epoch": 2.332869080779944,
      "grad_norm": 4.090051651000977,
      "learning_rate": 5.4768334280841385e-05,
      "loss": 3.7153,
      "step": 1675
    },
    {
      "epoch": 2.3342618384401113,
      "grad_norm": 4.634399890899658,
      "learning_rate": 5.475838544627628e-05,
      "loss": 3.9084,
      "step": 1676
    },
    {
      "epoch": 2.3356545961002784,
      "grad_norm": 4.183865070343018,
      "learning_rate": 5.474843661171119e-05,
      "loss": 3.9903,
      "step": 1677
    },
    {
      "epoch": 2.337047353760446,
      "grad_norm": 4.400298118591309,
      "learning_rate": 5.47384877771461e-05,
      "loss": 3.9344,
      "step": 1678
    },
    {
      "epoch": 2.3384401114206126,
      "grad_norm": 4.828468322753906,
      "learning_rate": 5.472853894258101e-05,
      "loss": 4.3338,
      "step": 1679
    },
    {
      "epoch": 2.33983286908078,
      "grad_norm": 4.010335922241211,
      "learning_rate": 5.471859010801592e-05,
      "loss": 3.8026,
      "step": 1680
    },
    {
      "epoch": 2.3412256267409473,
      "grad_norm": 4.217126846313477,
      "learning_rate": 5.4708641273450816e-05,
      "loss": 4.0941,
      "step": 1681
    },
    {
      "epoch": 2.3426183844011144,
      "grad_norm": 11.835099220275879,
      "learning_rate": 5.469869243888573e-05,
      "loss": 4.6217,
      "step": 1682
    },
    {
      "epoch": 2.3440111420612815,
      "grad_norm": 4.50076961517334,
      "learning_rate": 5.468874360432063e-05,
      "loss": 3.9282,
      "step": 1683
    },
    {
      "epoch": 2.3454038997214486,
      "grad_norm": 4.27176570892334,
      "learning_rate": 5.467879476975554e-05,
      "loss": 3.771,
      "step": 1684
    },
    {
      "epoch": 2.3467966573816157,
      "grad_norm": 3.9045350551605225,
      "learning_rate": 5.4668845935190446e-05,
      "loss": 4.0347,
      "step": 1685
    },
    {
      "epoch": 2.348189415041783,
      "grad_norm": 4.14553165435791,
      "learning_rate": 5.465889710062535e-05,
      "loss": 3.8646,
      "step": 1686
    },
    {
      "epoch": 2.34958217270195,
      "grad_norm": 3.7285969257354736,
      "learning_rate": 5.4648948266060254e-05,
      "loss": 4.1016,
      "step": 1687
    },
    {
      "epoch": 2.350974930362117,
      "grad_norm": 3.9861230850219727,
      "learning_rate": 5.4638999431495165e-05,
      "loss": 3.8058,
      "step": 1688
    },
    {
      "epoch": 2.352367688022284,
      "grad_norm": 4.064573287963867,
      "learning_rate": 5.462905059693007e-05,
      "loss": 4.2472,
      "step": 1689
    },
    {
      "epoch": 2.3537604456824512,
      "grad_norm": 6.249232769012451,
      "learning_rate": 5.461910176236498e-05,
      "loss": 3.7758,
      "step": 1690
    },
    {
      "epoch": 2.3551532033426184,
      "grad_norm": 4.630408763885498,
      "learning_rate": 5.460915292779988e-05,
      "loss": 4.0598,
      "step": 1691
    },
    {
      "epoch": 2.3565459610027855,
      "grad_norm": 4.128727912902832,
      "learning_rate": 5.459920409323479e-05,
      "loss": 3.7818,
      "step": 1692
    },
    {
      "epoch": 2.3579387186629526,
      "grad_norm": 4.089692115783691,
      "learning_rate": 5.458925525866969e-05,
      "loss": 3.6243,
      "step": 1693
    },
    {
      "epoch": 2.3593314763231197,
      "grad_norm": 4.420183181762695,
      "learning_rate": 5.45793064241046e-05,
      "loss": 4.1857,
      "step": 1694
    },
    {
      "epoch": 2.360724233983287,
      "grad_norm": 12.535345077514648,
      "learning_rate": 5.456935758953951e-05,
      "loss": 4.0737,
      "step": 1695
    },
    {
      "epoch": 2.362116991643454,
      "grad_norm": 4.6984076499938965,
      "learning_rate": 5.455940875497441e-05,
      "loss": 3.5673,
      "step": 1696
    },
    {
      "epoch": 2.363509749303621,
      "grad_norm": 4.740118503570557,
      "learning_rate": 5.4549459920409316e-05,
      "loss": 4.0019,
      "step": 1697
    },
    {
      "epoch": 2.364902506963788,
      "grad_norm": 4.562565326690674,
      "learning_rate": 5.4539511085844226e-05,
      "loss": 3.9198,
      "step": 1698
    },
    {
      "epoch": 2.3662952646239557,
      "grad_norm": 4.111634254455566,
      "learning_rate": 5.452956225127913e-05,
      "loss": 3.8422,
      "step": 1699
    },
    {
      "epoch": 2.3676880222841223,
      "grad_norm": 4.0607171058654785,
      "learning_rate": 5.451961341671404e-05,
      "loss": 4.0549,
      "step": 1700
    },
    {
      "epoch": 2.36908077994429,
      "grad_norm": 5.0293378829956055,
      "learning_rate": 5.450966458214894e-05,
      "loss": 4.3338,
      "step": 1701
    },
    {
      "epoch": 2.370473537604457,
      "grad_norm": 4.337513446807861,
      "learning_rate": 5.449971574758385e-05,
      "loss": 4.1525,
      "step": 1702
    },
    {
      "epoch": 2.371866295264624,
      "grad_norm": 4.8907880783081055,
      "learning_rate": 5.4489766913018754e-05,
      "loss": 3.6912,
      "step": 1703
    },
    {
      "epoch": 2.3732590529247912,
      "grad_norm": 4.43142032623291,
      "learning_rate": 5.4479818078453664e-05,
      "loss": 3.9502,
      "step": 1704
    },
    {
      "epoch": 2.3746518105849583,
      "grad_norm": 7.17999792098999,
      "learning_rate": 5.4469869243888575e-05,
      "loss": 3.9594,
      "step": 1705
    },
    {
      "epoch": 2.3760445682451254,
      "grad_norm": 4.692847728729248,
      "learning_rate": 5.445992040932347e-05,
      "loss": 3.913,
      "step": 1706
    },
    {
      "epoch": 2.3774373259052926,
      "grad_norm": 4.352915287017822,
      "learning_rate": 5.4449971574758377e-05,
      "loss": 3.8952,
      "step": 1707
    },
    {
      "epoch": 2.3788300835654597,
      "grad_norm": 4.265293598175049,
      "learning_rate": 5.444002274019329e-05,
      "loss": 3.4984,
      "step": 1708
    },
    {
      "epoch": 2.3802228412256268,
      "grad_norm": 4.7866668701171875,
      "learning_rate": 5.44300739056282e-05,
      "loss": 3.5351,
      "step": 1709
    },
    {
      "epoch": 2.381615598885794,
      "grad_norm": 4.605996608734131,
      "learning_rate": 5.4420125071063096e-05,
      "loss": 4.3709,
      "step": 1710
    },
    {
      "epoch": 2.383008356545961,
      "grad_norm": 4.188661098480225,
      "learning_rate": 5.4410176236498006e-05,
      "loss": 3.954,
      "step": 1711
    },
    {
      "epoch": 2.384401114206128,
      "grad_norm": 5.439994812011719,
      "learning_rate": 5.440022740193291e-05,
      "loss": 4.3375,
      "step": 1712
    },
    {
      "epoch": 2.385793871866295,
      "grad_norm": 4.298741817474365,
      "learning_rate": 5.439027856736782e-05,
      "loss": 3.3614,
      "step": 1713
    },
    {
      "epoch": 2.3871866295264623,
      "grad_norm": 3.844583511352539,
      "learning_rate": 5.4380329732802725e-05,
      "loss": 4.1435,
      "step": 1714
    },
    {
      "epoch": 2.3885793871866294,
      "grad_norm": 4.180552959442139,
      "learning_rate": 5.437038089823763e-05,
      "loss": 3.3169,
      "step": 1715
    },
    {
      "epoch": 2.3899721448467965,
      "grad_norm": 5.350617408752441,
      "learning_rate": 5.4360432063672534e-05,
      "loss": 3.6092,
      "step": 1716
    },
    {
      "epoch": 2.3913649025069637,
      "grad_norm": 4.727535724639893,
      "learning_rate": 5.4350483229107444e-05,
      "loss": 3.7643,
      "step": 1717
    },
    {
      "epoch": 2.3927576601671308,
      "grad_norm": 4.809422492980957,
      "learning_rate": 5.434053439454235e-05,
      "loss": 4.3711,
      "step": 1718
    },
    {
      "epoch": 2.394150417827298,
      "grad_norm": 4.07980489730835,
      "learning_rate": 5.433058555997726e-05,
      "loss": 3.738,
      "step": 1719
    },
    {
      "epoch": 2.3955431754874654,
      "grad_norm": 3.852039098739624,
      "learning_rate": 5.432063672541216e-05,
      "loss": 3.6097,
      "step": 1720
    },
    {
      "epoch": 2.396935933147632,
      "grad_norm": 5.284004211425781,
      "learning_rate": 5.431068789084707e-05,
      "loss": 3.5813,
      "step": 1721
    },
    {
      "epoch": 2.3983286908077996,
      "grad_norm": 4.828900337219238,
      "learning_rate": 5.430073905628197e-05,
      "loss": 3.8873,
      "step": 1722
    },
    {
      "epoch": 2.3997214484679668,
      "grad_norm": 3.7450292110443115,
      "learning_rate": 5.429079022171688e-05,
      "loss": 4.0325,
      "step": 1723
    },
    {
      "epoch": 2.401114206128134,
      "grad_norm": 4.9826765060424805,
      "learning_rate": 5.4280841387151787e-05,
      "loss": 3.7105,
      "step": 1724
    },
    {
      "epoch": 2.402506963788301,
      "grad_norm": 4.0178141593933105,
      "learning_rate": 5.427089255258669e-05,
      "loss": 3.2191,
      "step": 1725
    },
    {
      "epoch": 2.403899721448468,
      "grad_norm": 4.679317951202393,
      "learning_rate": 5.4260943718021595e-05,
      "loss": 3.6095,
      "step": 1726
    },
    {
      "epoch": 2.405292479108635,
      "grad_norm": 5.885893821716309,
      "learning_rate": 5.4250994883456506e-05,
      "loss": 3.7421,
      "step": 1727
    },
    {
      "epoch": 2.4066852367688023,
      "grad_norm": 4.902598857879639,
      "learning_rate": 5.424104604889141e-05,
      "loss": 3.624,
      "step": 1728
    },
    {
      "epoch": 2.4080779944289694,
      "grad_norm": 3.8110177516937256,
      "learning_rate": 5.423109721432632e-05,
      "loss": 3.6195,
      "step": 1729
    },
    {
      "epoch": 2.4094707520891365,
      "grad_norm": 4.584103584289551,
      "learning_rate": 5.422114837976122e-05,
      "loss": 3.7773,
      "step": 1730
    },
    {
      "epoch": 2.4108635097493036,
      "grad_norm": 4.206098556518555,
      "learning_rate": 5.421119954519613e-05,
      "loss": 4.3632,
      "step": 1731
    },
    {
      "epoch": 2.4122562674094707,
      "grad_norm": 4.200994491577148,
      "learning_rate": 5.420125071063103e-05,
      "loss": 3.4025,
      "step": 1732
    },
    {
      "epoch": 2.413649025069638,
      "grad_norm": 4.245271682739258,
      "learning_rate": 5.4191301876065944e-05,
      "loss": 3.7452,
      "step": 1733
    },
    {
      "epoch": 2.415041782729805,
      "grad_norm": 4.359995365142822,
      "learning_rate": 5.4181353041500854e-05,
      "loss": 4.0386,
      "step": 1734
    },
    {
      "epoch": 2.416434540389972,
      "grad_norm": 4.1396484375,
      "learning_rate": 5.417140420693575e-05,
      "loss": 4.4054,
      "step": 1735
    },
    {
      "epoch": 2.417827298050139,
      "grad_norm": 3.848989248275757,
      "learning_rate": 5.4161455372370656e-05,
      "loss": 3.7372,
      "step": 1736
    },
    {
      "epoch": 2.4192200557103063,
      "grad_norm": 3.9558351039886475,
      "learning_rate": 5.4151506537805567e-05,
      "loss": 3.6096,
      "step": 1737
    },
    {
      "epoch": 2.4206128133704734,
      "grad_norm": 4.3685526847839355,
      "learning_rate": 5.414155770324048e-05,
      "loss": 4.0534,
      "step": 1738
    },
    {
      "epoch": 2.4220055710306405,
      "grad_norm": 4.487792491912842,
      "learning_rate": 5.413160886867538e-05,
      "loss": 3.5896,
      "step": 1739
    },
    {
      "epoch": 2.4233983286908076,
      "grad_norm": 4.340232849121094,
      "learning_rate": 5.4121660034110286e-05,
      "loss": 3.6079,
      "step": 1740
    },
    {
      "epoch": 2.424791086350975,
      "grad_norm": 4.081694602966309,
      "learning_rate": 5.411171119954519e-05,
      "loss": 3.8478,
      "step": 1741
    },
    {
      "epoch": 2.426183844011142,
      "grad_norm": 4.9409708976745605,
      "learning_rate": 5.41017623649801e-05,
      "loss": 3.682,
      "step": 1742
    },
    {
      "epoch": 2.4275766016713094,
      "grad_norm": 4.346573352813721,
      "learning_rate": 5.4091813530415005e-05,
      "loss": 3.8987,
      "step": 1743
    },
    {
      "epoch": 2.4289693593314765,
      "grad_norm": 4.81184720993042,
      "learning_rate": 5.4081864695849915e-05,
      "loss": 3.8868,
      "step": 1744
    },
    {
      "epoch": 2.4303621169916436,
      "grad_norm": 3.792619466781616,
      "learning_rate": 5.407191586128481e-05,
      "loss": 3.4401,
      "step": 1745
    },
    {
      "epoch": 2.4317548746518107,
      "grad_norm": 5.124619483947754,
      "learning_rate": 5.4061967026719724e-05,
      "loss": 4.2587,
      "step": 1746
    },
    {
      "epoch": 2.433147632311978,
      "grad_norm": 3.9748077392578125,
      "learning_rate": 5.405201819215463e-05,
      "loss": 3.7289,
      "step": 1747
    },
    {
      "epoch": 2.434540389972145,
      "grad_norm": 4.482245922088623,
      "learning_rate": 5.404206935758954e-05,
      "loss": 3.7506,
      "step": 1748
    },
    {
      "epoch": 2.435933147632312,
      "grad_norm": 4.309085845947266,
      "learning_rate": 5.403212052302444e-05,
      "loss": 3.3056,
      "step": 1749
    },
    {
      "epoch": 2.437325905292479,
      "grad_norm": 4.857260227203369,
      "learning_rate": 5.402217168845935e-05,
      "loss": 3.9873,
      "step": 1750
    },
    {
      "epoch": 2.4387186629526463,
      "grad_norm": 4.169651985168457,
      "learning_rate": 5.401222285389425e-05,
      "loss": 3.5028,
      "step": 1751
    },
    {
      "epoch": 2.4401114206128134,
      "grad_norm": 4.409386157989502,
      "learning_rate": 5.400227401932916e-05,
      "loss": 4.1041,
      "step": 1752
    },
    {
      "epoch": 2.4415041782729805,
      "grad_norm": 4.25468111038208,
      "learning_rate": 5.3992325184764066e-05,
      "loss": 3.7783,
      "step": 1753
    },
    {
      "epoch": 2.4428969359331476,
      "grad_norm": 4.455057144165039,
      "learning_rate": 5.398237635019897e-05,
      "loss": 3.7907,
      "step": 1754
    },
    {
      "epoch": 2.4442896935933147,
      "grad_norm": 4.924750328063965,
      "learning_rate": 5.3972427515633874e-05,
      "loss": 4.0321,
      "step": 1755
    },
    {
      "epoch": 2.445682451253482,
      "grad_norm": 4.541840076446533,
      "learning_rate": 5.3962478681068785e-05,
      "loss": 4.1903,
      "step": 1756
    },
    {
      "epoch": 2.447075208913649,
      "grad_norm": 5.0196614265441895,
      "learning_rate": 5.395252984650369e-05,
      "loss": 3.9073,
      "step": 1757
    },
    {
      "epoch": 2.448467966573816,
      "grad_norm": 5.129353046417236,
      "learning_rate": 5.39425810119386e-05,
      "loss": 3.5858,
      "step": 1758
    },
    {
      "epoch": 2.449860724233983,
      "grad_norm": 4.817824363708496,
      "learning_rate": 5.39326321773735e-05,
      "loss": 4.1497,
      "step": 1759
    },
    {
      "epoch": 2.4512534818941503,
      "grad_norm": 3.773118734359741,
      "learning_rate": 5.392268334280841e-05,
      "loss": 3.655,
      "step": 1760
    },
    {
      "epoch": 2.4526462395543174,
      "grad_norm": 4.434882164001465,
      "learning_rate": 5.391273450824331e-05,
      "loss": 3.9889,
      "step": 1761
    },
    {
      "epoch": 2.4540389972144845,
      "grad_norm": 3.593027114868164,
      "learning_rate": 5.390278567367822e-05,
      "loss": 3.4343,
      "step": 1762
    },
    {
      "epoch": 2.4554317548746516,
      "grad_norm": 7.124451637268066,
      "learning_rate": 5.3892836839113134e-05,
      "loss": 3.7918,
      "step": 1763
    },
    {
      "epoch": 2.456824512534819,
      "grad_norm": 5.207165241241455,
      "learning_rate": 5.388288800454803e-05,
      "loss": 3.9079,
      "step": 1764
    },
    {
      "epoch": 2.4582172701949863,
      "grad_norm": 4.422651290893555,
      "learning_rate": 5.3872939169982935e-05,
      "loss": 4.0438,
      "step": 1765
    },
    {
      "epoch": 2.4596100278551534,
      "grad_norm": 4.428305625915527,
      "learning_rate": 5.3862990335417846e-05,
      "loss": 3.8374,
      "step": 1766
    },
    {
      "epoch": 2.4610027855153205,
      "grad_norm": 4.756076812744141,
      "learning_rate": 5.3853041500852757e-05,
      "loss": 3.7995,
      "step": 1767
    },
    {
      "epoch": 2.4623955431754876,
      "grad_norm": 4.255223274230957,
      "learning_rate": 5.384309266628766e-05,
      "loss": 3.9773,
      "step": 1768
    },
    {
      "epoch": 2.4637883008356547,
      "grad_norm": 4.941119194030762,
      "learning_rate": 5.3833143831722565e-05,
      "loss": 3.708,
      "step": 1769
    },
    {
      "epoch": 2.465181058495822,
      "grad_norm": 4.289640426635742,
      "learning_rate": 5.382319499715747e-05,
      "loss": 4.0151,
      "step": 1770
    },
    {
      "epoch": 2.466573816155989,
      "grad_norm": 4.04726505279541,
      "learning_rate": 5.381324616259238e-05,
      "loss": 3.9457,
      "step": 1771
    },
    {
      "epoch": 2.467966573816156,
      "grad_norm": 4.4292449951171875,
      "learning_rate": 5.3803297328027284e-05,
      "loss": 3.8842,
      "step": 1772
    },
    {
      "epoch": 2.469359331476323,
      "grad_norm": 4.514986991882324,
      "learning_rate": 5.3793348493462195e-05,
      "loss": 4.1711,
      "step": 1773
    },
    {
      "epoch": 2.4707520891364902,
      "grad_norm": 4.214544296264648,
      "learning_rate": 5.378339965889709e-05,
      "loss": 3.409,
      "step": 1774
    },
    {
      "epoch": 2.4721448467966574,
      "grad_norm": 4.354898452758789,
      "learning_rate": 5.3773450824332e-05,
      "loss": 3.914,
      "step": 1775
    },
    {
      "epoch": 2.4735376044568245,
      "grad_norm": 5.114170074462891,
      "learning_rate": 5.376350198976691e-05,
      "loss": 3.7704,
      "step": 1776
    },
    {
      "epoch": 2.4749303621169916,
      "grad_norm": 4.445180416107178,
      "learning_rate": 5.375355315520182e-05,
      "loss": 4.1139,
      "step": 1777
    },
    {
      "epoch": 2.4763231197771587,
      "grad_norm": 4.674963474273682,
      "learning_rate": 5.374360432063672e-05,
      "loss": 3.4318,
      "step": 1778
    },
    {
      "epoch": 2.477715877437326,
      "grad_norm": 3.9881012439727783,
      "learning_rate": 5.3733655486071626e-05,
      "loss": 3.5508,
      "step": 1779
    },
    {
      "epoch": 2.479108635097493,
      "grad_norm": 4.56321382522583,
      "learning_rate": 5.372370665150653e-05,
      "loss": 3.7149,
      "step": 1780
    },
    {
      "epoch": 2.48050139275766,
      "grad_norm": 4.433288097381592,
      "learning_rate": 5.371375781694144e-05,
      "loss": 3.9132,
      "step": 1781
    },
    {
      "epoch": 2.481894150417827,
      "grad_norm": 4.969962120056152,
      "learning_rate": 5.3703808982376345e-05,
      "loss": 4.5837,
      "step": 1782
    },
    {
      "epoch": 2.4832869080779942,
      "grad_norm": 4.592645645141602,
      "learning_rate": 5.3693860147811256e-05,
      "loss": 3.9482,
      "step": 1783
    },
    {
      "epoch": 2.4846796657381613,
      "grad_norm": 4.448911666870117,
      "learning_rate": 5.368391131324615e-05,
      "loss": 3.6117,
      "step": 1784
    },
    {
      "epoch": 2.486072423398329,
      "grad_norm": 4.952700614929199,
      "learning_rate": 5.3673962478681064e-05,
      "loss": 3.5696,
      "step": 1785
    },
    {
      "epoch": 2.487465181058496,
      "grad_norm": 4.145140171051025,
      "learning_rate": 5.366401364411597e-05,
      "loss": 4.1032,
      "step": 1786
    },
    {
      "epoch": 2.488857938718663,
      "grad_norm": 5.181249141693115,
      "learning_rate": 5.365406480955088e-05,
      "loss": 3.7582,
      "step": 1787
    },
    {
      "epoch": 2.4902506963788302,
      "grad_norm": 4.133391857147217,
      "learning_rate": 5.364411597498579e-05,
      "loss": 3.3871,
      "step": 1788
    },
    {
      "epoch": 2.4916434540389973,
      "grad_norm": 4.500240325927734,
      "learning_rate": 5.363416714042069e-05,
      "loss": 3.747,
      "step": 1789
    },
    {
      "epoch": 2.4930362116991645,
      "grad_norm": 4.199702262878418,
      "learning_rate": 5.362421830585559e-05,
      "loss": 4.1196,
      "step": 1790
    },
    {
      "epoch": 2.4944289693593316,
      "grad_norm": 4.462384223937988,
      "learning_rate": 5.36142694712905e-05,
      "loss": 3.7021,
      "step": 1791
    },
    {
      "epoch": 2.4958217270194987,
      "grad_norm": 4.20314359664917,
      "learning_rate": 5.360432063672541e-05,
      "loss": 3.3557,
      "step": 1792
    },
    {
      "epoch": 2.497214484679666,
      "grad_norm": 3.8399412631988525,
      "learning_rate": 5.359437180216032e-05,
      "loss": 3.3568,
      "step": 1793
    },
    {
      "epoch": 2.498607242339833,
      "grad_norm": 4.548267364501953,
      "learning_rate": 5.358442296759522e-05,
      "loss": 4.0814,
      "step": 1794
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.398299217224121,
      "learning_rate": 5.3574474133030125e-05,
      "loss": 3.4655,
      "step": 1795
    },
    {
      "epoch": 2.501392757660167,
      "grad_norm": 4.408416271209717,
      "learning_rate": 5.3564525298465036e-05,
      "loss": 3.5049,
      "step": 1796
    },
    {
      "epoch": 2.502785515320334,
      "grad_norm": 4.241742134094238,
      "learning_rate": 5.355457646389994e-05,
      "loss": 3.9685,
      "step": 1797
    },
    {
      "epoch": 2.5041782729805013,
      "grad_norm": 4.418053150177002,
      "learning_rate": 5.3544627629334844e-05,
      "loss": 4.0235,
      "step": 1798
    },
    {
      "epoch": 2.5055710306406684,
      "grad_norm": 4.750143527984619,
      "learning_rate": 5.353467879476975e-05,
      "loss": 3.5644,
      "step": 1799
    },
    {
      "epoch": 2.5069637883008355,
      "grad_norm": 4.102102279663086,
      "learning_rate": 5.352472996020466e-05,
      "loss": 3.7058,
      "step": 1800
    },
    {
      "epoch": 2.5083565459610027,
      "grad_norm": 4.465678691864014,
      "learning_rate": 5.351478112563956e-05,
      "loss": 3.4684,
      "step": 1801
    },
    {
      "epoch": 2.5097493036211698,
      "grad_norm": 5.520817756652832,
      "learning_rate": 5.3504832291074474e-05,
      "loss": 3.8512,
      "step": 1802
    },
    {
      "epoch": 2.511142061281337,
      "grad_norm": 4.454752445220947,
      "learning_rate": 5.349488345650937e-05,
      "loss": 4.0388,
      "step": 1803
    },
    {
      "epoch": 2.5125348189415044,
      "grad_norm": 3.9050180912017822,
      "learning_rate": 5.348493462194428e-05,
      "loss": 3.5277,
      "step": 1804
    },
    {
      "epoch": 2.513927576601671,
      "grad_norm": 4.071952819824219,
      "learning_rate": 5.3474985787379186e-05,
      "loss": 3.7311,
      "step": 1805
    },
    {
      "epoch": 2.5153203342618387,
      "grad_norm": 4.0145063400268555,
      "learning_rate": 5.34650369528141e-05,
      "loss": 3.4427,
      "step": 1806
    },
    {
      "epoch": 2.5167130919220053,
      "grad_norm": 8.699322700500488,
      "learning_rate": 5.3455088118249e-05,
      "loss": 3.5338,
      "step": 1807
    },
    {
      "epoch": 2.518105849582173,
      "grad_norm": 3.9393362998962402,
      "learning_rate": 5.3445139283683905e-05,
      "loss": 3.6937,
      "step": 1808
    },
    {
      "epoch": 2.51949860724234,
      "grad_norm": 5.14155912399292,
      "learning_rate": 5.343519044911881e-05,
      "loss": 3.9188,
      "step": 1809
    },
    {
      "epoch": 2.520891364902507,
      "grad_norm": 4.432671070098877,
      "learning_rate": 5.342524161455372e-05,
      "loss": 3.8052,
      "step": 1810
    },
    {
      "epoch": 2.522284122562674,
      "grad_norm": 4.019896030426025,
      "learning_rate": 5.3415292779988624e-05,
      "loss": 3.853,
      "step": 1811
    },
    {
      "epoch": 2.5236768802228413,
      "grad_norm": 3.7733829021453857,
      "learning_rate": 5.3405343945423535e-05,
      "loss": 3.1545,
      "step": 1812
    },
    {
      "epoch": 2.5250696378830084,
      "grad_norm": 4.388171195983887,
      "learning_rate": 5.339539511085843e-05,
      "loss": 3.7329,
      "step": 1813
    },
    {
      "epoch": 2.5264623955431755,
      "grad_norm": 4.052828311920166,
      "learning_rate": 5.338544627629334e-05,
      "loss": 3.4381,
      "step": 1814
    },
    {
      "epoch": 2.5278551532033426,
      "grad_norm": 4.079244136810303,
      "learning_rate": 5.337549744172825e-05,
      "loss": 3.7339,
      "step": 1815
    },
    {
      "epoch": 2.5292479108635098,
      "grad_norm": 5.9524946212768555,
      "learning_rate": 5.336554860716316e-05,
      "loss": 3.6238,
      "step": 1816
    },
    {
      "epoch": 2.530640668523677,
      "grad_norm": 4.794691562652588,
      "learning_rate": 5.335559977259807e-05,
      "loss": 3.9297,
      "step": 1817
    },
    {
      "epoch": 2.532033426183844,
      "grad_norm": 4.042095184326172,
      "learning_rate": 5.3345650938032966e-05,
      "loss": 3.5151,
      "step": 1818
    },
    {
      "epoch": 2.533426183844011,
      "grad_norm": 5.037164211273193,
      "learning_rate": 5.333570210346787e-05,
      "loss": 3.9337,
      "step": 1819
    },
    {
      "epoch": 2.534818941504178,
      "grad_norm": 4.347875118255615,
      "learning_rate": 5.332575326890278e-05,
      "loss": 4.1536,
      "step": 1820
    },
    {
      "epoch": 2.5362116991643453,
      "grad_norm": 5.549806594848633,
      "learning_rate": 5.331580443433769e-05,
      "loss": 3.4958,
      "step": 1821
    },
    {
      "epoch": 2.5376044568245124,
      "grad_norm": 5.315230369567871,
      "learning_rate": 5.3305855599772596e-05,
      "loss": 4.1105,
      "step": 1822
    },
    {
      "epoch": 2.5389972144846795,
      "grad_norm": 4.203426361083984,
      "learning_rate": 5.32959067652075e-05,
      "loss": 3.3643,
      "step": 1823
    },
    {
      "epoch": 2.5403899721448466,
      "grad_norm": 4.214809417724609,
      "learning_rate": 5.3285957930642404e-05,
      "loss": 3.9121,
      "step": 1824
    },
    {
      "epoch": 2.541782729805014,
      "grad_norm": 5.017365455627441,
      "learning_rate": 5.3276009096077315e-05,
      "loss": 3.7667,
      "step": 1825
    },
    {
      "epoch": 2.543175487465181,
      "grad_norm": 4.704203128814697,
      "learning_rate": 5.326606026151222e-05,
      "loss": 3.5719,
      "step": 1826
    },
    {
      "epoch": 2.5445682451253484,
      "grad_norm": 4.263749599456787,
      "learning_rate": 5.325611142694713e-05,
      "loss": 3.8484,
      "step": 1827
    },
    {
      "epoch": 2.545961002785515,
      "grad_norm": 4.997926235198975,
      "learning_rate": 5.324616259238203e-05,
      "loss": 3.5273,
      "step": 1828
    },
    {
      "epoch": 2.5473537604456826,
      "grad_norm": 4.328037738800049,
      "learning_rate": 5.323621375781694e-05,
      "loss": 3.6637,
      "step": 1829
    },
    {
      "epoch": 2.5487465181058497,
      "grad_norm": 4.116519927978516,
      "learning_rate": 5.322626492325184e-05,
      "loss": 3.5132,
      "step": 1830
    },
    {
      "epoch": 2.550139275766017,
      "grad_norm": 4.875148296356201,
      "learning_rate": 5.321631608868675e-05,
      "loss": 3.6665,
      "step": 1831
    },
    {
      "epoch": 2.551532033426184,
      "grad_norm": 4.478740215301514,
      "learning_rate": 5.320636725412166e-05,
      "loss": 3.4473,
      "step": 1832
    },
    {
      "epoch": 2.552924791086351,
      "grad_norm": 6.236715316772461,
      "learning_rate": 5.319641841955656e-05,
      "loss": 3.518,
      "step": 1833
    },
    {
      "epoch": 2.554317548746518,
      "grad_norm": 4.370816707611084,
      "learning_rate": 5.3186469584991465e-05,
      "loss": 3.5055,
      "step": 1834
    },
    {
      "epoch": 2.5557103064066853,
      "grad_norm": 4.396291255950928,
      "learning_rate": 5.3176520750426376e-05,
      "loss": 3.7455,
      "step": 1835
    },
    {
      "epoch": 2.5571030640668524,
      "grad_norm": 5.377126216888428,
      "learning_rate": 5.316657191586128e-05,
      "loss": 3.8579,
      "step": 1836
    },
    {
      "epoch": 2.5584958217270195,
      "grad_norm": 4.379935264587402,
      "learning_rate": 5.315662308129619e-05,
      "loss": 3.9947,
      "step": 1837
    },
    {
      "epoch": 2.5598885793871866,
      "grad_norm": 3.9767608642578125,
      "learning_rate": 5.314667424673109e-05,
      "loss": 3.6255,
      "step": 1838
    },
    {
      "epoch": 2.5612813370473537,
      "grad_norm": 5.672046661376953,
      "learning_rate": 5.3136725412166e-05,
      "loss": 3.6386,
      "step": 1839
    },
    {
      "epoch": 2.562674094707521,
      "grad_norm": 4.641458034515381,
      "learning_rate": 5.31267765776009e-05,
      "loss": 4.0477,
      "step": 1840
    },
    {
      "epoch": 2.564066852367688,
      "grad_norm": 4.017055511474609,
      "learning_rate": 5.3116827743035814e-05,
      "loss": 3.8583,
      "step": 1841
    },
    {
      "epoch": 2.565459610027855,
      "grad_norm": 4.255927085876465,
      "learning_rate": 5.310687890847071e-05,
      "loss": 3.5727,
      "step": 1842
    },
    {
      "epoch": 2.566852367688022,
      "grad_norm": 5.480453968048096,
      "learning_rate": 5.309693007390562e-05,
      "loss": 3.3779,
      "step": 1843
    },
    {
      "epoch": 2.5682451253481893,
      "grad_norm": 3.9198381900787354,
      "learning_rate": 5.3086981239340526e-05,
      "loss": 4.0592,
      "step": 1844
    },
    {
      "epoch": 2.5696378830083564,
      "grad_norm": 4.344557285308838,
      "learning_rate": 5.307703240477544e-05,
      "loss": 3.9158,
      "step": 1845
    },
    {
      "epoch": 2.571030640668524,
      "grad_norm": 4.110029220581055,
      "learning_rate": 5.306708357021035e-05,
      "loss": 4.1278,
      "step": 1846
    },
    {
      "epoch": 2.5724233983286906,
      "grad_norm": 3.936121940612793,
      "learning_rate": 5.3057134735645245e-05,
      "loss": 3.5425,
      "step": 1847
    },
    {
      "epoch": 2.573816155988858,
      "grad_norm": 3.695356607437134,
      "learning_rate": 5.304718590108015e-05,
      "loss": 3.4381,
      "step": 1848
    },
    {
      "epoch": 2.575208913649025,
      "grad_norm": 8.275914192199707,
      "learning_rate": 5.303723706651506e-05,
      "loss": 3.4775,
      "step": 1849
    },
    {
      "epoch": 2.5766016713091924,
      "grad_norm": 5.141213417053223,
      "learning_rate": 5.302728823194997e-05,
      "loss": 3.695,
      "step": 1850
    },
    {
      "epoch": 2.5779944289693595,
      "grad_norm": 4.1147332191467285,
      "learning_rate": 5.3017339397384875e-05,
      "loss": 3.3717,
      "step": 1851
    },
    {
      "epoch": 2.5793871866295266,
      "grad_norm": 5.368847846984863,
      "learning_rate": 5.300739056281978e-05,
      "loss": 3.7957,
      "step": 1852
    },
    {
      "epoch": 2.5807799442896937,
      "grad_norm": 4.1998395919799805,
      "learning_rate": 5.299744172825468e-05,
      "loss": 3.9779,
      "step": 1853
    },
    {
      "epoch": 2.582172701949861,
      "grad_norm": 3.5952372550964355,
      "learning_rate": 5.2987492893689594e-05,
      "loss": 3.3549,
      "step": 1854
    },
    {
      "epoch": 2.583565459610028,
      "grad_norm": 4.264059543609619,
      "learning_rate": 5.29775440591245e-05,
      "loss": 3.5971,
      "step": 1855
    },
    {
      "epoch": 2.584958217270195,
      "grad_norm": 4.747745037078857,
      "learning_rate": 5.296759522455941e-05,
      "loss": 3.5043,
      "step": 1856
    },
    {
      "epoch": 2.586350974930362,
      "grad_norm": 4.186683654785156,
      "learning_rate": 5.2957646389994306e-05,
      "loss": 3.9387,
      "step": 1857
    },
    {
      "epoch": 2.5877437325905293,
      "grad_norm": 5.196044921875,
      "learning_rate": 5.294769755542922e-05,
      "loss": 3.9363,
      "step": 1858
    },
    {
      "epoch": 2.5891364902506964,
      "grad_norm": 4.816099166870117,
      "learning_rate": 5.293774872086412e-05,
      "loss": 3.5953,
      "step": 1859
    },
    {
      "epoch": 2.5905292479108635,
      "grad_norm": 4.190717697143555,
      "learning_rate": 5.292779988629903e-05,
      "loss": 3.8588,
      "step": 1860
    },
    {
      "epoch": 2.5919220055710306,
      "grad_norm": 4.201430320739746,
      "learning_rate": 5.2917851051733936e-05,
      "loss": 4.0909,
      "step": 1861
    },
    {
      "epoch": 2.5933147632311977,
      "grad_norm": 4.05234956741333,
      "learning_rate": 5.290790221716884e-05,
      "loss": 3.6215,
      "step": 1862
    },
    {
      "epoch": 2.594707520891365,
      "grad_norm": 3.9277195930480957,
      "learning_rate": 5.2897953382603744e-05,
      "loss": 3.5398,
      "step": 1863
    },
    {
      "epoch": 2.596100278551532,
      "grad_norm": 4.1085591316223145,
      "learning_rate": 5.2888004548038655e-05,
      "loss": 4.4546,
      "step": 1864
    },
    {
      "epoch": 2.597493036211699,
      "grad_norm": 5.142462730407715,
      "learning_rate": 5.287805571347356e-05,
      "loss": 3.4016,
      "step": 1865
    },
    {
      "epoch": 2.598885793871866,
      "grad_norm": 5.263397693634033,
      "learning_rate": 5.286810687890847e-05,
      "loss": 3.6781,
      "step": 1866
    },
    {
      "epoch": 2.6002785515320337,
      "grad_norm": 4.9327497482299805,
      "learning_rate": 5.285815804434337e-05,
      "loss": 3.7238,
      "step": 1867
    },
    {
      "epoch": 2.6016713091922004,
      "grad_norm": 3.7192282676696777,
      "learning_rate": 5.284820920977828e-05,
      "loss": 3.5964,
      "step": 1868
    },
    {
      "epoch": 2.603064066852368,
      "grad_norm": 7.3771257400512695,
      "learning_rate": 5.283826037521318e-05,
      "loss": 3.53,
      "step": 1869
    },
    {
      "epoch": 2.6044568245125346,
      "grad_norm": 3.78019118309021,
      "learning_rate": 5.282831154064809e-05,
      "loss": 3.4685,
      "step": 1870
    },
    {
      "epoch": 2.605849582172702,
      "grad_norm": 4.528641223907471,
      "learning_rate": 5.2818362706083e-05,
      "loss": 3.5435,
      "step": 1871
    },
    {
      "epoch": 2.6072423398328692,
      "grad_norm": 6.675357818603516,
      "learning_rate": 5.28084138715179e-05,
      "loss": 3.4348,
      "step": 1872
    },
    {
      "epoch": 2.6086350974930363,
      "grad_norm": 4.7266130447387695,
      "learning_rate": 5.2798465036952805e-05,
      "loss": 3.5718,
      "step": 1873
    },
    {
      "epoch": 2.6100278551532035,
      "grad_norm": 4.423274040222168,
      "learning_rate": 5.2788516202387716e-05,
      "loss": 3.1636,
      "step": 1874
    },
    {
      "epoch": 2.6114206128133706,
      "grad_norm": 4.587574481964111,
      "learning_rate": 5.277856736782263e-05,
      "loss": 3.8414,
      "step": 1875
    },
    {
      "epoch": 2.6128133704735377,
      "grad_norm": 3.895914316177368,
      "learning_rate": 5.276861853325753e-05,
      "loss": 4.0406,
      "step": 1876
    },
    {
      "epoch": 2.614206128133705,
      "grad_norm": 4.607090950012207,
      "learning_rate": 5.275866969869243e-05,
      "loss": 4.121,
      "step": 1877
    },
    {
      "epoch": 2.615598885793872,
      "grad_norm": 9.816665649414062,
      "learning_rate": 5.274872086412734e-05,
      "loss": 3.8753,
      "step": 1878
    },
    {
      "epoch": 2.616991643454039,
      "grad_norm": 5.306396484375,
      "learning_rate": 5.273877202956225e-05,
      "loss": 3.7716,
      "step": 1879
    },
    {
      "epoch": 2.618384401114206,
      "grad_norm": 5.104506492614746,
      "learning_rate": 5.2728823194997154e-05,
      "loss": 3.6848,
      "step": 1880
    },
    {
      "epoch": 2.6197771587743732,
      "grad_norm": 4.126043796539307,
      "learning_rate": 5.2718874360432065e-05,
      "loss": 3.7632,
      "step": 1881
    },
    {
      "epoch": 2.6211699164345403,
      "grad_norm": 3.921090602874756,
      "learning_rate": 5.270892552586696e-05,
      "loss": 3.8218,
      "step": 1882
    },
    {
      "epoch": 2.6225626740947074,
      "grad_norm": 4.4806952476501465,
      "learning_rate": 5.269897669130187e-05,
      "loss": 3.6962,
      "step": 1883
    },
    {
      "epoch": 2.6239554317548746,
      "grad_norm": 5.107631683349609,
      "learning_rate": 5.268902785673678e-05,
      "loss": 3.8599,
      "step": 1884
    },
    {
      "epoch": 2.6253481894150417,
      "grad_norm": 4.719388961791992,
      "learning_rate": 5.267907902217169e-05,
      "loss": 3.8638,
      "step": 1885
    },
    {
      "epoch": 2.6267409470752088,
      "grad_norm": 5.310430526733398,
      "learning_rate": 5.2669130187606585e-05,
      "loss": 3.7588,
      "step": 1886
    },
    {
      "epoch": 2.628133704735376,
      "grad_norm": 4.737289905548096,
      "learning_rate": 5.2659181353041496e-05,
      "loss": 3.6632,
      "step": 1887
    },
    {
      "epoch": 2.6295264623955434,
      "grad_norm": 4.28202486038208,
      "learning_rate": 5.26492325184764e-05,
      "loss": 3.5808,
      "step": 1888
    },
    {
      "epoch": 2.63091922005571,
      "grad_norm": 4.6112494468688965,
      "learning_rate": 5.263928368391131e-05,
      "loss": 4.0221,
      "step": 1889
    },
    {
      "epoch": 2.6323119777158777,
      "grad_norm": 3.8772616386413574,
      "learning_rate": 5.2629334849346215e-05,
      "loss": 3.6025,
      "step": 1890
    },
    {
      "epoch": 2.6337047353760443,
      "grad_norm": 4.3526105880737305,
      "learning_rate": 5.261938601478112e-05,
      "loss": 3.611,
      "step": 1891
    },
    {
      "epoch": 2.635097493036212,
      "grad_norm": 4.766617298126221,
      "learning_rate": 5.2609437180216023e-05,
      "loss": 3.4089,
      "step": 1892
    },
    {
      "epoch": 2.636490250696379,
      "grad_norm": 4.573330879211426,
      "learning_rate": 5.2599488345650934e-05,
      "loss": 3.8785,
      "step": 1893
    },
    {
      "epoch": 2.637883008356546,
      "grad_norm": 4.703368186950684,
      "learning_rate": 5.258953951108584e-05,
      "loss": 3.6822,
      "step": 1894
    },
    {
      "epoch": 2.639275766016713,
      "grad_norm": 4.373324871063232,
      "learning_rate": 5.257959067652075e-05,
      "loss": 3.6716,
      "step": 1895
    },
    {
      "epoch": 2.6406685236768803,
      "grad_norm": 5.367077350616455,
      "learning_rate": 5.2569641841955647e-05,
      "loss": 4.0347,
      "step": 1896
    },
    {
      "epoch": 2.6420612813370474,
      "grad_norm": 5.086892127990723,
      "learning_rate": 5.255969300739056e-05,
      "loss": 3.6172,
      "step": 1897
    },
    {
      "epoch": 2.6434540389972145,
      "grad_norm": 4.16098690032959,
      "learning_rate": 5.254974417282546e-05,
      "loss": 3.7007,
      "step": 1898
    },
    {
      "epoch": 2.6448467966573816,
      "grad_norm": 4.16826057434082,
      "learning_rate": 5.253979533826037e-05,
      "loss": 4.066,
      "step": 1899
    },
    {
      "epoch": 2.6462395543175488,
      "grad_norm": 4.992652893066406,
      "learning_rate": 5.252984650369528e-05,
      "loss": 3.7062,
      "step": 1900
    },
    {
      "epoch": 2.647632311977716,
      "grad_norm": 4.250000953674316,
      "learning_rate": 5.251989766913018e-05,
      "loss": 3.1994,
      "step": 1901
    },
    {
      "epoch": 2.649025069637883,
      "grad_norm": 4.818694591522217,
      "learning_rate": 5.2509948834565085e-05,
      "loss": 4.2614,
      "step": 1902
    },
    {
      "epoch": 2.65041782729805,
      "grad_norm": 4.342881202697754,
      "learning_rate": 5.2499999999999995e-05,
      "loss": 3.6534,
      "step": 1903
    },
    {
      "epoch": 2.651810584958217,
      "grad_norm": 4.088788032531738,
      "learning_rate": 5.2490051165434906e-05,
      "loss": 3.302,
      "step": 1904
    },
    {
      "epoch": 2.6532033426183843,
      "grad_norm": 4.922445774078369,
      "learning_rate": 5.248010233086981e-05,
      "loss": 3.4726,
      "step": 1905
    },
    {
      "epoch": 2.6545961002785514,
      "grad_norm": 3.9478759765625,
      "learning_rate": 5.247015349630471e-05,
      "loss": 3.6109,
      "step": 1906
    },
    {
      "epoch": 2.6559888579387185,
      "grad_norm": 4.833628177642822,
      "learning_rate": 5.246020466173962e-05,
      "loss": 3.974,
      "step": 1907
    },
    {
      "epoch": 2.6573816155988856,
      "grad_norm": 3.918222427368164,
      "learning_rate": 5.245025582717453e-05,
      "loss": 3.327,
      "step": 1908
    },
    {
      "epoch": 2.658774373259053,
      "grad_norm": 4.314101219177246,
      "learning_rate": 5.2440306992609433e-05,
      "loss": 3.7502,
      "step": 1909
    },
    {
      "epoch": 2.66016713091922,
      "grad_norm": 4.25202751159668,
      "learning_rate": 5.2430358158044344e-05,
      "loss": 3.861,
      "step": 1910
    },
    {
      "epoch": 2.6615598885793874,
      "grad_norm": 3.9286015033721924,
      "learning_rate": 5.242040932347924e-05,
      "loss": 3.4853,
      "step": 1911
    },
    {
      "epoch": 2.662952646239554,
      "grad_norm": 4.644457817077637,
      "learning_rate": 5.241046048891415e-05,
      "loss": 3.548,
      "step": 1912
    },
    {
      "epoch": 2.6643454038997216,
      "grad_norm": 4.0190229415893555,
      "learning_rate": 5.2400511654349056e-05,
      "loss": 3.3824,
      "step": 1913
    },
    {
      "epoch": 2.6657381615598887,
      "grad_norm": 4.23594856262207,
      "learning_rate": 5.239056281978397e-05,
      "loss": 3.1528,
      "step": 1914
    },
    {
      "epoch": 2.667130919220056,
      "grad_norm": 4.653962135314941,
      "learning_rate": 5.238061398521887e-05,
      "loss": 3.5549,
      "step": 1915
    },
    {
      "epoch": 2.668523676880223,
      "grad_norm": 4.541530132293701,
      "learning_rate": 5.2370665150653775e-05,
      "loss": 3.427,
      "step": 1916
    },
    {
      "epoch": 2.66991643454039,
      "grad_norm": 4.232297897338867,
      "learning_rate": 5.236071631608868e-05,
      "loss": 3.5785,
      "step": 1917
    },
    {
      "epoch": 2.671309192200557,
      "grad_norm": 5.324038982391357,
      "learning_rate": 5.235076748152359e-05,
      "loss": 3.8548,
      "step": 1918
    },
    {
      "epoch": 2.6727019498607243,
      "grad_norm": 6.1439080238342285,
      "learning_rate": 5.2340818646958494e-05,
      "loss": 3.6788,
      "step": 1919
    },
    {
      "epoch": 2.6740947075208914,
      "grad_norm": 3.6627092361450195,
      "learning_rate": 5.2330869812393405e-05,
      "loss": 3.436,
      "step": 1920
    },
    {
      "epoch": 2.6754874651810585,
      "grad_norm": 4.146340370178223,
      "learning_rate": 5.23209209778283e-05,
      "loss": 3.6203,
      "step": 1921
    },
    {
      "epoch": 2.6768802228412256,
      "grad_norm": 4.556893348693848,
      "learning_rate": 5.2310972143263213e-05,
      "loss": 3.4084,
      "step": 1922
    },
    {
      "epoch": 2.6782729805013927,
      "grad_norm": 4.068876266479492,
      "learning_rate": 5.230102330869812e-05,
      "loss": 3.5732,
      "step": 1923
    },
    {
      "epoch": 2.67966573816156,
      "grad_norm": 4.500432968139648,
      "learning_rate": 5.229107447413303e-05,
      "loss": 3.8973,
      "step": 1924
    },
    {
      "epoch": 2.681058495821727,
      "grad_norm": 7.038255214691162,
      "learning_rate": 5.2281125639567926e-05,
      "loss": 3.7381,
      "step": 1925
    },
    {
      "epoch": 2.682451253481894,
      "grad_norm": 4.796656131744385,
      "learning_rate": 5.2271176805002837e-05,
      "loss": 3.7804,
      "step": 1926
    },
    {
      "epoch": 2.683844011142061,
      "grad_norm": 4.580173492431641,
      "learning_rate": 5.226122797043774e-05,
      "loss": 3.7832,
      "step": 1927
    },
    {
      "epoch": 2.6852367688022283,
      "grad_norm": 4.559717655181885,
      "learning_rate": 5.225127913587265e-05,
      "loss": 3.5402,
      "step": 1928
    },
    {
      "epoch": 2.6866295264623954,
      "grad_norm": 4.069858551025391,
      "learning_rate": 5.224133030130756e-05,
      "loss": 3.7922,
      "step": 1929
    },
    {
      "epoch": 2.688022284122563,
      "grad_norm": 4.148495197296143,
      "learning_rate": 5.223138146674246e-05,
      "loss": 3.5519,
      "step": 1930
    },
    {
      "epoch": 2.6894150417827296,
      "grad_norm": 4.800426483154297,
      "learning_rate": 5.2221432632177364e-05,
      "loss": 3.7697,
      "step": 1931
    },
    {
      "epoch": 2.690807799442897,
      "grad_norm": 9.859857559204102,
      "learning_rate": 5.2211483797612275e-05,
      "loss": 3.2831,
      "step": 1932
    },
    {
      "epoch": 2.692200557103064,
      "grad_norm": 4.45426607131958,
      "learning_rate": 5.2201534963047185e-05,
      "loss": 3.8314,
      "step": 1933
    },
    {
      "epoch": 2.6935933147632314,
      "grad_norm": 4.1094465255737305,
      "learning_rate": 5.219158612848209e-05,
      "loss": 3.389,
      "step": 1934
    },
    {
      "epoch": 2.6949860724233985,
      "grad_norm": 4.332690238952637,
      "learning_rate": 5.2181637293916994e-05,
      "loss": 4.0012,
      "step": 1935
    },
    {
      "epoch": 2.6963788300835656,
      "grad_norm": 4.643312454223633,
      "learning_rate": 5.21716884593519e-05,
      "loss": 3.3735,
      "step": 1936
    },
    {
      "epoch": 2.6977715877437327,
      "grad_norm": 5.339315414428711,
      "learning_rate": 5.216173962478681e-05,
      "loss": 3.2747,
      "step": 1937
    },
    {
      "epoch": 2.6991643454039,
      "grad_norm": 4.450771808624268,
      "learning_rate": 5.215179079022171e-05,
      "loss": 3.4963,
      "step": 1938
    },
    {
      "epoch": 2.700557103064067,
      "grad_norm": 4.209102630615234,
      "learning_rate": 5.2141841955656623e-05,
      "loss": 3.3696,
      "step": 1939
    },
    {
      "epoch": 2.701949860724234,
      "grad_norm": 4.320992469787598,
      "learning_rate": 5.213189312109152e-05,
      "loss": 3.6478,
      "step": 1940
    },
    {
      "epoch": 2.703342618384401,
      "grad_norm": 4.924124717712402,
      "learning_rate": 5.212194428652643e-05,
      "loss": 3.8642,
      "step": 1941
    },
    {
      "epoch": 2.7047353760445683,
      "grad_norm": 4.025848865509033,
      "learning_rate": 5.2111995451961336e-05,
      "loss": 3.1843,
      "step": 1942
    },
    {
      "epoch": 2.7061281337047354,
      "grad_norm": 5.065040111541748,
      "learning_rate": 5.2102046617396246e-05,
      "loss": 3.7114,
      "step": 1943
    },
    {
      "epoch": 2.7075208913649025,
      "grad_norm": 3.9760985374450684,
      "learning_rate": 5.209209778283115e-05,
      "loss": 3.402,
      "step": 1944
    },
    {
      "epoch": 2.7089136490250696,
      "grad_norm": 4.428731441497803,
      "learning_rate": 5.2082148948266055e-05,
      "loss": 3.6306,
      "step": 1945
    },
    {
      "epoch": 2.7103064066852367,
      "grad_norm": 3.8420968055725098,
      "learning_rate": 5.207220011370096e-05,
      "loss": 3.0539,
      "step": 1946
    },
    {
      "epoch": 2.711699164345404,
      "grad_norm": 4.728443622589111,
      "learning_rate": 5.206225127913587e-05,
      "loss": 4.1489,
      "step": 1947
    },
    {
      "epoch": 2.713091922005571,
      "grad_norm": 4.3703765869140625,
      "learning_rate": 5.2052302444570774e-05,
      "loss": 3.4861,
      "step": 1948
    },
    {
      "epoch": 2.714484679665738,
      "grad_norm": 4.901739120483398,
      "learning_rate": 5.2042353610005684e-05,
      "loss": 3.2697,
      "step": 1949
    },
    {
      "epoch": 2.715877437325905,
      "grad_norm": 4.627745628356934,
      "learning_rate": 5.203240477544058e-05,
      "loss": 3.3958,
      "step": 1950
    },
    {
      "epoch": 2.7172701949860727,
      "grad_norm": 5.152009963989258,
      "learning_rate": 5.202245594087549e-05,
      "loss": 2.9311,
      "step": 1951
    },
    {
      "epoch": 2.7186629526462394,
      "grad_norm": 4.1321587562561035,
      "learning_rate": 5.20125071063104e-05,
      "loss": 3.2103,
      "step": 1952
    },
    {
      "epoch": 2.720055710306407,
      "grad_norm": 4.1402740478515625,
      "learning_rate": 5.200255827174531e-05,
      "loss": 3.6239,
      "step": 1953
    },
    {
      "epoch": 2.7214484679665736,
      "grad_norm": 5.09497594833374,
      "learning_rate": 5.199260943718021e-05,
      "loss": 3.589,
      "step": 1954
    },
    {
      "epoch": 2.722841225626741,
      "grad_norm": 4.253708362579346,
      "learning_rate": 5.1982660602615116e-05,
      "loss": 3.2043,
      "step": 1955
    },
    {
      "epoch": 2.724233983286908,
      "grad_norm": 4.034773826599121,
      "learning_rate": 5.197271176805002e-05,
      "loss": 3.555,
      "step": 1956
    },
    {
      "epoch": 2.7256267409470754,
      "grad_norm": 4.14765739440918,
      "learning_rate": 5.196276293348493e-05,
      "loss": 3.5134,
      "step": 1957
    },
    {
      "epoch": 2.7270194986072425,
      "grad_norm": 4.335793972015381,
      "learning_rate": 5.195281409891984e-05,
      "loss": 3.3533,
      "step": 1958
    },
    {
      "epoch": 2.7284122562674096,
      "grad_norm": 5.370896339416504,
      "learning_rate": 5.1942865264354746e-05,
      "loss": 3.7931,
      "step": 1959
    },
    {
      "epoch": 2.7298050139275767,
      "grad_norm": 4.351439476013184,
      "learning_rate": 5.193291642978964e-05,
      "loss": 3.7763,
      "step": 1960
    },
    {
      "epoch": 2.731197771587744,
      "grad_norm": 4.499411106109619,
      "learning_rate": 5.1922967595224554e-05,
      "loss": 3.453,
      "step": 1961
    },
    {
      "epoch": 2.732590529247911,
      "grad_norm": 6.719456672668457,
      "learning_rate": 5.1913018760659465e-05,
      "loss": 4.2591,
      "step": 1962
    },
    {
      "epoch": 2.733983286908078,
      "grad_norm": 4.187103271484375,
      "learning_rate": 5.190306992609437e-05,
      "loss": 3.712,
      "step": 1963
    },
    {
      "epoch": 2.735376044568245,
      "grad_norm": 4.134088516235352,
      "learning_rate": 5.189312109152928e-05,
      "loss": 3.4247,
      "step": 1964
    },
    {
      "epoch": 2.7367688022284122,
      "grad_norm": 3.8819053173065186,
      "learning_rate": 5.188317225696418e-05,
      "loss": 3.1055,
      "step": 1965
    },
    {
      "epoch": 2.7381615598885793,
      "grad_norm": 4.02744197845459,
      "learning_rate": 5.187322342239909e-05,
      "loss": 3.665,
      "step": 1966
    },
    {
      "epoch": 2.7395543175487465,
      "grad_norm": 3.618467330932617,
      "learning_rate": 5.186327458783399e-05,
      "loss": 3.2742,
      "step": 1967
    },
    {
      "epoch": 2.7409470752089136,
      "grad_norm": 4.1958537101745605,
      "learning_rate": 5.18533257532689e-05,
      "loss": 3.7362,
      "step": 1968
    },
    {
      "epoch": 2.7423398328690807,
      "grad_norm": 4.144404411315918,
      "learning_rate": 5.18433769187038e-05,
      "loss": 3.7999,
      "step": 1969
    },
    {
      "epoch": 2.743732590529248,
      "grad_norm": 4.595558166503906,
      "learning_rate": 5.183342808413871e-05,
      "loss": 3.6315,
      "step": 1970
    },
    {
      "epoch": 2.745125348189415,
      "grad_norm": 3.7925376892089844,
      "learning_rate": 5.1823479249573615e-05,
      "loss": 2.9733,
      "step": 1971
    },
    {
      "epoch": 2.7465181058495824,
      "grad_norm": 3.7566452026367188,
      "learning_rate": 5.1813530415008526e-05,
      "loss": 3.198,
      "step": 1972
    },
    {
      "epoch": 2.747910863509749,
      "grad_norm": 4.324394226074219,
      "learning_rate": 5.180358158044343e-05,
      "loss": 3.6722,
      "step": 1973
    },
    {
      "epoch": 2.7493036211699167,
      "grad_norm": 5.738814353942871,
      "learning_rate": 5.1793632745878334e-05,
      "loss": 3.5951,
      "step": 1974
    },
    {
      "epoch": 2.7506963788300833,
      "grad_norm": 4.4775567054748535,
      "learning_rate": 5.178368391131324e-05,
      "loss": 3.524,
      "step": 1975
    },
    {
      "epoch": 2.752089136490251,
      "grad_norm": 4.070470333099365,
      "learning_rate": 5.177373507674815e-05,
      "loss": 3.2178,
      "step": 1976
    },
    {
      "epoch": 2.7534818941504176,
      "grad_norm": 4.293017864227295,
      "learning_rate": 5.176378624218305e-05,
      "loss": 3.8558,
      "step": 1977
    },
    {
      "epoch": 2.754874651810585,
      "grad_norm": 4.2213640213012695,
      "learning_rate": 5.1753837407617964e-05,
      "loss": 3.5929,
      "step": 1978
    },
    {
      "epoch": 2.756267409470752,
      "grad_norm": 4.38236665725708,
      "learning_rate": 5.174388857305286e-05,
      "loss": 3.4067,
      "step": 1979
    },
    {
      "epoch": 2.7576601671309193,
      "grad_norm": 4.706287384033203,
      "learning_rate": 5.173393973848777e-05,
      "loss": 3.7148,
      "step": 1980
    },
    {
      "epoch": 2.7590529247910864,
      "grad_norm": 4.4334516525268555,
      "learning_rate": 5.1723990903922676e-05,
      "loss": 3.5652,
      "step": 1981
    },
    {
      "epoch": 2.7604456824512535,
      "grad_norm": 4.028783798217773,
      "learning_rate": 5.171404206935759e-05,
      "loss": 3.4242,
      "step": 1982
    },
    {
      "epoch": 2.7618384401114207,
      "grad_norm": 4.340991020202637,
      "learning_rate": 5.170409323479249e-05,
      "loss": 3.861,
      "step": 1983
    },
    {
      "epoch": 2.7632311977715878,
      "grad_norm": 4.295785427093506,
      "learning_rate": 5.1694144400227395e-05,
      "loss": 3.5997,
      "step": 1984
    },
    {
      "epoch": 2.764623955431755,
      "grad_norm": 6.218203544616699,
      "learning_rate": 5.16841955656623e-05,
      "loss": 3.4819,
      "step": 1985
    },
    {
      "epoch": 2.766016713091922,
      "grad_norm": 4.786428451538086,
      "learning_rate": 5.167424673109721e-05,
      "loss": 3.01,
      "step": 1986
    },
    {
      "epoch": 2.767409470752089,
      "grad_norm": 4.1766252517700195,
      "learning_rate": 5.166429789653212e-05,
      "loss": 3.3219,
      "step": 1987
    },
    {
      "epoch": 2.768802228412256,
      "grad_norm": 4.175639629364014,
      "learning_rate": 5.1654349061967025e-05,
      "loss": 3.8378,
      "step": 1988
    },
    {
      "epoch": 2.7701949860724233,
      "grad_norm": 5.461728096008301,
      "learning_rate": 5.164440022740192e-05,
      "loss": 3.6676,
      "step": 1989
    },
    {
      "epoch": 2.7715877437325904,
      "grad_norm": 4.121284008026123,
      "learning_rate": 5.163445139283683e-05,
      "loss": 3.7472,
      "step": 1990
    },
    {
      "epoch": 2.7729805013927575,
      "grad_norm": 4.15053653717041,
      "learning_rate": 5.1624502558271744e-05,
      "loss": 3.297,
      "step": 1991
    },
    {
      "epoch": 2.7743732590529246,
      "grad_norm": 4.42463493347168,
      "learning_rate": 5.161455372370665e-05,
      "loss": 3.6092,
      "step": 1992
    },
    {
      "epoch": 2.775766016713092,
      "grad_norm": 3.930119752883911,
      "learning_rate": 5.160460488914156e-05,
      "loss": 3.3793,
      "step": 1993
    },
    {
      "epoch": 2.777158774373259,
      "grad_norm": 4.388279438018799,
      "learning_rate": 5.1594656054576456e-05,
      "loss": 3.472,
      "step": 1994
    },
    {
      "epoch": 2.7785515320334264,
      "grad_norm": 5.646612167358398,
      "learning_rate": 5.158470722001137e-05,
      "loss": 3.2413,
      "step": 1995
    },
    {
      "epoch": 2.779944289693593,
      "grad_norm": 4.878632545471191,
      "learning_rate": 5.157475838544627e-05,
      "loss": 3.386,
      "step": 1996
    },
    {
      "epoch": 2.7813370473537606,
      "grad_norm": 4.885819911956787,
      "learning_rate": 5.156480955088118e-05,
      "loss": 3.8848,
      "step": 1997
    },
    {
      "epoch": 2.7827298050139273,
      "grad_norm": 4.037961959838867,
      "learning_rate": 5.1554860716316086e-05,
      "loss": 3.2975,
      "step": 1998
    },
    {
      "epoch": 2.784122562674095,
      "grad_norm": 4.697548866271973,
      "learning_rate": 5.154491188175099e-05,
      "loss": 4.0258,
      "step": 1999
    },
    {
      "epoch": 2.785515320334262,
      "grad_norm": 5.877390384674072,
      "learning_rate": 5.1534963047185894e-05,
      "loss": 3.9183,
      "step": 2000
    },
    {
      "epoch": 2.786908077994429,
      "grad_norm": 4.240438461303711,
      "learning_rate": 5.1525014212620805e-05,
      "loss": 3.7199,
      "step": 2001
    },
    {
      "epoch": 2.788300835654596,
      "grad_norm": 3.9093611240386963,
      "learning_rate": 5.151506537805571e-05,
      "loss": 3.2419,
      "step": 2002
    },
    {
      "epoch": 2.7896935933147633,
      "grad_norm": 4.7583699226379395,
      "learning_rate": 5.150511654349062e-05,
      "loss": 3.2971,
      "step": 2003
    },
    {
      "epoch": 2.7910863509749304,
      "grad_norm": 9.54155445098877,
      "learning_rate": 5.149516770892552e-05,
      "loss": 3.7672,
      "step": 2004
    },
    {
      "epoch": 2.7924791086350975,
      "grad_norm": 4.426088333129883,
      "learning_rate": 5.148521887436043e-05,
      "loss": 3.5291,
      "step": 2005
    },
    {
      "epoch": 2.7938718662952646,
      "grad_norm": 5.26630163192749,
      "learning_rate": 5.147527003979533e-05,
      "loss": 3.2663,
      "step": 2006
    },
    {
      "epoch": 2.7952646239554317,
      "grad_norm": 4.465476036071777,
      "learning_rate": 5.146532120523024e-05,
      "loss": 3.1189,
      "step": 2007
    },
    {
      "epoch": 2.796657381615599,
      "grad_norm": 5.240188121795654,
      "learning_rate": 5.145537237066515e-05,
      "loss": 3.5407,
      "step": 2008
    },
    {
      "epoch": 2.798050139275766,
      "grad_norm": 3.9314498901367188,
      "learning_rate": 5.144542353610005e-05,
      "loss": 3.2046,
      "step": 2009
    },
    {
      "epoch": 2.799442896935933,
      "grad_norm": 5.52628231048584,
      "learning_rate": 5.1435474701534955e-05,
      "loss": 3.216,
      "step": 2010
    },
    {
      "epoch": 2.8008356545961,
      "grad_norm": 3.9009246826171875,
      "learning_rate": 5.1425525866969866e-05,
      "loss": 3.3723,
      "step": 2011
    },
    {
      "epoch": 2.8022284122562673,
      "grad_norm": 4.057968616485596,
      "learning_rate": 5.141557703240477e-05,
      "loss": 3.6863,
      "step": 2012
    },
    {
      "epoch": 2.8036211699164344,
      "grad_norm": 4.20530891418457,
      "learning_rate": 5.1405628197839674e-05,
      "loss": 3.3286,
      "step": 2013
    },
    {
      "epoch": 2.8050139275766015,
      "grad_norm": 3.844698905944824,
      "learning_rate": 5.139567936327458e-05,
      "loss": 3.5184,
      "step": 2014
    },
    {
      "epoch": 2.8064066852367686,
      "grad_norm": 4.140641689300537,
      "learning_rate": 5.138573052870949e-05,
      "loss": 3.512,
      "step": 2015
    },
    {
      "epoch": 2.807799442896936,
      "grad_norm": 5.2917985916137695,
      "learning_rate": 5.13757816941444e-05,
      "loss": 3.1181,
      "step": 2016
    },
    {
      "epoch": 2.809192200557103,
      "grad_norm": 4.944843292236328,
      "learning_rate": 5.1365832859579304e-05,
      "loss": 4.0601,
      "step": 2017
    },
    {
      "epoch": 2.8105849582172704,
      "grad_norm": 4.644077777862549,
      "learning_rate": 5.13558840250142e-05,
      "loss": 3.1175,
      "step": 2018
    },
    {
      "epoch": 2.811977715877437,
      "grad_norm": 4.8350677490234375,
      "learning_rate": 5.134593519044911e-05,
      "loss": 3.7456,
      "step": 2019
    },
    {
      "epoch": 2.8133704735376046,
      "grad_norm": 4.3641510009765625,
      "learning_rate": 5.133598635588402e-05,
      "loss": 3.3445,
      "step": 2020
    },
    {
      "epoch": 2.8147632311977717,
      "grad_norm": 4.085535049438477,
      "learning_rate": 5.132603752131893e-05,
      "loss": 3.5831,
      "step": 2021
    },
    {
      "epoch": 2.816155988857939,
      "grad_norm": 6.408960819244385,
      "learning_rate": 5.131608868675384e-05,
      "loss": 3.8303,
      "step": 2022
    },
    {
      "epoch": 2.817548746518106,
      "grad_norm": 4.139590263366699,
      "learning_rate": 5.1306139852188735e-05,
      "loss": 3.2063,
      "step": 2023
    },
    {
      "epoch": 2.818941504178273,
      "grad_norm": 3.946959972381592,
      "learning_rate": 5.1296191017623646e-05,
      "loss": 3.3457,
      "step": 2024
    },
    {
      "epoch": 2.82033426183844,
      "grad_norm": 4.58073091506958,
      "learning_rate": 5.128624218305855e-05,
      "loss": 4.1379,
      "step": 2025
    },
    {
      "epoch": 2.8217270194986073,
      "grad_norm": 3.9743804931640625,
      "learning_rate": 5.127629334849346e-05,
      "loss": 3.5062,
      "step": 2026
    },
    {
      "epoch": 2.8231197771587744,
      "grad_norm": 4.281818866729736,
      "learning_rate": 5.1266344513928365e-05,
      "loss": 3.447,
      "step": 2027
    },
    {
      "epoch": 2.8245125348189415,
      "grad_norm": 4.268107891082764,
      "learning_rate": 5.125639567936327e-05,
      "loss": 3.0485,
      "step": 2028
    },
    {
      "epoch": 2.8259052924791086,
      "grad_norm": 4.711485862731934,
      "learning_rate": 5.124644684479817e-05,
      "loss": 3.8763,
      "step": 2029
    },
    {
      "epoch": 2.8272980501392757,
      "grad_norm": 4.70322322845459,
      "learning_rate": 5.1236498010233084e-05,
      "loss": 3.6961,
      "step": 2030
    },
    {
      "epoch": 2.828690807799443,
      "grad_norm": 6.191211700439453,
      "learning_rate": 5.122654917566799e-05,
      "loss": 3.2545,
      "step": 2031
    },
    {
      "epoch": 2.83008356545961,
      "grad_norm": 4.219863414764404,
      "learning_rate": 5.12166003411029e-05,
      "loss": 3.268,
      "step": 2032
    },
    {
      "epoch": 2.831476323119777,
      "grad_norm": 4.127409934997559,
      "learning_rate": 5.1206651506537796e-05,
      "loss": 3.1905,
      "step": 2033
    },
    {
      "epoch": 2.832869080779944,
      "grad_norm": 4.068781852722168,
      "learning_rate": 5.119670267197271e-05,
      "loss": 2.9591,
      "step": 2034
    },
    {
      "epoch": 2.8342618384401113,
      "grad_norm": 3.95893931388855,
      "learning_rate": 5.118675383740761e-05,
      "loss": 2.8394,
      "step": 2035
    },
    {
      "epoch": 2.8356545961002784,
      "grad_norm": 3.969543933868408,
      "learning_rate": 5.117680500284252e-05,
      "loss": 3.4354,
      "step": 2036
    },
    {
      "epoch": 2.837047353760446,
      "grad_norm": 3.9751126766204834,
      "learning_rate": 5.1166856168277426e-05,
      "loss": 3.3789,
      "step": 2037
    },
    {
      "epoch": 2.8384401114206126,
      "grad_norm": 4.16307258605957,
      "learning_rate": 5.115690733371233e-05,
      "loss": 3.5744,
      "step": 2038
    },
    {
      "epoch": 2.83983286908078,
      "grad_norm": 4.139352798461914,
      "learning_rate": 5.1146958499147234e-05,
      "loss": 3.4809,
      "step": 2039
    },
    {
      "epoch": 2.841225626740947,
      "grad_norm": 4.255465507507324,
      "learning_rate": 5.1137009664582145e-05,
      "loss": 3.2271,
      "step": 2040
    },
    {
      "epoch": 2.8426183844011144,
      "grad_norm": 5.983844757080078,
      "learning_rate": 5.1127060830017056e-05,
      "loss": 3.0895,
      "step": 2041
    },
    {
      "epoch": 2.8440111420612815,
      "grad_norm": 5.0439910888671875,
      "learning_rate": 5.111711199545196e-05,
      "loss": 3.5335,
      "step": 2042
    },
    {
      "epoch": 2.8454038997214486,
      "grad_norm": 3.8901827335357666,
      "learning_rate": 5.110716316088686e-05,
      "loss": 3.5002,
      "step": 2043
    },
    {
      "epoch": 2.8467966573816157,
      "grad_norm": 4.606709957122803,
      "learning_rate": 5.109721432632177e-05,
      "loss": 3.2298,
      "step": 2044
    },
    {
      "epoch": 2.848189415041783,
      "grad_norm": 5.750291347503662,
      "learning_rate": 5.108726549175668e-05,
      "loss": 3.3431,
      "step": 2045
    },
    {
      "epoch": 2.84958217270195,
      "grad_norm": 4.332377910614014,
      "learning_rate": 5.107731665719158e-05,
      "loss": 3.8921,
      "step": 2046
    },
    {
      "epoch": 2.850974930362117,
      "grad_norm": 3.8708927631378174,
      "learning_rate": 5.1067367822626494e-05,
      "loss": 3.5135,
      "step": 2047
    },
    {
      "epoch": 2.852367688022284,
      "grad_norm": 3.9907753467559814,
      "learning_rate": 5.105741898806139e-05,
      "loss": 3.1292,
      "step": 2048
    },
    {
      "epoch": 2.8537604456824512,
      "grad_norm": 4.350218772888184,
      "learning_rate": 5.10474701534963e-05,
      "loss": 3.5061,
      "step": 2049
    },
    {
      "epoch": 2.8551532033426184,
      "grad_norm": 4.249178409576416,
      "learning_rate": 5.1037521318931206e-05,
      "loss": 3.537,
      "step": 2050
    },
    {
      "epoch": 2.8565459610027855,
      "grad_norm": 4.3957061767578125,
      "learning_rate": 5.102757248436612e-05,
      "loss": 3.0038,
      "step": 2051
    },
    {
      "epoch": 2.8579387186629526,
      "grad_norm": 4.325258731842041,
      "learning_rate": 5.101762364980102e-05,
      "loss": 3.8054,
      "step": 2052
    },
    {
      "epoch": 2.8593314763231197,
      "grad_norm": 4.3150153160095215,
      "learning_rate": 5.1007674815235925e-05,
      "loss": 3.2175,
      "step": 2053
    },
    {
      "epoch": 2.860724233983287,
      "grad_norm": 4.273927211761475,
      "learning_rate": 5.099772598067083e-05,
      "loss": 3.3386,
      "step": 2054
    },
    {
      "epoch": 2.862116991643454,
      "grad_norm": 4.8481903076171875,
      "learning_rate": 5.098777714610574e-05,
      "loss": 3.4993,
      "step": 2055
    },
    {
      "epoch": 2.863509749303621,
      "grad_norm": 3.8663220405578613,
      "learning_rate": 5.0977828311540644e-05,
      "loss": 3.2226,
      "step": 2056
    },
    {
      "epoch": 2.864902506963788,
      "grad_norm": 4.236097812652588,
      "learning_rate": 5.096787947697555e-05,
      "loss": 3.1875,
      "step": 2057
    },
    {
      "epoch": 2.8662952646239557,
      "grad_norm": 5.388729572296143,
      "learning_rate": 5.095793064241045e-05,
      "loss": 3.0379,
      "step": 2058
    },
    {
      "epoch": 2.8676880222841223,
      "grad_norm": 3.7443923950195312,
      "learning_rate": 5.094798180784536e-05,
      "loss": 3.1198,
      "step": 2059
    },
    {
      "epoch": 2.86908077994429,
      "grad_norm": 3.982881546020508,
      "learning_rate": 5.093803297328027e-05,
      "loss": 3.639,
      "step": 2060
    },
    {
      "epoch": 2.8704735376044566,
      "grad_norm": 5.106269836425781,
      "learning_rate": 5.092808413871518e-05,
      "loss": 3.3559,
      "step": 2061
    },
    {
      "epoch": 2.871866295264624,
      "grad_norm": 4.366353511810303,
      "learning_rate": 5.0918135304150075e-05,
      "loss": 3.5405,
      "step": 2062
    },
    {
      "epoch": 2.8732590529247912,
      "grad_norm": 4.137770652770996,
      "learning_rate": 5.0908186469584986e-05,
      "loss": 3.283,
      "step": 2063
    },
    {
      "epoch": 2.8746518105849583,
      "grad_norm": 4.251780986785889,
      "learning_rate": 5.089823763501989e-05,
      "loss": 3.3503,
      "step": 2064
    },
    {
      "epoch": 2.8760445682451254,
      "grad_norm": 4.264405250549316,
      "learning_rate": 5.08882888004548e-05,
      "loss": 3.3128,
      "step": 2065
    },
    {
      "epoch": 2.8774373259052926,
      "grad_norm": 4.219046115875244,
      "learning_rate": 5.0878339965889705e-05,
      "loss": 3.2743,
      "step": 2066
    },
    {
      "epoch": 2.8788300835654597,
      "grad_norm": 4.376394271850586,
      "learning_rate": 5.086839113132461e-05,
      "loss": 3.7509,
      "step": 2067
    },
    {
      "epoch": 2.8802228412256268,
      "grad_norm": 3.643519401550293,
      "learning_rate": 5.085844229675951e-05,
      "loss": 3.1443,
      "step": 2068
    },
    {
      "epoch": 2.881615598885794,
      "grad_norm": 4.250158309936523,
      "learning_rate": 5.0848493462194424e-05,
      "loss": 3.3089,
      "step": 2069
    },
    {
      "epoch": 2.883008356545961,
      "grad_norm": 4.211243152618408,
      "learning_rate": 5.0838544627629335e-05,
      "loss": 3.6829,
      "step": 2070
    },
    {
      "epoch": 2.884401114206128,
      "grad_norm": 3.8670225143432617,
      "learning_rate": 5.082859579306424e-05,
      "loss": 3.1512,
      "step": 2071
    },
    {
      "epoch": 2.885793871866295,
      "grad_norm": 4.404150485992432,
      "learning_rate": 5.0818646958499136e-05,
      "loss": 3.4532,
      "step": 2072
    },
    {
      "epoch": 2.8871866295264623,
      "grad_norm": 5.2248101234436035,
      "learning_rate": 5.080869812393405e-05,
      "loss": 3.2853,
      "step": 2073
    },
    {
      "epoch": 2.8885793871866294,
      "grad_norm": 4.03637170791626,
      "learning_rate": 5.079874928936896e-05,
      "loss": 3.4545,
      "step": 2074
    },
    {
      "epoch": 2.8899721448467965,
      "grad_norm": 4.957915306091309,
      "learning_rate": 5.078880045480386e-05,
      "loss": 3.2843,
      "step": 2075
    },
    {
      "epoch": 2.8913649025069637,
      "grad_norm": 4.652523517608643,
      "learning_rate": 5.077885162023877e-05,
      "loss": 3.7638,
      "step": 2076
    },
    {
      "epoch": 2.8927576601671308,
      "grad_norm": 5.459745407104492,
      "learning_rate": 5.076890278567367e-05,
      "loss": 3.1137,
      "step": 2077
    },
    {
      "epoch": 2.894150417827298,
      "grad_norm": 4.156223773956299,
      "learning_rate": 5.075895395110858e-05,
      "loss": 3.0294,
      "step": 2078
    },
    {
      "epoch": 2.8955431754874654,
      "grad_norm": 4.485111236572266,
      "learning_rate": 5.0749005116543485e-05,
      "loss": 3.6693,
      "step": 2079
    },
    {
      "epoch": 2.896935933147632,
      "grad_norm": 4.370935440063477,
      "learning_rate": 5.0739056281978396e-05,
      "loss": 3.4302,
      "step": 2080
    },
    {
      "epoch": 2.8983286908077996,
      "grad_norm": 4.037283897399902,
      "learning_rate": 5.07291074474133e-05,
      "loss": 3.3939,
      "step": 2081
    },
    {
      "epoch": 2.8997214484679663,
      "grad_norm": 4.518899440765381,
      "learning_rate": 5.0719158612848204e-05,
      "loss": 3.8476,
      "step": 2082
    },
    {
      "epoch": 2.901114206128134,
      "grad_norm": 5.807830810546875,
      "learning_rate": 5.070920977828311e-05,
      "loss": 3.7524,
      "step": 2083
    },
    {
      "epoch": 2.902506963788301,
      "grad_norm": 5.00191593170166,
      "learning_rate": 5.069926094371802e-05,
      "loss": 3.457,
      "step": 2084
    },
    {
      "epoch": 2.903899721448468,
      "grad_norm": 4.1947503089904785,
      "learning_rate": 5.068931210915292e-05,
      "loss": 3.9833,
      "step": 2085
    },
    {
      "epoch": 2.905292479108635,
      "grad_norm": 4.5223259925842285,
      "learning_rate": 5.0679363274587834e-05,
      "loss": 3.3464,
      "step": 2086
    },
    {
      "epoch": 2.9066852367688023,
      "grad_norm": 4.826055526733398,
      "learning_rate": 5.066941444002273e-05,
      "loss": 3.7102,
      "step": 2087
    },
    {
      "epoch": 2.9080779944289694,
      "grad_norm": 3.685760498046875,
      "learning_rate": 5.065946560545764e-05,
      "loss": 3.3121,
      "step": 2088
    },
    {
      "epoch": 2.9094707520891365,
      "grad_norm": 4.318071365356445,
      "learning_rate": 5.0649516770892546e-05,
      "loss": 3.7514,
      "step": 2089
    },
    {
      "epoch": 2.9108635097493036,
      "grad_norm": 3.9932713508605957,
      "learning_rate": 5.063956793632746e-05,
      "loss": 3.6518,
      "step": 2090
    },
    {
      "epoch": 2.9122562674094707,
      "grad_norm": 5.094545841217041,
      "learning_rate": 5.062961910176236e-05,
      "loss": 3.3245,
      "step": 2091
    },
    {
      "epoch": 2.913649025069638,
      "grad_norm": 4.561112403869629,
      "learning_rate": 5.0619670267197265e-05,
      "loss": 3.7716,
      "step": 2092
    },
    {
      "epoch": 2.915041782729805,
      "grad_norm": 3.9744319915771484,
      "learning_rate": 5.060972143263217e-05,
      "loss": 2.7508,
      "step": 2093
    },
    {
      "epoch": 2.916434540389972,
      "grad_norm": 4.323774337768555,
      "learning_rate": 5.059977259806708e-05,
      "loss": 3.3937,
      "step": 2094
    },
    {
      "epoch": 2.917827298050139,
      "grad_norm": 4.414550304412842,
      "learning_rate": 5.0589823763501984e-05,
      "loss": 3.7095,
      "step": 2095
    },
    {
      "epoch": 2.9192200557103063,
      "grad_norm": 4.5342912673950195,
      "learning_rate": 5.0579874928936895e-05,
      "loss": 3.3317,
      "step": 2096
    },
    {
      "epoch": 2.9206128133704734,
      "grad_norm": 6.246424674987793,
      "learning_rate": 5.056992609437179e-05,
      "loss": 3.6733,
      "step": 2097
    },
    {
      "epoch": 2.9220055710306405,
      "grad_norm": 6.608816146850586,
      "learning_rate": 5.05599772598067e-05,
      "loss": 3.3776,
      "step": 2098
    },
    {
      "epoch": 2.9233983286908076,
      "grad_norm": 4.423455238342285,
      "learning_rate": 5.0550028425241614e-05,
      "loss": 3.4785,
      "step": 2099
    },
    {
      "epoch": 2.924791086350975,
      "grad_norm": 7.79443359375,
      "learning_rate": 5.054007959067652e-05,
      "loss": 3.7376,
      "step": 2100
    },
    {
      "epoch": 2.926183844011142,
      "grad_norm": 4.41956901550293,
      "learning_rate": 5.0530130756111416e-05,
      "loss": 3.5199,
      "step": 2101
    },
    {
      "epoch": 2.9275766016713094,
      "grad_norm": 4.584619522094727,
      "learning_rate": 5.0520181921546326e-05,
      "loss": 3.8453,
      "step": 2102
    },
    {
      "epoch": 2.928969359331476,
      "grad_norm": 4.029069900512695,
      "learning_rate": 5.051023308698124e-05,
      "loss": 3.2393,
      "step": 2103
    },
    {
      "epoch": 2.9303621169916436,
      "grad_norm": 4.323663711547852,
      "learning_rate": 5.050028425241614e-05,
      "loss": 3.7648,
      "step": 2104
    },
    {
      "epoch": 2.9317548746518107,
      "grad_norm": 4.1720404624938965,
      "learning_rate": 5.049033541785105e-05,
      "loss": 2.879,
      "step": 2105
    },
    {
      "epoch": 2.933147632311978,
      "grad_norm": 4.187463760375977,
      "learning_rate": 5.048038658328595e-05,
      "loss": 2.8754,
      "step": 2106
    },
    {
      "epoch": 2.934540389972145,
      "grad_norm": 4.060224533081055,
      "learning_rate": 5.047043774872086e-05,
      "loss": 3.123,
      "step": 2107
    },
    {
      "epoch": 2.935933147632312,
      "grad_norm": 4.4883904457092285,
      "learning_rate": 5.0460488914155764e-05,
      "loss": 3.6887,
      "step": 2108
    },
    {
      "epoch": 2.937325905292479,
      "grad_norm": 4.690547943115234,
      "learning_rate": 5.0450540079590675e-05,
      "loss": 3.657,
      "step": 2109
    },
    {
      "epoch": 2.9387186629526463,
      "grad_norm": 5.70568323135376,
      "learning_rate": 5.044059124502558e-05,
      "loss": 3.3645,
      "step": 2110
    },
    {
      "epoch": 2.9401114206128134,
      "grad_norm": 5.02766227722168,
      "learning_rate": 5.0430642410460483e-05,
      "loss": 3.3286,
      "step": 2111
    },
    {
      "epoch": 2.9415041782729805,
      "grad_norm": 4.652088165283203,
      "learning_rate": 5.042069357589539e-05,
      "loss": 3.1985,
      "step": 2112
    },
    {
      "epoch": 2.9428969359331476,
      "grad_norm": 4.628662586212158,
      "learning_rate": 5.04107447413303e-05,
      "loss": 3.6984,
      "step": 2113
    },
    {
      "epoch": 2.9442896935933147,
      "grad_norm": 3.9817137718200684,
      "learning_rate": 5.04007959067652e-05,
      "loss": 2.973,
      "step": 2114
    },
    {
      "epoch": 2.945682451253482,
      "grad_norm": 4.922610282897949,
      "learning_rate": 5.039084707220011e-05,
      "loss": 3.0331,
      "step": 2115
    },
    {
      "epoch": 2.947075208913649,
      "grad_norm": 3.9922170639038086,
      "learning_rate": 5.038089823763501e-05,
      "loss": 3.1564,
      "step": 2116
    },
    {
      "epoch": 2.948467966573816,
      "grad_norm": 4.425002574920654,
      "learning_rate": 5.037094940306992e-05,
      "loss": 3.6502,
      "step": 2117
    },
    {
      "epoch": 2.949860724233983,
      "grad_norm": 4.016625881195068,
      "learning_rate": 5.0361000568504826e-05,
      "loss": 3.3882,
      "step": 2118
    },
    {
      "epoch": 2.9512534818941503,
      "grad_norm": 4.493213653564453,
      "learning_rate": 5.0351051733939736e-05,
      "loss": 3.1698,
      "step": 2119
    },
    {
      "epoch": 2.9526462395543174,
      "grad_norm": 4.914352893829346,
      "learning_rate": 5.034110289937464e-05,
      "loss": 3.6761,
      "step": 2120
    },
    {
      "epoch": 2.954038997214485,
      "grad_norm": 5.080784320831299,
      "learning_rate": 5.0331154064809545e-05,
      "loss": 3.7529,
      "step": 2121
    },
    {
      "epoch": 2.9554317548746516,
      "grad_norm": 4.149218559265137,
      "learning_rate": 5.032120523024445e-05,
      "loss": 3.1979,
      "step": 2122
    },
    {
      "epoch": 2.956824512534819,
      "grad_norm": 9.52330493927002,
      "learning_rate": 5.031125639567936e-05,
      "loss": 2.9479,
      "step": 2123
    },
    {
      "epoch": 2.958217270194986,
      "grad_norm": 5.114642143249512,
      "learning_rate": 5.0301307561114264e-05,
      "loss": 3.9038,
      "step": 2124
    },
    {
      "epoch": 2.9596100278551534,
      "grad_norm": 4.347400665283203,
      "learning_rate": 5.0291358726549174e-05,
      "loss": 3.6515,
      "step": 2125
    },
    {
      "epoch": 2.9610027855153205,
      "grad_norm": 5.055492877960205,
      "learning_rate": 5.028140989198407e-05,
      "loss": 3.9448,
      "step": 2126
    },
    {
      "epoch": 2.9623955431754876,
      "grad_norm": 5.06788444519043,
      "learning_rate": 5.027146105741898e-05,
      "loss": 3.5084,
      "step": 2127
    },
    {
      "epoch": 2.9637883008356547,
      "grad_norm": 4.336512565612793,
      "learning_rate": 5.026151222285389e-05,
      "loss": 3.4349,
      "step": 2128
    },
    {
      "epoch": 2.965181058495822,
      "grad_norm": 3.7679922580718994,
      "learning_rate": 5.02515633882888e-05,
      "loss": 3.2156,
      "step": 2129
    },
    {
      "epoch": 2.966573816155989,
      "grad_norm": 5.449789047241211,
      "learning_rate": 5.024161455372371e-05,
      "loss": 3.5185,
      "step": 2130
    },
    {
      "epoch": 2.967966573816156,
      "grad_norm": 3.7605178356170654,
      "learning_rate": 5.0231665719158606e-05,
      "loss": 3.0766,
      "step": 2131
    },
    {
      "epoch": 2.969359331476323,
      "grad_norm": 3.99495267868042,
      "learning_rate": 5.0221716884593516e-05,
      "loss": 3.2622,
      "step": 2132
    },
    {
      "epoch": 2.9707520891364902,
      "grad_norm": 4.847777366638184,
      "learning_rate": 5.021176805002842e-05,
      "loss": 3.2505,
      "step": 2133
    },
    {
      "epoch": 2.9721448467966574,
      "grad_norm": 4.027635097503662,
      "learning_rate": 5.020181921546333e-05,
      "loss": 3.1319,
      "step": 2134
    },
    {
      "epoch": 2.9735376044568245,
      "grad_norm": 4.147630214691162,
      "learning_rate": 5.0191870380898235e-05,
      "loss": 3.1158,
      "step": 2135
    },
    {
      "epoch": 2.9749303621169916,
      "grad_norm": 4.148489475250244,
      "learning_rate": 5.018192154633314e-05,
      "loss": 3.3226,
      "step": 2136
    },
    {
      "epoch": 2.9763231197771587,
      "grad_norm": 4.217398643493652,
      "learning_rate": 5.0171972711768044e-05,
      "loss": 3.651,
      "step": 2137
    },
    {
      "epoch": 2.977715877437326,
      "grad_norm": 5.1949381828308105,
      "learning_rate": 5.0162023877202954e-05,
      "loss": 3.4924,
      "step": 2138
    },
    {
      "epoch": 2.979108635097493,
      "grad_norm": 4.664876461029053,
      "learning_rate": 5.015207504263786e-05,
      "loss": 3.4569,
      "step": 2139
    },
    {
      "epoch": 2.98050139275766,
      "grad_norm": 4.591869354248047,
      "learning_rate": 5.014212620807276e-05,
      "loss": 3.2384,
      "step": 2140
    },
    {
      "epoch": 2.981894150417827,
      "grad_norm": 4.423793792724609,
      "learning_rate": 5.013217737350767e-05,
      "loss": 3.1595,
      "step": 2141
    },
    {
      "epoch": 2.9832869080779947,
      "grad_norm": 4.307295322418213,
      "learning_rate": 5.012222853894258e-05,
      "loss": 3.0333,
      "step": 2142
    },
    {
      "epoch": 2.9846796657381613,
      "grad_norm": 4.102986812591553,
      "learning_rate": 5.011227970437748e-05,
      "loss": 3.3809,
      "step": 2143
    },
    {
      "epoch": 2.986072423398329,
      "grad_norm": 4.3870768547058105,
      "learning_rate": 5.010233086981239e-05,
      "loss": 3.4217,
      "step": 2144
    },
    {
      "epoch": 2.9874651810584956,
      "grad_norm": 3.7008109092712402,
      "learning_rate": 5.009238203524729e-05,
      "loss": 2.917,
      "step": 2145
    },
    {
      "epoch": 2.988857938718663,
      "grad_norm": 5.337913513183594,
      "learning_rate": 5.00824332006822e-05,
      "loss": 3.0094,
      "step": 2146
    },
    {
      "epoch": 2.9902506963788302,
      "grad_norm": 5.179196834564209,
      "learning_rate": 5.0072484366117105e-05,
      "loss": 3.2392,
      "step": 2147
    },
    {
      "epoch": 2.9916434540389973,
      "grad_norm": 5.839849472045898,
      "learning_rate": 5.0062535531552016e-05,
      "loss": 3.3579,
      "step": 2148
    },
    {
      "epoch": 2.9930362116991645,
      "grad_norm": 4.230762481689453,
      "learning_rate": 5.005258669698692e-05,
      "loss": 3.1051,
      "step": 2149
    },
    {
      "epoch": 2.9944289693593316,
      "grad_norm": 3.6010777950286865,
      "learning_rate": 5.0042637862421824e-05,
      "loss": 3.1442,
      "step": 2150
    },
    {
      "epoch": 2.9958217270194987,
      "grad_norm": 4.144538879394531,
      "learning_rate": 5.003268902785673e-05,
      "loss": 3.5637,
      "step": 2151
    },
    {
      "epoch": 2.997214484679666,
      "grad_norm": 3.8748812675476074,
      "learning_rate": 5.002274019329164e-05,
      "loss": 3.5695,
      "step": 2152
    },
    {
      "epoch": 2.998607242339833,
      "grad_norm": 3.9042489528656006,
      "learning_rate": 5.001279135872655e-05,
      "loss": 3.1174,
      "step": 2153
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.9408371448516846,
      "learning_rate": 5.0002842524161454e-05,
      "loss": 3.4697,
      "step": 2154
    },
    {
      "epoch": 3.001392757660167,
      "grad_norm": 3.8024237155914307,
      "learning_rate": 4.999289368959635e-05,
      "loss": 3.0165,
      "step": 2155
    },
    {
      "epoch": 3.002785515320334,
      "grad_norm": 3.8605546951293945,
      "learning_rate": 4.998294485503126e-05,
      "loss": 2.6671,
      "step": 2156
    },
    {
      "epoch": 3.0041782729805013,
      "grad_norm": 4.001540660858154,
      "learning_rate": 4.997299602046617e-05,
      "loss": 2.9679,
      "step": 2157
    },
    {
      "epoch": 3.0055710306406684,
      "grad_norm": 3.791510581970215,
      "learning_rate": 4.9963047185901077e-05,
      "loss": 2.7511,
      "step": 2158
    },
    {
      "epoch": 3.0069637883008355,
      "grad_norm": 3.9625966548919678,
      "learning_rate": 4.995309835133599e-05,
      "loss": 2.7843,
      "step": 2159
    },
    {
      "epoch": 3.0083565459610027,
      "grad_norm": 4.191686630249023,
      "learning_rate": 4.9943149516770885e-05,
      "loss": 2.6372,
      "step": 2160
    },
    {
      "epoch": 3.0097493036211698,
      "grad_norm": 4.725612163543701,
      "learning_rate": 4.9933200682205796e-05,
      "loss": 2.6615,
      "step": 2161
    },
    {
      "epoch": 3.011142061281337,
      "grad_norm": 3.901984214782715,
      "learning_rate": 4.99232518476407e-05,
      "loss": 3.2304,
      "step": 2162
    },
    {
      "epoch": 3.012534818941504,
      "grad_norm": 4.5678229331970215,
      "learning_rate": 4.991330301307561e-05,
      "loss": 2.8722,
      "step": 2163
    },
    {
      "epoch": 3.013927576601671,
      "grad_norm": 3.9938695430755615,
      "learning_rate": 4.9903354178510515e-05,
      "loss": 2.9263,
      "step": 2164
    },
    {
      "epoch": 3.0153203342618387,
      "grad_norm": 4.176566123962402,
      "learning_rate": 4.989340534394542e-05,
      "loss": 2.9044,
      "step": 2165
    },
    {
      "epoch": 3.0167130919220058,
      "grad_norm": 4.377198696136475,
      "learning_rate": 4.988345650938032e-05,
      "loss": 3.1145,
      "step": 2166
    },
    {
      "epoch": 3.018105849582173,
      "grad_norm": 3.675611972808838,
      "learning_rate": 4.9873507674815234e-05,
      "loss": 3.1403,
      "step": 2167
    },
    {
      "epoch": 3.01949860724234,
      "grad_norm": 3.7560067176818848,
      "learning_rate": 4.986355884025014e-05,
      "loss": 3.1056,
      "step": 2168
    },
    {
      "epoch": 3.020891364902507,
      "grad_norm": 3.496814012527466,
      "learning_rate": 4.985361000568505e-05,
      "loss": 2.5276,
      "step": 2169
    },
    {
      "epoch": 3.022284122562674,
      "grad_norm": 3.7299299240112305,
      "learning_rate": 4.9843661171119946e-05,
      "loss": 2.7589,
      "step": 2170
    },
    {
      "epoch": 3.0236768802228413,
      "grad_norm": 4.047593593597412,
      "learning_rate": 4.983371233655486e-05,
      "loss": 3.0481,
      "step": 2171
    },
    {
      "epoch": 3.0250696378830084,
      "grad_norm": 3.7720863819122314,
      "learning_rate": 4.982376350198976e-05,
      "loss": 3.2035,
      "step": 2172
    },
    {
      "epoch": 3.0264623955431755,
      "grad_norm": 4.240333080291748,
      "learning_rate": 4.981381466742467e-05,
      "loss": 2.4098,
      "step": 2173
    },
    {
      "epoch": 3.0278551532033426,
      "grad_norm": 3.621859550476074,
      "learning_rate": 4.9803865832859576e-05,
      "loss": 3.319,
      "step": 2174
    },
    {
      "epoch": 3.0292479108635098,
      "grad_norm": 3.9396166801452637,
      "learning_rate": 4.979391699829448e-05,
      "loss": 2.6429,
      "step": 2175
    },
    {
      "epoch": 3.030640668523677,
      "grad_norm": 4.1023077964782715,
      "learning_rate": 4.9783968163729384e-05,
      "loss": 2.5816,
      "step": 2176
    },
    {
      "epoch": 3.032033426183844,
      "grad_norm": 3.9348316192626953,
      "learning_rate": 4.9774019329164295e-05,
      "loss": 3.1643,
      "step": 2177
    },
    {
      "epoch": 3.033426183844011,
      "grad_norm": 6.0089592933654785,
      "learning_rate": 4.97640704945992e-05,
      "loss": 3.0826,
      "step": 2178
    },
    {
      "epoch": 3.034818941504178,
      "grad_norm": 3.588162899017334,
      "learning_rate": 4.975412166003411e-05,
      "loss": 2.942,
      "step": 2179
    },
    {
      "epoch": 3.0362116991643453,
      "grad_norm": 5.2281718254089355,
      "learning_rate": 4.974417282546901e-05,
      "loss": 2.4846,
      "step": 2180
    },
    {
      "epoch": 3.0376044568245124,
      "grad_norm": 3.4828572273254395,
      "learning_rate": 4.973422399090392e-05,
      "loss": 2.6165,
      "step": 2181
    },
    {
      "epoch": 3.0389972144846795,
      "grad_norm": 4.738771438598633,
      "learning_rate": 4.972427515633883e-05,
      "loss": 3.1826,
      "step": 2182
    },
    {
      "epoch": 3.0403899721448466,
      "grad_norm": 3.785372018814087,
      "learning_rate": 4.971432632177373e-05,
      "loss": 2.872,
      "step": 2183
    },
    {
      "epoch": 3.0417827298050137,
      "grad_norm": 3.934426784515381,
      "learning_rate": 4.970437748720863e-05,
      "loss": 3.3355,
      "step": 2184
    },
    {
      "epoch": 3.043175487465181,
      "grad_norm": 4.093682765960693,
      "learning_rate": 4.969442865264354e-05,
      "loss": 3.0227,
      "step": 2185
    },
    {
      "epoch": 3.0445682451253484,
      "grad_norm": 3.8305344581604004,
      "learning_rate": 4.968447981807845e-05,
      "loss": 2.8497,
      "step": 2186
    },
    {
      "epoch": 3.0459610027855155,
      "grad_norm": 3.768501043319702,
      "learning_rate": 4.9674530983513356e-05,
      "loss": 3.252,
      "step": 2187
    },
    {
      "epoch": 3.0473537604456826,
      "grad_norm": 3.5746545791625977,
      "learning_rate": 4.966458214894827e-05,
      "loss": 2.6726,
      "step": 2188
    },
    {
      "epoch": 3.0487465181058497,
      "grad_norm": 3.641993999481201,
      "learning_rate": 4.9654633314383164e-05,
      "loss": 2.8944,
      "step": 2189
    },
    {
      "epoch": 3.050139275766017,
      "grad_norm": 3.740527629852295,
      "learning_rate": 4.9644684479818075e-05,
      "loss": 2.858,
      "step": 2190
    },
    {
      "epoch": 3.051532033426184,
      "grad_norm": 3.6628808975219727,
      "learning_rate": 4.963473564525298e-05,
      "loss": 2.5151,
      "step": 2191
    },
    {
      "epoch": 3.052924791086351,
      "grad_norm": 3.681525707244873,
      "learning_rate": 4.962478681068789e-05,
      "loss": 2.8237,
      "step": 2192
    },
    {
      "epoch": 3.054317548746518,
      "grad_norm": 4.520201206207275,
      "learning_rate": 4.9614837976122794e-05,
      "loss": 2.6566,
      "step": 2193
    },
    {
      "epoch": 3.0557103064066853,
      "grad_norm": 4.108491897583008,
      "learning_rate": 4.96048891415577e-05,
      "loss": 3.2071,
      "step": 2194
    },
    {
      "epoch": 3.0571030640668524,
      "grad_norm": 4.114054203033447,
      "learning_rate": 4.95949403069926e-05,
      "loss": 2.6665,
      "step": 2195
    },
    {
      "epoch": 3.0584958217270195,
      "grad_norm": 4.240559101104736,
      "learning_rate": 4.958499147242751e-05,
      "loss": 2.9796,
      "step": 2196
    },
    {
      "epoch": 3.0598885793871866,
      "grad_norm": 3.4781413078308105,
      "learning_rate": 4.957504263786242e-05,
      "loss": 2.5954,
      "step": 2197
    },
    {
      "epoch": 3.0612813370473537,
      "grad_norm": 3.296037435531616,
      "learning_rate": 4.956509380329733e-05,
      "loss": 2.4479,
      "step": 2198
    },
    {
      "epoch": 3.062674094707521,
      "grad_norm": 4.22749662399292,
      "learning_rate": 4.9555144968732225e-05,
      "loss": 2.7733,
      "step": 2199
    },
    {
      "epoch": 3.064066852367688,
      "grad_norm": 5.45635461807251,
      "learning_rate": 4.9545196134167136e-05,
      "loss": 2.6946,
      "step": 2200
    },
    {
      "epoch": 3.065459610027855,
      "grad_norm": 3.8924145698547363,
      "learning_rate": 4.953524729960204e-05,
      "loss": 3.0797,
      "step": 2201
    },
    {
      "epoch": 3.066852367688022,
      "grad_norm": 3.5238921642303467,
      "learning_rate": 4.952529846503695e-05,
      "loss": 2.869,
      "step": 2202
    },
    {
      "epoch": 3.0682451253481893,
      "grad_norm": 4.308964729309082,
      "learning_rate": 4.9515349630471855e-05,
      "loss": 3.0236,
      "step": 2203
    },
    {
      "epoch": 3.0696378830083564,
      "grad_norm": 4.2634477615356445,
      "learning_rate": 4.950540079590676e-05,
      "loss": 2.9298,
      "step": 2204
    },
    {
      "epoch": 3.0710306406685235,
      "grad_norm": 3.6639902591705322,
      "learning_rate": 4.949545196134166e-05,
      "loss": 2.8863,
      "step": 2205
    },
    {
      "epoch": 3.0724233983286906,
      "grad_norm": 3.3267507553100586,
      "learning_rate": 4.9485503126776574e-05,
      "loss": 2.7208,
      "step": 2206
    },
    {
      "epoch": 3.073816155988858,
      "grad_norm": 4.104797840118408,
      "learning_rate": 4.947555429221148e-05,
      "loss": 3.1285,
      "step": 2207
    },
    {
      "epoch": 3.0752089136490253,
      "grad_norm": 3.7130515575408936,
      "learning_rate": 4.946560545764639e-05,
      "loss": 2.8338,
      "step": 2208
    },
    {
      "epoch": 3.0766016713091924,
      "grad_norm": 4.037384510040283,
      "learning_rate": 4.9455656623081286e-05,
      "loss": 2.7723,
      "step": 2209
    },
    {
      "epoch": 3.0779944289693595,
      "grad_norm": 3.7449655532836914,
      "learning_rate": 4.94457077885162e-05,
      "loss": 2.7966,
      "step": 2210
    },
    {
      "epoch": 3.0793871866295266,
      "grad_norm": 3.6848769187927246,
      "learning_rate": 4.943575895395111e-05,
      "loss": 2.9968,
      "step": 2211
    },
    {
      "epoch": 3.0807799442896937,
      "grad_norm": 3.8089780807495117,
      "learning_rate": 4.942581011938601e-05,
      "loss": 2.7478,
      "step": 2212
    },
    {
      "epoch": 3.082172701949861,
      "grad_norm": 5.400775909423828,
      "learning_rate": 4.941586128482092e-05,
      "loss": 2.8524,
      "step": 2213
    },
    {
      "epoch": 3.083565459610028,
      "grad_norm": 4.017539978027344,
      "learning_rate": 4.940591245025582e-05,
      "loss": 2.6039,
      "step": 2214
    },
    {
      "epoch": 3.084958217270195,
      "grad_norm": 3.5601894855499268,
      "learning_rate": 4.939596361569073e-05,
      "loss": 2.9378,
      "step": 2215
    },
    {
      "epoch": 3.086350974930362,
      "grad_norm": 3.6896731853485107,
      "learning_rate": 4.9386014781125635e-05,
      "loss": 2.8304,
      "step": 2216
    },
    {
      "epoch": 3.0877437325905293,
      "grad_norm": 4.51615047454834,
      "learning_rate": 4.9376065946560546e-05,
      "loss": 3.2497,
      "step": 2217
    },
    {
      "epoch": 3.0891364902506964,
      "grad_norm": 4.1750102043151855,
      "learning_rate": 4.936611711199545e-05,
      "loss": 3.1508,
      "step": 2218
    },
    {
      "epoch": 3.0905292479108635,
      "grad_norm": 4.524855136871338,
      "learning_rate": 4.9356168277430354e-05,
      "loss": 2.818,
      "step": 2219
    },
    {
      "epoch": 3.0919220055710306,
      "grad_norm": 4.195107936859131,
      "learning_rate": 4.934621944286526e-05,
      "loss": 3.273,
      "step": 2220
    },
    {
      "epoch": 3.0933147632311977,
      "grad_norm": 5.05075740814209,
      "learning_rate": 4.933627060830017e-05,
      "loss": 2.6328,
      "step": 2221
    },
    {
      "epoch": 3.094707520891365,
      "grad_norm": 3.962237596511841,
      "learning_rate": 4.932632177373507e-05,
      "loss": 2.7848,
      "step": 2222
    },
    {
      "epoch": 3.096100278551532,
      "grad_norm": 4.436080455780029,
      "learning_rate": 4.9316372939169984e-05,
      "loss": 2.9761,
      "step": 2223
    },
    {
      "epoch": 3.097493036211699,
      "grad_norm": 4.083108901977539,
      "learning_rate": 4.930642410460488e-05,
      "loss": 3.057,
      "step": 2224
    },
    {
      "epoch": 3.098885793871866,
      "grad_norm": 3.324781894683838,
      "learning_rate": 4.929647527003979e-05,
      "loss": 2.4656,
      "step": 2225
    },
    {
      "epoch": 3.1002785515320332,
      "grad_norm": 3.7265758514404297,
      "learning_rate": 4.9286526435474696e-05,
      "loss": 2.6025,
      "step": 2226
    },
    {
      "epoch": 3.1016713091922004,
      "grad_norm": 4.01738166809082,
      "learning_rate": 4.927657760090961e-05,
      "loss": 2.8352,
      "step": 2227
    },
    {
      "epoch": 3.103064066852368,
      "grad_norm": 4.062448978424072,
      "learning_rate": 4.9266628766344504e-05,
      "loss": 3.2861,
      "step": 2228
    },
    {
      "epoch": 3.104456824512535,
      "grad_norm": 3.786773443222046,
      "learning_rate": 4.9256679931779415e-05,
      "loss": 2.6291,
      "step": 2229
    },
    {
      "epoch": 3.105849582172702,
      "grad_norm": 4.09287166595459,
      "learning_rate": 4.924673109721432e-05,
      "loss": 2.8467,
      "step": 2230
    },
    {
      "epoch": 3.1072423398328692,
      "grad_norm": 4.1100311279296875,
      "learning_rate": 4.923678226264923e-05,
      "loss": 2.9087,
      "step": 2231
    },
    {
      "epoch": 3.1086350974930363,
      "grad_norm": 3.9571471214294434,
      "learning_rate": 4.9226833428084134e-05,
      "loss": 2.4647,
      "step": 2232
    },
    {
      "epoch": 3.1100278551532035,
      "grad_norm": 4.0034260749816895,
      "learning_rate": 4.921688459351904e-05,
      "loss": 2.7365,
      "step": 2233
    },
    {
      "epoch": 3.1114206128133706,
      "grad_norm": 3.7767553329467773,
      "learning_rate": 4.920693575895394e-05,
      "loss": 2.9574,
      "step": 2234
    },
    {
      "epoch": 3.1128133704735377,
      "grad_norm": 6.777519702911377,
      "learning_rate": 4.919698692438885e-05,
      "loss": 2.9007,
      "step": 2235
    },
    {
      "epoch": 3.114206128133705,
      "grad_norm": 3.5743494033813477,
      "learning_rate": 4.918703808982376e-05,
      "loss": 2.678,
      "step": 2236
    },
    {
      "epoch": 3.115598885793872,
      "grad_norm": 3.626544237136841,
      "learning_rate": 4.917708925525867e-05,
      "loss": 2.7042,
      "step": 2237
    },
    {
      "epoch": 3.116991643454039,
      "grad_norm": 3.875890016555786,
      "learning_rate": 4.9167140420693565e-05,
      "loss": 3.0001,
      "step": 2238
    },
    {
      "epoch": 3.118384401114206,
      "grad_norm": 3.7919387817382812,
      "learning_rate": 4.9157191586128476e-05,
      "loss": 2.9725,
      "step": 2239
    },
    {
      "epoch": 3.1197771587743732,
      "grad_norm": 3.46844744682312,
      "learning_rate": 4.914724275156339e-05,
      "loss": 2.8531,
      "step": 2240
    },
    {
      "epoch": 3.1211699164345403,
      "grad_norm": 4.005001068115234,
      "learning_rate": 4.913729391699829e-05,
      "loss": 2.8883,
      "step": 2241
    },
    {
      "epoch": 3.1225626740947074,
      "grad_norm": 4.705630779266357,
      "learning_rate": 4.91273450824332e-05,
      "loss": 3.0957,
      "step": 2242
    },
    {
      "epoch": 3.1239554317548746,
      "grad_norm": 3.739565372467041,
      "learning_rate": 4.91173962478681e-05,
      "loss": 2.4562,
      "step": 2243
    },
    {
      "epoch": 3.1253481894150417,
      "grad_norm": 3.5640506744384766,
      "learning_rate": 4.910744741330301e-05,
      "loss": 3.0103,
      "step": 2244
    },
    {
      "epoch": 3.1267409470752088,
      "grad_norm": 3.751476764678955,
      "learning_rate": 4.9097498578737914e-05,
      "loss": 2.7388,
      "step": 2245
    },
    {
      "epoch": 3.128133704735376,
      "grad_norm": 3.8549933433532715,
      "learning_rate": 4.9087549744172825e-05,
      "loss": 2.7139,
      "step": 2246
    },
    {
      "epoch": 3.129526462395543,
      "grad_norm": 4.479499340057373,
      "learning_rate": 4.907760090960773e-05,
      "loss": 2.7457,
      "step": 2247
    },
    {
      "epoch": 3.13091922005571,
      "grad_norm": 3.822279691696167,
      "learning_rate": 4.906765207504263e-05,
      "loss": 2.8294,
      "step": 2248
    },
    {
      "epoch": 3.1323119777158777,
      "grad_norm": 3.825222969055176,
      "learning_rate": 4.905770324047754e-05,
      "loss": 2.8442,
      "step": 2249
    },
    {
      "epoch": 3.1337047353760448,
      "grad_norm": 4.0725531578063965,
      "learning_rate": 4.904775440591245e-05,
      "loss": 2.6,
      "step": 2250
    },
    {
      "epoch": 3.135097493036212,
      "grad_norm": 3.854177236557007,
      "learning_rate": 4.903780557134735e-05,
      "loss": 2.6668,
      "step": 2251
    },
    {
      "epoch": 3.136490250696379,
      "grad_norm": 3.6467463970184326,
      "learning_rate": 4.902785673678226e-05,
      "loss": 2.7115,
      "step": 2252
    },
    {
      "epoch": 3.137883008356546,
      "grad_norm": 3.687175989151001,
      "learning_rate": 4.901790790221716e-05,
      "loss": 2.8751,
      "step": 2253
    },
    {
      "epoch": 3.139275766016713,
      "grad_norm": 3.6928493976593018,
      "learning_rate": 4.900795906765207e-05,
      "loss": 2.992,
      "step": 2254
    },
    {
      "epoch": 3.1406685236768803,
      "grad_norm": 4.653807163238525,
      "learning_rate": 4.8998010233086975e-05,
      "loss": 2.9187,
      "step": 2255
    },
    {
      "epoch": 3.1420612813370474,
      "grad_norm": 3.7770557403564453,
      "learning_rate": 4.8988061398521886e-05,
      "loss": 3.0289,
      "step": 2256
    },
    {
      "epoch": 3.1434540389972145,
      "grad_norm": 3.6280288696289062,
      "learning_rate": 4.897811256395679e-05,
      "loss": 2.6832,
      "step": 2257
    },
    {
      "epoch": 3.1448467966573816,
      "grad_norm": 4.064112186431885,
      "learning_rate": 4.8968163729391694e-05,
      "loss": 3.05,
      "step": 2258
    },
    {
      "epoch": 3.1462395543175488,
      "grad_norm": 3.938138961791992,
      "learning_rate": 4.89582148948266e-05,
      "loss": 2.9412,
      "step": 2259
    },
    {
      "epoch": 3.147632311977716,
      "grad_norm": 3.529313802719116,
      "learning_rate": 4.894826606026151e-05,
      "loss": 2.9496,
      "step": 2260
    },
    {
      "epoch": 3.149025069637883,
      "grad_norm": 3.544926643371582,
      "learning_rate": 4.893831722569641e-05,
      "loss": 2.7199,
      "step": 2261
    },
    {
      "epoch": 3.15041782729805,
      "grad_norm": 3.8904011249542236,
      "learning_rate": 4.8928368391131324e-05,
      "loss": 3.0123,
      "step": 2262
    },
    {
      "epoch": 3.151810584958217,
      "grad_norm": 3.769857883453369,
      "learning_rate": 4.891841955656622e-05,
      "loss": 2.856,
      "step": 2263
    },
    {
      "epoch": 3.1532033426183843,
      "grad_norm": 3.689384698867798,
      "learning_rate": 4.890847072200113e-05,
      "loss": 2.8636,
      "step": 2264
    },
    {
      "epoch": 3.1545961002785514,
      "grad_norm": 3.706634044647217,
      "learning_rate": 4.8898521887436036e-05,
      "loss": 2.7097,
      "step": 2265
    },
    {
      "epoch": 3.1559888579387185,
      "grad_norm": 3.793800115585327,
      "learning_rate": 4.888857305287095e-05,
      "loss": 3.2398,
      "step": 2266
    },
    {
      "epoch": 3.1573816155988856,
      "grad_norm": 3.964437246322632,
      "learning_rate": 4.887862421830586e-05,
      "loss": 2.64,
      "step": 2267
    },
    {
      "epoch": 3.1587743732590527,
      "grad_norm": 3.7391197681427,
      "learning_rate": 4.8868675383740755e-05,
      "loss": 2.8607,
      "step": 2268
    },
    {
      "epoch": 3.16016713091922,
      "grad_norm": 4.067811012268066,
      "learning_rate": 4.8858726549175666e-05,
      "loss": 2.6981,
      "step": 2269
    },
    {
      "epoch": 3.1615598885793874,
      "grad_norm": 4.685571670532227,
      "learning_rate": 4.884877771461057e-05,
      "loss": 2.9726,
      "step": 2270
    },
    {
      "epoch": 3.1629526462395545,
      "grad_norm": 4.283031463623047,
      "learning_rate": 4.883882888004548e-05,
      "loss": 3.1427,
      "step": 2271
    },
    {
      "epoch": 3.1643454038997216,
      "grad_norm": 4.66605281829834,
      "learning_rate": 4.882888004548038e-05,
      "loss": 2.9865,
      "step": 2272
    },
    {
      "epoch": 3.1657381615598887,
      "grad_norm": 4.3901166915893555,
      "learning_rate": 4.881893121091529e-05,
      "loss": 3.0573,
      "step": 2273
    },
    {
      "epoch": 3.167130919220056,
      "grad_norm": 3.78541898727417,
      "learning_rate": 4.880898237635019e-05,
      "loss": 2.9784,
      "step": 2274
    },
    {
      "epoch": 3.168523676880223,
      "grad_norm": 3.5710551738739014,
      "learning_rate": 4.8799033541785104e-05,
      "loss": 2.7992,
      "step": 2275
    },
    {
      "epoch": 3.16991643454039,
      "grad_norm": 3.791133165359497,
      "learning_rate": 4.878908470722001e-05,
      "loss": 2.9119,
      "step": 2276
    },
    {
      "epoch": 3.171309192200557,
      "grad_norm": 3.72632098197937,
      "learning_rate": 4.877913587265491e-05,
      "loss": 3.1021,
      "step": 2277
    },
    {
      "epoch": 3.1727019498607243,
      "grad_norm": 3.671421766281128,
      "learning_rate": 4.8769187038089816e-05,
      "loss": 2.7327,
      "step": 2278
    },
    {
      "epoch": 3.1740947075208914,
      "grad_norm": 3.593618869781494,
      "learning_rate": 4.875923820352473e-05,
      "loss": 2.7006,
      "step": 2279
    },
    {
      "epoch": 3.1754874651810585,
      "grad_norm": 4.296664237976074,
      "learning_rate": 4.874928936895963e-05,
      "loss": 2.6943,
      "step": 2280
    },
    {
      "epoch": 3.1768802228412256,
      "grad_norm": 3.8954293727874756,
      "learning_rate": 4.873934053439454e-05,
      "loss": 2.8782,
      "step": 2281
    },
    {
      "epoch": 3.1782729805013927,
      "grad_norm": 6.061985969543457,
      "learning_rate": 4.872939169982944e-05,
      "loss": 2.8432,
      "step": 2282
    },
    {
      "epoch": 3.17966573816156,
      "grad_norm": 4.803857326507568,
      "learning_rate": 4.871944286526435e-05,
      "loss": 3.7857,
      "step": 2283
    },
    {
      "epoch": 3.181058495821727,
      "grad_norm": 4.052546501159668,
      "learning_rate": 4.8709494030699254e-05,
      "loss": 2.9245,
      "step": 2284
    },
    {
      "epoch": 3.182451253481894,
      "grad_norm": 6.619697570800781,
      "learning_rate": 4.8699545196134165e-05,
      "loss": 3.0243,
      "step": 2285
    },
    {
      "epoch": 3.183844011142061,
      "grad_norm": 4.285874843597412,
      "learning_rate": 4.868959636156907e-05,
      "loss": 2.9593,
      "step": 2286
    },
    {
      "epoch": 3.1852367688022283,
      "grad_norm": 3.6087124347686768,
      "learning_rate": 4.867964752700397e-05,
      "loss": 2.9436,
      "step": 2287
    },
    {
      "epoch": 3.1866295264623954,
      "grad_norm": 3.975787878036499,
      "learning_rate": 4.866969869243888e-05,
      "loss": 2.801,
      "step": 2288
    },
    {
      "epoch": 3.1880222841225625,
      "grad_norm": 4.284610271453857,
      "learning_rate": 4.865974985787379e-05,
      "loss": 2.9373,
      "step": 2289
    },
    {
      "epoch": 3.1894150417827296,
      "grad_norm": 4.059457778930664,
      "learning_rate": 4.864980102330869e-05,
      "loss": 2.5689,
      "step": 2290
    },
    {
      "epoch": 3.190807799442897,
      "grad_norm": 3.7924251556396484,
      "learning_rate": 4.86398521887436e-05,
      "loss": 2.8105,
      "step": 2291
    },
    {
      "epoch": 3.1922005571030643,
      "grad_norm": 4.193859577178955,
      "learning_rate": 4.86299033541785e-05,
      "loss": 2.7084,
      "step": 2292
    },
    {
      "epoch": 3.1935933147632314,
      "grad_norm": 3.9864838123321533,
      "learning_rate": 4.861995451961341e-05,
      "loss": 2.9286,
      "step": 2293
    },
    {
      "epoch": 3.1949860724233985,
      "grad_norm": 4.141392707824707,
      "learning_rate": 4.861000568504832e-05,
      "loss": 2.7396,
      "step": 2294
    },
    {
      "epoch": 3.1963788300835656,
      "grad_norm": 4.376644134521484,
      "learning_rate": 4.8600056850483226e-05,
      "loss": 3.1385,
      "step": 2295
    },
    {
      "epoch": 3.1977715877437327,
      "grad_norm": 3.9887776374816895,
      "learning_rate": 4.859010801591814e-05,
      "loss": 2.8945,
      "step": 2296
    },
    {
      "epoch": 3.1991643454039,
      "grad_norm": 4.6822004318237305,
      "learning_rate": 4.8580159181353034e-05,
      "loss": 3.0602,
      "step": 2297
    },
    {
      "epoch": 3.200557103064067,
      "grad_norm": 4.151376247406006,
      "learning_rate": 4.8570210346787945e-05,
      "loss": 2.9018,
      "step": 2298
    },
    {
      "epoch": 3.201949860724234,
      "grad_norm": 4.49393892288208,
      "learning_rate": 4.856026151222285e-05,
      "loss": 3.2492,
      "step": 2299
    },
    {
      "epoch": 3.203342618384401,
      "grad_norm": 3.828725814819336,
      "learning_rate": 4.855031267765776e-05,
      "loss": 2.7855,
      "step": 2300
    },
    {
      "epoch": 3.2047353760445683,
      "grad_norm": 3.9831230640411377,
      "learning_rate": 4.8540363843092664e-05,
      "loss": 3.0632,
      "step": 2301
    },
    {
      "epoch": 3.2061281337047354,
      "grad_norm": 4.14377498626709,
      "learning_rate": 4.853041500852757e-05,
      "loss": 3.1884,
      "step": 2302
    },
    {
      "epoch": 3.2075208913649025,
      "grad_norm": 3.670644521713257,
      "learning_rate": 4.852046617396247e-05,
      "loss": 3.0383,
      "step": 2303
    },
    {
      "epoch": 3.2089136490250696,
      "grad_norm": 3.681523323059082,
      "learning_rate": 4.851051733939738e-05,
      "loss": 2.7217,
      "step": 2304
    },
    {
      "epoch": 3.2103064066852367,
      "grad_norm": 3.5931689739227295,
      "learning_rate": 4.850056850483229e-05,
      "loss": 2.8416,
      "step": 2305
    },
    {
      "epoch": 3.211699164345404,
      "grad_norm": 4.240254878997803,
      "learning_rate": 4.84906196702672e-05,
      "loss": 2.602,
      "step": 2306
    },
    {
      "epoch": 3.213091922005571,
      "grad_norm": 4.246129035949707,
      "learning_rate": 4.8480670835702095e-05,
      "loss": 3.0062,
      "step": 2307
    },
    {
      "epoch": 3.214484679665738,
      "grad_norm": 4.037646770477295,
      "learning_rate": 4.8470722001137006e-05,
      "loss": 2.8875,
      "step": 2308
    },
    {
      "epoch": 3.215877437325905,
      "grad_norm": 4.305235385894775,
      "learning_rate": 4.846077316657191e-05,
      "loss": 3.1422,
      "step": 2309
    },
    {
      "epoch": 3.2172701949860723,
      "grad_norm": 3.8170394897460938,
      "learning_rate": 4.845082433200682e-05,
      "loss": 2.6108,
      "step": 2310
    },
    {
      "epoch": 3.2186629526462394,
      "grad_norm": 4.461042404174805,
      "learning_rate": 4.844087549744172e-05,
      "loss": 2.9229,
      "step": 2311
    },
    {
      "epoch": 3.220055710306407,
      "grad_norm": 6.392706871032715,
      "learning_rate": 4.843092666287663e-05,
      "loss": 3.0824,
      "step": 2312
    },
    {
      "epoch": 3.2214484679665736,
      "grad_norm": 5.156744956970215,
      "learning_rate": 4.8420977828311533e-05,
      "loss": 2.6974,
      "step": 2313
    },
    {
      "epoch": 3.222841225626741,
      "grad_norm": 4.121723651885986,
      "learning_rate": 4.8411028993746444e-05,
      "loss": 2.751,
      "step": 2314
    },
    {
      "epoch": 3.2242339832869082,
      "grad_norm": 4.308473110198975,
      "learning_rate": 4.840108015918135e-05,
      "loss": 2.9132,
      "step": 2315
    },
    {
      "epoch": 3.2256267409470754,
      "grad_norm": 4.490592956542969,
      "learning_rate": 4.839113132461625e-05,
      "loss": 3.0115,
      "step": 2316
    },
    {
      "epoch": 3.2270194986072425,
      "grad_norm": 4.198948860168457,
      "learning_rate": 4.8381182490051157e-05,
      "loss": 3.1584,
      "step": 2317
    },
    {
      "epoch": 3.2284122562674096,
      "grad_norm": 4.588038921356201,
      "learning_rate": 4.837123365548607e-05,
      "loss": 3.1666,
      "step": 2318
    },
    {
      "epoch": 3.2298050139275767,
      "grad_norm": 4.044364929199219,
      "learning_rate": 4.836128482092097e-05,
      "loss": 2.7106,
      "step": 2319
    },
    {
      "epoch": 3.231197771587744,
      "grad_norm": 3.7738356590270996,
      "learning_rate": 4.835133598635588e-05,
      "loss": 2.654,
      "step": 2320
    },
    {
      "epoch": 3.232590529247911,
      "grad_norm": 4.125843048095703,
      "learning_rate": 4.834138715179078e-05,
      "loss": 3.1454,
      "step": 2321
    },
    {
      "epoch": 3.233983286908078,
      "grad_norm": 3.8941538333892822,
      "learning_rate": 4.833143831722569e-05,
      "loss": 2.7563,
      "step": 2322
    },
    {
      "epoch": 3.235376044568245,
      "grad_norm": 3.7710914611816406,
      "learning_rate": 4.83214894826606e-05,
      "loss": 2.5314,
      "step": 2323
    },
    {
      "epoch": 3.2367688022284122,
      "grad_norm": 3.6377623081207275,
      "learning_rate": 4.8311540648095505e-05,
      "loss": 3.0108,
      "step": 2324
    },
    {
      "epoch": 3.2381615598885793,
      "grad_norm": 4.007593631744385,
      "learning_rate": 4.8301591813530416e-05,
      "loss": 2.9897,
      "step": 2325
    },
    {
      "epoch": 3.2395543175487465,
      "grad_norm": 3.7874231338500977,
      "learning_rate": 4.8291642978965314e-05,
      "loss": 2.3803,
      "step": 2326
    },
    {
      "epoch": 3.2409470752089136,
      "grad_norm": 4.07582426071167,
      "learning_rate": 4.8281694144400224e-05,
      "loss": 3.0804,
      "step": 2327
    },
    {
      "epoch": 3.2423398328690807,
      "grad_norm": 4.140418529510498,
      "learning_rate": 4.827174530983513e-05,
      "loss": 2.8582,
      "step": 2328
    },
    {
      "epoch": 3.243732590529248,
      "grad_norm": 3.6024129390716553,
      "learning_rate": 4.826179647527004e-05,
      "loss": 2.9104,
      "step": 2329
    },
    {
      "epoch": 3.245125348189415,
      "grad_norm": 5.591879367828369,
      "learning_rate": 4.8251847640704943e-05,
      "loss": 2.6721,
      "step": 2330
    },
    {
      "epoch": 3.246518105849582,
      "grad_norm": 4.41704797744751,
      "learning_rate": 4.824189880613985e-05,
      "loss": 2.9928,
      "step": 2331
    },
    {
      "epoch": 3.247910863509749,
      "grad_norm": 4.577050685882568,
      "learning_rate": 4.823194997157475e-05,
      "loss": 2.8255,
      "step": 2332
    },
    {
      "epoch": 3.2493036211699167,
      "grad_norm": 4.430422782897949,
      "learning_rate": 4.822200113700966e-05,
      "loss": 3.1678,
      "step": 2333
    },
    {
      "epoch": 3.2506963788300833,
      "grad_norm": 5.460785865783691,
      "learning_rate": 4.8212052302444566e-05,
      "loss": 2.6286,
      "step": 2334
    },
    {
      "epoch": 3.252089136490251,
      "grad_norm": 5.823190212249756,
      "learning_rate": 4.820210346787948e-05,
      "loss": 2.568,
      "step": 2335
    },
    {
      "epoch": 3.253481894150418,
      "grad_norm": 3.726700782775879,
      "learning_rate": 4.8192154633314375e-05,
      "loss": 3.0072,
      "step": 2336
    },
    {
      "epoch": 3.254874651810585,
      "grad_norm": 3.914973735809326,
      "learning_rate": 4.8182205798749285e-05,
      "loss": 2.8886,
      "step": 2337
    },
    {
      "epoch": 3.256267409470752,
      "grad_norm": 4.194888114929199,
      "learning_rate": 4.817225696418419e-05,
      "loss": 2.8432,
      "step": 2338
    },
    {
      "epoch": 3.2576601671309193,
      "grad_norm": 3.9118857383728027,
      "learning_rate": 4.81623081296191e-05,
      "loss": 2.8313,
      "step": 2339
    },
    {
      "epoch": 3.2590529247910864,
      "grad_norm": 3.5121490955352783,
      "learning_rate": 4.8152359295054004e-05,
      "loss": 2.5016,
      "step": 2340
    },
    {
      "epoch": 3.2604456824512535,
      "grad_norm": 4.009439468383789,
      "learning_rate": 4.814241046048891e-05,
      "loss": 2.7574,
      "step": 2341
    },
    {
      "epoch": 3.2618384401114207,
      "grad_norm": 7.137579917907715,
      "learning_rate": 4.813246162592381e-05,
      "loss": 3.2454,
      "step": 2342
    },
    {
      "epoch": 3.2632311977715878,
      "grad_norm": 3.6958656311035156,
      "learning_rate": 4.8122512791358723e-05,
      "loss": 2.8281,
      "step": 2343
    },
    {
      "epoch": 3.264623955431755,
      "grad_norm": 3.7219202518463135,
      "learning_rate": 4.811256395679363e-05,
      "loss": 2.6826,
      "step": 2344
    },
    {
      "epoch": 3.266016713091922,
      "grad_norm": 3.782970905303955,
      "learning_rate": 4.810261512222854e-05,
      "loss": 2.6493,
      "step": 2345
    },
    {
      "epoch": 3.267409470752089,
      "grad_norm": 3.8960306644439697,
      "learning_rate": 4.8092666287663436e-05,
      "loss": 3.3011,
      "step": 2346
    },
    {
      "epoch": 3.268802228412256,
      "grad_norm": 4.1278252601623535,
      "learning_rate": 4.8082717453098347e-05,
      "loss": 3.2684,
      "step": 2347
    },
    {
      "epoch": 3.2701949860724233,
      "grad_norm": 3.858278512954712,
      "learning_rate": 4.807276861853325e-05,
      "loss": 2.7097,
      "step": 2348
    },
    {
      "epoch": 3.2715877437325904,
      "grad_norm": 4.56193208694458,
      "learning_rate": 4.806281978396816e-05,
      "loss": 2.8349,
      "step": 2349
    },
    {
      "epoch": 3.2729805013927575,
      "grad_norm": 4.361737251281738,
      "learning_rate": 4.805287094940307e-05,
      "loss": 2.8348,
      "step": 2350
    },
    {
      "epoch": 3.2743732590529246,
      "grad_norm": 3.825388193130493,
      "learning_rate": 4.804292211483797e-05,
      "loss": 2.6212,
      "step": 2351
    },
    {
      "epoch": 3.2757660167130918,
      "grad_norm": 4.141374588012695,
      "learning_rate": 4.803297328027288e-05,
      "loss": 2.9403,
      "step": 2352
    },
    {
      "epoch": 3.277158774373259,
      "grad_norm": 3.733556032180786,
      "learning_rate": 4.8023024445707785e-05,
      "loss": 2.7187,
      "step": 2353
    },
    {
      "epoch": 3.2785515320334264,
      "grad_norm": 4.972504138946533,
      "learning_rate": 4.8013075611142695e-05,
      "loss": 2.8249,
      "step": 2354
    },
    {
      "epoch": 3.279944289693593,
      "grad_norm": 4.3905839920043945,
      "learning_rate": 4.800312677657759e-05,
      "loss": 2.6179,
      "step": 2355
    },
    {
      "epoch": 3.2813370473537606,
      "grad_norm": 4.249562740325928,
      "learning_rate": 4.7993177942012504e-05,
      "loss": 2.8928,
      "step": 2356
    },
    {
      "epoch": 3.2827298050139277,
      "grad_norm": 4.65199613571167,
      "learning_rate": 4.798322910744741e-05,
      "loss": 2.766,
      "step": 2357
    },
    {
      "epoch": 3.284122562674095,
      "grad_norm": 3.633455514907837,
      "learning_rate": 4.797328027288232e-05,
      "loss": 2.6158,
      "step": 2358
    },
    {
      "epoch": 3.285515320334262,
      "grad_norm": 3.461538791656494,
      "learning_rate": 4.796333143831722e-05,
      "loss": 2.3918,
      "step": 2359
    },
    {
      "epoch": 3.286908077994429,
      "grad_norm": 3.6453847885131836,
      "learning_rate": 4.795338260375213e-05,
      "loss": 2.8247,
      "step": 2360
    },
    {
      "epoch": 3.288300835654596,
      "grad_norm": 4.0668158531188965,
      "learning_rate": 4.794343376918703e-05,
      "loss": 2.8784,
      "step": 2361
    },
    {
      "epoch": 3.2896935933147633,
      "grad_norm": 3.821699857711792,
      "learning_rate": 4.793348493462194e-05,
      "loss": 2.9337,
      "step": 2362
    },
    {
      "epoch": 3.2910863509749304,
      "grad_norm": 4.014733791351318,
      "learning_rate": 4.7923536100056846e-05,
      "loss": 3.0672,
      "step": 2363
    },
    {
      "epoch": 3.2924791086350975,
      "grad_norm": 3.88468599319458,
      "learning_rate": 4.7913587265491757e-05,
      "loss": 2.5454,
      "step": 2364
    },
    {
      "epoch": 3.2938718662952646,
      "grad_norm": 3.78873872756958,
      "learning_rate": 4.7903638430926654e-05,
      "loss": 2.8817,
      "step": 2365
    },
    {
      "epoch": 3.2952646239554317,
      "grad_norm": 4.950428485870361,
      "learning_rate": 4.7893689596361565e-05,
      "loss": 2.8254,
      "step": 2366
    },
    {
      "epoch": 3.296657381615599,
      "grad_norm": 3.688955545425415,
      "learning_rate": 4.788374076179647e-05,
      "loss": 2.8069,
      "step": 2367
    },
    {
      "epoch": 3.298050139275766,
      "grad_norm": 4.237142562866211,
      "learning_rate": 4.787379192723138e-05,
      "loss": 3.0093,
      "step": 2368
    },
    {
      "epoch": 3.299442896935933,
      "grad_norm": 3.334515333175659,
      "learning_rate": 4.7863843092666284e-05,
      "loss": 2.3511,
      "step": 2369
    },
    {
      "epoch": 3.3008356545961,
      "grad_norm": 3.505284309387207,
      "learning_rate": 4.785389425810119e-05,
      "loss": 2.2776,
      "step": 2370
    },
    {
      "epoch": 3.3022284122562673,
      "grad_norm": 4.798534870147705,
      "learning_rate": 4.784394542353609e-05,
      "loss": 3.3179,
      "step": 2371
    },
    {
      "epoch": 3.3036211699164344,
      "grad_norm": 4.9942240715026855,
      "learning_rate": 4.7833996588971e-05,
      "loss": 2.6916,
      "step": 2372
    },
    {
      "epoch": 3.3050139275766015,
      "grad_norm": 3.5696606636047363,
      "learning_rate": 4.782404775440591e-05,
      "loss": 2.7702,
      "step": 2373
    },
    {
      "epoch": 3.3064066852367686,
      "grad_norm": 3.6908276081085205,
      "learning_rate": 4.781409891984082e-05,
      "loss": 2.6051,
      "step": 2374
    },
    {
      "epoch": 3.307799442896936,
      "grad_norm": 4.1333394050598145,
      "learning_rate": 4.7804150085275715e-05,
      "loss": 2.917,
      "step": 2375
    },
    {
      "epoch": 3.309192200557103,
      "grad_norm": 4.748862266540527,
      "learning_rate": 4.7794201250710626e-05,
      "loss": 2.6613,
      "step": 2376
    },
    {
      "epoch": 3.3105849582172704,
      "grad_norm": 4.2038254737854,
      "learning_rate": 4.778425241614553e-05,
      "loss": 2.8604,
      "step": 2377
    },
    {
      "epoch": 3.3119777158774375,
      "grad_norm": 4.11339807510376,
      "learning_rate": 4.777430358158044e-05,
      "loss": 2.9838,
      "step": 2378
    },
    {
      "epoch": 3.3133704735376046,
      "grad_norm": 3.785566806793213,
      "learning_rate": 4.776435474701535e-05,
      "loss": 2.4237,
      "step": 2379
    },
    {
      "epoch": 3.3147632311977717,
      "grad_norm": 3.6763899326324463,
      "learning_rate": 4.775440591245025e-05,
      "loss": 2.7251,
      "step": 2380
    },
    {
      "epoch": 3.316155988857939,
      "grad_norm": 3.59271502494812,
      "learning_rate": 4.774445707788516e-05,
      "loss": 2.3965,
      "step": 2381
    },
    {
      "epoch": 3.317548746518106,
      "grad_norm": 4.120853424072266,
      "learning_rate": 4.7734508243320064e-05,
      "loss": 2.7869,
      "step": 2382
    },
    {
      "epoch": 3.318941504178273,
      "grad_norm": 3.6389341354370117,
      "learning_rate": 4.7724559408754975e-05,
      "loss": 2.7214,
      "step": 2383
    },
    {
      "epoch": 3.32033426183844,
      "grad_norm": 10.967327117919922,
      "learning_rate": 4.771461057418988e-05,
      "loss": 2.7669,
      "step": 2384
    },
    {
      "epoch": 3.3217270194986073,
      "grad_norm": 3.867356061935425,
      "learning_rate": 4.770466173962478e-05,
      "loss": 2.6242,
      "step": 2385
    },
    {
      "epoch": 3.3231197771587744,
      "grad_norm": 3.5651192665100098,
      "learning_rate": 4.769471290505969e-05,
      "loss": 2.4967,
      "step": 2386
    },
    {
      "epoch": 3.3245125348189415,
      "grad_norm": 3.5389902591705322,
      "learning_rate": 4.76847640704946e-05,
      "loss": 2.7493,
      "step": 2387
    },
    {
      "epoch": 3.3259052924791086,
      "grad_norm": 6.526866912841797,
      "learning_rate": 4.76748152359295e-05,
      "loss": 2.8418,
      "step": 2388
    },
    {
      "epoch": 3.3272980501392757,
      "grad_norm": 4.658627510070801,
      "learning_rate": 4.766486640136441e-05,
      "loss": 2.858,
      "step": 2389
    },
    {
      "epoch": 3.328690807799443,
      "grad_norm": 3.680142641067505,
      "learning_rate": 4.765491756679931e-05,
      "loss": 2.7746,
      "step": 2390
    },
    {
      "epoch": 3.33008356545961,
      "grad_norm": 4.72170877456665,
      "learning_rate": 4.764496873223422e-05,
      "loss": 2.8055,
      "step": 2391
    },
    {
      "epoch": 3.331476323119777,
      "grad_norm": 4.868217468261719,
      "learning_rate": 4.7635019897669125e-05,
      "loss": 2.6525,
      "step": 2392
    },
    {
      "epoch": 3.332869080779944,
      "grad_norm": 4.732217311859131,
      "learning_rate": 4.7625071063104036e-05,
      "loss": 2.8297,
      "step": 2393
    },
    {
      "epoch": 3.3342618384401113,
      "grad_norm": 3.6755013465881348,
      "learning_rate": 4.761512222853894e-05,
      "loss": 2.8107,
      "step": 2394
    },
    {
      "epoch": 3.3356545961002784,
      "grad_norm": 4.290006160736084,
      "learning_rate": 4.7605173393973844e-05,
      "loss": 2.7585,
      "step": 2395
    },
    {
      "epoch": 3.337047353760446,
      "grad_norm": 3.6859662532806396,
      "learning_rate": 4.759522455940875e-05,
      "loss": 2.469,
      "step": 2396
    },
    {
      "epoch": 3.3384401114206126,
      "grad_norm": 3.754054069519043,
      "learning_rate": 4.758527572484366e-05,
      "loss": 2.9326,
      "step": 2397
    },
    {
      "epoch": 3.33983286908078,
      "grad_norm": 4.631336688995361,
      "learning_rate": 4.757532689027856e-05,
      "loss": 3.0036,
      "step": 2398
    },
    {
      "epoch": 3.3412256267409473,
      "grad_norm": 3.606992244720459,
      "learning_rate": 4.756537805571347e-05,
      "loss": 2.7054,
      "step": 2399
    },
    {
      "epoch": 3.3426183844011144,
      "grad_norm": 3.855879545211792,
      "learning_rate": 4.755542922114837e-05,
      "loss": 2.8321,
      "step": 2400
    },
    {
      "epoch": 3.3440111420612815,
      "grad_norm": 3.460355520248413,
      "learning_rate": 4.754548038658328e-05,
      "loss": 2.5688,
      "step": 2401
    },
    {
      "epoch": 3.3454038997214486,
      "grad_norm": 4.121874809265137,
      "learning_rate": 4.7535531552018186e-05,
      "loss": 2.8504,
      "step": 2402
    },
    {
      "epoch": 3.3467966573816157,
      "grad_norm": 4.156680583953857,
      "learning_rate": 4.75255827174531e-05,
      "loss": 2.8232,
      "step": 2403
    },
    {
      "epoch": 3.348189415041783,
      "grad_norm": 3.8161020278930664,
      "learning_rate": 4.7515633882887994e-05,
      "loss": 2.6313,
      "step": 2404
    },
    {
      "epoch": 3.34958217270195,
      "grad_norm": 4.528938293457031,
      "learning_rate": 4.7505685048322905e-05,
      "loss": 2.9876,
      "step": 2405
    },
    {
      "epoch": 3.350974930362117,
      "grad_norm": 3.8600189685821533,
      "learning_rate": 4.749573621375781e-05,
      "loss": 2.6049,
      "step": 2406
    },
    {
      "epoch": 3.352367688022284,
      "grad_norm": 5.053176403045654,
      "learning_rate": 4.748578737919272e-05,
      "loss": 3.0565,
      "step": 2407
    },
    {
      "epoch": 3.3537604456824512,
      "grad_norm": 4.079180717468262,
      "learning_rate": 4.747583854462763e-05,
      "loss": 2.7893,
      "step": 2408
    },
    {
      "epoch": 3.3551532033426184,
      "grad_norm": 3.8171560764312744,
      "learning_rate": 4.746588971006253e-05,
      "loss": 2.7338,
      "step": 2409
    },
    {
      "epoch": 3.3565459610027855,
      "grad_norm": 3.907892942428589,
      "learning_rate": 4.745594087549744e-05,
      "loss": 2.8174,
      "step": 2410
    },
    {
      "epoch": 3.3579387186629526,
      "grad_norm": 7.531633377075195,
      "learning_rate": 4.744599204093234e-05,
      "loss": 3.2603,
      "step": 2411
    },
    {
      "epoch": 3.3593314763231197,
      "grad_norm": 4.269933223724365,
      "learning_rate": 4.7436043206367254e-05,
      "loss": 2.9665,
      "step": 2412
    },
    {
      "epoch": 3.360724233983287,
      "grad_norm": 3.293668508529663,
      "learning_rate": 4.742609437180216e-05,
      "loss": 2.3873,
      "step": 2413
    },
    {
      "epoch": 3.362116991643454,
      "grad_norm": 4.437375545501709,
      "learning_rate": 4.741614553723706e-05,
      "loss": 2.9374,
      "step": 2414
    },
    {
      "epoch": 3.363509749303621,
      "grad_norm": 3.826282024383545,
      "learning_rate": 4.7406196702671966e-05,
      "loss": 2.9442,
      "step": 2415
    },
    {
      "epoch": 3.364902506963788,
      "grad_norm": 3.6247544288635254,
      "learning_rate": 4.739624786810688e-05,
      "loss": 2.9575,
      "step": 2416
    },
    {
      "epoch": 3.3662952646239557,
      "grad_norm": 3.819434404373169,
      "learning_rate": 4.738629903354178e-05,
      "loss": 2.6576,
      "step": 2417
    },
    {
      "epoch": 3.3676880222841223,
      "grad_norm": 4.170529365539551,
      "learning_rate": 4.737635019897669e-05,
      "loss": 2.7199,
      "step": 2418
    },
    {
      "epoch": 3.36908077994429,
      "grad_norm": 3.944037914276123,
      "learning_rate": 4.736640136441159e-05,
      "loss": 2.5178,
      "step": 2419
    },
    {
      "epoch": 3.370473537604457,
      "grad_norm": 4.178440570831299,
      "learning_rate": 4.73564525298465e-05,
      "loss": 2.5889,
      "step": 2420
    },
    {
      "epoch": 3.371866295264624,
      "grad_norm": 4.817466735839844,
      "learning_rate": 4.7346503695281404e-05,
      "loss": 2.7761,
      "step": 2421
    },
    {
      "epoch": 3.3732590529247912,
      "grad_norm": 4.657197952270508,
      "learning_rate": 4.7336554860716315e-05,
      "loss": 2.6743,
      "step": 2422
    },
    {
      "epoch": 3.3746518105849583,
      "grad_norm": 4.162113189697266,
      "learning_rate": 4.732660602615122e-05,
      "loss": 2.6157,
      "step": 2423
    },
    {
      "epoch": 3.3760445682451254,
      "grad_norm": 3.604059934616089,
      "learning_rate": 4.731665719158612e-05,
      "loss": 2.6157,
      "step": 2424
    },
    {
      "epoch": 3.3774373259052926,
      "grad_norm": 4.041855812072754,
      "learning_rate": 4.730670835702103e-05,
      "loss": 2.8309,
      "step": 2425
    },
    {
      "epoch": 3.3788300835654597,
      "grad_norm": 3.923802614212036,
      "learning_rate": 4.729675952245594e-05,
      "loss": 3.0207,
      "step": 2426
    },
    {
      "epoch": 3.3802228412256268,
      "grad_norm": 4.154032230377197,
      "learning_rate": 4.728681068789084e-05,
      "loss": 3.1268,
      "step": 2427
    },
    {
      "epoch": 3.381615598885794,
      "grad_norm": 3.595344066619873,
      "learning_rate": 4.727686185332575e-05,
      "loss": 2.6528,
      "step": 2428
    },
    {
      "epoch": 3.383008356545961,
      "grad_norm": 4.08375883102417,
      "learning_rate": 4.726691301876065e-05,
      "loss": 2.7236,
      "step": 2429
    },
    {
      "epoch": 3.384401114206128,
      "grad_norm": 4.021947383880615,
      "learning_rate": 4.725696418419556e-05,
      "loss": 2.6887,
      "step": 2430
    },
    {
      "epoch": 3.385793871866295,
      "grad_norm": 4.529158115386963,
      "learning_rate": 4.7247015349630465e-05,
      "loss": 3.2312,
      "step": 2431
    },
    {
      "epoch": 3.3871866295264623,
      "grad_norm": 4.145749568939209,
      "learning_rate": 4.7237066515065376e-05,
      "loss": 2.9076,
      "step": 2432
    },
    {
      "epoch": 3.3885793871866294,
      "grad_norm": 3.9654574394226074,
      "learning_rate": 4.722711768050029e-05,
      "loss": 2.9585,
      "step": 2433
    },
    {
      "epoch": 3.3899721448467965,
      "grad_norm": 4.129126071929932,
      "learning_rate": 4.7217168845935184e-05,
      "loss": 2.7876,
      "step": 2434
    },
    {
      "epoch": 3.3913649025069637,
      "grad_norm": 4.0180182456970215,
      "learning_rate": 4.7207220011370095e-05,
      "loss": 2.8569,
      "step": 2435
    },
    {
      "epoch": 3.3927576601671308,
      "grad_norm": 6.775938510894775,
      "learning_rate": 4.7197271176805e-05,
      "loss": 2.8111,
      "step": 2436
    },
    {
      "epoch": 3.394150417827298,
      "grad_norm": 4.824405193328857,
      "learning_rate": 4.718732234223991e-05,
      "loss": 2.7421,
      "step": 2437
    },
    {
      "epoch": 3.3955431754874654,
      "grad_norm": 4.232151985168457,
      "learning_rate": 4.7177373507674814e-05,
      "loss": 2.6584,
      "step": 2438
    },
    {
      "epoch": 3.396935933147632,
      "grad_norm": 3.910310745239258,
      "learning_rate": 4.716742467310972e-05,
      "loss": 2.7436,
      "step": 2439
    },
    {
      "epoch": 3.3983286908077996,
      "grad_norm": 3.6283116340637207,
      "learning_rate": 4.715747583854462e-05,
      "loss": 2.9569,
      "step": 2440
    },
    {
      "epoch": 3.3997214484679668,
      "grad_norm": 4.020989894866943,
      "learning_rate": 4.714752700397953e-05,
      "loss": 2.662,
      "step": 2441
    },
    {
      "epoch": 3.401114206128134,
      "grad_norm": 4.190573215484619,
      "learning_rate": 4.713757816941444e-05,
      "loss": 2.8246,
      "step": 2442
    },
    {
      "epoch": 3.402506963788301,
      "grad_norm": 4.062912464141846,
      "learning_rate": 4.712762933484934e-05,
      "loss": 2.9073,
      "step": 2443
    },
    {
      "epoch": 3.403899721448468,
      "grad_norm": 4.1267170906066895,
      "learning_rate": 4.7117680500284245e-05,
      "loss": 3.395,
      "step": 2444
    },
    {
      "epoch": 3.405292479108635,
      "grad_norm": 3.808474063873291,
      "learning_rate": 4.7107731665719156e-05,
      "loss": 2.4232,
      "step": 2445
    },
    {
      "epoch": 3.4066852367688023,
      "grad_norm": 4.739579677581787,
      "learning_rate": 4.709778283115406e-05,
      "loss": 2.6484,
      "step": 2446
    },
    {
      "epoch": 3.4080779944289694,
      "grad_norm": 3.631610155105591,
      "learning_rate": 4.708783399658897e-05,
      "loss": 2.6488,
      "step": 2447
    },
    {
      "epoch": 3.4094707520891365,
      "grad_norm": 3.909418821334839,
      "learning_rate": 4.707788516202387e-05,
      "loss": 2.7904,
      "step": 2448
    },
    {
      "epoch": 3.4108635097493036,
      "grad_norm": 3.8568201065063477,
      "learning_rate": 4.706793632745878e-05,
      "loss": 2.4642,
      "step": 2449
    },
    {
      "epoch": 3.4122562674094707,
      "grad_norm": 3.8794257640838623,
      "learning_rate": 4.705798749289368e-05,
      "loss": 2.9015,
      "step": 2450
    },
    {
      "epoch": 3.413649025069638,
      "grad_norm": 4.410316467285156,
      "learning_rate": 4.7048038658328594e-05,
      "loss": 2.4628,
      "step": 2451
    },
    {
      "epoch": 3.415041782729805,
      "grad_norm": 4.268566131591797,
      "learning_rate": 4.70380898237635e-05,
      "loss": 2.8573,
      "step": 2452
    },
    {
      "epoch": 3.416434540389972,
      "grad_norm": 3.9342827796936035,
      "learning_rate": 4.70281409891984e-05,
      "loss": 2.292,
      "step": 2453
    },
    {
      "epoch": 3.417827298050139,
      "grad_norm": 4.145666599273682,
      "learning_rate": 4.7018192154633306e-05,
      "loss": 3.166,
      "step": 2454
    },
    {
      "epoch": 3.4192200557103063,
      "grad_norm": 5.339357376098633,
      "learning_rate": 4.700824332006822e-05,
      "loss": 2.6585,
      "step": 2455
    },
    {
      "epoch": 3.4206128133704734,
      "grad_norm": 3.945842981338501,
      "learning_rate": 4.699829448550312e-05,
      "loss": 2.5021,
      "step": 2456
    },
    {
      "epoch": 3.4220055710306405,
      "grad_norm": 3.6610825061798096,
      "learning_rate": 4.698834565093803e-05,
      "loss": 2.8762,
      "step": 2457
    },
    {
      "epoch": 3.4233983286908076,
      "grad_norm": 3.8504912853240967,
      "learning_rate": 4.697839681637293e-05,
      "loss": 2.8399,
      "step": 2458
    },
    {
      "epoch": 3.424791086350975,
      "grad_norm": 5.630061626434326,
      "learning_rate": 4.696844798180784e-05,
      "loss": 2.7062,
      "step": 2459
    },
    {
      "epoch": 3.426183844011142,
      "grad_norm": 4.1890692710876465,
      "learning_rate": 4.6958499147242744e-05,
      "loss": 2.9676,
      "step": 2460
    },
    {
      "epoch": 3.4275766016713094,
      "grad_norm": 3.9940645694732666,
      "learning_rate": 4.6948550312677655e-05,
      "loss": 2.5669,
      "step": 2461
    },
    {
      "epoch": 3.4289693593314765,
      "grad_norm": 3.853699207305908,
      "learning_rate": 4.6938601478112566e-05,
      "loss": 2.9325,
      "step": 2462
    },
    {
      "epoch": 3.4303621169916436,
      "grad_norm": 3.8538451194763184,
      "learning_rate": 4.692865264354746e-05,
      "loss": 2.9975,
      "step": 2463
    },
    {
      "epoch": 3.4317548746518107,
      "grad_norm": 4.6403069496154785,
      "learning_rate": 4.6918703808982374e-05,
      "loss": 3.2556,
      "step": 2464
    },
    {
      "epoch": 3.433147632311978,
      "grad_norm": 4.27023983001709,
      "learning_rate": 4.690875497441728e-05,
      "loss": 2.9309,
      "step": 2465
    },
    {
      "epoch": 3.434540389972145,
      "grad_norm": 3.9356207847595215,
      "learning_rate": 4.689880613985219e-05,
      "loss": 2.5403,
      "step": 2466
    },
    {
      "epoch": 3.435933147632312,
      "grad_norm": 4.530642986297607,
      "learning_rate": 4.688885730528709e-05,
      "loss": 2.6984,
      "step": 2467
    },
    {
      "epoch": 3.437325905292479,
      "grad_norm": 3.673515558242798,
      "learning_rate": 4.6878908470722e-05,
      "loss": 2.4632,
      "step": 2468
    },
    {
      "epoch": 3.4387186629526463,
      "grad_norm": 3.577712059020996,
      "learning_rate": 4.68689596361569e-05,
      "loss": 2.5888,
      "step": 2469
    },
    {
      "epoch": 3.4401114206128134,
      "grad_norm": 3.673774242401123,
      "learning_rate": 4.685901080159181e-05,
      "loss": 2.502,
      "step": 2470
    },
    {
      "epoch": 3.4415041782729805,
      "grad_norm": 4.250201225280762,
      "learning_rate": 4.6849061967026716e-05,
      "loss": 2.9086,
      "step": 2471
    },
    {
      "epoch": 3.4428969359331476,
      "grad_norm": 4.001884937286377,
      "learning_rate": 4.683911313246163e-05,
      "loss": 2.6975,
      "step": 2472
    },
    {
      "epoch": 3.4442896935933147,
      "grad_norm": 4.061591148376465,
      "learning_rate": 4.6829164297896524e-05,
      "loss": 2.492,
      "step": 2473
    },
    {
      "epoch": 3.445682451253482,
      "grad_norm": 3.9153151512145996,
      "learning_rate": 4.6819215463331435e-05,
      "loss": 2.8333,
      "step": 2474
    },
    {
      "epoch": 3.447075208913649,
      "grad_norm": 3.529660940170288,
      "learning_rate": 4.680926662876634e-05,
      "loss": 2.6082,
      "step": 2475
    },
    {
      "epoch": 3.448467966573816,
      "grad_norm": 3.9232723712921143,
      "learning_rate": 4.679931779420125e-05,
      "loss": 2.8638,
      "step": 2476
    },
    {
      "epoch": 3.449860724233983,
      "grad_norm": 3.741406202316284,
      "learning_rate": 4.6789368959636154e-05,
      "loss": 2.4546,
      "step": 2477
    },
    {
      "epoch": 3.4512534818941503,
      "grad_norm": 3.93326997756958,
      "learning_rate": 4.677942012507106e-05,
      "loss": 3.0198,
      "step": 2478
    },
    {
      "epoch": 3.4526462395543174,
      "grad_norm": 4.116133213043213,
      "learning_rate": 4.676947129050596e-05,
      "loss": 2.5921,
      "step": 2479
    },
    {
      "epoch": 3.4540389972144845,
      "grad_norm": 4.393657684326172,
      "learning_rate": 4.675952245594087e-05,
      "loss": 2.7356,
      "step": 2480
    },
    {
      "epoch": 3.4554317548746516,
      "grad_norm": 4.12740421295166,
      "learning_rate": 4.674957362137578e-05,
      "loss": 2.7773,
      "step": 2481
    },
    {
      "epoch": 3.456824512534819,
      "grad_norm": 4.11367654800415,
      "learning_rate": 4.673962478681069e-05,
      "loss": 3.455,
      "step": 2482
    },
    {
      "epoch": 3.4582172701949863,
      "grad_norm": 4.02070426940918,
      "learning_rate": 4.6729675952245585e-05,
      "loss": 3.1256,
      "step": 2483
    },
    {
      "epoch": 3.4596100278551534,
      "grad_norm": 4.215537071228027,
      "learning_rate": 4.6719727117680496e-05,
      "loss": 2.6202,
      "step": 2484
    },
    {
      "epoch": 3.4610027855153205,
      "grad_norm": 3.678616523742676,
      "learning_rate": 4.67097782831154e-05,
      "loss": 2.9246,
      "step": 2485
    },
    {
      "epoch": 3.4623955431754876,
      "grad_norm": 5.492940902709961,
      "learning_rate": 4.669982944855031e-05,
      "loss": 2.6919,
      "step": 2486
    },
    {
      "epoch": 3.4637883008356547,
      "grad_norm": 4.754304885864258,
      "learning_rate": 4.668988061398521e-05,
      "loss": 2.8382,
      "step": 2487
    },
    {
      "epoch": 3.465181058495822,
      "grad_norm": 3.8414769172668457,
      "learning_rate": 4.667993177942012e-05,
      "loss": 2.2719,
      "step": 2488
    },
    {
      "epoch": 3.466573816155989,
      "grad_norm": 3.788646697998047,
      "learning_rate": 4.666998294485502e-05,
      "loss": 2.2803,
      "step": 2489
    },
    {
      "epoch": 3.467966573816156,
      "grad_norm": 3.948223114013672,
      "learning_rate": 4.6660034110289934e-05,
      "loss": 2.5775,
      "step": 2490
    },
    {
      "epoch": 3.469359331476323,
      "grad_norm": 4.155910968780518,
      "learning_rate": 4.6650085275724845e-05,
      "loss": 2.9139,
      "step": 2491
    },
    {
      "epoch": 3.4707520891364902,
      "grad_norm": 6.739853858947754,
      "learning_rate": 4.664013644115974e-05,
      "loss": 2.9066,
      "step": 2492
    },
    {
      "epoch": 3.4721448467966574,
      "grad_norm": 4.00689172744751,
      "learning_rate": 4.663018760659465e-05,
      "loss": 2.856,
      "step": 2493
    },
    {
      "epoch": 3.4735376044568245,
      "grad_norm": 3.6261701583862305,
      "learning_rate": 4.662023877202956e-05,
      "loss": 2.8077,
      "step": 2494
    },
    {
      "epoch": 3.4749303621169916,
      "grad_norm": 3.793961524963379,
      "learning_rate": 4.661028993746447e-05,
      "loss": 2.6904,
      "step": 2495
    },
    {
      "epoch": 3.4763231197771587,
      "grad_norm": 4.114292621612549,
      "learning_rate": 4.660034110289937e-05,
      "loss": 3.1544,
      "step": 2496
    },
    {
      "epoch": 3.477715877437326,
      "grad_norm": 3.8729336261749268,
      "learning_rate": 4.6590392268334276e-05,
      "loss": 2.6785,
      "step": 2497
    },
    {
      "epoch": 3.479108635097493,
      "grad_norm": 4.672178268432617,
      "learning_rate": 4.658044343376918e-05,
      "loss": 2.5692,
      "step": 2498
    },
    {
      "epoch": 3.48050139275766,
      "grad_norm": 3.973593235015869,
      "learning_rate": 4.657049459920409e-05,
      "loss": 2.5247,
      "step": 2499
    },
    {
      "epoch": 3.481894150417827,
      "grad_norm": 4.204497814178467,
      "learning_rate": 4.6560545764638995e-05,
      "loss": 3.3036,
      "step": 2500
    },
    {
      "epoch": 3.4832869080779942,
      "grad_norm": 3.5338776111602783,
      "learning_rate": 4.6550596930073906e-05,
      "loss": 2.2795,
      "step": 2501
    },
    {
      "epoch": 3.4846796657381613,
      "grad_norm": 3.7186532020568848,
      "learning_rate": 4.6540648095508803e-05,
      "loss": 2.4927,
      "step": 2502
    },
    {
      "epoch": 3.486072423398329,
      "grad_norm": 4.113531112670898,
      "learning_rate": 4.6530699260943714e-05,
      "loss": 2.3812,
      "step": 2503
    },
    {
      "epoch": 3.487465181058496,
      "grad_norm": 3.9955506324768066,
      "learning_rate": 4.652075042637862e-05,
      "loss": 3.0143,
      "step": 2504
    },
    {
      "epoch": 3.488857938718663,
      "grad_norm": 3.6784656047821045,
      "learning_rate": 4.651080159181353e-05,
      "loss": 2.3338,
      "step": 2505
    },
    {
      "epoch": 3.4902506963788302,
      "grad_norm": 4.05626916885376,
      "learning_rate": 4.650085275724843e-05,
      "loss": 2.8512,
      "step": 2506
    },
    {
      "epoch": 3.4916434540389973,
      "grad_norm": 4.049886226654053,
      "learning_rate": 4.649090392268334e-05,
      "loss": 2.8965,
      "step": 2507
    },
    {
      "epoch": 3.4930362116991645,
      "grad_norm": 4.836441993713379,
      "learning_rate": 4.648095508811824e-05,
      "loss": 2.6381,
      "step": 2508
    },
    {
      "epoch": 3.4944289693593316,
      "grad_norm": 4.007538318634033,
      "learning_rate": 4.647100625355315e-05,
      "loss": 2.3314,
      "step": 2509
    },
    {
      "epoch": 3.4958217270194987,
      "grad_norm": 5.4083662033081055,
      "learning_rate": 4.6461057418988056e-05,
      "loss": 2.4072,
      "step": 2510
    },
    {
      "epoch": 3.497214484679666,
      "grad_norm": 3.6691737174987793,
      "learning_rate": 4.645110858442297e-05,
      "loss": 2.3187,
      "step": 2511
    },
    {
      "epoch": 3.498607242339833,
      "grad_norm": 5.101102828979492,
      "learning_rate": 4.6441159749857865e-05,
      "loss": 2.8893,
      "step": 2512
    },
    {
      "epoch": 3.5,
      "grad_norm": 3.9152889251708984,
      "learning_rate": 4.6431210915292775e-05,
      "loss": 2.4472,
      "step": 2513
    },
    {
      "epoch": 3.501392757660167,
      "grad_norm": 4.1873555183410645,
      "learning_rate": 4.642126208072768e-05,
      "loss": 2.7775,
      "step": 2514
    },
    {
      "epoch": 3.502785515320334,
      "grad_norm": 3.612882614135742,
      "learning_rate": 4.641131324616259e-05,
      "loss": 2.703,
      "step": 2515
    },
    {
      "epoch": 3.5041782729805013,
      "grad_norm": 4.355597972869873,
      "learning_rate": 4.64013644115975e-05,
      "loss": 2.9082,
      "step": 2516
    },
    {
      "epoch": 3.5055710306406684,
      "grad_norm": 3.797044277191162,
      "learning_rate": 4.63914155770324e-05,
      "loss": 2.6401,
      "step": 2517
    },
    {
      "epoch": 3.5069637883008355,
      "grad_norm": 3.89890456199646,
      "learning_rate": 4.63814667424673e-05,
      "loss": 2.5919,
      "step": 2518
    },
    {
      "epoch": 3.5083565459610027,
      "grad_norm": 3.6076433658599854,
      "learning_rate": 4.637151790790221e-05,
      "loss": 2.422,
      "step": 2519
    },
    {
      "epoch": 3.5097493036211698,
      "grad_norm": 3.984490156173706,
      "learning_rate": 4.6361569073337124e-05,
      "loss": 2.6891,
      "step": 2520
    },
    {
      "epoch": 3.511142061281337,
      "grad_norm": 3.5938799381256104,
      "learning_rate": 4.635162023877203e-05,
      "loss": 2.7263,
      "step": 2521
    },
    {
      "epoch": 3.5125348189415044,
      "grad_norm": 4.015036106109619,
      "learning_rate": 4.634167140420693e-05,
      "loss": 2.8112,
      "step": 2522
    },
    {
      "epoch": 3.513927576601671,
      "grad_norm": 3.9240949153900146,
      "learning_rate": 4.6331722569641836e-05,
      "loss": 2.7094,
      "step": 2523
    },
    {
      "epoch": 3.5153203342618387,
      "grad_norm": 4.509311676025391,
      "learning_rate": 4.632177373507675e-05,
      "loss": 3.0363,
      "step": 2524
    },
    {
      "epoch": 3.5167130919220053,
      "grad_norm": 4.242368221282959,
      "learning_rate": 4.631182490051165e-05,
      "loss": 2.9884,
      "step": 2525
    },
    {
      "epoch": 3.518105849582173,
      "grad_norm": 5.659360885620117,
      "learning_rate": 4.6301876065946555e-05,
      "loss": 2.782,
      "step": 2526
    },
    {
      "epoch": 3.51949860724234,
      "grad_norm": 4.399890422821045,
      "learning_rate": 4.629192723138146e-05,
      "loss": 2.7044,
      "step": 2527
    },
    {
      "epoch": 3.520891364902507,
      "grad_norm": 4.138412952423096,
      "learning_rate": 4.628197839681637e-05,
      "loss": 2.7621,
      "step": 2528
    },
    {
      "epoch": 3.522284122562674,
      "grad_norm": 4.267613887786865,
      "learning_rate": 4.6272029562251274e-05,
      "loss": 2.5129,
      "step": 2529
    },
    {
      "epoch": 3.5236768802228413,
      "grad_norm": 3.9596104621887207,
      "learning_rate": 4.6262080727686185e-05,
      "loss": 2.6278,
      "step": 2530
    },
    {
      "epoch": 3.5250696378830084,
      "grad_norm": 3.815171003341675,
      "learning_rate": 4.625213189312108e-05,
      "loss": 2.4257,
      "step": 2531
    },
    {
      "epoch": 3.5264623955431755,
      "grad_norm": 5.524709224700928,
      "learning_rate": 4.6242183058555993e-05,
      "loss": 2.7696,
      "step": 2532
    },
    {
      "epoch": 3.5278551532033426,
      "grad_norm": 4.13335657119751,
      "learning_rate": 4.62322342239909e-05,
      "loss": 3.011,
      "step": 2533
    },
    {
      "epoch": 3.5292479108635098,
      "grad_norm": 4.052757263183594,
      "learning_rate": 4.622228538942581e-05,
      "loss": 2.6474,
      "step": 2534
    },
    {
      "epoch": 3.530640668523677,
      "grad_norm": 3.7928733825683594,
      "learning_rate": 4.621233655486071e-05,
      "loss": 2.4821,
      "step": 2535
    },
    {
      "epoch": 3.532033426183844,
      "grad_norm": 3.940757989883423,
      "learning_rate": 4.6202387720295617e-05,
      "loss": 2.5799,
      "step": 2536
    },
    {
      "epoch": 3.533426183844011,
      "grad_norm": 4.096225738525391,
      "learning_rate": 4.619243888573052e-05,
      "loss": 2.8428,
      "step": 2537
    },
    {
      "epoch": 3.534818941504178,
      "grad_norm": 4.849141597747803,
      "learning_rate": 4.618249005116543e-05,
      "loss": 2.3243,
      "step": 2538
    },
    {
      "epoch": 3.5362116991643453,
      "grad_norm": 4.354057312011719,
      "learning_rate": 4.6172541216600336e-05,
      "loss": 2.5553,
      "step": 2539
    },
    {
      "epoch": 3.5376044568245124,
      "grad_norm": 4.655442714691162,
      "learning_rate": 4.6162592382035246e-05,
      "loss": 2.5603,
      "step": 2540
    },
    {
      "epoch": 3.5389972144846795,
      "grad_norm": 3.835197687149048,
      "learning_rate": 4.6152643547470144e-05,
      "loss": 2.3249,
      "step": 2541
    },
    {
      "epoch": 3.5403899721448466,
      "grad_norm": 3.771993398666382,
      "learning_rate": 4.6142694712905055e-05,
      "loss": 2.9533,
      "step": 2542
    },
    {
      "epoch": 3.541782729805014,
      "grad_norm": 4.148719310760498,
      "learning_rate": 4.613274587833996e-05,
      "loss": 2.7036,
      "step": 2543
    },
    {
      "epoch": 3.543175487465181,
      "grad_norm": 4.081655025482178,
      "learning_rate": 4.612279704377487e-05,
      "loss": 2.9815,
      "step": 2544
    },
    {
      "epoch": 3.5445682451253484,
      "grad_norm": 4.131512641906738,
      "learning_rate": 4.611284820920978e-05,
      "loss": 2.6722,
      "step": 2545
    },
    {
      "epoch": 3.545961002785515,
      "grad_norm": 4.983314514160156,
      "learning_rate": 4.610289937464468e-05,
      "loss": 2.4037,
      "step": 2546
    },
    {
      "epoch": 3.5473537604456826,
      "grad_norm": 4.0679097175598145,
      "learning_rate": 4.609295054007959e-05,
      "loss": 2.5493,
      "step": 2547
    },
    {
      "epoch": 3.5487465181058497,
      "grad_norm": 3.8854141235351562,
      "learning_rate": 4.608300170551449e-05,
      "loss": 2.9201,
      "step": 2548
    },
    {
      "epoch": 3.550139275766017,
      "grad_norm": 4.404861927032471,
      "learning_rate": 4.6073052870949403e-05,
      "loss": 2.3592,
      "step": 2549
    },
    {
      "epoch": 3.551532033426184,
      "grad_norm": 4.275940895080566,
      "learning_rate": 4.606310403638431e-05,
      "loss": 2.4952,
      "step": 2550
    },
    {
      "epoch": 3.552924791086351,
      "grad_norm": 4.227753162384033,
      "learning_rate": 4.605315520181921e-05,
      "loss": 2.7161,
      "step": 2551
    },
    {
      "epoch": 3.554317548746518,
      "grad_norm": 4.181053638458252,
      "learning_rate": 4.6043206367254116e-05,
      "loss": 2.8687,
      "step": 2552
    },
    {
      "epoch": 3.5557103064066853,
      "grad_norm": 3.793886661529541,
      "learning_rate": 4.6033257532689026e-05,
      "loss": 2.7349,
      "step": 2553
    },
    {
      "epoch": 3.5571030640668524,
      "grad_norm": 3.7882189750671387,
      "learning_rate": 4.602330869812393e-05,
      "loss": 2.308,
      "step": 2554
    },
    {
      "epoch": 3.5584958217270195,
      "grad_norm": 3.880540370941162,
      "learning_rate": 4.601335986355884e-05,
      "loss": 2.4352,
      "step": 2555
    },
    {
      "epoch": 3.5598885793871866,
      "grad_norm": 3.4791016578674316,
      "learning_rate": 4.600341102899374e-05,
      "loss": 2.4757,
      "step": 2556
    },
    {
      "epoch": 3.5612813370473537,
      "grad_norm": 3.6199772357940674,
      "learning_rate": 4.599346219442865e-05,
      "loss": 2.4552,
      "step": 2557
    },
    {
      "epoch": 3.562674094707521,
      "grad_norm": 4.231076240539551,
      "learning_rate": 4.5983513359863554e-05,
      "loss": 2.8545,
      "step": 2558
    },
    {
      "epoch": 3.564066852367688,
      "grad_norm": 3.7487900257110596,
      "learning_rate": 4.5973564525298464e-05,
      "loss": 2.5915,
      "step": 2559
    },
    {
      "epoch": 3.565459610027855,
      "grad_norm": 4.234863758087158,
      "learning_rate": 4.596361569073337e-05,
      "loss": 2.4,
      "step": 2560
    },
    {
      "epoch": 3.566852367688022,
      "grad_norm": 4.43113374710083,
      "learning_rate": 4.595366685616827e-05,
      "loss": 2.569,
      "step": 2561
    },
    {
      "epoch": 3.5682451253481893,
      "grad_norm": 4.24310827255249,
      "learning_rate": 4.594371802160318e-05,
      "loss": 2.8166,
      "step": 2562
    },
    {
      "epoch": 3.5696378830083564,
      "grad_norm": 4.14475679397583,
      "learning_rate": 4.593376918703809e-05,
      "loss": 2.6704,
      "step": 2563
    },
    {
      "epoch": 3.571030640668524,
      "grad_norm": 3.697046995162964,
      "learning_rate": 4.592382035247299e-05,
      "loss": 2.403,
      "step": 2564
    },
    {
      "epoch": 3.5724233983286906,
      "grad_norm": 3.756159543991089,
      "learning_rate": 4.59138715179079e-05,
      "loss": 2.9002,
      "step": 2565
    },
    {
      "epoch": 3.573816155988858,
      "grad_norm": 3.5814194679260254,
      "learning_rate": 4.59039226833428e-05,
      "loss": 2.3633,
      "step": 2566
    },
    {
      "epoch": 3.575208913649025,
      "grad_norm": 3.5276687145233154,
      "learning_rate": 4.589397384877771e-05,
      "loss": 2.3979,
      "step": 2567
    },
    {
      "epoch": 3.5766016713091924,
      "grad_norm": 3.988581418991089,
      "learning_rate": 4.5884025014212615e-05,
      "loss": 2.6663,
      "step": 2568
    },
    {
      "epoch": 3.5779944289693595,
      "grad_norm": 3.8402259349823,
      "learning_rate": 4.5874076179647526e-05,
      "loss": 2.734,
      "step": 2569
    },
    {
      "epoch": 3.5793871866295266,
      "grad_norm": 4.858250141143799,
      "learning_rate": 4.586412734508242e-05,
      "loss": 2.404,
      "step": 2570
    },
    {
      "epoch": 3.5807799442896937,
      "grad_norm": 4.333425998687744,
      "learning_rate": 4.5854178510517334e-05,
      "loss": 3.1859,
      "step": 2571
    },
    {
      "epoch": 3.582172701949861,
      "grad_norm": 4.770310878753662,
      "learning_rate": 4.584422967595224e-05,
      "loss": 2.8104,
      "step": 2572
    },
    {
      "epoch": 3.583565459610028,
      "grad_norm": 4.867776870727539,
      "learning_rate": 4.583428084138715e-05,
      "loss": 2.8107,
      "step": 2573
    },
    {
      "epoch": 3.584958217270195,
      "grad_norm": 3.9517855644226074,
      "learning_rate": 4.582433200682206e-05,
      "loss": 2.9689,
      "step": 2574
    },
    {
      "epoch": 3.586350974930362,
      "grad_norm": 3.9247171878814697,
      "learning_rate": 4.581438317225696e-05,
      "loss": 2.4604,
      "step": 2575
    },
    {
      "epoch": 3.5877437325905293,
      "grad_norm": 3.239447832107544,
      "learning_rate": 4.580443433769187e-05,
      "loss": 2.2073,
      "step": 2576
    },
    {
      "epoch": 3.5891364902506964,
      "grad_norm": 3.8720920085906982,
      "learning_rate": 4.579448550312677e-05,
      "loss": 2.6184,
      "step": 2577
    },
    {
      "epoch": 3.5905292479108635,
      "grad_norm": 4.217468738555908,
      "learning_rate": 4.578453666856168e-05,
      "loss": 2.8072,
      "step": 2578
    },
    {
      "epoch": 3.5919220055710306,
      "grad_norm": 4.635210990905762,
      "learning_rate": 4.577458783399659e-05,
      "loss": 2.4138,
      "step": 2579
    },
    {
      "epoch": 3.5933147632311977,
      "grad_norm": 4.103201389312744,
      "learning_rate": 4.576463899943149e-05,
      "loss": 2.5774,
      "step": 2580
    },
    {
      "epoch": 3.594707520891365,
      "grad_norm": 4.105129241943359,
      "learning_rate": 4.5754690164866395e-05,
      "loss": 2.5725,
      "step": 2581
    },
    {
      "epoch": 3.596100278551532,
      "grad_norm": 4.445010185241699,
      "learning_rate": 4.5744741330301306e-05,
      "loss": 2.3895,
      "step": 2582
    },
    {
      "epoch": 3.597493036211699,
      "grad_norm": 3.87135910987854,
      "learning_rate": 4.573479249573621e-05,
      "loss": 2.5712,
      "step": 2583
    },
    {
      "epoch": 3.598885793871866,
      "grad_norm": 4.35500955581665,
      "learning_rate": 4.572484366117112e-05,
      "loss": 2.7363,
      "step": 2584
    },
    {
      "epoch": 3.6002785515320337,
      "grad_norm": 4.664234638214111,
      "learning_rate": 4.571489482660602e-05,
      "loss": 2.9104,
      "step": 2585
    },
    {
      "epoch": 3.6016713091922004,
      "grad_norm": 4.105116844177246,
      "learning_rate": 4.570494599204093e-05,
      "loss": 2.3681,
      "step": 2586
    },
    {
      "epoch": 3.603064066852368,
      "grad_norm": 3.8014397621154785,
      "learning_rate": 4.569499715747583e-05,
      "loss": 2.685,
      "step": 2587
    },
    {
      "epoch": 3.6044568245125346,
      "grad_norm": 4.089648723602295,
      "learning_rate": 4.5685048322910744e-05,
      "loss": 2.4354,
      "step": 2588
    },
    {
      "epoch": 3.605849582172702,
      "grad_norm": 4.380812168121338,
      "learning_rate": 4.567509948834565e-05,
      "loss": 2.6941,
      "step": 2589
    },
    {
      "epoch": 3.6072423398328692,
      "grad_norm": 4.799055576324463,
      "learning_rate": 4.566515065378055e-05,
      "loss": 3.1246,
      "step": 2590
    },
    {
      "epoch": 3.6086350974930363,
      "grad_norm": 4.490792274475098,
      "learning_rate": 4.5655201819215456e-05,
      "loss": 2.8987,
      "step": 2591
    },
    {
      "epoch": 3.6100278551532035,
      "grad_norm": 3.71956467628479,
      "learning_rate": 4.564525298465037e-05,
      "loss": 2.7376,
      "step": 2592
    },
    {
      "epoch": 3.6114206128133706,
      "grad_norm": 3.9480626583099365,
      "learning_rate": 4.563530415008527e-05,
      "loss": 2.6768,
      "step": 2593
    },
    {
      "epoch": 3.6128133704735377,
      "grad_norm": 5.132822513580322,
      "learning_rate": 4.562535531552018e-05,
      "loss": 2.8901,
      "step": 2594
    },
    {
      "epoch": 3.614206128133705,
      "grad_norm": 4.388399600982666,
      "learning_rate": 4.561540648095508e-05,
      "loss": 2.8447,
      "step": 2595
    },
    {
      "epoch": 3.615598885793872,
      "grad_norm": 3.8116705417633057,
      "learning_rate": 4.560545764638999e-05,
      "loss": 2.5332,
      "step": 2596
    },
    {
      "epoch": 3.616991643454039,
      "grad_norm": 6.826498985290527,
      "learning_rate": 4.5595508811824894e-05,
      "loss": 2.6834,
      "step": 2597
    },
    {
      "epoch": 3.618384401114206,
      "grad_norm": 3.922801971435547,
      "learning_rate": 4.5585559977259805e-05,
      "loss": 2.9515,
      "step": 2598
    },
    {
      "epoch": 3.6197771587743732,
      "grad_norm": 4.208038806915283,
      "learning_rate": 4.5575611142694716e-05,
      "loss": 2.5785,
      "step": 2599
    },
    {
      "epoch": 3.6211699164345403,
      "grad_norm": 3.956136465072632,
      "learning_rate": 4.556566230812961e-05,
      "loss": 2.4827,
      "step": 2600
    },
    {
      "epoch": 3.6225626740947074,
      "grad_norm": 4.067508697509766,
      "learning_rate": 4.555571347356452e-05,
      "loss": 2.242,
      "step": 2601
    },
    {
      "epoch": 3.6239554317548746,
      "grad_norm": 4.087154388427734,
      "learning_rate": 4.554576463899943e-05,
      "loss": 2.6914,
      "step": 2602
    },
    {
      "epoch": 3.6253481894150417,
      "grad_norm": 3.8476922512054443,
      "learning_rate": 4.553581580443434e-05,
      "loss": 2.582,
      "step": 2603
    },
    {
      "epoch": 3.6267409470752088,
      "grad_norm": 3.9368791580200195,
      "learning_rate": 4.552586696986924e-05,
      "loss": 2.1679,
      "step": 2604
    },
    {
      "epoch": 3.628133704735376,
      "grad_norm": 4.296043395996094,
      "learning_rate": 4.551591813530415e-05,
      "loss": 2.8044,
      "step": 2605
    },
    {
      "epoch": 3.6295264623955434,
      "grad_norm": 4.281557559967041,
      "learning_rate": 4.550596930073905e-05,
      "loss": 2.8132,
      "step": 2606
    },
    {
      "epoch": 3.63091922005571,
      "grad_norm": 3.6667840480804443,
      "learning_rate": 4.549602046617396e-05,
      "loss": 2.562,
      "step": 2607
    },
    {
      "epoch": 3.6323119777158777,
      "grad_norm": 3.9428160190582275,
      "learning_rate": 4.5486071631608866e-05,
      "loss": 2.726,
      "step": 2608
    },
    {
      "epoch": 3.6337047353760443,
      "grad_norm": 4.410280227661133,
      "learning_rate": 4.547612279704378e-05,
      "loss": 2.5515,
      "step": 2609
    },
    {
      "epoch": 3.635097493036212,
      "grad_norm": 3.685770034790039,
      "learning_rate": 4.5466173962478674e-05,
      "loss": 2.566,
      "step": 2610
    },
    {
      "epoch": 3.636490250696379,
      "grad_norm": 5.172934055328369,
      "learning_rate": 4.5456225127913585e-05,
      "loss": 2.1931,
      "step": 2611
    },
    {
      "epoch": 3.637883008356546,
      "grad_norm": 3.7498984336853027,
      "learning_rate": 4.544627629334849e-05,
      "loss": 2.2915,
      "step": 2612
    },
    {
      "epoch": 3.639275766016713,
      "grad_norm": 3.7322449684143066,
      "learning_rate": 4.54363274587834e-05,
      "loss": 2.2766,
      "step": 2613
    },
    {
      "epoch": 3.6406685236768803,
      "grad_norm": 4.4291090965271,
      "learning_rate": 4.54263786242183e-05,
      "loss": 2.5531,
      "step": 2614
    },
    {
      "epoch": 3.6420612813370474,
      "grad_norm": 4.109873294830322,
      "learning_rate": 4.541642978965321e-05,
      "loss": 2.727,
      "step": 2615
    },
    {
      "epoch": 3.6434540389972145,
      "grad_norm": 3.9789352416992188,
      "learning_rate": 4.540648095508811e-05,
      "loss": 2.346,
      "step": 2616
    },
    {
      "epoch": 3.6448467966573816,
      "grad_norm": 3.961202383041382,
      "learning_rate": 4.539653212052302e-05,
      "loss": 2.4444,
      "step": 2617
    },
    {
      "epoch": 3.6462395543175488,
      "grad_norm": 3.8635852336883545,
      "learning_rate": 4.538658328595793e-05,
      "loss": 2.6535,
      "step": 2618
    },
    {
      "epoch": 3.647632311977716,
      "grad_norm": 3.8671016693115234,
      "learning_rate": 4.537663445139283e-05,
      "loss": 2.5828,
      "step": 2619
    },
    {
      "epoch": 3.649025069637883,
      "grad_norm": 3.676393985748291,
      "learning_rate": 4.5366685616827735e-05,
      "loss": 2.4813,
      "step": 2620
    },
    {
      "epoch": 3.65041782729805,
      "grad_norm": 4.046565532684326,
      "learning_rate": 4.5356736782262646e-05,
      "loss": 2.5864,
      "step": 2621
    },
    {
      "epoch": 3.651810584958217,
      "grad_norm": 3.296393394470215,
      "learning_rate": 4.534678794769755e-05,
      "loss": 2.194,
      "step": 2622
    },
    {
      "epoch": 3.6532033426183843,
      "grad_norm": 3.507030725479126,
      "learning_rate": 4.533683911313246e-05,
      "loss": 2.6079,
      "step": 2623
    },
    {
      "epoch": 3.6545961002785514,
      "grad_norm": 4.159409523010254,
      "learning_rate": 4.532689027856736e-05,
      "loss": 2.7253,
      "step": 2624
    },
    {
      "epoch": 3.6559888579387185,
      "grad_norm": 3.7661807537078857,
      "learning_rate": 4.531694144400227e-05,
      "loss": 2.7663,
      "step": 2625
    },
    {
      "epoch": 3.6573816155988856,
      "grad_norm": 4.4413323402404785,
      "learning_rate": 4.530699260943717e-05,
      "loss": 2.7119,
      "step": 2626
    },
    {
      "epoch": 3.658774373259053,
      "grad_norm": 4.101630210876465,
      "learning_rate": 4.5297043774872084e-05,
      "loss": 2.089,
      "step": 2627
    },
    {
      "epoch": 3.66016713091922,
      "grad_norm": 4.0640034675598145,
      "learning_rate": 4.5287094940306995e-05,
      "loss": 2.997,
      "step": 2628
    },
    {
      "epoch": 3.6615598885793874,
      "grad_norm": 3.9334235191345215,
      "learning_rate": 4.527714610574189e-05,
      "loss": 2.6158,
      "step": 2629
    },
    {
      "epoch": 3.662952646239554,
      "grad_norm": 4.678755760192871,
      "learning_rate": 4.5267197271176796e-05,
      "loss": 2.5875,
      "step": 2630
    },
    {
      "epoch": 3.6643454038997216,
      "grad_norm": 3.4286279678344727,
      "learning_rate": 4.525724843661171e-05,
      "loss": 2.301,
      "step": 2631
    },
    {
      "epoch": 3.6657381615598887,
      "grad_norm": 4.978024959564209,
      "learning_rate": 4.524729960204662e-05,
      "loss": 3.3852,
      "step": 2632
    },
    {
      "epoch": 3.667130919220056,
      "grad_norm": 3.922593116760254,
      "learning_rate": 4.523735076748152e-05,
      "loss": 2.5403,
      "step": 2633
    },
    {
      "epoch": 3.668523676880223,
      "grad_norm": 3.587224245071411,
      "learning_rate": 4.5227401932916426e-05,
      "loss": 2.6945,
      "step": 2634
    },
    {
      "epoch": 3.66991643454039,
      "grad_norm": 3.586855173110962,
      "learning_rate": 4.521745309835133e-05,
      "loss": 2.4708,
      "step": 2635
    },
    {
      "epoch": 3.671309192200557,
      "grad_norm": 4.244228839874268,
      "learning_rate": 4.520750426378624e-05,
      "loss": 2.5793,
      "step": 2636
    },
    {
      "epoch": 3.6727019498607243,
      "grad_norm": 3.6994094848632812,
      "learning_rate": 4.5197555429221145e-05,
      "loss": 2.2796,
      "step": 2637
    },
    {
      "epoch": 3.6740947075208914,
      "grad_norm": 3.9471349716186523,
      "learning_rate": 4.5187606594656056e-05,
      "loss": 2.9239,
      "step": 2638
    },
    {
      "epoch": 3.6754874651810585,
      "grad_norm": 3.8247509002685547,
      "learning_rate": 4.517765776009095e-05,
      "loss": 2.3052,
      "step": 2639
    },
    {
      "epoch": 3.6768802228412256,
      "grad_norm": 4.448849678039551,
      "learning_rate": 4.5167708925525864e-05,
      "loss": 2.5112,
      "step": 2640
    },
    {
      "epoch": 3.6782729805013927,
      "grad_norm": 4.322947978973389,
      "learning_rate": 4.515776009096077e-05,
      "loss": 2.663,
      "step": 2641
    },
    {
      "epoch": 3.67966573816156,
      "grad_norm": 3.362961769104004,
      "learning_rate": 4.514781125639568e-05,
      "loss": 2.422,
      "step": 2642
    },
    {
      "epoch": 3.681058495821727,
      "grad_norm": 3.8426859378814697,
      "learning_rate": 4.513786242183058e-05,
      "loss": 2.6184,
      "step": 2643
    },
    {
      "epoch": 3.682451253481894,
      "grad_norm": 4.057262897491455,
      "learning_rate": 4.512791358726549e-05,
      "loss": 2.7178,
      "step": 2644
    },
    {
      "epoch": 3.683844011142061,
      "grad_norm": 4.93907356262207,
      "learning_rate": 4.511796475270039e-05,
      "loss": 2.766,
      "step": 2645
    },
    {
      "epoch": 3.6852367688022283,
      "grad_norm": 4.774403095245361,
      "learning_rate": 4.51080159181353e-05,
      "loss": 2.6246,
      "step": 2646
    },
    {
      "epoch": 3.6866295264623954,
      "grad_norm": 3.8961191177368164,
      "learning_rate": 4.5098067083570206e-05,
      "loss": 2.6065,
      "step": 2647
    },
    {
      "epoch": 3.688022284122563,
      "grad_norm": 4.014467716217041,
      "learning_rate": 4.508811824900512e-05,
      "loss": 2.3052,
      "step": 2648
    },
    {
      "epoch": 3.6894150417827296,
      "grad_norm": 3.5372679233551025,
      "learning_rate": 4.5078169414440014e-05,
      "loss": 2.2911,
      "step": 2649
    },
    {
      "epoch": 3.690807799442897,
      "grad_norm": 4.601160526275635,
      "learning_rate": 4.5068220579874925e-05,
      "loss": 2.5624,
      "step": 2650
    },
    {
      "epoch": 3.692200557103064,
      "grad_norm": 4.832505226135254,
      "learning_rate": 4.505827174530983e-05,
      "loss": 2.713,
      "step": 2651
    },
    {
      "epoch": 3.6935933147632314,
      "grad_norm": 3.792067527770996,
      "learning_rate": 4.504832291074474e-05,
      "loss": 2.8528,
      "step": 2652
    },
    {
      "epoch": 3.6949860724233985,
      "grad_norm": 8.431668281555176,
      "learning_rate": 4.503837407617965e-05,
      "loss": 2.4823,
      "step": 2653
    },
    {
      "epoch": 3.6963788300835656,
      "grad_norm": 4.345920562744141,
      "learning_rate": 4.502842524161455e-05,
      "loss": 3.0058,
      "step": 2654
    },
    {
      "epoch": 3.6977715877437327,
      "grad_norm": 3.7541372776031494,
      "learning_rate": 4.501847640704945e-05,
      "loss": 2.2991,
      "step": 2655
    },
    {
      "epoch": 3.6991643454039,
      "grad_norm": 3.7721927165985107,
      "learning_rate": 4.500852757248436e-05,
      "loss": 2.5565,
      "step": 2656
    },
    {
      "epoch": 3.700557103064067,
      "grad_norm": 4.094515800476074,
      "learning_rate": 4.4998578737919274e-05,
      "loss": 2.4353,
      "step": 2657
    },
    {
      "epoch": 3.701949860724234,
      "grad_norm": 3.893754005432129,
      "learning_rate": 4.498862990335417e-05,
      "loss": 2.3503,
      "step": 2658
    },
    {
      "epoch": 3.703342618384401,
      "grad_norm": 3.969780206680298,
      "learning_rate": 4.4978681068789075e-05,
      "loss": 2.6403,
      "step": 2659
    },
    {
      "epoch": 3.7047353760445683,
      "grad_norm": 3.9899258613586426,
      "learning_rate": 4.4968732234223986e-05,
      "loss": 2.6126,
      "step": 2660
    },
    {
      "epoch": 3.7061281337047354,
      "grad_norm": 4.702632904052734,
      "learning_rate": 4.49587833996589e-05,
      "loss": 2.6044,
      "step": 2661
    },
    {
      "epoch": 3.7075208913649025,
      "grad_norm": 3.7289204597473145,
      "learning_rate": 4.49488345650938e-05,
      "loss": 2.4549,
      "step": 2662
    },
    {
      "epoch": 3.7089136490250696,
      "grad_norm": 3.6642167568206787,
      "learning_rate": 4.4938885730528705e-05,
      "loss": 2.7396,
      "step": 2663
    },
    {
      "epoch": 3.7103064066852367,
      "grad_norm": 3.549678325653076,
      "learning_rate": 4.492893689596361e-05,
      "loss": 2.2628,
      "step": 2664
    },
    {
      "epoch": 3.711699164345404,
      "grad_norm": 4.3354997634887695,
      "learning_rate": 4.491898806139852e-05,
      "loss": 2.9193,
      "step": 2665
    },
    {
      "epoch": 3.713091922005571,
      "grad_norm": 4.62516975402832,
      "learning_rate": 4.4909039226833424e-05,
      "loss": 2.2609,
      "step": 2666
    },
    {
      "epoch": 3.714484679665738,
      "grad_norm": 3.9241013526916504,
      "learning_rate": 4.4899090392268335e-05,
      "loss": 2.3211,
      "step": 2667
    },
    {
      "epoch": 3.715877437325905,
      "grad_norm": 3.7552497386932373,
      "learning_rate": 4.488914155770323e-05,
      "loss": 2.1628,
      "step": 2668
    },
    {
      "epoch": 3.7172701949860727,
      "grad_norm": 3.472870349884033,
      "learning_rate": 4.487919272313814e-05,
      "loss": 2.4897,
      "step": 2669
    },
    {
      "epoch": 3.7186629526462394,
      "grad_norm": 4.359379291534424,
      "learning_rate": 4.486924388857305e-05,
      "loss": 2.8167,
      "step": 2670
    },
    {
      "epoch": 3.720055710306407,
      "grad_norm": 3.691875696182251,
      "learning_rate": 4.485929505400796e-05,
      "loss": 2.5296,
      "step": 2671
    },
    {
      "epoch": 3.7214484679665736,
      "grad_norm": 3.495535373687744,
      "learning_rate": 4.484934621944286e-05,
      "loss": 2.241,
      "step": 2672
    },
    {
      "epoch": 3.722841225626741,
      "grad_norm": 4.101537704467773,
      "learning_rate": 4.4839397384877766e-05,
      "loss": 2.2998,
      "step": 2673
    },
    {
      "epoch": 3.724233983286908,
      "grad_norm": 3.879869222640991,
      "learning_rate": 4.482944855031267e-05,
      "loss": 2.7232,
      "step": 2674
    },
    {
      "epoch": 3.7256267409470754,
      "grad_norm": 3.9464526176452637,
      "learning_rate": 4.481949971574758e-05,
      "loss": 2.5505,
      "step": 2675
    },
    {
      "epoch": 3.7270194986072425,
      "grad_norm": 4.150542736053467,
      "learning_rate": 4.4809550881182485e-05,
      "loss": 2.5747,
      "step": 2676
    },
    {
      "epoch": 3.7284122562674096,
      "grad_norm": 4.016899108886719,
      "learning_rate": 4.4799602046617396e-05,
      "loss": 2.0973,
      "step": 2677
    },
    {
      "epoch": 3.7298050139275767,
      "grad_norm": 4.207693576812744,
      "learning_rate": 4.478965321205229e-05,
      "loss": 2.2592,
      "step": 2678
    },
    {
      "epoch": 3.731197771587744,
      "grad_norm": 4.351635456085205,
      "learning_rate": 4.4779704377487204e-05,
      "loss": 2.7811,
      "step": 2679
    },
    {
      "epoch": 3.732590529247911,
      "grad_norm": 4.195699691772461,
      "learning_rate": 4.476975554292211e-05,
      "loss": 2.8685,
      "step": 2680
    },
    {
      "epoch": 3.733983286908078,
      "grad_norm": 4.452206611633301,
      "learning_rate": 4.475980670835702e-05,
      "loss": 2.619,
      "step": 2681
    },
    {
      "epoch": 3.735376044568245,
      "grad_norm": 3.9670653343200684,
      "learning_rate": 4.474985787379193e-05,
      "loss": 2.5058,
      "step": 2682
    },
    {
      "epoch": 3.7367688022284122,
      "grad_norm": 5.118613243103027,
      "learning_rate": 4.473990903922683e-05,
      "loss": 2.8105,
      "step": 2683
    },
    {
      "epoch": 3.7381615598885793,
      "grad_norm": 4.750264644622803,
      "learning_rate": 4.472996020466173e-05,
      "loss": 2.6768,
      "step": 2684
    },
    {
      "epoch": 3.7395543175487465,
      "grad_norm": 3.6492106914520264,
      "learning_rate": 4.472001137009664e-05,
      "loss": 2.5859,
      "step": 2685
    },
    {
      "epoch": 3.7409470752089136,
      "grad_norm": 3.87990403175354,
      "learning_rate": 4.471006253553155e-05,
      "loss": 2.1777,
      "step": 2686
    },
    {
      "epoch": 3.7423398328690807,
      "grad_norm": 3.9534242153167725,
      "learning_rate": 4.470011370096646e-05,
      "loss": 2.5173,
      "step": 2687
    },
    {
      "epoch": 3.743732590529248,
      "grad_norm": 6.62573766708374,
      "learning_rate": 4.469016486640136e-05,
      "loss": 2.5524,
      "step": 2688
    },
    {
      "epoch": 3.745125348189415,
      "grad_norm": 3.6964545249938965,
      "learning_rate": 4.4680216031836265e-05,
      "loss": 2.6479,
      "step": 2689
    },
    {
      "epoch": 3.7465181058495824,
      "grad_norm": 4.097850799560547,
      "learning_rate": 4.4670267197271176e-05,
      "loss": 2.2815,
      "step": 2690
    },
    {
      "epoch": 3.747910863509749,
      "grad_norm": 3.6248035430908203,
      "learning_rate": 4.466031836270608e-05,
      "loss": 2.5696,
      "step": 2691
    },
    {
      "epoch": 3.7493036211699167,
      "grad_norm": 4.852518558502197,
      "learning_rate": 4.465036952814099e-05,
      "loss": 2.3997,
      "step": 2692
    },
    {
      "epoch": 3.7506963788300833,
      "grad_norm": 3.2907462120056152,
      "learning_rate": 4.464042069357589e-05,
      "loss": 2.3746,
      "step": 2693
    },
    {
      "epoch": 3.752089136490251,
      "grad_norm": 4.214056968688965,
      "learning_rate": 4.46304718590108e-05,
      "loss": 2.8967,
      "step": 2694
    },
    {
      "epoch": 3.7534818941504176,
      "grad_norm": 3.662506341934204,
      "learning_rate": 4.46205230244457e-05,
      "loss": 2.5207,
      "step": 2695
    },
    {
      "epoch": 3.754874651810585,
      "grad_norm": 3.9540629386901855,
      "learning_rate": 4.4610574189880614e-05,
      "loss": 2.7571,
      "step": 2696
    },
    {
      "epoch": 3.756267409470752,
      "grad_norm": 3.5481412410736084,
      "learning_rate": 4.460062535531551e-05,
      "loss": 2.4145,
      "step": 2697
    },
    {
      "epoch": 3.7576601671309193,
      "grad_norm": 4.129096984863281,
      "learning_rate": 4.459067652075042e-05,
      "loss": 2.9039,
      "step": 2698
    },
    {
      "epoch": 3.7590529247910864,
      "grad_norm": 3.8313071727752686,
      "learning_rate": 4.4580727686185326e-05,
      "loss": 2.2129,
      "step": 2699
    },
    {
      "epoch": 3.7604456824512535,
      "grad_norm": 3.844362258911133,
      "learning_rate": 4.457077885162024e-05,
      "loss": 2.4664,
      "step": 2700
    },
    {
      "epoch": 3.7618384401114207,
      "grad_norm": 3.797652006149292,
      "learning_rate": 4.456083001705514e-05,
      "loss": 2.2317,
      "step": 2701
    },
    {
      "epoch": 3.7632311977715878,
      "grad_norm": 4.749051094055176,
      "learning_rate": 4.4550881182490045e-05,
      "loss": 2.5312,
      "step": 2702
    },
    {
      "epoch": 3.764623955431755,
      "grad_norm": 4.275303840637207,
      "learning_rate": 4.454093234792495e-05,
      "loss": 2.9626,
      "step": 2703
    },
    {
      "epoch": 3.766016713091922,
      "grad_norm": 4.067491054534912,
      "learning_rate": 4.453098351335986e-05,
      "loss": 2.4986,
      "step": 2704
    },
    {
      "epoch": 3.767409470752089,
      "grad_norm": 4.040477275848389,
      "learning_rate": 4.4521034678794764e-05,
      "loss": 2.6475,
      "step": 2705
    },
    {
      "epoch": 3.768802228412256,
      "grad_norm": 3.455617666244507,
      "learning_rate": 4.4511085844229675e-05,
      "loss": 2.3625,
      "step": 2706
    },
    {
      "epoch": 3.7701949860724233,
      "grad_norm": 3.8546557426452637,
      "learning_rate": 4.450113700966457e-05,
      "loss": 2.3496,
      "step": 2707
    },
    {
      "epoch": 3.7715877437325904,
      "grad_norm": 3.701049566268921,
      "learning_rate": 4.449118817509948e-05,
      "loss": 2.3824,
      "step": 2708
    },
    {
      "epoch": 3.7729805013927575,
      "grad_norm": 4.046526908874512,
      "learning_rate": 4.448123934053439e-05,
      "loss": 2.5241,
      "step": 2709
    },
    {
      "epoch": 3.7743732590529246,
      "grad_norm": 4.742768287658691,
      "learning_rate": 4.44712905059693e-05,
      "loss": 2.7906,
      "step": 2710
    },
    {
      "epoch": 3.775766016713092,
      "grad_norm": 5.611998558044434,
      "learning_rate": 4.446134167140421e-05,
      "loss": 2.3346,
      "step": 2711
    },
    {
      "epoch": 3.777158774373259,
      "grad_norm": 4.185859203338623,
      "learning_rate": 4.4451392836839106e-05,
      "loss": 2.4558,
      "step": 2712
    },
    {
      "epoch": 3.7785515320334264,
      "grad_norm": 3.4776322841644287,
      "learning_rate": 4.444144400227401e-05,
      "loss": 2.036,
      "step": 2713
    },
    {
      "epoch": 3.779944289693593,
      "grad_norm": 3.5825843811035156,
      "learning_rate": 4.443149516770892e-05,
      "loss": 2.5068,
      "step": 2714
    },
    {
      "epoch": 3.7813370473537606,
      "grad_norm": 3.74259352684021,
      "learning_rate": 4.442154633314383e-05,
      "loss": 2.3882,
      "step": 2715
    },
    {
      "epoch": 3.7827298050139273,
      "grad_norm": 4.258945465087891,
      "learning_rate": 4.4411597498578736e-05,
      "loss": 2.7586,
      "step": 2716
    },
    {
      "epoch": 3.784122562674095,
      "grad_norm": 3.828115940093994,
      "learning_rate": 4.440164866401364e-05,
      "loss": 2.3588,
      "step": 2717
    },
    {
      "epoch": 3.785515320334262,
      "grad_norm": 4.125650405883789,
      "learning_rate": 4.4391699829448544e-05,
      "loss": 2.6631,
      "step": 2718
    },
    {
      "epoch": 3.786908077994429,
      "grad_norm": 4.53554630279541,
      "learning_rate": 4.4381750994883455e-05,
      "loss": 2.772,
      "step": 2719
    },
    {
      "epoch": 3.788300835654596,
      "grad_norm": 4.092480182647705,
      "learning_rate": 4.437180216031836e-05,
      "loss": 2.4638,
      "step": 2720
    },
    {
      "epoch": 3.7896935933147633,
      "grad_norm": 4.31044864654541,
      "learning_rate": 4.436185332575327e-05,
      "loss": 3.0333,
      "step": 2721
    },
    {
      "epoch": 3.7910863509749304,
      "grad_norm": 5.6687421798706055,
      "learning_rate": 4.435190449118817e-05,
      "loss": 2.5189,
      "step": 2722
    },
    {
      "epoch": 3.7924791086350975,
      "grad_norm": 3.9867589473724365,
      "learning_rate": 4.434195565662308e-05,
      "loss": 1.9807,
      "step": 2723
    },
    {
      "epoch": 3.7938718662952646,
      "grad_norm": 4.122042655944824,
      "learning_rate": 4.433200682205798e-05,
      "loss": 2.2054,
      "step": 2724
    },
    {
      "epoch": 3.7952646239554317,
      "grad_norm": 9.375899314880371,
      "learning_rate": 4.432205798749289e-05,
      "loss": 2.737,
      "step": 2725
    },
    {
      "epoch": 3.796657381615599,
      "grad_norm": 4.639389514923096,
      "learning_rate": 4.43121091529278e-05,
      "loss": 2.355,
      "step": 2726
    },
    {
      "epoch": 3.798050139275766,
      "grad_norm": 4.087195873260498,
      "learning_rate": 4.43021603183627e-05,
      "loss": 2.8663,
      "step": 2727
    },
    {
      "epoch": 3.799442896935933,
      "grad_norm": 4.093467712402344,
      "learning_rate": 4.4292211483797605e-05,
      "loss": 2.6564,
      "step": 2728
    },
    {
      "epoch": 3.8008356545961,
      "grad_norm": 4.882555961608887,
      "learning_rate": 4.4282262649232516e-05,
      "loss": 2.361,
      "step": 2729
    },
    {
      "epoch": 3.8022284122562673,
      "grad_norm": 3.6350321769714355,
      "learning_rate": 4.427231381466742e-05,
      "loss": 2.6936,
      "step": 2730
    },
    {
      "epoch": 3.8036211699164344,
      "grad_norm": 3.577855348587036,
      "learning_rate": 4.426236498010233e-05,
      "loss": 2.1247,
      "step": 2731
    },
    {
      "epoch": 3.8050139275766015,
      "grad_norm": 4.559157848358154,
      "learning_rate": 4.425241614553723e-05,
      "loss": 2.8159,
      "step": 2732
    },
    {
      "epoch": 3.8064066852367686,
      "grad_norm": 4.250999927520752,
      "learning_rate": 4.424246731097214e-05,
      "loss": 2.4847,
      "step": 2733
    },
    {
      "epoch": 3.807799442896936,
      "grad_norm": 3.956256151199341,
      "learning_rate": 4.4232518476407044e-05,
      "loss": 2.5316,
      "step": 2734
    },
    {
      "epoch": 3.809192200557103,
      "grad_norm": 4.939937114715576,
      "learning_rate": 4.4222569641841954e-05,
      "loss": 2.788,
      "step": 2735
    },
    {
      "epoch": 3.8105849582172704,
      "grad_norm": 3.885408401489258,
      "learning_rate": 4.421262080727686e-05,
      "loss": 2.5442,
      "step": 2736
    },
    {
      "epoch": 3.811977715877437,
      "grad_norm": 3.944845676422119,
      "learning_rate": 4.420267197271176e-05,
      "loss": 2.247,
      "step": 2737
    },
    {
      "epoch": 3.8133704735376046,
      "grad_norm": 4.235049247741699,
      "learning_rate": 4.4192723138146667e-05,
      "loss": 2.3736,
      "step": 2738
    },
    {
      "epoch": 3.8147632311977717,
      "grad_norm": 3.783020257949829,
      "learning_rate": 4.418277430358158e-05,
      "loss": 2.8785,
      "step": 2739
    },
    {
      "epoch": 3.816155988857939,
      "grad_norm": 3.18996524810791,
      "learning_rate": 4.417282546901649e-05,
      "loss": 2.0317,
      "step": 2740
    },
    {
      "epoch": 3.817548746518106,
      "grad_norm": 3.9531784057617188,
      "learning_rate": 4.4162876634451386e-05,
      "loss": 2.6962,
      "step": 2741
    },
    {
      "epoch": 3.818941504178273,
      "grad_norm": 5.225775241851807,
      "learning_rate": 4.415292779988629e-05,
      "loss": 2.593,
      "step": 2742
    },
    {
      "epoch": 3.82033426183844,
      "grad_norm": 4.643198490142822,
      "learning_rate": 4.41429789653212e-05,
      "loss": 2.5271,
      "step": 2743
    },
    {
      "epoch": 3.8217270194986073,
      "grad_norm": 4.245351791381836,
      "learning_rate": 4.413303013075611e-05,
      "loss": 2.7188,
      "step": 2744
    },
    {
      "epoch": 3.8231197771587744,
      "grad_norm": 3.899301767349243,
      "learning_rate": 4.4123081296191015e-05,
      "loss": 2.2749,
      "step": 2745
    },
    {
      "epoch": 3.8245125348189415,
      "grad_norm": 3.4520652294158936,
      "learning_rate": 4.411313246162592e-05,
      "loss": 2.0981,
      "step": 2746
    },
    {
      "epoch": 3.8259052924791086,
      "grad_norm": 4.36779260635376,
      "learning_rate": 4.4103183627060824e-05,
      "loss": 2.4021,
      "step": 2747
    },
    {
      "epoch": 3.8272980501392757,
      "grad_norm": 4.111340045928955,
      "learning_rate": 4.4093234792495734e-05,
      "loss": 2.5331,
      "step": 2748
    },
    {
      "epoch": 3.828690807799443,
      "grad_norm": 4.206274032592773,
      "learning_rate": 4.408328595793064e-05,
      "loss": 2.6779,
      "step": 2749
    },
    {
      "epoch": 3.83008356545961,
      "grad_norm": 3.6885695457458496,
      "learning_rate": 4.407333712336555e-05,
      "loss": 2.2924,
      "step": 2750
    },
    {
      "epoch": 3.831476323119777,
      "grad_norm": 4.14463472366333,
      "learning_rate": 4.406338828880045e-05,
      "loss": 2.731,
      "step": 2751
    },
    {
      "epoch": 3.832869080779944,
      "grad_norm": 4.981577396392822,
      "learning_rate": 4.405343945423536e-05,
      "loss": 2.5577,
      "step": 2752
    },
    {
      "epoch": 3.8342618384401113,
      "grad_norm": 4.107276439666748,
      "learning_rate": 4.404349061967026e-05,
      "loss": 2.5428,
      "step": 2753
    },
    {
      "epoch": 3.8356545961002784,
      "grad_norm": 3.483412027359009,
      "learning_rate": 4.403354178510517e-05,
      "loss": 2.1633,
      "step": 2754
    },
    {
      "epoch": 3.837047353760446,
      "grad_norm": 3.5000691413879395,
      "learning_rate": 4.4023592950540077e-05,
      "loss": 2.2543,
      "step": 2755
    },
    {
      "epoch": 3.8384401114206126,
      "grad_norm": 3.731545925140381,
      "learning_rate": 4.401364411597498e-05,
      "loss": 2.678,
      "step": 2756
    },
    {
      "epoch": 3.83983286908078,
      "grad_norm": 3.8793838024139404,
      "learning_rate": 4.4003695281409885e-05,
      "loss": 2.1981,
      "step": 2757
    },
    {
      "epoch": 3.841225626740947,
      "grad_norm": 3.9608139991760254,
      "learning_rate": 4.3993746446844796e-05,
      "loss": 2.4132,
      "step": 2758
    },
    {
      "epoch": 3.8426183844011144,
      "grad_norm": 5.246201515197754,
      "learning_rate": 4.39837976122797e-05,
      "loss": 2.4739,
      "step": 2759
    },
    {
      "epoch": 3.8440111420612815,
      "grad_norm": 3.9054718017578125,
      "learning_rate": 4.397384877771461e-05,
      "loss": 2.4072,
      "step": 2760
    },
    {
      "epoch": 3.8454038997214486,
      "grad_norm": 3.5004072189331055,
      "learning_rate": 4.396389994314951e-05,
      "loss": 2.2678,
      "step": 2761
    },
    {
      "epoch": 3.8467966573816157,
      "grad_norm": 4.606179237365723,
      "learning_rate": 4.395395110858442e-05,
      "loss": 2.3299,
      "step": 2762
    },
    {
      "epoch": 3.848189415041783,
      "grad_norm": 3.8552868366241455,
      "learning_rate": 4.394400227401932e-05,
      "loss": 2.9338,
      "step": 2763
    },
    {
      "epoch": 3.84958217270195,
      "grad_norm": 3.881601095199585,
      "learning_rate": 4.3934053439454234e-05,
      "loss": 2.5753,
      "step": 2764
    },
    {
      "epoch": 3.850974930362117,
      "grad_norm": 4.5721659660339355,
      "learning_rate": 4.392410460488914e-05,
      "loss": 2.6409,
      "step": 2765
    },
    {
      "epoch": 3.852367688022284,
      "grad_norm": 4.485108852386475,
      "learning_rate": 4.391415577032404e-05,
      "loss": 2.3553,
      "step": 2766
    },
    {
      "epoch": 3.8537604456824512,
      "grad_norm": 4.302353858947754,
      "learning_rate": 4.3904206935758946e-05,
      "loss": 2.3465,
      "step": 2767
    },
    {
      "epoch": 3.8551532033426184,
      "grad_norm": 3.646881580352783,
      "learning_rate": 4.3894258101193857e-05,
      "loss": 2.2946,
      "step": 2768
    },
    {
      "epoch": 3.8565459610027855,
      "grad_norm": 4.009352207183838,
      "learning_rate": 4.388430926662877e-05,
      "loss": 2.5868,
      "step": 2769
    },
    {
      "epoch": 3.8579387186629526,
      "grad_norm": 4.348059177398682,
      "learning_rate": 4.387436043206367e-05,
      "loss": 2.1755,
      "step": 2770
    },
    {
      "epoch": 3.8593314763231197,
      "grad_norm": 3.8434622287750244,
      "learning_rate": 4.386441159749857e-05,
      "loss": 2.3478,
      "step": 2771
    },
    {
      "epoch": 3.860724233983287,
      "grad_norm": 5.061528205871582,
      "learning_rate": 4.385446276293348e-05,
      "loss": 2.6413,
      "step": 2772
    },
    {
      "epoch": 3.862116991643454,
      "grad_norm": 4.642963886260986,
      "learning_rate": 4.384451392836839e-05,
      "loss": 2.2953,
      "step": 2773
    },
    {
      "epoch": 3.863509749303621,
      "grad_norm": 4.183655261993408,
      "learning_rate": 4.3834565093803295e-05,
      "loss": 2.6933,
      "step": 2774
    },
    {
      "epoch": 3.864902506963788,
      "grad_norm": 3.5605030059814453,
      "learning_rate": 4.3824616259238205e-05,
      "loss": 2.4551,
      "step": 2775
    },
    {
      "epoch": 3.8662952646239557,
      "grad_norm": 4.208330154418945,
      "learning_rate": 4.38146674246731e-05,
      "loss": 2.7612,
      "step": 2776
    },
    {
      "epoch": 3.8676880222841223,
      "grad_norm": 4.467424392700195,
      "learning_rate": 4.3804718590108014e-05,
      "loss": 2.3246,
      "step": 2777
    },
    {
      "epoch": 3.86908077994429,
      "grad_norm": 3.943737268447876,
      "learning_rate": 4.379476975554292e-05,
      "loss": 2.2271,
      "step": 2778
    },
    {
      "epoch": 3.8704735376044566,
      "grad_norm": 3.7545974254608154,
      "learning_rate": 4.378482092097783e-05,
      "loss": 2.4962,
      "step": 2779
    },
    {
      "epoch": 3.871866295264624,
      "grad_norm": 4.182112693786621,
      "learning_rate": 4.377487208641273e-05,
      "loss": 2.8764,
      "step": 2780
    },
    {
      "epoch": 3.8732590529247912,
      "grad_norm": 4.237875461578369,
      "learning_rate": 4.376492325184764e-05,
      "loss": 2.3709,
      "step": 2781
    },
    {
      "epoch": 3.8746518105849583,
      "grad_norm": 4.057324409484863,
      "learning_rate": 4.375497441728254e-05,
      "loss": 2.5617,
      "step": 2782
    },
    {
      "epoch": 3.8760445682451254,
      "grad_norm": 4.075103282928467,
      "learning_rate": 4.374502558271745e-05,
      "loss": 2.3773,
      "step": 2783
    },
    {
      "epoch": 3.8774373259052926,
      "grad_norm": 4.05328893661499,
      "learning_rate": 4.3735076748152356e-05,
      "loss": 2.5776,
      "step": 2784
    },
    {
      "epoch": 3.8788300835654597,
      "grad_norm": 3.459596633911133,
      "learning_rate": 4.372512791358726e-05,
      "loss": 2.2137,
      "step": 2785
    },
    {
      "epoch": 3.8802228412256268,
      "grad_norm": 3.943903923034668,
      "learning_rate": 4.3715179079022164e-05,
      "loss": 2.2072,
      "step": 2786
    },
    {
      "epoch": 3.881615598885794,
      "grad_norm": 3.6659719944000244,
      "learning_rate": 4.3705230244457075e-05,
      "loss": 2.2368,
      "step": 2787
    },
    {
      "epoch": 3.883008356545961,
      "grad_norm": 3.9039340019226074,
      "learning_rate": 4.369528140989198e-05,
      "loss": 2.3746,
      "step": 2788
    },
    {
      "epoch": 3.884401114206128,
      "grad_norm": 3.675025701522827,
      "learning_rate": 4.368533257532689e-05,
      "loss": 2.4561,
      "step": 2789
    },
    {
      "epoch": 3.885793871866295,
      "grad_norm": 4.125492572784424,
      "learning_rate": 4.367538374076179e-05,
      "loss": 2.2744,
      "step": 2790
    },
    {
      "epoch": 3.8871866295264623,
      "grad_norm": 5.663262844085693,
      "learning_rate": 4.36654349061967e-05,
      "loss": 2.8396,
      "step": 2791
    },
    {
      "epoch": 3.8885793871866294,
      "grad_norm": 3.851743459701538,
      "learning_rate": 4.36554860716316e-05,
      "loss": 2.2684,
      "step": 2792
    },
    {
      "epoch": 3.8899721448467965,
      "grad_norm": 3.9266207218170166,
      "learning_rate": 4.364553723706651e-05,
      "loss": 2.4336,
      "step": 2793
    },
    {
      "epoch": 3.8913649025069637,
      "grad_norm": 3.5865793228149414,
      "learning_rate": 4.3635588402501424e-05,
      "loss": 2.214,
      "step": 2794
    },
    {
      "epoch": 3.8927576601671308,
      "grad_norm": 3.5342860221862793,
      "learning_rate": 4.362563956793632e-05,
      "loss": 2.462,
      "step": 2795
    },
    {
      "epoch": 3.894150417827298,
      "grad_norm": 3.487074375152588,
      "learning_rate": 4.3615690733371225e-05,
      "loss": 2.2586,
      "step": 2796
    },
    {
      "epoch": 3.8955431754874654,
      "grad_norm": 3.9496841430664062,
      "learning_rate": 4.3605741898806136e-05,
      "loss": 2.6275,
      "step": 2797
    },
    {
      "epoch": 3.896935933147632,
      "grad_norm": 4.085710048675537,
      "learning_rate": 4.3595793064241047e-05,
      "loss": 2.3676,
      "step": 2798
    },
    {
      "epoch": 3.8983286908077996,
      "grad_norm": 4.594913959503174,
      "learning_rate": 4.358584422967595e-05,
      "loss": 2.553,
      "step": 2799
    },
    {
      "epoch": 3.8997214484679663,
      "grad_norm": 3.9324839115142822,
      "learning_rate": 4.3575895395110855e-05,
      "loss": 2.8682,
      "step": 2800
    },
    {
      "epoch": 3.901114206128134,
      "grad_norm": 3.9619557857513428,
      "learning_rate": 4.356594656054576e-05,
      "loss": 2.2718,
      "step": 2801
    },
    {
      "epoch": 3.902506963788301,
      "grad_norm": 4.3043084144592285,
      "learning_rate": 4.355599772598067e-05,
      "loss": 2.6799,
      "step": 2802
    },
    {
      "epoch": 3.903899721448468,
      "grad_norm": 5.833071231842041,
      "learning_rate": 4.3546048891415574e-05,
      "loss": 2.2675,
      "step": 2803
    },
    {
      "epoch": 3.905292479108635,
      "grad_norm": 4.207306385040283,
      "learning_rate": 4.3536100056850485e-05,
      "loss": 2.7466,
      "step": 2804
    },
    {
      "epoch": 3.9066852367688023,
      "grad_norm": 3.897665500640869,
      "learning_rate": 4.352615122228538e-05,
      "loss": 2.5291,
      "step": 2805
    },
    {
      "epoch": 3.9080779944289694,
      "grad_norm": 4.754603862762451,
      "learning_rate": 4.351620238772029e-05,
      "loss": 2.1749,
      "step": 2806
    },
    {
      "epoch": 3.9094707520891365,
      "grad_norm": 6.453915119171143,
      "learning_rate": 4.35062535531552e-05,
      "loss": 2.2067,
      "step": 2807
    },
    {
      "epoch": 3.9108635097493036,
      "grad_norm": 3.9593875408172607,
      "learning_rate": 4.349630471859011e-05,
      "loss": 2.1644,
      "step": 2808
    },
    {
      "epoch": 3.9122562674094707,
      "grad_norm": 3.7252540588378906,
      "learning_rate": 4.348635588402501e-05,
      "loss": 2.5891,
      "step": 2809
    },
    {
      "epoch": 3.913649025069638,
      "grad_norm": 3.7336575984954834,
      "learning_rate": 4.3476407049459916e-05,
      "loss": 2.6783,
      "step": 2810
    },
    {
      "epoch": 3.915041782729805,
      "grad_norm": 3.7686209678649902,
      "learning_rate": 4.346645821489482e-05,
      "loss": 2.2132,
      "step": 2811
    },
    {
      "epoch": 3.916434540389972,
      "grad_norm": 4.698162078857422,
      "learning_rate": 4.345650938032973e-05,
      "loss": 2.2761,
      "step": 2812
    },
    {
      "epoch": 3.917827298050139,
      "grad_norm": 4.409482955932617,
      "learning_rate": 4.3446560545764635e-05,
      "loss": 2.3576,
      "step": 2813
    },
    {
      "epoch": 3.9192200557103063,
      "grad_norm": 4.060924053192139,
      "learning_rate": 4.3436611711199546e-05,
      "loss": 2.5578,
      "step": 2814
    },
    {
      "epoch": 3.9206128133704734,
      "grad_norm": 3.9536235332489014,
      "learning_rate": 4.342666287663444e-05,
      "loss": 2.3978,
      "step": 2815
    },
    {
      "epoch": 3.9220055710306405,
      "grad_norm": 3.7282447814941406,
      "learning_rate": 4.3416714042069354e-05,
      "loss": 2.2678,
      "step": 2816
    },
    {
      "epoch": 3.9233983286908076,
      "grad_norm": 3.6149067878723145,
      "learning_rate": 4.340676520750426e-05,
      "loss": 2.371,
      "step": 2817
    },
    {
      "epoch": 3.924791086350975,
      "grad_norm": 3.990464210510254,
      "learning_rate": 4.339681637293917e-05,
      "loss": 2.5115,
      "step": 2818
    },
    {
      "epoch": 3.926183844011142,
      "grad_norm": 3.926725387573242,
      "learning_rate": 4.338686753837407e-05,
      "loss": 2.4013,
      "step": 2819
    },
    {
      "epoch": 3.9275766016713094,
      "grad_norm": 4.2266645431518555,
      "learning_rate": 4.337691870380898e-05,
      "loss": 2.46,
      "step": 2820
    },
    {
      "epoch": 3.928969359331476,
      "grad_norm": 4.060005187988281,
      "learning_rate": 4.336696986924388e-05,
      "loss": 2.6188,
      "step": 2821
    },
    {
      "epoch": 3.9303621169916436,
      "grad_norm": 4.332579612731934,
      "learning_rate": 4.335702103467879e-05,
      "loss": 2.2021,
      "step": 2822
    },
    {
      "epoch": 3.9317548746518107,
      "grad_norm": 3.5953142642974854,
      "learning_rate": 4.33470722001137e-05,
      "loss": 2.3435,
      "step": 2823
    },
    {
      "epoch": 3.933147632311978,
      "grad_norm": 4.64898681640625,
      "learning_rate": 4.333712336554861e-05,
      "loss": 2.4961,
      "step": 2824
    },
    {
      "epoch": 3.934540389972145,
      "grad_norm": 4.039651393890381,
      "learning_rate": 4.3327174530983504e-05,
      "loss": 2.0873,
      "step": 2825
    },
    {
      "epoch": 3.935933147632312,
      "grad_norm": 3.3444809913635254,
      "learning_rate": 4.3317225696418415e-05,
      "loss": 1.8563,
      "step": 2826
    },
    {
      "epoch": 3.937325905292479,
      "grad_norm": 3.111057996749878,
      "learning_rate": 4.3307276861853326e-05,
      "loss": 2.0835,
      "step": 2827
    },
    {
      "epoch": 3.9387186629526463,
      "grad_norm": 3.7342870235443115,
      "learning_rate": 4.329732802728823e-05,
      "loss": 2.1274,
      "step": 2828
    },
    {
      "epoch": 3.9401114206128134,
      "grad_norm": 4.010530948638916,
      "learning_rate": 4.3287379192723134e-05,
      "loss": 2.5928,
      "step": 2829
    },
    {
      "epoch": 3.9415041782729805,
      "grad_norm": 4.297623157501221,
      "learning_rate": 4.327743035815804e-05,
      "loss": 2.4479,
      "step": 2830
    },
    {
      "epoch": 3.9428969359331476,
      "grad_norm": 3.8704824447631836,
      "learning_rate": 4.326748152359295e-05,
      "loss": 2.3586,
      "step": 2831
    },
    {
      "epoch": 3.9442896935933147,
      "grad_norm": 3.575561761856079,
      "learning_rate": 4.325753268902785e-05,
      "loss": 2.1189,
      "step": 2832
    },
    {
      "epoch": 3.945682451253482,
      "grad_norm": 4.465643405914307,
      "learning_rate": 4.3247583854462764e-05,
      "loss": 1.9393,
      "step": 2833
    },
    {
      "epoch": 3.947075208913649,
      "grad_norm": 4.366245746612549,
      "learning_rate": 4.323763501989766e-05,
      "loss": 2.1547,
      "step": 2834
    },
    {
      "epoch": 3.948467966573816,
      "grad_norm": 4.428537845611572,
      "learning_rate": 4.322768618533257e-05,
      "loss": 2.7843,
      "step": 2835
    },
    {
      "epoch": 3.949860724233983,
      "grad_norm": 4.058773994445801,
      "learning_rate": 4.3217737350767476e-05,
      "loss": 2.3258,
      "step": 2836
    },
    {
      "epoch": 3.9512534818941503,
      "grad_norm": 3.489189386367798,
      "learning_rate": 4.320778851620239e-05,
      "loss": 2.3345,
      "step": 2837
    },
    {
      "epoch": 3.9526462395543174,
      "grad_norm": 5.146334648132324,
      "learning_rate": 4.319783968163729e-05,
      "loss": 2.5709,
      "step": 2838
    },
    {
      "epoch": 3.954038997214485,
      "grad_norm": 4.094646453857422,
      "learning_rate": 4.3187890847072195e-05,
      "loss": 2.3741,
      "step": 2839
    },
    {
      "epoch": 3.9554317548746516,
      "grad_norm": 3.887098789215088,
      "learning_rate": 4.31779420125071e-05,
      "loss": 2.1513,
      "step": 2840
    },
    {
      "epoch": 3.956824512534819,
      "grad_norm": 4.42372465133667,
      "learning_rate": 4.316799317794201e-05,
      "loss": 2.3141,
      "step": 2841
    },
    {
      "epoch": 3.958217270194986,
      "grad_norm": 4.165892601013184,
      "learning_rate": 4.3158044343376914e-05,
      "loss": 2.6135,
      "step": 2842
    },
    {
      "epoch": 3.9596100278551534,
      "grad_norm": 4.015235900878906,
      "learning_rate": 4.3148095508811825e-05,
      "loss": 2.4057,
      "step": 2843
    },
    {
      "epoch": 3.9610027855153205,
      "grad_norm": 4.624741554260254,
      "learning_rate": 4.313814667424672e-05,
      "loss": 2.7105,
      "step": 2844
    },
    {
      "epoch": 3.9623955431754876,
      "grad_norm": 3.801917314529419,
      "learning_rate": 4.312819783968163e-05,
      "loss": 2.603,
      "step": 2845
    },
    {
      "epoch": 3.9637883008356547,
      "grad_norm": 3.7009623050689697,
      "learning_rate": 4.311824900511654e-05,
      "loss": 2.2056,
      "step": 2846
    },
    {
      "epoch": 3.965181058495822,
      "grad_norm": 4.001925468444824,
      "learning_rate": 4.310830017055145e-05,
      "loss": 2.0946,
      "step": 2847
    },
    {
      "epoch": 3.966573816155989,
      "grad_norm": 4.103342533111572,
      "learning_rate": 4.309835133598635e-05,
      "loss": 2.8683,
      "step": 2848
    },
    {
      "epoch": 3.967966573816156,
      "grad_norm": 4.002389430999756,
      "learning_rate": 4.3088402501421256e-05,
      "loss": 2.6372,
      "step": 2849
    },
    {
      "epoch": 3.969359331476323,
      "grad_norm": 3.70853590965271,
      "learning_rate": 4.307845366685616e-05,
      "loss": 1.8587,
      "step": 2850
    },
    {
      "epoch": 3.9707520891364902,
      "grad_norm": 3.8358755111694336,
      "learning_rate": 4.306850483229107e-05,
      "loss": 2.5599,
      "step": 2851
    },
    {
      "epoch": 3.9721448467966574,
      "grad_norm": 4.751933574676514,
      "learning_rate": 4.305855599772598e-05,
      "loss": 2.4604,
      "step": 2852
    },
    {
      "epoch": 3.9735376044568245,
      "grad_norm": 3.9108285903930664,
      "learning_rate": 4.3048607163160886e-05,
      "loss": 2.5878,
      "step": 2853
    },
    {
      "epoch": 3.9749303621169916,
      "grad_norm": 4.267486095428467,
      "learning_rate": 4.303865832859578e-05,
      "loss": 2.4386,
      "step": 2854
    },
    {
      "epoch": 3.9763231197771587,
      "grad_norm": 4.0073323249816895,
      "learning_rate": 4.3028709494030694e-05,
      "loss": 2.6417,
      "step": 2855
    },
    {
      "epoch": 3.977715877437326,
      "grad_norm": 3.7361462116241455,
      "learning_rate": 4.3018760659465605e-05,
      "loss": 2.5317,
      "step": 2856
    },
    {
      "epoch": 3.979108635097493,
      "grad_norm": 5.070940017700195,
      "learning_rate": 4.300881182490051e-05,
      "loss": 2.4261,
      "step": 2857
    },
    {
      "epoch": 3.98050139275766,
      "grad_norm": 7.305459976196289,
      "learning_rate": 4.299886299033542e-05,
      "loss": 2.2394,
      "step": 2858
    },
    {
      "epoch": 3.981894150417827,
      "grad_norm": 5.064977169036865,
      "learning_rate": 4.298891415577032e-05,
      "loss": 2.8809,
      "step": 2859
    },
    {
      "epoch": 3.9832869080779947,
      "grad_norm": 4.64074182510376,
      "learning_rate": 4.297896532120523e-05,
      "loss": 2.3022,
      "step": 2860
    },
    {
      "epoch": 3.9846796657381613,
      "grad_norm": 4.33528995513916,
      "learning_rate": 4.296901648664013e-05,
      "loss": 2.1313,
      "step": 2861
    },
    {
      "epoch": 3.986072423398329,
      "grad_norm": 3.887798547744751,
      "learning_rate": 4.295906765207504e-05,
      "loss": 2.212,
      "step": 2862
    },
    {
      "epoch": 3.9874651810584956,
      "grad_norm": 3.91961407661438,
      "learning_rate": 4.294911881750995e-05,
      "loss": 2.2057,
      "step": 2863
    },
    {
      "epoch": 3.988857938718663,
      "grad_norm": 3.9067459106445312,
      "learning_rate": 4.293916998294485e-05,
      "loss": 2.5234,
      "step": 2864
    },
    {
      "epoch": 3.9902506963788302,
      "grad_norm": 3.912951707839966,
      "learning_rate": 4.2929221148379755e-05,
      "loss": 2.2181,
      "step": 2865
    },
    {
      "epoch": 3.9916434540389973,
      "grad_norm": 3.513366937637329,
      "learning_rate": 4.2919272313814666e-05,
      "loss": 2.3435,
      "step": 2866
    },
    {
      "epoch": 3.9930362116991645,
      "grad_norm": 3.928295135498047,
      "learning_rate": 4.290932347924957e-05,
      "loss": 2.4312,
      "step": 2867
    },
    {
      "epoch": 3.9944289693593316,
      "grad_norm": 4.228482246398926,
      "learning_rate": 4.289937464468448e-05,
      "loss": 2.7527,
      "step": 2868
    },
    {
      "epoch": 3.9958217270194987,
      "grad_norm": 4.201807022094727,
      "learning_rate": 4.288942581011938e-05,
      "loss": 2.5804,
      "step": 2869
    },
    {
      "epoch": 3.997214484679666,
      "grad_norm": 3.8392558097839355,
      "learning_rate": 4.287947697555429e-05,
      "loss": 2.3846,
      "step": 2870
    },
    {
      "epoch": 3.998607242339833,
      "grad_norm": 4.189210891723633,
      "learning_rate": 4.286952814098919e-05,
      "loss": 2.2174,
      "step": 2871
    },
    {
      "epoch": 4.0,
      "grad_norm": 4.876611232757568,
      "learning_rate": 4.2859579306424104e-05,
      "loss": 2.778,
      "step": 2872
    },
    {
      "epoch": 4.0013927576601676,
      "grad_norm": 3.852813720703125,
      "learning_rate": 4.2849630471859e-05,
      "loss": 2.2284,
      "step": 2873
    },
    {
      "epoch": 4.002785515320334,
      "grad_norm": 3.3349533081054688,
      "learning_rate": 4.283968163729391e-05,
      "loss": 2.0552,
      "step": 2874
    },
    {
      "epoch": 4.004178272980502,
      "grad_norm": 3.858365058898926,
      "learning_rate": 4.2829732802728816e-05,
      "loss": 2.1243,
      "step": 2875
    },
    {
      "epoch": 4.005571030640668,
      "grad_norm": 3.3642642498016357,
      "learning_rate": 4.281978396816373e-05,
      "loss": 1.8814,
      "step": 2876
    },
    {
      "epoch": 4.006963788300836,
      "grad_norm": 4.753489971160889,
      "learning_rate": 4.280983513359863e-05,
      "loss": 2.1598,
      "step": 2877
    },
    {
      "epoch": 4.008356545961003,
      "grad_norm": 4.414871692657471,
      "learning_rate": 4.2799886299033535e-05,
      "loss": 1.9104,
      "step": 2878
    },
    {
      "epoch": 4.00974930362117,
      "grad_norm": 3.775564432144165,
      "learning_rate": 4.278993746446844e-05,
      "loss": 1.8956,
      "step": 2879
    },
    {
      "epoch": 4.011142061281337,
      "grad_norm": 3.8523037433624268,
      "learning_rate": 4.277998862990335e-05,
      "loss": 1.9274,
      "step": 2880
    },
    {
      "epoch": 4.012534818941504,
      "grad_norm": 3.4568839073181152,
      "learning_rate": 4.277003979533826e-05,
      "loss": 1.9917,
      "step": 2881
    },
    {
      "epoch": 4.013927576601671,
      "grad_norm": 3.492802381515503,
      "learning_rate": 4.2760090960773165e-05,
      "loss": 2.1702,
      "step": 2882
    },
    {
      "epoch": 4.015320334261839,
      "grad_norm": 3.5977847576141357,
      "learning_rate": 4.275014212620806e-05,
      "loss": 2.2774,
      "step": 2883
    },
    {
      "epoch": 4.016713091922005,
      "grad_norm": 4.024469375610352,
      "learning_rate": 4.274019329164297e-05,
      "loss": 2.1418,
      "step": 2884
    },
    {
      "epoch": 4.018105849582173,
      "grad_norm": 3.895038366317749,
      "learning_rate": 4.2730244457077884e-05,
      "loss": 2.3048,
      "step": 2885
    },
    {
      "epoch": 4.0194986072423395,
      "grad_norm": 3.8219165802001953,
      "learning_rate": 4.272029562251279e-05,
      "loss": 1.807,
      "step": 2886
    },
    {
      "epoch": 4.020891364902507,
      "grad_norm": 3.387883424758911,
      "learning_rate": 4.27103467879477e-05,
      "loss": 1.8544,
      "step": 2887
    },
    {
      "epoch": 4.022284122562674,
      "grad_norm": 3.87650203704834,
      "learning_rate": 4.2700397953382596e-05,
      "loss": 2.0758,
      "step": 2888
    },
    {
      "epoch": 4.023676880222841,
      "grad_norm": 3.484063148498535,
      "learning_rate": 4.269044911881751e-05,
      "loss": 2.0685,
      "step": 2889
    },
    {
      "epoch": 4.025069637883008,
      "grad_norm": 3.4695568084716797,
      "learning_rate": 4.268050028425241e-05,
      "loss": 2.0025,
      "step": 2890
    },
    {
      "epoch": 4.0264623955431755,
      "grad_norm": 3.7375874519348145,
      "learning_rate": 4.267055144968732e-05,
      "loss": 2.1755,
      "step": 2891
    },
    {
      "epoch": 4.027855153203342,
      "grad_norm": 3.4818079471588135,
      "learning_rate": 4.2660602615122226e-05,
      "loss": 1.8245,
      "step": 2892
    },
    {
      "epoch": 4.02924791086351,
      "grad_norm": 3.2014174461364746,
      "learning_rate": 4.265065378055713e-05,
      "loss": 1.7686,
      "step": 2893
    },
    {
      "epoch": 4.030640668523677,
      "grad_norm": 3.6586625576019287,
      "learning_rate": 4.2640704945992034e-05,
      "loss": 2.2697,
      "step": 2894
    },
    {
      "epoch": 4.032033426183844,
      "grad_norm": 4.191351890563965,
      "learning_rate": 4.2630756111426945e-05,
      "loss": 2.1766,
      "step": 2895
    },
    {
      "epoch": 4.0334261838440115,
      "grad_norm": 3.260627031326294,
      "learning_rate": 4.262080727686185e-05,
      "loss": 2.201,
      "step": 2896
    },
    {
      "epoch": 4.034818941504178,
      "grad_norm": 3.9480907917022705,
      "learning_rate": 4.261085844229676e-05,
      "loss": 2.0782,
      "step": 2897
    },
    {
      "epoch": 4.036211699164346,
      "grad_norm": 3.4172251224517822,
      "learning_rate": 4.260090960773166e-05,
      "loss": 1.9945,
      "step": 2898
    },
    {
      "epoch": 4.037604456824512,
      "grad_norm": 3.900606393814087,
      "learning_rate": 4.259096077316657e-05,
      "loss": 2.163,
      "step": 2899
    },
    {
      "epoch": 4.03899721448468,
      "grad_norm": 4.304780006408691,
      "learning_rate": 4.258101193860147e-05,
      "loss": 2.3168,
      "step": 2900
    },
    {
      "epoch": 4.040389972144847,
      "grad_norm": 3.3004045486450195,
      "learning_rate": 4.257106310403638e-05,
      "loss": 2.1544,
      "step": 2901
    },
    {
      "epoch": 4.041782729805014,
      "grad_norm": 3.5081892013549805,
      "learning_rate": 4.256111426947129e-05,
      "loss": 2.0612,
      "step": 2902
    },
    {
      "epoch": 4.043175487465181,
      "grad_norm": 3.855876922607422,
      "learning_rate": 4.255116543490619e-05,
      "loss": 2.1881,
      "step": 2903
    },
    {
      "epoch": 4.044568245125348,
      "grad_norm": 3.5717360973358154,
      "learning_rate": 4.2541216600341095e-05,
      "loss": 1.8758,
      "step": 2904
    },
    {
      "epoch": 4.045961002785515,
      "grad_norm": 3.3564541339874268,
      "learning_rate": 4.2531267765776006e-05,
      "loss": 1.6998,
      "step": 2905
    },
    {
      "epoch": 4.047353760445683,
      "grad_norm": 3.6321773529052734,
      "learning_rate": 4.252131893121092e-05,
      "loss": 1.789,
      "step": 2906
    },
    {
      "epoch": 4.048746518105849,
      "grad_norm": 3.646332263946533,
      "learning_rate": 4.251137009664582e-05,
      "loss": 1.8159,
      "step": 2907
    },
    {
      "epoch": 4.050139275766017,
      "grad_norm": 3.75654673576355,
      "learning_rate": 4.250142126208072e-05,
      "loss": 1.9863,
      "step": 2908
    },
    {
      "epoch": 4.0515320334261835,
      "grad_norm": 3.1938226222991943,
      "learning_rate": 4.249147242751563e-05,
      "loss": 1.7361,
      "step": 2909
    },
    {
      "epoch": 4.052924791086351,
      "grad_norm": 4.553583145141602,
      "learning_rate": 4.248152359295054e-05,
      "loss": 2.097,
      "step": 2910
    },
    {
      "epoch": 4.054317548746518,
      "grad_norm": 3.4492368698120117,
      "learning_rate": 4.2471574758385444e-05,
      "loss": 2.1103,
      "step": 2911
    },
    {
      "epoch": 4.055710306406685,
      "grad_norm": 3.943769693374634,
      "learning_rate": 4.246162592382034e-05,
      "loss": 2.1472,
      "step": 2912
    },
    {
      "epoch": 4.057103064066852,
      "grad_norm": 3.1672894954681396,
      "learning_rate": 4.245167708925525e-05,
      "loss": 1.9289,
      "step": 2913
    },
    {
      "epoch": 4.0584958217270195,
      "grad_norm": 3.6658668518066406,
      "learning_rate": 4.244172825469016e-05,
      "loss": 2.1979,
      "step": 2914
    },
    {
      "epoch": 4.059888579387187,
      "grad_norm": 3.32881236076355,
      "learning_rate": 4.243177942012507e-05,
      "loss": 1.8956,
      "step": 2915
    },
    {
      "epoch": 4.061281337047354,
      "grad_norm": 3.316603660583496,
      "learning_rate": 4.242183058555998e-05,
      "loss": 1.9835,
      "step": 2916
    },
    {
      "epoch": 4.062674094707521,
      "grad_norm": 3.3299334049224854,
      "learning_rate": 4.2411881750994875e-05,
      "loss": 1.7492,
      "step": 2917
    },
    {
      "epoch": 4.064066852367688,
      "grad_norm": 4.63602876663208,
      "learning_rate": 4.2401932916429786e-05,
      "loss": 2.1778,
      "step": 2918
    },
    {
      "epoch": 4.0654596100278555,
      "grad_norm": 3.0447895526885986,
      "learning_rate": 4.239198408186469e-05,
      "loss": 1.9583,
      "step": 2919
    },
    {
      "epoch": 4.066852367688022,
      "grad_norm": 3.6520590782165527,
      "learning_rate": 4.23820352472996e-05,
      "loss": 2.0538,
      "step": 2920
    },
    {
      "epoch": 4.06824512534819,
      "grad_norm": 4.1695146560668945,
      "learning_rate": 4.2372086412734505e-05,
      "loss": 2.5048,
      "step": 2921
    },
    {
      "epoch": 4.069637883008356,
      "grad_norm": 3.669248104095459,
      "learning_rate": 4.236213757816941e-05,
      "loss": 2.1274,
      "step": 2922
    },
    {
      "epoch": 4.071030640668524,
      "grad_norm": 3.8280062675476074,
      "learning_rate": 4.2352188743604313e-05,
      "loss": 2.0354,
      "step": 2923
    },
    {
      "epoch": 4.072423398328691,
      "grad_norm": 3.502394199371338,
      "learning_rate": 4.2342239909039224e-05,
      "loss": 2.1068,
      "step": 2924
    },
    {
      "epoch": 4.073816155988858,
      "grad_norm": 3.5648293495178223,
      "learning_rate": 4.233229107447413e-05,
      "loss": 2.4372,
      "step": 2925
    },
    {
      "epoch": 4.075208913649025,
      "grad_norm": 6.297248840332031,
      "learning_rate": 4.232234223990904e-05,
      "loss": 2.0153,
      "step": 2926
    },
    {
      "epoch": 4.076601671309192,
      "grad_norm": 3.234280586242676,
      "learning_rate": 4.2312393405343937e-05,
      "loss": 1.8531,
      "step": 2927
    },
    {
      "epoch": 4.077994428969359,
      "grad_norm": 7.501010894775391,
      "learning_rate": 4.230244457077885e-05,
      "loss": 1.9979,
      "step": 2928
    },
    {
      "epoch": 4.079387186629527,
      "grad_norm": 4.206099510192871,
      "learning_rate": 4.229249573621375e-05,
      "loss": 2.0006,
      "step": 2929
    },
    {
      "epoch": 4.080779944289693,
      "grad_norm": 3.550398349761963,
      "learning_rate": 4.228254690164866e-05,
      "loss": 2.3617,
      "step": 2930
    },
    {
      "epoch": 4.082172701949861,
      "grad_norm": 3.0064449310302734,
      "learning_rate": 4.2272598067083566e-05,
      "loss": 1.71,
      "step": 2931
    },
    {
      "epoch": 4.0835654596100275,
      "grad_norm": 3.389880418777466,
      "learning_rate": 4.226264923251847e-05,
      "loss": 1.7805,
      "step": 2932
    },
    {
      "epoch": 4.084958217270195,
      "grad_norm": 3.2349741458892822,
      "learning_rate": 4.2252700397953375e-05,
      "loss": 1.8118,
      "step": 2933
    },
    {
      "epoch": 4.086350974930362,
      "grad_norm": 4.4300432205200195,
      "learning_rate": 4.2242751563388285e-05,
      "loss": 2.3677,
      "step": 2934
    },
    {
      "epoch": 4.087743732590529,
      "grad_norm": 3.7126057147979736,
      "learning_rate": 4.2232802728823196e-05,
      "loss": 2.4539,
      "step": 2935
    },
    {
      "epoch": 4.089136490250697,
      "grad_norm": 3.426689386367798,
      "learning_rate": 4.22228538942581e-05,
      "loss": 2.1186,
      "step": 2936
    },
    {
      "epoch": 4.0905292479108635,
      "grad_norm": 3.7912368774414062,
      "learning_rate": 4.2212905059693e-05,
      "loss": 1.9035,
      "step": 2937
    },
    {
      "epoch": 4.091922005571031,
      "grad_norm": 3.738041877746582,
      "learning_rate": 4.220295622512791e-05,
      "loss": 2.0842,
      "step": 2938
    },
    {
      "epoch": 4.093314763231198,
      "grad_norm": 5.30759334564209,
      "learning_rate": 4.219300739056282e-05,
      "loss": 1.9049,
      "step": 2939
    },
    {
      "epoch": 4.094707520891365,
      "grad_norm": 5.949930191040039,
      "learning_rate": 4.2183058555997723e-05,
      "loss": 2.1455,
      "step": 2940
    },
    {
      "epoch": 4.096100278551532,
      "grad_norm": 3.648623466491699,
      "learning_rate": 4.2173109721432634e-05,
      "loss": 2.1864,
      "step": 2941
    },
    {
      "epoch": 4.0974930362116995,
      "grad_norm": 5.2805328369140625,
      "learning_rate": 4.216316088686753e-05,
      "loss": 1.8307,
      "step": 2942
    },
    {
      "epoch": 4.098885793871866,
      "grad_norm": 3.3273491859436035,
      "learning_rate": 4.215321205230244e-05,
      "loss": 1.9127,
      "step": 2943
    },
    {
      "epoch": 4.100278551532034,
      "grad_norm": 3.942098379135132,
      "learning_rate": 4.2143263217737346e-05,
      "loss": 2.0882,
      "step": 2944
    },
    {
      "epoch": 4.1016713091922,
      "grad_norm": 3.925662040710449,
      "learning_rate": 4.213331438317226e-05,
      "loss": 2.1373,
      "step": 2945
    },
    {
      "epoch": 4.103064066852368,
      "grad_norm": 3.572674036026001,
      "learning_rate": 4.212336554860716e-05,
      "loss": 2.2118,
      "step": 2946
    },
    {
      "epoch": 4.104456824512535,
      "grad_norm": 3.682919979095459,
      "learning_rate": 4.2113416714042065e-05,
      "loss": 2.02,
      "step": 2947
    },
    {
      "epoch": 4.105849582172702,
      "grad_norm": 3.69718337059021,
      "learning_rate": 4.210346787947697e-05,
      "loss": 1.9632,
      "step": 2948
    },
    {
      "epoch": 4.107242339832869,
      "grad_norm": 3.8436195850372314,
      "learning_rate": 4.209351904491188e-05,
      "loss": 2.0702,
      "step": 2949
    },
    {
      "epoch": 4.108635097493036,
      "grad_norm": 3.8712940216064453,
      "learning_rate": 4.2083570210346784e-05,
      "loss": 2.3238,
      "step": 2950
    },
    {
      "epoch": 4.110027855153203,
      "grad_norm": 3.6358802318573,
      "learning_rate": 4.2073621375781695e-05,
      "loss": 1.9997,
      "step": 2951
    },
    {
      "epoch": 4.111420612813371,
      "grad_norm": 3.9602253437042236,
      "learning_rate": 4.206367254121659e-05,
      "loss": 2.311,
      "step": 2952
    },
    {
      "epoch": 4.112813370473537,
      "grad_norm": 3.319279193878174,
      "learning_rate": 4.2053723706651503e-05,
      "loss": 2.0274,
      "step": 2953
    },
    {
      "epoch": 4.114206128133705,
      "grad_norm": 3.474858045578003,
      "learning_rate": 4.204377487208641e-05,
      "loss": 2.1625,
      "step": 2954
    },
    {
      "epoch": 4.1155988857938715,
      "grad_norm": 3.547170400619507,
      "learning_rate": 4.203382603752132e-05,
      "loss": 2.1221,
      "step": 2955
    },
    {
      "epoch": 4.116991643454039,
      "grad_norm": 3.437105417251587,
      "learning_rate": 4.2023877202956216e-05,
      "loss": 2.1375,
      "step": 2956
    },
    {
      "epoch": 4.118384401114207,
      "grad_norm": 3.3093979358673096,
      "learning_rate": 4.2013928368391127e-05,
      "loss": 1.9692,
      "step": 2957
    },
    {
      "epoch": 4.119777158774373,
      "grad_norm": 3.6160924434661865,
      "learning_rate": 4.200397953382603e-05,
      "loss": 2.1094,
      "step": 2958
    },
    {
      "epoch": 4.121169916434541,
      "grad_norm": 3.928183078765869,
      "learning_rate": 4.199403069926094e-05,
      "loss": 1.9048,
      "step": 2959
    },
    {
      "epoch": 4.1225626740947074,
      "grad_norm": 3.6659960746765137,
      "learning_rate": 4.1984081864695846e-05,
      "loss": 1.847,
      "step": 2960
    },
    {
      "epoch": 4.123955431754875,
      "grad_norm": 4.066314697265625,
      "learning_rate": 4.197413303013075e-05,
      "loss": 2.4269,
      "step": 2961
    },
    {
      "epoch": 4.125348189415042,
      "grad_norm": 3.5907700061798096,
      "learning_rate": 4.1964184195565654e-05,
      "loss": 1.8727,
      "step": 2962
    },
    {
      "epoch": 4.126740947075209,
      "grad_norm": 3.6240313053131104,
      "learning_rate": 4.1954235361000565e-05,
      "loss": 2.1292,
      "step": 2963
    },
    {
      "epoch": 4.128133704735376,
      "grad_norm": 3.639030933380127,
      "learning_rate": 4.1944286526435475e-05,
      "loss": 1.9945,
      "step": 2964
    },
    {
      "epoch": 4.129526462395543,
      "grad_norm": 3.69741153717041,
      "learning_rate": 4.193433769187038e-05,
      "loss": 2.123,
      "step": 2965
    },
    {
      "epoch": 4.13091922005571,
      "grad_norm": 4.221983432769775,
      "learning_rate": 4.192438885730528e-05,
      "loss": 2.3641,
      "step": 2966
    },
    {
      "epoch": 4.132311977715878,
      "grad_norm": 3.8189964294433594,
      "learning_rate": 4.191444002274019e-05,
      "loss": 2.0556,
      "step": 2967
    },
    {
      "epoch": 4.133704735376044,
      "grad_norm": 3.6743671894073486,
      "learning_rate": 4.19044911881751e-05,
      "loss": 2.0531,
      "step": 2968
    },
    {
      "epoch": 4.135097493036212,
      "grad_norm": 3.03293776512146,
      "learning_rate": 4.189454235361e-05,
      "loss": 1.8302,
      "step": 2969
    },
    {
      "epoch": 4.1364902506963785,
      "grad_norm": 4.1291046142578125,
      "learning_rate": 4.1884593519044913e-05,
      "loss": 1.7203,
      "step": 2970
    },
    {
      "epoch": 4.137883008356546,
      "grad_norm": 4.686734676361084,
      "learning_rate": 4.187464468447981e-05,
      "loss": 2.0674,
      "step": 2971
    },
    {
      "epoch": 4.139275766016713,
      "grad_norm": 3.4521591663360596,
      "learning_rate": 4.186469584991472e-05,
      "loss": 1.8709,
      "step": 2972
    },
    {
      "epoch": 4.14066852367688,
      "grad_norm": 5.810756206512451,
      "learning_rate": 4.1854747015349626e-05,
      "loss": 2.2401,
      "step": 2973
    },
    {
      "epoch": 4.142061281337047,
      "grad_norm": 3.6473703384399414,
      "learning_rate": 4.1844798180784536e-05,
      "loss": 2.1024,
      "step": 2974
    },
    {
      "epoch": 4.1434540389972145,
      "grad_norm": 3.4020042419433594,
      "learning_rate": 4.183484934621944e-05,
      "loss": 1.8226,
      "step": 2975
    },
    {
      "epoch": 4.144846796657381,
      "grad_norm": 3.425886869430542,
      "learning_rate": 4.1824900511654345e-05,
      "loss": 2.0591,
      "step": 2976
    },
    {
      "epoch": 4.146239554317549,
      "grad_norm": 3.797455310821533,
      "learning_rate": 4.181495167708925e-05,
      "loss": 2.2521,
      "step": 2977
    },
    {
      "epoch": 4.147632311977716,
      "grad_norm": 3.6039366722106934,
      "learning_rate": 4.180500284252416e-05,
      "loss": 2.2441,
      "step": 2978
    },
    {
      "epoch": 4.149025069637883,
      "grad_norm": 3.46376633644104,
      "learning_rate": 4.1795054007959064e-05,
      "loss": 2.1358,
      "step": 2979
    },
    {
      "epoch": 4.1504178272980505,
      "grad_norm": 3.632885217666626,
      "learning_rate": 4.1785105173393974e-05,
      "loss": 2.1073,
      "step": 2980
    },
    {
      "epoch": 4.151810584958217,
      "grad_norm": 3.925652265548706,
      "learning_rate": 4.177515633882887e-05,
      "loss": 2.0397,
      "step": 2981
    },
    {
      "epoch": 4.153203342618385,
      "grad_norm": 3.7493999004364014,
      "learning_rate": 4.176520750426378e-05,
      "loss": 2.0389,
      "step": 2982
    },
    {
      "epoch": 4.154596100278551,
      "grad_norm": 4.511785984039307,
      "learning_rate": 4.175525866969869e-05,
      "loss": 2.3307,
      "step": 2983
    },
    {
      "epoch": 4.155988857938719,
      "grad_norm": 3.2160871028900146,
      "learning_rate": 4.17453098351336e-05,
      "loss": 1.6668,
      "step": 2984
    },
    {
      "epoch": 4.157381615598886,
      "grad_norm": 3.8907697200775146,
      "learning_rate": 4.17353610005685e-05,
      "loss": 1.6878,
      "step": 2985
    },
    {
      "epoch": 4.158774373259053,
      "grad_norm": 4.556327819824219,
      "learning_rate": 4.1725412166003406e-05,
      "loss": 1.9801,
      "step": 2986
    },
    {
      "epoch": 4.16016713091922,
      "grad_norm": 3.8720271587371826,
      "learning_rate": 4.171546333143831e-05,
      "loss": 2.238,
      "step": 2987
    },
    {
      "epoch": 4.161559888579387,
      "grad_norm": 3.6748290061950684,
      "learning_rate": 4.170551449687322e-05,
      "loss": 2.2768,
      "step": 2988
    },
    {
      "epoch": 4.162952646239554,
      "grad_norm": 7.88837194442749,
      "learning_rate": 4.1695565662308125e-05,
      "loss": 2.2326,
      "step": 2989
    },
    {
      "epoch": 4.164345403899722,
      "grad_norm": 3.624276638031006,
      "learning_rate": 4.1685616827743036e-05,
      "loss": 2.2305,
      "step": 2990
    },
    {
      "epoch": 4.165738161559888,
      "grad_norm": 4.0302414894104,
      "learning_rate": 4.167566799317793e-05,
      "loss": 1.9653,
      "step": 2991
    },
    {
      "epoch": 4.167130919220056,
      "grad_norm": 3.4902541637420654,
      "learning_rate": 4.1665719158612844e-05,
      "loss": 2.1037,
      "step": 2992
    },
    {
      "epoch": 4.1685236768802225,
      "grad_norm": 3.871030569076538,
      "learning_rate": 4.1655770324047755e-05,
      "loss": 2.134,
      "step": 2993
    },
    {
      "epoch": 4.16991643454039,
      "grad_norm": 3.963440418243408,
      "learning_rate": 4.164582148948266e-05,
      "loss": 2.197,
      "step": 2994
    },
    {
      "epoch": 4.171309192200557,
      "grad_norm": 3.3340349197387695,
      "learning_rate": 4.163587265491757e-05,
      "loss": 1.8162,
      "step": 2995
    },
    {
      "epoch": 4.172701949860724,
      "grad_norm": 3.550037145614624,
      "learning_rate": 4.162592382035247e-05,
      "loss": 1.7161,
      "step": 2996
    },
    {
      "epoch": 4.174094707520891,
      "grad_norm": 3.3833820819854736,
      "learning_rate": 4.161597498578738e-05,
      "loss": 1.789,
      "step": 2997
    },
    {
      "epoch": 4.1754874651810585,
      "grad_norm": 3.274717330932617,
      "learning_rate": 4.160602615122228e-05,
      "loss": 1.7684,
      "step": 2998
    },
    {
      "epoch": 4.176880222841225,
      "grad_norm": 3.4522273540496826,
      "learning_rate": 4.159607731665719e-05,
      "loss": 1.9979,
      "step": 2999
    },
    {
      "epoch": 4.178272980501393,
      "grad_norm": 3.6612470149993896,
      "learning_rate": 4.158612848209209e-05,
      "loss": 1.9038,
      "step": 3000
    },
    {
      "epoch": 4.17966573816156,
      "grad_norm": 3.563570022583008,
      "learning_rate": 4.1576179647527e-05,
      "loss": 2.0586,
      "step": 3001
    },
    {
      "epoch": 4.181058495821727,
      "grad_norm": 3.3609519004821777,
      "learning_rate": 4.1566230812961905e-05,
      "loss": 1.6657,
      "step": 3002
    },
    {
      "epoch": 4.1824512534818945,
      "grad_norm": 3.595749616622925,
      "learning_rate": 4.1556281978396816e-05,
      "loss": 1.984,
      "step": 3003
    },
    {
      "epoch": 4.183844011142061,
      "grad_norm": 3.685802698135376,
      "learning_rate": 4.154633314383172e-05,
      "loss": 1.9852,
      "step": 3004
    },
    {
      "epoch": 4.185236768802229,
      "grad_norm": 3.6919596195220947,
      "learning_rate": 4.1536384309266624e-05,
      "loss": 2.2896,
      "step": 3005
    },
    {
      "epoch": 4.186629526462395,
      "grad_norm": 4.045046329498291,
      "learning_rate": 4.152643547470153e-05,
      "loss": 1.9941,
      "step": 3006
    },
    {
      "epoch": 4.188022284122563,
      "grad_norm": 4.211788654327393,
      "learning_rate": 4.151648664013644e-05,
      "loss": 1.8658,
      "step": 3007
    },
    {
      "epoch": 4.18941504178273,
      "grad_norm": 3.4465973377227783,
      "learning_rate": 4.150653780557134e-05,
      "loss": 1.9987,
      "step": 3008
    },
    {
      "epoch": 4.190807799442897,
      "grad_norm": 3.3546183109283447,
      "learning_rate": 4.1496588971006254e-05,
      "loss": 1.9247,
      "step": 3009
    },
    {
      "epoch": 4.192200557103064,
      "grad_norm": 3.7048683166503906,
      "learning_rate": 4.148664013644115e-05,
      "loss": 2.0285,
      "step": 3010
    },
    {
      "epoch": 4.193593314763231,
      "grad_norm": 3.253375768661499,
      "learning_rate": 4.147669130187606e-05,
      "loss": 1.9872,
      "step": 3011
    },
    {
      "epoch": 4.194986072423398,
      "grad_norm": 3.39601469039917,
      "learning_rate": 4.1466742467310966e-05,
      "loss": 1.7865,
      "step": 3012
    },
    {
      "epoch": 4.196378830083566,
      "grad_norm": 3.560274362564087,
      "learning_rate": 4.145679363274588e-05,
      "loss": 1.9759,
      "step": 3013
    },
    {
      "epoch": 4.197771587743732,
      "grad_norm": 3.3260228633880615,
      "learning_rate": 4.144684479818078e-05,
      "loss": 1.7974,
      "step": 3014
    },
    {
      "epoch": 4.1991643454039,
      "grad_norm": 3.7586898803710938,
      "learning_rate": 4.1436895963615685e-05,
      "loss": 2.3483,
      "step": 3015
    },
    {
      "epoch": 4.2005571030640665,
      "grad_norm": 3.291304349899292,
      "learning_rate": 4.142694712905059e-05,
      "loss": 1.8448,
      "step": 3016
    },
    {
      "epoch": 4.201949860724234,
      "grad_norm": 3.7757081985473633,
      "learning_rate": 4.14169982944855e-05,
      "loss": 1.7119,
      "step": 3017
    },
    {
      "epoch": 4.203342618384401,
      "grad_norm": 3.5498621463775635,
      "learning_rate": 4.1407049459920404e-05,
      "loss": 1.9627,
      "step": 3018
    },
    {
      "epoch": 4.204735376044568,
      "grad_norm": 3.449716567993164,
      "learning_rate": 4.1397100625355315e-05,
      "loss": 1.9399,
      "step": 3019
    },
    {
      "epoch": 4.206128133704736,
      "grad_norm": 3.383650779724121,
      "learning_rate": 4.138715179079021e-05,
      "loss": 1.8659,
      "step": 3020
    },
    {
      "epoch": 4.2075208913649025,
      "grad_norm": 3.405827522277832,
      "learning_rate": 4.137720295622512e-05,
      "loss": 1.6832,
      "step": 3021
    },
    {
      "epoch": 4.20891364902507,
      "grad_norm": 3.6824240684509277,
      "learning_rate": 4.1367254121660034e-05,
      "loss": 2.4785,
      "step": 3022
    },
    {
      "epoch": 4.210306406685237,
      "grad_norm": 3.528207778930664,
      "learning_rate": 4.135730528709494e-05,
      "loss": 2.0306,
      "step": 3023
    },
    {
      "epoch": 4.211699164345404,
      "grad_norm": 3.8610434532165527,
      "learning_rate": 4.134735645252985e-05,
      "loss": 2.5256,
      "step": 3024
    },
    {
      "epoch": 4.213091922005571,
      "grad_norm": 3.419217586517334,
      "learning_rate": 4.1337407617964746e-05,
      "loss": 1.9272,
      "step": 3025
    },
    {
      "epoch": 4.2144846796657385,
      "grad_norm": 3.739485502243042,
      "learning_rate": 4.132745878339966e-05,
      "loss": 2.0173,
      "step": 3026
    },
    {
      "epoch": 4.215877437325905,
      "grad_norm": 3.5592610836029053,
      "learning_rate": 4.131750994883456e-05,
      "loss": 2.1957,
      "step": 3027
    },
    {
      "epoch": 4.217270194986073,
      "grad_norm": 3.525094747543335,
      "learning_rate": 4.130756111426947e-05,
      "loss": 2.2607,
      "step": 3028
    },
    {
      "epoch": 4.218662952646239,
      "grad_norm": 4.066737174987793,
      "learning_rate": 4.1297612279704376e-05,
      "loss": 2.1214,
      "step": 3029
    },
    {
      "epoch": 4.220055710306407,
      "grad_norm": 5.838018894195557,
      "learning_rate": 4.128766344513928e-05,
      "loss": 2.3155,
      "step": 3030
    },
    {
      "epoch": 4.221448467966574,
      "grad_norm": 3.724910259246826,
      "learning_rate": 4.1277714610574184e-05,
      "loss": 2.2864,
      "step": 3031
    },
    {
      "epoch": 4.222841225626741,
      "grad_norm": 3.427699089050293,
      "learning_rate": 4.1267765776009095e-05,
      "loss": 1.9764,
      "step": 3032
    },
    {
      "epoch": 4.224233983286908,
      "grad_norm": 3.9147324562072754,
      "learning_rate": 4.1257816941444e-05,
      "loss": 2.2344,
      "step": 3033
    },
    {
      "epoch": 4.225626740947075,
      "grad_norm": 3.2432050704956055,
      "learning_rate": 4.124786810687891e-05,
      "loss": 1.5073,
      "step": 3034
    },
    {
      "epoch": 4.227019498607242,
      "grad_norm": 4.5403032302856445,
      "learning_rate": 4.123791927231381e-05,
      "loss": 1.8843,
      "step": 3035
    },
    {
      "epoch": 4.22841225626741,
      "grad_norm": 3.7331345081329346,
      "learning_rate": 4.122797043774872e-05,
      "loss": 2.126,
      "step": 3036
    },
    {
      "epoch": 4.229805013927576,
      "grad_norm": 3.9318816661834717,
      "learning_rate": 4.121802160318362e-05,
      "loss": 2.052,
      "step": 3037
    },
    {
      "epoch": 4.231197771587744,
      "grad_norm": 4.157036304473877,
      "learning_rate": 4.120807276861853e-05,
      "loss": 1.9879,
      "step": 3038
    },
    {
      "epoch": 4.2325905292479105,
      "grad_norm": 3.6946725845336914,
      "learning_rate": 4.119812393405344e-05,
      "loss": 1.6204,
      "step": 3039
    },
    {
      "epoch": 4.233983286908078,
      "grad_norm": 3.3176159858703613,
      "learning_rate": 4.118817509948834e-05,
      "loss": 1.7824,
      "step": 3040
    },
    {
      "epoch": 4.235376044568245,
      "grad_norm": 3.661743402481079,
      "learning_rate": 4.1178226264923245e-05,
      "loss": 2.0289,
      "step": 3041
    },
    {
      "epoch": 4.236768802228412,
      "grad_norm": 3.5172345638275146,
      "learning_rate": 4.1168277430358156e-05,
      "loss": 1.8457,
      "step": 3042
    },
    {
      "epoch": 4.23816155988858,
      "grad_norm": 3.398505210876465,
      "learning_rate": 4.115832859579306e-05,
      "loss": 1.7284,
      "step": 3043
    },
    {
      "epoch": 4.2395543175487465,
      "grad_norm": 3.7780115604400635,
      "learning_rate": 4.1148379761227964e-05,
      "loss": 2.1162,
      "step": 3044
    },
    {
      "epoch": 4.240947075208914,
      "grad_norm": 4.354976654052734,
      "learning_rate": 4.113843092666287e-05,
      "loss": 2.555,
      "step": 3045
    },
    {
      "epoch": 4.242339832869081,
      "grad_norm": 3.9801361560821533,
      "learning_rate": 4.112848209209778e-05,
      "loss": 2.3081,
      "step": 3046
    },
    {
      "epoch": 4.243732590529248,
      "grad_norm": 3.451056718826294,
      "learning_rate": 4.111853325753269e-05,
      "loss": 2.1083,
      "step": 3047
    },
    {
      "epoch": 4.245125348189415,
      "grad_norm": 3.6292426586151123,
      "learning_rate": 4.1108584422967594e-05,
      "loss": 1.8113,
      "step": 3048
    },
    {
      "epoch": 4.2465181058495824,
      "grad_norm": 3.9774768352508545,
      "learning_rate": 4.109863558840249e-05,
      "loss": 1.9119,
      "step": 3049
    },
    {
      "epoch": 4.247910863509749,
      "grad_norm": 3.4504611492156982,
      "learning_rate": 4.10886867538374e-05,
      "loss": 1.8894,
      "step": 3050
    },
    {
      "epoch": 4.249303621169917,
      "grad_norm": 4.225854873657227,
      "learning_rate": 4.107873791927231e-05,
      "loss": 2.3452,
      "step": 3051
    },
    {
      "epoch": 4.250696378830083,
      "grad_norm": 3.4578373432159424,
      "learning_rate": 4.106878908470722e-05,
      "loss": 1.7064,
      "step": 3052
    },
    {
      "epoch": 4.252089136490251,
      "grad_norm": 4.071245193481445,
      "learning_rate": 4.105884025014213e-05,
      "loss": 2.2979,
      "step": 3053
    },
    {
      "epoch": 4.2534818941504176,
      "grad_norm": 3.548414707183838,
      "learning_rate": 4.1048891415577025e-05,
      "loss": 2.2977,
      "step": 3054
    },
    {
      "epoch": 4.254874651810585,
      "grad_norm": 3.9203896522521973,
      "learning_rate": 4.1038942581011936e-05,
      "loss": 1.9007,
      "step": 3055
    },
    {
      "epoch": 4.256267409470752,
      "grad_norm": 3.354689359664917,
      "learning_rate": 4.102899374644684e-05,
      "loss": 2.0673,
      "step": 3056
    },
    {
      "epoch": 4.257660167130919,
      "grad_norm": 3.854431629180908,
      "learning_rate": 4.101904491188175e-05,
      "loss": 2.1364,
      "step": 3057
    },
    {
      "epoch": 4.259052924791086,
      "grad_norm": 3.1632230281829834,
      "learning_rate": 4.1009096077316655e-05,
      "loss": 1.9589,
      "step": 3058
    },
    {
      "epoch": 4.2604456824512535,
      "grad_norm": 3.4056951999664307,
      "learning_rate": 4.099914724275156e-05,
      "loss": 1.7211,
      "step": 3059
    },
    {
      "epoch": 4.26183844011142,
      "grad_norm": 3.9283065795898438,
      "learning_rate": 4.098919840818646e-05,
      "loss": 2.3037,
      "step": 3060
    },
    {
      "epoch": 4.263231197771588,
      "grad_norm": 3.8917815685272217,
      "learning_rate": 4.0979249573621374e-05,
      "loss": 2.0805,
      "step": 3061
    },
    {
      "epoch": 4.264623955431755,
      "grad_norm": 3.466789960861206,
      "learning_rate": 4.096930073905628e-05,
      "loss": 1.7961,
      "step": 3062
    },
    {
      "epoch": 4.266016713091922,
      "grad_norm": 3.8265340328216553,
      "learning_rate": 4.095935190449119e-05,
      "loss": 2.1436,
      "step": 3063
    },
    {
      "epoch": 4.2674094707520895,
      "grad_norm": 3.4796509742736816,
      "learning_rate": 4.0949403069926086e-05,
      "loss": 1.6483,
      "step": 3064
    },
    {
      "epoch": 4.268802228412256,
      "grad_norm": 3.108628988265991,
      "learning_rate": 4.0939454235361e-05,
      "loss": 1.7464,
      "step": 3065
    },
    {
      "epoch": 4.270194986072424,
      "grad_norm": 3.9219963550567627,
      "learning_rate": 4.09295054007959e-05,
      "loss": 1.8372,
      "step": 3066
    },
    {
      "epoch": 4.27158774373259,
      "grad_norm": 4.510400295257568,
      "learning_rate": 4.091955656623081e-05,
      "loss": 1.7882,
      "step": 3067
    },
    {
      "epoch": 4.272980501392758,
      "grad_norm": 3.6423346996307373,
      "learning_rate": 4.0909607731665716e-05,
      "loss": 2.1274,
      "step": 3068
    },
    {
      "epoch": 4.274373259052925,
      "grad_norm": 3.5593934059143066,
      "learning_rate": 4.089965889710062e-05,
      "loss": 1.8293,
      "step": 3069
    },
    {
      "epoch": 4.275766016713092,
      "grad_norm": 3.4503164291381836,
      "learning_rate": 4.0889710062535524e-05,
      "loss": 1.9578,
      "step": 3070
    },
    {
      "epoch": 4.277158774373259,
      "grad_norm": 3.593991994857788,
      "learning_rate": 4.0879761227970435e-05,
      "loss": 1.9559,
      "step": 3071
    },
    {
      "epoch": 4.278551532033426,
      "grad_norm": 3.898189067840576,
      "learning_rate": 4.086981239340534e-05,
      "loss": 2.4211,
      "step": 3072
    },
    {
      "epoch": 4.279944289693593,
      "grad_norm": 3.5171823501586914,
      "learning_rate": 4.085986355884025e-05,
      "loss": 1.7952,
      "step": 3073
    },
    {
      "epoch": 4.281337047353761,
      "grad_norm": 3.7271249294281006,
      "learning_rate": 4.084991472427515e-05,
      "loss": 2.1002,
      "step": 3074
    },
    {
      "epoch": 4.282729805013927,
      "grad_norm": 4.111434459686279,
      "learning_rate": 4.083996588971006e-05,
      "loss": 2.3111,
      "step": 3075
    },
    {
      "epoch": 4.284122562674095,
      "grad_norm": 3.5992376804351807,
      "learning_rate": 4.083001705514497e-05,
      "loss": 1.8289,
      "step": 3076
    },
    {
      "epoch": 4.2855153203342615,
      "grad_norm": 3.5889201164245605,
      "learning_rate": 4.082006822057987e-05,
      "loss": 2.1275,
      "step": 3077
    },
    {
      "epoch": 4.286908077994429,
      "grad_norm": 4.682888984680176,
      "learning_rate": 4.0810119386014784e-05,
      "loss": 2.1404,
      "step": 3078
    },
    {
      "epoch": 4.288300835654596,
      "grad_norm": 3.8084707260131836,
      "learning_rate": 4.080017055144968e-05,
      "loss": 1.8622,
      "step": 3079
    },
    {
      "epoch": 4.289693593314763,
      "grad_norm": 3.434199094772339,
      "learning_rate": 4.079022171688459e-05,
      "loss": 1.909,
      "step": 3080
    },
    {
      "epoch": 4.29108635097493,
      "grad_norm": 3.2674553394317627,
      "learning_rate": 4.0780272882319496e-05,
      "loss": 1.8136,
      "step": 3081
    },
    {
      "epoch": 4.2924791086350975,
      "grad_norm": 3.6555733680725098,
      "learning_rate": 4.077032404775441e-05,
      "loss": 1.9099,
      "step": 3082
    },
    {
      "epoch": 4.293871866295264,
      "grad_norm": 6.769938945770264,
      "learning_rate": 4.0760375213189304e-05,
      "loss": 1.9184,
      "step": 3083
    },
    {
      "epoch": 4.295264623955432,
      "grad_norm": 3.5074851512908936,
      "learning_rate": 4.0750426378624215e-05,
      "loss": 1.9444,
      "step": 3084
    },
    {
      "epoch": 4.296657381615599,
      "grad_norm": 4.210463523864746,
      "learning_rate": 4.074047754405912e-05,
      "loss": 1.9365,
      "step": 3085
    },
    {
      "epoch": 4.298050139275766,
      "grad_norm": 3.9172286987304688,
      "learning_rate": 4.073052870949403e-05,
      "loss": 2.022,
      "step": 3086
    },
    {
      "epoch": 4.2994428969359335,
      "grad_norm": 3.6236064434051514,
      "learning_rate": 4.0720579874928934e-05,
      "loss": 2.1497,
      "step": 3087
    },
    {
      "epoch": 4.3008356545961,
      "grad_norm": 4.024149417877197,
      "learning_rate": 4.071063104036384e-05,
      "loss": 2.3488,
      "step": 3088
    },
    {
      "epoch": 4.302228412256268,
      "grad_norm": 5.756144046783447,
      "learning_rate": 4.070068220579874e-05,
      "loss": 1.8647,
      "step": 3089
    },
    {
      "epoch": 4.303621169916434,
      "grad_norm": 3.4565234184265137,
      "learning_rate": 4.069073337123365e-05,
      "loss": 2.3004,
      "step": 3090
    },
    {
      "epoch": 4.305013927576602,
      "grad_norm": 3.532681941986084,
      "learning_rate": 4.068078453666856e-05,
      "loss": 1.7751,
      "step": 3091
    },
    {
      "epoch": 4.306406685236769,
      "grad_norm": 4.027726650238037,
      "learning_rate": 4.067083570210347e-05,
      "loss": 2.2313,
      "step": 3092
    },
    {
      "epoch": 4.307799442896936,
      "grad_norm": 3.7110137939453125,
      "learning_rate": 4.0660886867538365e-05,
      "loss": 2.0806,
      "step": 3093
    },
    {
      "epoch": 4.309192200557103,
      "grad_norm": 4.137944221496582,
      "learning_rate": 4.0650938032973276e-05,
      "loss": 2.1601,
      "step": 3094
    },
    {
      "epoch": 4.31058495821727,
      "grad_norm": 3.6126034259796143,
      "learning_rate": 4.064098919840818e-05,
      "loss": 2.1295,
      "step": 3095
    },
    {
      "epoch": 4.311977715877437,
      "grad_norm": 4.281980514526367,
      "learning_rate": 4.063104036384309e-05,
      "loss": 2.3458,
      "step": 3096
    },
    {
      "epoch": 4.313370473537605,
      "grad_norm": 3.6331682205200195,
      "learning_rate": 4.0621091529277995e-05,
      "loss": 1.6873,
      "step": 3097
    },
    {
      "epoch": 4.314763231197771,
      "grad_norm": 3.5378000736236572,
      "learning_rate": 4.06111426947129e-05,
      "loss": 1.5119,
      "step": 3098
    },
    {
      "epoch": 4.316155988857939,
      "grad_norm": 3.42341685295105,
      "learning_rate": 4.06011938601478e-05,
      "loss": 1.7202,
      "step": 3099
    },
    {
      "epoch": 4.3175487465181055,
      "grad_norm": 3.846165418624878,
      "learning_rate": 4.0591245025582714e-05,
      "loss": 1.9793,
      "step": 3100
    },
    {
      "epoch": 4.318941504178273,
      "grad_norm": 3.754512071609497,
      "learning_rate": 4.058129619101762e-05,
      "loss": 2.1661,
      "step": 3101
    },
    {
      "epoch": 4.32033426183844,
      "grad_norm": 6.259832859039307,
      "learning_rate": 4.057134735645253e-05,
      "loss": 2.3145,
      "step": 3102
    },
    {
      "epoch": 4.321727019498607,
      "grad_norm": 3.235311985015869,
      "learning_rate": 4.0561398521887426e-05,
      "loss": 1.8265,
      "step": 3103
    },
    {
      "epoch": 4.323119777158775,
      "grad_norm": 3.2759287357330322,
      "learning_rate": 4.055144968732234e-05,
      "loss": 1.8827,
      "step": 3104
    },
    {
      "epoch": 4.3245125348189415,
      "grad_norm": 3.4988794326782227,
      "learning_rate": 4.054150085275725e-05,
      "loss": 1.925,
      "step": 3105
    },
    {
      "epoch": 4.325905292479109,
      "grad_norm": 3.346482038497925,
      "learning_rate": 4.053155201819215e-05,
      "loss": 1.8684,
      "step": 3106
    },
    {
      "epoch": 4.327298050139276,
      "grad_norm": 3.441643714904785,
      "learning_rate": 4.052160318362706e-05,
      "loss": 1.7618,
      "step": 3107
    },
    {
      "epoch": 4.328690807799443,
      "grad_norm": 3.765227794647217,
      "learning_rate": 4.051165434906196e-05,
      "loss": 1.8764,
      "step": 3108
    },
    {
      "epoch": 4.33008356545961,
      "grad_norm": 4.57131814956665,
      "learning_rate": 4.050170551449687e-05,
      "loss": 2.296,
      "step": 3109
    },
    {
      "epoch": 4.3314763231197775,
      "grad_norm": 3.5450220108032227,
      "learning_rate": 4.0491756679931775e-05,
      "loss": 1.8936,
      "step": 3110
    },
    {
      "epoch": 4.332869080779944,
      "grad_norm": 3.6709892749786377,
      "learning_rate": 4.0481807845366686e-05,
      "loss": 2.3217,
      "step": 3111
    },
    {
      "epoch": 4.334261838440112,
      "grad_norm": 7.978917121887207,
      "learning_rate": 4.047185901080159e-05,
      "loss": 2.1545,
      "step": 3112
    },
    {
      "epoch": 4.335654596100278,
      "grad_norm": 3.426018238067627,
      "learning_rate": 4.0461910176236494e-05,
      "loss": 1.7577,
      "step": 3113
    },
    {
      "epoch": 4.337047353760446,
      "grad_norm": 5.184499740600586,
      "learning_rate": 4.04519613416714e-05,
      "loss": 1.9331,
      "step": 3114
    },
    {
      "epoch": 4.338440111420613,
      "grad_norm": 3.6694085597991943,
      "learning_rate": 4.044201250710631e-05,
      "loss": 2.2532,
      "step": 3115
    },
    {
      "epoch": 4.33983286908078,
      "grad_norm": 3.379873275756836,
      "learning_rate": 4.043206367254121e-05,
      "loss": 1.7379,
      "step": 3116
    },
    {
      "epoch": 4.341225626740947,
      "grad_norm": 3.5996804237365723,
      "learning_rate": 4.0422114837976124e-05,
      "loss": 1.9487,
      "step": 3117
    },
    {
      "epoch": 4.342618384401114,
      "grad_norm": 4.125095367431641,
      "learning_rate": 4.041216600341102e-05,
      "loss": 2.3051,
      "step": 3118
    },
    {
      "epoch": 4.344011142061281,
      "grad_norm": 3.487342596054077,
      "learning_rate": 4.040221716884593e-05,
      "loss": 2.0359,
      "step": 3119
    },
    {
      "epoch": 4.345403899721449,
      "grad_norm": 4.149247646331787,
      "learning_rate": 4.0392268334280836e-05,
      "loss": 1.8801,
      "step": 3120
    },
    {
      "epoch": 4.346796657381615,
      "grad_norm": 3.596158981323242,
      "learning_rate": 4.038231949971575e-05,
      "loss": 1.9599,
      "step": 3121
    },
    {
      "epoch": 4.348189415041783,
      "grad_norm": 3.6659107208251953,
      "learning_rate": 4.037237066515065e-05,
      "loss": 2.0527,
      "step": 3122
    },
    {
      "epoch": 4.3495821727019495,
      "grad_norm": 3.5319089889526367,
      "learning_rate": 4.0362421830585555e-05,
      "loss": 1.9435,
      "step": 3123
    },
    {
      "epoch": 4.350974930362117,
      "grad_norm": 3.8539369106292725,
      "learning_rate": 4.035247299602046e-05,
      "loss": 1.7566,
      "step": 3124
    },
    {
      "epoch": 4.352367688022284,
      "grad_norm": 3.5454413890838623,
      "learning_rate": 4.034252416145537e-05,
      "loss": 1.8269,
      "step": 3125
    },
    {
      "epoch": 4.353760445682451,
      "grad_norm": 5.290493965148926,
      "learning_rate": 4.0332575326890274e-05,
      "loss": 2.0032,
      "step": 3126
    },
    {
      "epoch": 4.355153203342619,
      "grad_norm": 3.8820858001708984,
      "learning_rate": 4.032262649232518e-05,
      "loss": 2.0953,
      "step": 3127
    },
    {
      "epoch": 4.3565459610027855,
      "grad_norm": 3.353656768798828,
      "learning_rate": 4.031267765776008e-05,
      "loss": 2.0047,
      "step": 3128
    },
    {
      "epoch": 4.357938718662953,
      "grad_norm": 3.3565492630004883,
      "learning_rate": 4.030272882319499e-05,
      "loss": 2.1301,
      "step": 3129
    },
    {
      "epoch": 4.35933147632312,
      "grad_norm": 3.818831205368042,
      "learning_rate": 4.02927799886299e-05,
      "loss": 1.9086,
      "step": 3130
    },
    {
      "epoch": 4.360724233983287,
      "grad_norm": 3.3351616859436035,
      "learning_rate": 4.028283115406481e-05,
      "loss": 1.6885,
      "step": 3131
    },
    {
      "epoch": 4.362116991643454,
      "grad_norm": 3.724071979522705,
      "learning_rate": 4.0272882319499706e-05,
      "loss": 2.3188,
      "step": 3132
    },
    {
      "epoch": 4.3635097493036215,
      "grad_norm": 4.343799591064453,
      "learning_rate": 4.0262933484934616e-05,
      "loss": 2.209,
      "step": 3133
    },
    {
      "epoch": 4.364902506963788,
      "grad_norm": 4.293625831604004,
      "learning_rate": 4.025298465036953e-05,
      "loss": 2.315,
      "step": 3134
    },
    {
      "epoch": 4.366295264623956,
      "grad_norm": 3.800335645675659,
      "learning_rate": 4.024303581580443e-05,
      "loss": 1.8779,
      "step": 3135
    },
    {
      "epoch": 4.367688022284122,
      "grad_norm": 4.923582553863525,
      "learning_rate": 4.023308698123934e-05,
      "loss": 2.2346,
      "step": 3136
    },
    {
      "epoch": 4.36908077994429,
      "grad_norm": 3.8185031414031982,
      "learning_rate": 4.022313814667424e-05,
      "loss": 1.7602,
      "step": 3137
    },
    {
      "epoch": 4.370473537604457,
      "grad_norm": 3.3094518184661865,
      "learning_rate": 4.021318931210915e-05,
      "loss": 1.8159,
      "step": 3138
    },
    {
      "epoch": 4.371866295264624,
      "grad_norm": 3.3837151527404785,
      "learning_rate": 4.0203240477544054e-05,
      "loss": 2.0494,
      "step": 3139
    },
    {
      "epoch": 4.373259052924791,
      "grad_norm": 3.692568302154541,
      "learning_rate": 4.0193291642978965e-05,
      "loss": 1.951,
      "step": 3140
    },
    {
      "epoch": 4.374651810584958,
      "grad_norm": 3.4474568367004395,
      "learning_rate": 4.018334280841387e-05,
      "loss": 2.0262,
      "step": 3141
    },
    {
      "epoch": 4.376044568245125,
      "grad_norm": 3.624847888946533,
      "learning_rate": 4.0173393973848773e-05,
      "loss": 2.3257,
      "step": 3142
    },
    {
      "epoch": 4.3774373259052926,
      "grad_norm": 3.9856884479522705,
      "learning_rate": 4.016344513928368e-05,
      "loss": 2.4661,
      "step": 3143
    },
    {
      "epoch": 4.378830083565459,
      "grad_norm": 3.9437685012817383,
      "learning_rate": 4.015349630471859e-05,
      "loss": 2.1859,
      "step": 3144
    },
    {
      "epoch": 4.380222841225627,
      "grad_norm": 3.850909948348999,
      "learning_rate": 4.014354747015349e-05,
      "loss": 2.142,
      "step": 3145
    },
    {
      "epoch": 4.381615598885794,
      "grad_norm": 3.9388673305511475,
      "learning_rate": 4.01335986355884e-05,
      "loss": 2.2247,
      "step": 3146
    },
    {
      "epoch": 4.383008356545961,
      "grad_norm": 5.086613655090332,
      "learning_rate": 4.01236498010233e-05,
      "loss": 2.4244,
      "step": 3147
    },
    {
      "epoch": 4.3844011142061285,
      "grad_norm": 3.144426107406616,
      "learning_rate": 4.011370096645821e-05,
      "loss": 1.3181,
      "step": 3148
    },
    {
      "epoch": 4.385793871866295,
      "grad_norm": 3.7608373165130615,
      "learning_rate": 4.0103752131893116e-05,
      "loss": 1.9655,
      "step": 3149
    },
    {
      "epoch": 4.387186629526463,
      "grad_norm": 3.7845873832702637,
      "learning_rate": 4.0093803297328026e-05,
      "loss": 2.1247,
      "step": 3150
    },
    {
      "epoch": 4.388579387186629,
      "grad_norm": 3.580233573913574,
      "learning_rate": 4.008385446276293e-05,
      "loss": 1.903,
      "step": 3151
    },
    {
      "epoch": 4.389972144846797,
      "grad_norm": 4.054161071777344,
      "learning_rate": 4.0073905628197835e-05,
      "loss": 1.8935,
      "step": 3152
    },
    {
      "epoch": 4.391364902506964,
      "grad_norm": 3.250823974609375,
      "learning_rate": 4.006395679363274e-05,
      "loss": 1.6088,
      "step": 3153
    },
    {
      "epoch": 4.392757660167131,
      "grad_norm": 3.544598340988159,
      "learning_rate": 4.005400795906765e-05,
      "loss": 1.9608,
      "step": 3154
    },
    {
      "epoch": 4.394150417827298,
      "grad_norm": 4.119276523590088,
      "learning_rate": 4.0044059124502554e-05,
      "loss": 2.116,
      "step": 3155
    },
    {
      "epoch": 4.395543175487465,
      "grad_norm": 6.200865745544434,
      "learning_rate": 4.0034110289937464e-05,
      "loss": 1.8136,
      "step": 3156
    },
    {
      "epoch": 4.396935933147632,
      "grad_norm": 3.2142770290374756,
      "learning_rate": 4.002416145537236e-05,
      "loss": 1.5655,
      "step": 3157
    },
    {
      "epoch": 4.3983286908078,
      "grad_norm": 4.196826934814453,
      "learning_rate": 4.001421262080727e-05,
      "loss": 2.0261,
      "step": 3158
    },
    {
      "epoch": 4.399721448467966,
      "grad_norm": 3.638808488845825,
      "learning_rate": 4.000426378624218e-05,
      "loss": 1.9075,
      "step": 3159
    },
    {
      "epoch": 4.401114206128134,
      "grad_norm": 3.1467158794403076,
      "learning_rate": 3.999431495167709e-05,
      "loss": 2.0249,
      "step": 3160
    },
    {
      "epoch": 4.4025069637883005,
      "grad_norm": 3.6672751903533936,
      "learning_rate": 3.9984366117112e-05,
      "loss": 1.9514,
      "step": 3161
    },
    {
      "epoch": 4.403899721448468,
      "grad_norm": 3.508639335632324,
      "learning_rate": 3.9974417282546896e-05,
      "loss": 1.762,
      "step": 3162
    },
    {
      "epoch": 4.405292479108635,
      "grad_norm": 3.3780128955841064,
      "learning_rate": 3.9964468447981806e-05,
      "loss": 2.0071,
      "step": 3163
    },
    {
      "epoch": 4.406685236768802,
      "grad_norm": 3.5957813262939453,
      "learning_rate": 3.995451961341671e-05,
      "loss": 2.0229,
      "step": 3164
    },
    {
      "epoch": 4.408077994428969,
      "grad_norm": 4.51296854019165,
      "learning_rate": 3.994457077885162e-05,
      "loss": 1.9209,
      "step": 3165
    },
    {
      "epoch": 4.4094707520891365,
      "grad_norm": 3.6258111000061035,
      "learning_rate": 3.9934621944286525e-05,
      "loss": 2.0767,
      "step": 3166
    },
    {
      "epoch": 4.410863509749303,
      "grad_norm": 3.485344886779785,
      "learning_rate": 3.992467310972143e-05,
      "loss": 1.8335,
      "step": 3167
    },
    {
      "epoch": 4.412256267409471,
      "grad_norm": 3.6054396629333496,
      "learning_rate": 3.9914724275156334e-05,
      "loss": 1.8713,
      "step": 3168
    },
    {
      "epoch": 4.413649025069638,
      "grad_norm": 3.383699893951416,
      "learning_rate": 3.9904775440591244e-05,
      "loss": 1.7211,
      "step": 3169
    },
    {
      "epoch": 4.415041782729805,
      "grad_norm": 3.8874316215515137,
      "learning_rate": 3.989482660602615e-05,
      "loss": 2.1207,
      "step": 3170
    },
    {
      "epoch": 4.4164345403899725,
      "grad_norm": 6.061794757843018,
      "learning_rate": 3.988487777146105e-05,
      "loss": 1.5361,
      "step": 3171
    },
    {
      "epoch": 4.417827298050139,
      "grad_norm": 3.5344083309173584,
      "learning_rate": 3.987492893689596e-05,
      "loss": 2.1361,
      "step": 3172
    },
    {
      "epoch": 4.419220055710307,
      "grad_norm": 3.9785499572753906,
      "learning_rate": 3.986498010233087e-05,
      "loss": 2.3867,
      "step": 3173
    },
    {
      "epoch": 4.420612813370473,
      "grad_norm": 3.3845369815826416,
      "learning_rate": 3.985503126776577e-05,
      "loss": 1.5828,
      "step": 3174
    },
    {
      "epoch": 4.422005571030641,
      "grad_norm": 3.99165940284729,
      "learning_rate": 3.984508243320068e-05,
      "loss": 1.9207,
      "step": 3175
    },
    {
      "epoch": 4.423398328690808,
      "grad_norm": 3.9511852264404297,
      "learning_rate": 3.983513359863558e-05,
      "loss": 2.1961,
      "step": 3176
    },
    {
      "epoch": 4.424791086350975,
      "grad_norm": 4.167506694793701,
      "learning_rate": 3.982518476407049e-05,
      "loss": 2.242,
      "step": 3177
    },
    {
      "epoch": 4.426183844011142,
      "grad_norm": 4.409764289855957,
      "learning_rate": 3.9815235929505395e-05,
      "loss": 2.0475,
      "step": 3178
    },
    {
      "epoch": 4.427576601671309,
      "grad_norm": 3.4161460399627686,
      "learning_rate": 3.9805287094940306e-05,
      "loss": 1.8625,
      "step": 3179
    },
    {
      "epoch": 4.428969359331476,
      "grad_norm": 3.3184471130371094,
      "learning_rate": 3.979533826037521e-05,
      "loss": 1.7828,
      "step": 3180
    },
    {
      "epoch": 4.430362116991644,
      "grad_norm": 4.11037015914917,
      "learning_rate": 3.9785389425810114e-05,
      "loss": 1.5766,
      "step": 3181
    },
    {
      "epoch": 4.43175487465181,
      "grad_norm": 4.095383644104004,
      "learning_rate": 3.977544059124502e-05,
      "loss": 1.8509,
      "step": 3182
    },
    {
      "epoch": 4.433147632311978,
      "grad_norm": 4.078642845153809,
      "learning_rate": 3.976549175667993e-05,
      "loss": 1.9048,
      "step": 3183
    },
    {
      "epoch": 4.4345403899721445,
      "grad_norm": 3.6307432651519775,
      "learning_rate": 3.975554292211483e-05,
      "loss": 1.7649,
      "step": 3184
    },
    {
      "epoch": 4.435933147632312,
      "grad_norm": 3.682129383087158,
      "learning_rate": 3.9745594087549744e-05,
      "loss": 1.8686,
      "step": 3185
    },
    {
      "epoch": 4.437325905292479,
      "grad_norm": 3.911226987838745,
      "learning_rate": 3.973564525298464e-05,
      "loss": 2.2978,
      "step": 3186
    },
    {
      "epoch": 4.438718662952646,
      "grad_norm": 3.18542742729187,
      "learning_rate": 3.972569641841955e-05,
      "loss": 1.5903,
      "step": 3187
    },
    {
      "epoch": 4.440111420612814,
      "grad_norm": 3.3501884937286377,
      "learning_rate": 3.971574758385446e-05,
      "loss": 1.3564,
      "step": 3188
    },
    {
      "epoch": 4.4415041782729805,
      "grad_norm": 4.470577716827393,
      "learning_rate": 3.9705798749289367e-05,
      "loss": 1.8826,
      "step": 3189
    },
    {
      "epoch": 4.442896935933147,
      "grad_norm": 3.6222612857818604,
      "learning_rate": 3.969584991472428e-05,
      "loss": 1.8286,
      "step": 3190
    },
    {
      "epoch": 4.444289693593315,
      "grad_norm": 3.79026460647583,
      "learning_rate": 3.9685901080159175e-05,
      "loss": 2.1936,
      "step": 3191
    },
    {
      "epoch": 4.445682451253482,
      "grad_norm": 3.869800329208374,
      "learning_rate": 3.9675952245594086e-05,
      "loss": 2.1218,
      "step": 3192
    },
    {
      "epoch": 4.447075208913649,
      "grad_norm": 3.4443883895874023,
      "learning_rate": 3.966600341102899e-05,
      "loss": 2.0388,
      "step": 3193
    },
    {
      "epoch": 4.4484679665738165,
      "grad_norm": 3.4055368900299072,
      "learning_rate": 3.96560545764639e-05,
      "loss": 1.6828,
      "step": 3194
    },
    {
      "epoch": 4.449860724233983,
      "grad_norm": 3.665839910507202,
      "learning_rate": 3.9646105741898805e-05,
      "loss": 2.0586,
      "step": 3195
    },
    {
      "epoch": 4.451253481894151,
      "grad_norm": 3.516286849975586,
      "learning_rate": 3.963615690733371e-05,
      "loss": 2.0749,
      "step": 3196
    },
    {
      "epoch": 4.452646239554317,
      "grad_norm": 4.141451835632324,
      "learning_rate": 3.962620807276861e-05,
      "loss": 2.3455,
      "step": 3197
    },
    {
      "epoch": 4.454038997214485,
      "grad_norm": 4.327914237976074,
      "learning_rate": 3.9616259238203524e-05,
      "loss": 2.3216,
      "step": 3198
    },
    {
      "epoch": 4.455431754874652,
      "grad_norm": 3.882049560546875,
      "learning_rate": 3.960631040363843e-05,
      "loss": 2.0532,
      "step": 3199
    },
    {
      "epoch": 4.456824512534819,
      "grad_norm": 3.3833377361297607,
      "learning_rate": 3.959636156907334e-05,
      "loss": 1.5927,
      "step": 3200
    },
    {
      "epoch": 4.458217270194986,
      "grad_norm": 3.203939199447632,
      "learning_rate": 3.9586412734508236e-05,
      "loss": 1.6464,
      "step": 3201
    },
    {
      "epoch": 4.459610027855153,
      "grad_norm": 3.5028281211853027,
      "learning_rate": 3.957646389994315e-05,
      "loss": 1.9767,
      "step": 3202
    },
    {
      "epoch": 4.46100278551532,
      "grad_norm": 4.2450666427612305,
      "learning_rate": 3.956651506537805e-05,
      "loss": 1.7586,
      "step": 3203
    },
    {
      "epoch": 4.462395543175488,
      "grad_norm": 3.6107778549194336,
      "learning_rate": 3.955656623081296e-05,
      "loss": 1.8337,
      "step": 3204
    },
    {
      "epoch": 4.463788300835654,
      "grad_norm": 3.225759267807007,
      "learning_rate": 3.9546617396247866e-05,
      "loss": 1.7373,
      "step": 3205
    },
    {
      "epoch": 4.465181058495822,
      "grad_norm": 3.8535566329956055,
      "learning_rate": 3.953666856168277e-05,
      "loss": 1.9903,
      "step": 3206
    },
    {
      "epoch": 4.4665738161559885,
      "grad_norm": 3.146134376525879,
      "learning_rate": 3.9526719727117674e-05,
      "loss": 1.5497,
      "step": 3207
    },
    {
      "epoch": 4.467966573816156,
      "grad_norm": 3.726294994354248,
      "learning_rate": 3.9516770892552585e-05,
      "loss": 2.1184,
      "step": 3208
    },
    {
      "epoch": 4.469359331476323,
      "grad_norm": 3.102433681488037,
      "learning_rate": 3.950682205798749e-05,
      "loss": 1.513,
      "step": 3209
    },
    {
      "epoch": 4.47075208913649,
      "grad_norm": 4.219781398773193,
      "learning_rate": 3.94968732234224e-05,
      "loss": 1.9301,
      "step": 3210
    },
    {
      "epoch": 4.472144846796658,
      "grad_norm": 3.8631253242492676,
      "learning_rate": 3.94869243888573e-05,
      "loss": 1.9974,
      "step": 3211
    },
    {
      "epoch": 4.4735376044568245,
      "grad_norm": 3.584851026535034,
      "learning_rate": 3.947697555429221e-05,
      "loss": 1.9607,
      "step": 3212
    },
    {
      "epoch": 4.474930362116992,
      "grad_norm": 3.701464891433716,
      "learning_rate": 3.946702671972711e-05,
      "loss": 1.8906,
      "step": 3213
    },
    {
      "epoch": 4.476323119777159,
      "grad_norm": 3.6301727294921875,
      "learning_rate": 3.945707788516202e-05,
      "loss": 1.9891,
      "step": 3214
    },
    {
      "epoch": 4.477715877437326,
      "grad_norm": 3.6762855052948,
      "learning_rate": 3.944712905059692e-05,
      "loss": 1.6908,
      "step": 3215
    },
    {
      "epoch": 4.479108635097493,
      "grad_norm": 3.854891061782837,
      "learning_rate": 3.943718021603183e-05,
      "loss": 2.3549,
      "step": 3216
    },
    {
      "epoch": 4.4805013927576605,
      "grad_norm": 3.8298966884613037,
      "learning_rate": 3.942723138146674e-05,
      "loss": 1.9698,
      "step": 3217
    },
    {
      "epoch": 4.481894150417827,
      "grad_norm": 4.6079912185668945,
      "learning_rate": 3.9417282546901646e-05,
      "loss": 1.8695,
      "step": 3218
    },
    {
      "epoch": 4.483286908077995,
      "grad_norm": 3.379887580871582,
      "learning_rate": 3.940733371233656e-05,
      "loss": 1.6363,
      "step": 3219
    },
    {
      "epoch": 4.484679665738161,
      "grad_norm": 3.9019153118133545,
      "learning_rate": 3.9397384877771454e-05,
      "loss": 1.6475,
      "step": 3220
    },
    {
      "epoch": 4.486072423398329,
      "grad_norm": 4.2392897605896,
      "learning_rate": 3.9387436043206365e-05,
      "loss": 2.2761,
      "step": 3221
    },
    {
      "epoch": 4.487465181058496,
      "grad_norm": 3.652559995651245,
      "learning_rate": 3.937748720864127e-05,
      "loss": 2.1351,
      "step": 3222
    },
    {
      "epoch": 4.488857938718663,
      "grad_norm": 3.350752592086792,
      "learning_rate": 3.936753837407618e-05,
      "loss": 1.6802,
      "step": 3223
    },
    {
      "epoch": 4.49025069637883,
      "grad_norm": 3.8682892322540283,
      "learning_rate": 3.9357589539511084e-05,
      "loss": 2.0835,
      "step": 3224
    },
    {
      "epoch": 4.491643454038997,
      "grad_norm": 3.9659388065338135,
      "learning_rate": 3.934764070494599e-05,
      "loss": 2.0425,
      "step": 3225
    },
    {
      "epoch": 4.493036211699164,
      "grad_norm": 4.409570693969727,
      "learning_rate": 3.933769187038089e-05,
      "loss": 1.8561,
      "step": 3226
    },
    {
      "epoch": 4.494428969359332,
      "grad_norm": 3.5168051719665527,
      "learning_rate": 3.93277430358158e-05,
      "loss": 1.995,
      "step": 3227
    },
    {
      "epoch": 4.495821727019498,
      "grad_norm": 3.20039701461792,
      "learning_rate": 3.931779420125071e-05,
      "loss": 1.6086,
      "step": 3228
    },
    {
      "epoch": 4.497214484679666,
      "grad_norm": 3.4170303344726562,
      "learning_rate": 3.930784536668562e-05,
      "loss": 1.7782,
      "step": 3229
    },
    {
      "epoch": 4.498607242339833,
      "grad_norm": 3.441661834716797,
      "learning_rate": 3.9297896532120515e-05,
      "loss": 1.5542,
      "step": 3230
    },
    {
      "epoch": 4.5,
      "grad_norm": 4.464572906494141,
      "learning_rate": 3.9287947697555426e-05,
      "loss": 1.8985,
      "step": 3231
    },
    {
      "epoch": 4.501392757660167,
      "grad_norm": 3.6527247428894043,
      "learning_rate": 3.927799886299033e-05,
      "loss": 1.9854,
      "step": 3232
    },
    {
      "epoch": 4.502785515320334,
      "grad_norm": 3.33829927444458,
      "learning_rate": 3.926805002842524e-05,
      "loss": 1.8484,
      "step": 3233
    },
    {
      "epoch": 4.504178272980502,
      "grad_norm": 3.4014110565185547,
      "learning_rate": 3.9258101193860145e-05,
      "loss": 1.9452,
      "step": 3234
    },
    {
      "epoch": 4.505571030640668,
      "grad_norm": 3.632737398147583,
      "learning_rate": 3.924815235929505e-05,
      "loss": 1.8714,
      "step": 3235
    },
    {
      "epoch": 4.506963788300836,
      "grad_norm": 3.3304083347320557,
      "learning_rate": 3.923820352472995e-05,
      "loss": 1.8572,
      "step": 3236
    },
    {
      "epoch": 4.508356545961003,
      "grad_norm": 3.681295156478882,
      "learning_rate": 3.9228254690164864e-05,
      "loss": 1.8286,
      "step": 3237
    },
    {
      "epoch": 4.50974930362117,
      "grad_norm": 3.4436419010162354,
      "learning_rate": 3.921830585559977e-05,
      "loss": 1.8035,
      "step": 3238
    },
    {
      "epoch": 4.511142061281337,
      "grad_norm": 3.5060203075408936,
      "learning_rate": 3.920835702103468e-05,
      "loss": 1.9942,
      "step": 3239
    },
    {
      "epoch": 4.512534818941504,
      "grad_norm": 3.54351806640625,
      "learning_rate": 3.9198408186469576e-05,
      "loss": 1.8422,
      "step": 3240
    },
    {
      "epoch": 4.513927576601671,
      "grad_norm": 4.204930305480957,
      "learning_rate": 3.918845935190449e-05,
      "loss": 1.98,
      "step": 3241
    },
    {
      "epoch": 4.515320334261839,
      "grad_norm": 4.28328800201416,
      "learning_rate": 3.917851051733939e-05,
      "loss": 1.7543,
      "step": 3242
    },
    {
      "epoch": 4.516713091922005,
      "grad_norm": 3.403895378112793,
      "learning_rate": 3.91685616827743e-05,
      "loss": 1.797,
      "step": 3243
    },
    {
      "epoch": 4.518105849582173,
      "grad_norm": 4.003249645233154,
      "learning_rate": 3.915861284820921e-05,
      "loss": 1.915,
      "step": 3244
    },
    {
      "epoch": 4.5194986072423395,
      "grad_norm": 3.870506763458252,
      "learning_rate": 3.914866401364411e-05,
      "loss": 1.8718,
      "step": 3245
    },
    {
      "epoch": 4.520891364902507,
      "grad_norm": 3.469045877456665,
      "learning_rate": 3.913871517907902e-05,
      "loss": 1.9785,
      "step": 3246
    },
    {
      "epoch": 4.522284122562674,
      "grad_norm": 3.2671995162963867,
      "learning_rate": 3.9128766344513925e-05,
      "loss": 1.9109,
      "step": 3247
    },
    {
      "epoch": 4.523676880222841,
      "grad_norm": 3.3625526428222656,
      "learning_rate": 3.9118817509948836e-05,
      "loss": 1.7483,
      "step": 3248
    },
    {
      "epoch": 4.525069637883008,
      "grad_norm": 3.8905181884765625,
      "learning_rate": 3.910886867538374e-05,
      "loss": 2.4342,
      "step": 3249
    },
    {
      "epoch": 4.5264623955431755,
      "grad_norm": 4.0306243896484375,
      "learning_rate": 3.9098919840818644e-05,
      "loss": 2.2661,
      "step": 3250
    },
    {
      "epoch": 4.527855153203342,
      "grad_norm": 4.22361946105957,
      "learning_rate": 3.908897100625355e-05,
      "loss": 1.9663,
      "step": 3251
    },
    {
      "epoch": 4.52924791086351,
      "grad_norm": 4.0032806396484375,
      "learning_rate": 3.907902217168846e-05,
      "loss": 2.3843,
      "step": 3252
    },
    {
      "epoch": 4.530640668523677,
      "grad_norm": 3.5702474117279053,
      "learning_rate": 3.906907333712336e-05,
      "loss": 1.9604,
      "step": 3253
    },
    {
      "epoch": 4.532033426183844,
      "grad_norm": 4.1573710441589355,
      "learning_rate": 3.9059124502558274e-05,
      "loss": 2.1405,
      "step": 3254
    },
    {
      "epoch": 4.5334261838440115,
      "grad_norm": 3.67299485206604,
      "learning_rate": 3.904917566799317e-05,
      "loss": 1.682,
      "step": 3255
    },
    {
      "epoch": 4.534818941504178,
      "grad_norm": 3.53426456451416,
      "learning_rate": 3.903922683342808e-05,
      "loss": 1.9956,
      "step": 3256
    },
    {
      "epoch": 4.536211699164346,
      "grad_norm": 3.801340341567993,
      "learning_rate": 3.9029277998862986e-05,
      "loss": 2.4488,
      "step": 3257
    },
    {
      "epoch": 4.537604456824512,
      "grad_norm": 3.5699822902679443,
      "learning_rate": 3.90193291642979e-05,
      "loss": 1.9663,
      "step": 3258
    },
    {
      "epoch": 4.53899721448468,
      "grad_norm": 4.429647445678711,
      "learning_rate": 3.9009380329732794e-05,
      "loss": 1.8785,
      "step": 3259
    },
    {
      "epoch": 4.540389972144847,
      "grad_norm": 3.939802885055542,
      "learning_rate": 3.8999431495167705e-05,
      "loss": 2.0498,
      "step": 3260
    },
    {
      "epoch": 4.541782729805014,
      "grad_norm": 3.1633245944976807,
      "learning_rate": 3.898948266060261e-05,
      "loss": 1.4715,
      "step": 3261
    },
    {
      "epoch": 4.543175487465181,
      "grad_norm": 3.1759512424468994,
      "learning_rate": 3.897953382603752e-05,
      "loss": 1.6324,
      "step": 3262
    },
    {
      "epoch": 4.544568245125348,
      "grad_norm": 3.4950180053710938,
      "learning_rate": 3.8969584991472424e-05,
      "loss": 1.8991,
      "step": 3263
    },
    {
      "epoch": 4.545961002785515,
      "grad_norm": 3.6954591274261475,
      "learning_rate": 3.895963615690733e-05,
      "loss": 1.9564,
      "step": 3264
    },
    {
      "epoch": 4.547353760445683,
      "grad_norm": 3.3584558963775635,
      "learning_rate": 3.894968732234223e-05,
      "loss": 1.9633,
      "step": 3265
    },
    {
      "epoch": 4.548746518105849,
      "grad_norm": 4.469377040863037,
      "learning_rate": 3.893973848777714e-05,
      "loss": 1.4949,
      "step": 3266
    },
    {
      "epoch": 4.550139275766017,
      "grad_norm": 3.7999792098999023,
      "learning_rate": 3.892978965321205e-05,
      "loss": 1.5881,
      "step": 3267
    },
    {
      "epoch": 4.5515320334261835,
      "grad_norm": 3.7202396392822266,
      "learning_rate": 3.891984081864696e-05,
      "loss": 2.0198,
      "step": 3268
    },
    {
      "epoch": 4.552924791086351,
      "grad_norm": 3.222660541534424,
      "learning_rate": 3.8909891984081855e-05,
      "loss": 1.7424,
      "step": 3269
    },
    {
      "epoch": 4.554317548746518,
      "grad_norm": 3.1136035919189453,
      "learning_rate": 3.8899943149516766e-05,
      "loss": 1.6429,
      "step": 3270
    },
    {
      "epoch": 4.555710306406685,
      "grad_norm": 4.08917236328125,
      "learning_rate": 3.888999431495167e-05,
      "loss": 1.8382,
      "step": 3271
    },
    {
      "epoch": 4.557103064066853,
      "grad_norm": 3.8375301361083984,
      "learning_rate": 3.888004548038658e-05,
      "loss": 1.93,
      "step": 3272
    },
    {
      "epoch": 4.5584958217270195,
      "grad_norm": 3.7703723907470703,
      "learning_rate": 3.887009664582149e-05,
      "loss": 1.9536,
      "step": 3273
    },
    {
      "epoch": 4.559888579387186,
      "grad_norm": 3.5565130710601807,
      "learning_rate": 3.886014781125639e-05,
      "loss": 2.0665,
      "step": 3274
    },
    {
      "epoch": 4.561281337047354,
      "grad_norm": 3.986150026321411,
      "learning_rate": 3.88501989766913e-05,
      "loss": 1.7912,
      "step": 3275
    },
    {
      "epoch": 4.562674094707521,
      "grad_norm": 3.5278992652893066,
      "learning_rate": 3.8840250142126204e-05,
      "loss": 1.6781,
      "step": 3276
    },
    {
      "epoch": 4.564066852367688,
      "grad_norm": 5.309002876281738,
      "learning_rate": 3.8830301307561115e-05,
      "loss": 1.7468,
      "step": 3277
    },
    {
      "epoch": 4.5654596100278555,
      "grad_norm": 3.887995958328247,
      "learning_rate": 3.882035247299602e-05,
      "loss": 1.7888,
      "step": 3278
    },
    {
      "epoch": 4.566852367688022,
      "grad_norm": 2.9909744262695312,
      "learning_rate": 3.881040363843092e-05,
      "loss": 1.3848,
      "step": 3279
    },
    {
      "epoch": 4.56824512534819,
      "grad_norm": 3.7661995887756348,
      "learning_rate": 3.880045480386583e-05,
      "loss": 1.7164,
      "step": 3280
    },
    {
      "epoch": 4.569637883008356,
      "grad_norm": 4.145754337310791,
      "learning_rate": 3.879050596930074e-05,
      "loss": 1.9152,
      "step": 3281
    },
    {
      "epoch": 4.571030640668524,
      "grad_norm": 3.8252034187316895,
      "learning_rate": 3.878055713473564e-05,
      "loss": 2.0524,
      "step": 3282
    },
    {
      "epoch": 4.572423398328691,
      "grad_norm": 4.5574164390563965,
      "learning_rate": 3.877060830017055e-05,
      "loss": 1.9094,
      "step": 3283
    },
    {
      "epoch": 4.573816155988858,
      "grad_norm": 4.0055012702941895,
      "learning_rate": 3.876065946560545e-05,
      "loss": 2.1733,
      "step": 3284
    },
    {
      "epoch": 4.575208913649025,
      "grad_norm": 3.6039435863494873,
      "learning_rate": 3.875071063104036e-05,
      "loss": 1.7209,
      "step": 3285
    },
    {
      "epoch": 4.576601671309192,
      "grad_norm": 3.5629451274871826,
      "learning_rate": 3.8740761796475265e-05,
      "loss": 2.0675,
      "step": 3286
    },
    {
      "epoch": 4.577994428969359,
      "grad_norm": 4.082374572753906,
      "learning_rate": 3.8730812961910176e-05,
      "loss": 1.9238,
      "step": 3287
    },
    {
      "epoch": 4.579387186629527,
      "grad_norm": 3.620668888092041,
      "learning_rate": 3.872086412734508e-05,
      "loss": 2.0418,
      "step": 3288
    },
    {
      "epoch": 4.580779944289693,
      "grad_norm": 3.4701499938964844,
      "learning_rate": 3.8710915292779984e-05,
      "loss": 1.7259,
      "step": 3289
    },
    {
      "epoch": 4.582172701949861,
      "grad_norm": 4.000033855438232,
      "learning_rate": 3.870096645821489e-05,
      "loss": 2.0742,
      "step": 3290
    },
    {
      "epoch": 4.5835654596100275,
      "grad_norm": 5.691669464111328,
      "learning_rate": 3.86910176236498e-05,
      "loss": 1.8358,
      "step": 3291
    },
    {
      "epoch": 4.584958217270195,
      "grad_norm": 3.2448883056640625,
      "learning_rate": 3.86810687890847e-05,
      "loss": 1.7106,
      "step": 3292
    },
    {
      "epoch": 4.586350974930362,
      "grad_norm": 3.6653659343719482,
      "learning_rate": 3.8671119954519614e-05,
      "loss": 1.8573,
      "step": 3293
    },
    {
      "epoch": 4.587743732590529,
      "grad_norm": 3.8329854011535645,
      "learning_rate": 3.866117111995451e-05,
      "loss": 2.0607,
      "step": 3294
    },
    {
      "epoch": 4.589136490250697,
      "grad_norm": 3.4483561515808105,
      "learning_rate": 3.865122228538942e-05,
      "loss": 1.7451,
      "step": 3295
    },
    {
      "epoch": 4.5905292479108635,
      "grad_norm": 3.327455520629883,
      "learning_rate": 3.8641273450824326e-05,
      "loss": 1.849,
      "step": 3296
    },
    {
      "epoch": 4.591922005571031,
      "grad_norm": 3.6674351692199707,
      "learning_rate": 3.863132461625924e-05,
      "loss": 1.687,
      "step": 3297
    },
    {
      "epoch": 4.593314763231198,
      "grad_norm": 4.475555419921875,
      "learning_rate": 3.8621375781694134e-05,
      "loss": 1.7948,
      "step": 3298
    },
    {
      "epoch": 4.594707520891365,
      "grad_norm": 3.7766876220703125,
      "learning_rate": 3.8611426947129045e-05,
      "loss": 1.87,
      "step": 3299
    },
    {
      "epoch": 4.596100278551532,
      "grad_norm": 3.52447247505188,
      "learning_rate": 3.8601478112563956e-05,
      "loss": 1.9438,
      "step": 3300
    },
    {
      "epoch": 4.5974930362116995,
      "grad_norm": 3.5671544075012207,
      "learning_rate": 3.859152927799886e-05,
      "loss": 2.0848,
      "step": 3301
    },
    {
      "epoch": 4.598885793871866,
      "grad_norm": 3.7626819610595703,
      "learning_rate": 3.858158044343377e-05,
      "loss": 1.7089,
      "step": 3302
    },
    {
      "epoch": 4.600278551532034,
      "grad_norm": 3.4325287342071533,
      "learning_rate": 3.857163160886867e-05,
      "loss": 1.6694,
      "step": 3303
    },
    {
      "epoch": 4.6016713091922,
      "grad_norm": 3.536877393722534,
      "learning_rate": 3.856168277430358e-05,
      "loss": 1.7793,
      "step": 3304
    },
    {
      "epoch": 4.603064066852368,
      "grad_norm": 3.5021097660064697,
      "learning_rate": 3.855173393973848e-05,
      "loss": 1.5806,
      "step": 3305
    },
    {
      "epoch": 4.604456824512535,
      "grad_norm": 4.428031921386719,
      "learning_rate": 3.8541785105173394e-05,
      "loss": 1.749,
      "step": 3306
    },
    {
      "epoch": 4.605849582172702,
      "grad_norm": 3.478275775909424,
      "learning_rate": 3.85318362706083e-05,
      "loss": 1.8083,
      "step": 3307
    },
    {
      "epoch": 4.607242339832869,
      "grad_norm": 3.6865408420562744,
      "learning_rate": 3.85218874360432e-05,
      "loss": 2.1574,
      "step": 3308
    },
    {
      "epoch": 4.608635097493036,
      "grad_norm": 3.195882558822632,
      "learning_rate": 3.8511938601478106e-05,
      "loss": 1.8088,
      "step": 3309
    },
    {
      "epoch": 4.610027855153203,
      "grad_norm": 3.710249662399292,
      "learning_rate": 3.850198976691302e-05,
      "loss": 2.3266,
      "step": 3310
    },
    {
      "epoch": 4.611420612813371,
      "grad_norm": 3.181318521499634,
      "learning_rate": 3.849204093234792e-05,
      "loss": 1.7057,
      "step": 3311
    },
    {
      "epoch": 4.612813370473537,
      "grad_norm": 3.261418342590332,
      "learning_rate": 3.848209209778283e-05,
      "loss": 2.0305,
      "step": 3312
    },
    {
      "epoch": 4.614206128133705,
      "grad_norm": 3.871584415435791,
      "learning_rate": 3.847214326321773e-05,
      "loss": 1.8123,
      "step": 3313
    },
    {
      "epoch": 4.615598885793872,
      "grad_norm": 3.5570790767669678,
      "learning_rate": 3.846219442865264e-05,
      "loss": 1.8704,
      "step": 3314
    },
    {
      "epoch": 4.616991643454039,
      "grad_norm": 3.5778913497924805,
      "learning_rate": 3.8452245594087544e-05,
      "loss": 2.0716,
      "step": 3315
    },
    {
      "epoch": 4.618384401114206,
      "grad_norm": 3.684006929397583,
      "learning_rate": 3.8442296759522455e-05,
      "loss": 1.622,
      "step": 3316
    },
    {
      "epoch": 4.619777158774373,
      "grad_norm": 4.059329032897949,
      "learning_rate": 3.843234792495736e-05,
      "loss": 2.3033,
      "step": 3317
    },
    {
      "epoch": 4.621169916434541,
      "grad_norm": 3.6581759452819824,
      "learning_rate": 3.842239909039226e-05,
      "loss": 2.1392,
      "step": 3318
    },
    {
      "epoch": 4.6225626740947074,
      "grad_norm": 3.5814528465270996,
      "learning_rate": 3.841245025582717e-05,
      "loss": 1.8765,
      "step": 3319
    },
    {
      "epoch": 4.623955431754875,
      "grad_norm": 3.860010862350464,
      "learning_rate": 3.840250142126208e-05,
      "loss": 2.1593,
      "step": 3320
    },
    {
      "epoch": 4.625348189415042,
      "grad_norm": 4.103090286254883,
      "learning_rate": 3.839255258669698e-05,
      "loss": 2.0104,
      "step": 3321
    },
    {
      "epoch": 4.626740947075209,
      "grad_norm": 3.645543098449707,
      "learning_rate": 3.838260375213189e-05,
      "loss": 1.7529,
      "step": 3322
    },
    {
      "epoch": 4.628133704735376,
      "grad_norm": 4.183781623840332,
      "learning_rate": 3.837265491756679e-05,
      "loss": 1.8564,
      "step": 3323
    },
    {
      "epoch": 4.629526462395543,
      "grad_norm": 3.8604724407196045,
      "learning_rate": 3.83627060830017e-05,
      "loss": 2.3319,
      "step": 3324
    },
    {
      "epoch": 4.63091922005571,
      "grad_norm": 3.996187210083008,
      "learning_rate": 3.8352757248436605e-05,
      "loss": 2.2144,
      "step": 3325
    },
    {
      "epoch": 4.632311977715878,
      "grad_norm": 3.186293125152588,
      "learning_rate": 3.8342808413871516e-05,
      "loss": 1.8631,
      "step": 3326
    },
    {
      "epoch": 4.633704735376044,
      "grad_norm": 3.9150609970092773,
      "learning_rate": 3.833285957930643e-05,
      "loss": 1.9915,
      "step": 3327
    },
    {
      "epoch": 4.635097493036212,
      "grad_norm": 3.5735533237457275,
      "learning_rate": 3.8322910744741324e-05,
      "loss": 1.8528,
      "step": 3328
    },
    {
      "epoch": 4.6364902506963785,
      "grad_norm": 3.7437288761138916,
      "learning_rate": 3.8312961910176235e-05,
      "loss": 2.0451,
      "step": 3329
    },
    {
      "epoch": 4.637883008356546,
      "grad_norm": 3.541717767715454,
      "learning_rate": 3.830301307561114e-05,
      "loss": 1.7844,
      "step": 3330
    },
    {
      "epoch": 4.639275766016713,
      "grad_norm": 3.573110818862915,
      "learning_rate": 3.829306424104605e-05,
      "loss": 1.8942,
      "step": 3331
    },
    {
      "epoch": 4.64066852367688,
      "grad_norm": 4.328987121582031,
      "learning_rate": 3.8283115406480954e-05,
      "loss": 2.1885,
      "step": 3332
    },
    {
      "epoch": 4.642061281337047,
      "grad_norm": 4.351795196533203,
      "learning_rate": 3.827316657191586e-05,
      "loss": 1.9701,
      "step": 3333
    },
    {
      "epoch": 4.6434540389972145,
      "grad_norm": 3.7214114665985107,
      "learning_rate": 3.826321773735076e-05,
      "loss": 1.8902,
      "step": 3334
    },
    {
      "epoch": 4.644846796657381,
      "grad_norm": 4.347647190093994,
      "learning_rate": 3.825326890278567e-05,
      "loss": 1.9213,
      "step": 3335
    },
    {
      "epoch": 4.646239554317549,
      "grad_norm": 4.3626909255981445,
      "learning_rate": 3.824332006822058e-05,
      "loss": 2.0552,
      "step": 3336
    },
    {
      "epoch": 4.647632311977716,
      "grad_norm": 5.106423854827881,
      "learning_rate": 3.823337123365549e-05,
      "loss": 2.0844,
      "step": 3337
    },
    {
      "epoch": 4.649025069637883,
      "grad_norm": 3.6310203075408936,
      "learning_rate": 3.8223422399090385e-05,
      "loss": 1.6776,
      "step": 3338
    },
    {
      "epoch": 4.65041782729805,
      "grad_norm": 4.02175235748291,
      "learning_rate": 3.8213473564525296e-05,
      "loss": 2.0913,
      "step": 3339
    },
    {
      "epoch": 4.651810584958217,
      "grad_norm": 3.364502191543579,
      "learning_rate": 3.82035247299602e-05,
      "loss": 1.6781,
      "step": 3340
    },
    {
      "epoch": 4.653203342618385,
      "grad_norm": 3.457348108291626,
      "learning_rate": 3.819357589539511e-05,
      "loss": 2.0024,
      "step": 3341
    },
    {
      "epoch": 4.654596100278551,
      "grad_norm": 3.8440089225769043,
      "learning_rate": 3.818362706083001e-05,
      "loss": 1.85,
      "step": 3342
    },
    {
      "epoch": 4.655988857938719,
      "grad_norm": 3.5063223838806152,
      "learning_rate": 3.817367822626492e-05,
      "loss": 1.8078,
      "step": 3343
    },
    {
      "epoch": 4.657381615598886,
      "grad_norm": 3.236264228820801,
      "learning_rate": 3.8163729391699823e-05,
      "loss": 1.6326,
      "step": 3344
    },
    {
      "epoch": 4.658774373259053,
      "grad_norm": 3.800360918045044,
      "learning_rate": 3.8153780557134734e-05,
      "loss": 2.0866,
      "step": 3345
    },
    {
      "epoch": 4.66016713091922,
      "grad_norm": 3.2970800399780273,
      "learning_rate": 3.814383172256964e-05,
      "loss": 1.8191,
      "step": 3346
    },
    {
      "epoch": 4.661559888579387,
      "grad_norm": 4.163506984710693,
      "learning_rate": 3.813388288800454e-05,
      "loss": 1.781,
      "step": 3347
    },
    {
      "epoch": 4.662952646239554,
      "grad_norm": 3.7520077228546143,
      "learning_rate": 3.8123934053439447e-05,
      "loss": 1.9486,
      "step": 3348
    },
    {
      "epoch": 4.664345403899722,
      "grad_norm": 3.782254934310913,
      "learning_rate": 3.811398521887436e-05,
      "loss": 2.0663,
      "step": 3349
    },
    {
      "epoch": 4.665738161559888,
      "grad_norm": 3.776536464691162,
      "learning_rate": 3.810403638430926e-05,
      "loss": 1.8538,
      "step": 3350
    },
    {
      "epoch": 4.667130919220056,
      "grad_norm": 3.2819957733154297,
      "learning_rate": 3.809408754974417e-05,
      "loss": 1.7891,
      "step": 3351
    },
    {
      "epoch": 4.6685236768802225,
      "grad_norm": 4.100096702575684,
      "learning_rate": 3.808413871517907e-05,
      "loss": 1.9173,
      "step": 3352
    },
    {
      "epoch": 4.66991643454039,
      "grad_norm": 4.292869567871094,
      "learning_rate": 3.807418988061398e-05,
      "loss": 1.9693,
      "step": 3353
    },
    {
      "epoch": 4.671309192200557,
      "grad_norm": 3.77502703666687,
      "learning_rate": 3.8064241046048885e-05,
      "loss": 1.8048,
      "step": 3354
    },
    {
      "epoch": 4.672701949860724,
      "grad_norm": 3.7283897399902344,
      "learning_rate": 3.8054292211483795e-05,
      "loss": 1.7996,
      "step": 3355
    },
    {
      "epoch": 4.674094707520892,
      "grad_norm": 3.706955671310425,
      "learning_rate": 3.8044343376918706e-05,
      "loss": 1.9962,
      "step": 3356
    },
    {
      "epoch": 4.6754874651810585,
      "grad_norm": 3.4003515243530273,
      "learning_rate": 3.8034394542353604e-05,
      "loss": 1.5745,
      "step": 3357
    },
    {
      "epoch": 4.676880222841225,
      "grad_norm": 3.4651854038238525,
      "learning_rate": 3.8024445707788514e-05,
      "loss": 1.9375,
      "step": 3358
    },
    {
      "epoch": 4.678272980501393,
      "grad_norm": 3.9952895641326904,
      "learning_rate": 3.801449687322342e-05,
      "loss": 1.9034,
      "step": 3359
    },
    {
      "epoch": 4.67966573816156,
      "grad_norm": 3.463488817214966,
      "learning_rate": 3.800454803865833e-05,
      "loss": 1.8098,
      "step": 3360
    },
    {
      "epoch": 4.681058495821727,
      "grad_norm": 3.772735834121704,
      "learning_rate": 3.7994599204093233e-05,
      "loss": 1.9488,
      "step": 3361
    },
    {
      "epoch": 4.6824512534818945,
      "grad_norm": 3.6313321590423584,
      "learning_rate": 3.798465036952814e-05,
      "loss": 1.8965,
      "step": 3362
    },
    {
      "epoch": 4.683844011142061,
      "grad_norm": 3.1754677295684814,
      "learning_rate": 3.797470153496304e-05,
      "loss": 1.7626,
      "step": 3363
    },
    {
      "epoch": 4.685236768802229,
      "grad_norm": 3.6511735916137695,
      "learning_rate": 3.796475270039795e-05,
      "loss": 1.8018,
      "step": 3364
    },
    {
      "epoch": 4.686629526462395,
      "grad_norm": 4.095413684844971,
      "learning_rate": 3.7954803865832856e-05,
      "loss": 2.2402,
      "step": 3365
    },
    {
      "epoch": 4.688022284122563,
      "grad_norm": 3.337965726852417,
      "learning_rate": 3.794485503126777e-05,
      "loss": 1.6362,
      "step": 3366
    },
    {
      "epoch": 4.68941504178273,
      "grad_norm": 3.9275214672088623,
      "learning_rate": 3.7934906196702665e-05,
      "loss": 1.5543,
      "step": 3367
    },
    {
      "epoch": 4.690807799442897,
      "grad_norm": 3.974381446838379,
      "learning_rate": 3.7924957362137575e-05,
      "loss": 1.8771,
      "step": 3368
    },
    {
      "epoch": 4.692200557103064,
      "grad_norm": 4.257464408874512,
      "learning_rate": 3.791500852757248e-05,
      "loss": 1.8199,
      "step": 3369
    },
    {
      "epoch": 4.693593314763231,
      "grad_norm": 3.9581267833709717,
      "learning_rate": 3.790505969300739e-05,
      "loss": 2.0797,
      "step": 3370
    },
    {
      "epoch": 4.694986072423398,
      "grad_norm": 4.050745964050293,
      "learning_rate": 3.7895110858442294e-05,
      "loss": 1.5836,
      "step": 3371
    },
    {
      "epoch": 4.696378830083566,
      "grad_norm": 3.8411006927490234,
      "learning_rate": 3.78851620238772e-05,
      "loss": 1.9422,
      "step": 3372
    },
    {
      "epoch": 4.697771587743732,
      "grad_norm": 3.5904810428619385,
      "learning_rate": 3.78752131893121e-05,
      "loss": 1.6732,
      "step": 3373
    },
    {
      "epoch": 4.6991643454039,
      "grad_norm": 4.09559965133667,
      "learning_rate": 3.7865264354747013e-05,
      "loss": 1.9492,
      "step": 3374
    },
    {
      "epoch": 4.7005571030640665,
      "grad_norm": 3.2541232109069824,
      "learning_rate": 3.785531552018192e-05,
      "loss": 1.856,
      "step": 3375
    },
    {
      "epoch": 4.701949860724234,
      "grad_norm": 3.921887159347534,
      "learning_rate": 3.784536668561683e-05,
      "loss": 1.9857,
      "step": 3376
    },
    {
      "epoch": 4.703342618384401,
      "grad_norm": 3.925410509109497,
      "learning_rate": 3.7835417851051726e-05,
      "loss": 2.043,
      "step": 3377
    },
    {
      "epoch": 4.704735376044568,
      "grad_norm": 3.9396636486053467,
      "learning_rate": 3.7825469016486637e-05,
      "loss": 1.8628,
      "step": 3378
    },
    {
      "epoch": 4.706128133704736,
      "grad_norm": 3.7705271244049072,
      "learning_rate": 3.781552018192154e-05,
      "loss": 1.8439,
      "step": 3379
    },
    {
      "epoch": 4.7075208913649025,
      "grad_norm": 4.013179302215576,
      "learning_rate": 3.780557134735645e-05,
      "loss": 1.8958,
      "step": 3380
    },
    {
      "epoch": 4.708913649025069,
      "grad_norm": 4.116738319396973,
      "learning_rate": 3.779562251279136e-05,
      "loss": 1.9555,
      "step": 3381
    },
    {
      "epoch": 4.710306406685237,
      "grad_norm": 3.614222764968872,
      "learning_rate": 3.778567367822626e-05,
      "loss": 1.8712,
      "step": 3382
    },
    {
      "epoch": 4.711699164345404,
      "grad_norm": 4.237809658050537,
      "learning_rate": 3.7775724843661164e-05,
      "loss": 1.8454,
      "step": 3383
    },
    {
      "epoch": 4.713091922005571,
      "grad_norm": 3.4724388122558594,
      "learning_rate": 3.7765776009096075e-05,
      "loss": 1.7271,
      "step": 3384
    },
    {
      "epoch": 4.7144846796657385,
      "grad_norm": 3.5229036808013916,
      "learning_rate": 3.7755827174530985e-05,
      "loss": 1.9329,
      "step": 3385
    },
    {
      "epoch": 4.715877437325905,
      "grad_norm": 3.386718988418579,
      "learning_rate": 3.774587833996588e-05,
      "loss": 1.9594,
      "step": 3386
    },
    {
      "epoch": 4.717270194986073,
      "grad_norm": 3.710028648376465,
      "learning_rate": 3.7735929505400794e-05,
      "loss": 1.7565,
      "step": 3387
    },
    {
      "epoch": 4.718662952646239,
      "grad_norm": 4.659442901611328,
      "learning_rate": 3.77259806708357e-05,
      "loss": 1.7809,
      "step": 3388
    },
    {
      "epoch": 4.720055710306407,
      "grad_norm": 3.3721377849578857,
      "learning_rate": 3.771603183627061e-05,
      "loss": 1.675,
      "step": 3389
    },
    {
      "epoch": 4.721448467966574,
      "grad_norm": 3.6961467266082764,
      "learning_rate": 3.770608300170551e-05,
      "loss": 1.9312,
      "step": 3390
    },
    {
      "epoch": 4.722841225626741,
      "grad_norm": 3.807015895843506,
      "learning_rate": 3.769613416714042e-05,
      "loss": 1.9269,
      "step": 3391
    },
    {
      "epoch": 4.724233983286908,
      "grad_norm": 4.433807849884033,
      "learning_rate": 3.768618533257532e-05,
      "loss": 1.7858,
      "step": 3392
    },
    {
      "epoch": 4.725626740947075,
      "grad_norm": 3.145347833633423,
      "learning_rate": 3.767623649801023e-05,
      "loss": 1.4089,
      "step": 3393
    },
    {
      "epoch": 4.727019498607242,
      "grad_norm": 3.8560850620269775,
      "learning_rate": 3.7666287663445136e-05,
      "loss": 2.0344,
      "step": 3394
    },
    {
      "epoch": 4.72841225626741,
      "grad_norm": 4.012540340423584,
      "learning_rate": 3.7656338828880046e-05,
      "loss": 1.768,
      "step": 3395
    },
    {
      "epoch": 4.729805013927576,
      "grad_norm": 3.9792680740356445,
      "learning_rate": 3.7646389994314944e-05,
      "loss": 1.7829,
      "step": 3396
    },
    {
      "epoch": 4.731197771587744,
      "grad_norm": 3.9786007404327393,
      "learning_rate": 3.7636441159749855e-05,
      "loss": 2.1115,
      "step": 3397
    },
    {
      "epoch": 4.732590529247911,
      "grad_norm": 3.971142053604126,
      "learning_rate": 3.762649232518476e-05,
      "loss": 1.7771,
      "step": 3398
    },
    {
      "epoch": 4.733983286908078,
      "grad_norm": 3.923093557357788,
      "learning_rate": 3.761654349061967e-05,
      "loss": 1.8994,
      "step": 3399
    },
    {
      "epoch": 4.735376044568245,
      "grad_norm": 3.5467283725738525,
      "learning_rate": 3.7606594656054574e-05,
      "loss": 1.6225,
      "step": 3400
    },
    {
      "epoch": 4.736768802228412,
      "grad_norm": 4.023221492767334,
      "learning_rate": 3.759664582148948e-05,
      "loss": 1.7095,
      "step": 3401
    },
    {
      "epoch": 4.73816155988858,
      "grad_norm": 4.696751594543457,
      "learning_rate": 3.758669698692438e-05,
      "loss": 1.7263,
      "step": 3402
    },
    {
      "epoch": 4.7395543175487465,
      "grad_norm": 3.790797472000122,
      "learning_rate": 3.757674815235929e-05,
      "loss": 1.845,
      "step": 3403
    },
    {
      "epoch": 4.740947075208914,
      "grad_norm": 3.5207347869873047,
      "learning_rate": 3.75667993177942e-05,
      "loss": 2.0653,
      "step": 3404
    },
    {
      "epoch": 4.742339832869081,
      "grad_norm": 3.7906086444854736,
      "learning_rate": 3.755685048322911e-05,
      "loss": 1.7273,
      "step": 3405
    },
    {
      "epoch": 4.743732590529248,
      "grad_norm": 3.459949493408203,
      "learning_rate": 3.7546901648664005e-05,
      "loss": 1.6935,
      "step": 3406
    },
    {
      "epoch": 4.745125348189415,
      "grad_norm": 7.457804203033447,
      "learning_rate": 3.7536952814098916e-05,
      "loss": 1.7552,
      "step": 3407
    },
    {
      "epoch": 4.7465181058495824,
      "grad_norm": 4.607371807098389,
      "learning_rate": 3.752700397953382e-05,
      "loss": 2.175,
      "step": 3408
    },
    {
      "epoch": 4.747910863509749,
      "grad_norm": 3.4577276706695557,
      "learning_rate": 3.751705514496873e-05,
      "loss": 1.6552,
      "step": 3409
    },
    {
      "epoch": 4.749303621169917,
      "grad_norm": 3.7358248233795166,
      "learning_rate": 3.750710631040364e-05,
      "loss": 1.8011,
      "step": 3410
    },
    {
      "epoch": 4.750696378830083,
      "grad_norm": 3.9303715229034424,
      "learning_rate": 3.749715747583854e-05,
      "loss": 1.7354,
      "step": 3411
    },
    {
      "epoch": 4.752089136490251,
      "grad_norm": 5.308157444000244,
      "learning_rate": 3.748720864127344e-05,
      "loss": 1.6196,
      "step": 3412
    },
    {
      "epoch": 4.7534818941504176,
      "grad_norm": 4.389355182647705,
      "learning_rate": 3.7477259806708354e-05,
      "loss": 1.7116,
      "step": 3413
    },
    {
      "epoch": 4.754874651810585,
      "grad_norm": 4.6615376472473145,
      "learning_rate": 3.7467310972143265e-05,
      "loss": 1.6443,
      "step": 3414
    },
    {
      "epoch": 4.756267409470752,
      "grad_norm": 3.608461380004883,
      "learning_rate": 3.745736213757817e-05,
      "loss": 2.0131,
      "step": 3415
    },
    {
      "epoch": 4.757660167130919,
      "grad_norm": 4.320363998413086,
      "learning_rate": 3.744741330301307e-05,
      "loss": 1.9642,
      "step": 3416
    },
    {
      "epoch": 4.759052924791086,
      "grad_norm": 3.189528226852417,
      "learning_rate": 3.743746446844798e-05,
      "loss": 1.6693,
      "step": 3417
    },
    {
      "epoch": 4.7604456824512535,
      "grad_norm": 3.9767041206359863,
      "learning_rate": 3.742751563388289e-05,
      "loss": 2.0967,
      "step": 3418
    },
    {
      "epoch": 4.76183844011142,
      "grad_norm": 3.6122829914093018,
      "learning_rate": 3.741756679931779e-05,
      "loss": 2.1152,
      "step": 3419
    },
    {
      "epoch": 4.763231197771588,
      "grad_norm": 3.6352224349975586,
      "learning_rate": 3.74076179647527e-05,
      "loss": 1.8194,
      "step": 3420
    },
    {
      "epoch": 4.764623955431755,
      "grad_norm": 3.21540904045105,
      "learning_rate": 3.73976691301876e-05,
      "loss": 1.6025,
      "step": 3421
    },
    {
      "epoch": 4.766016713091922,
      "grad_norm": 4.017698287963867,
      "learning_rate": 3.738772029562251e-05,
      "loss": 2.1644,
      "step": 3422
    },
    {
      "epoch": 4.767409470752089,
      "grad_norm": 3.6868584156036377,
      "learning_rate": 3.7377771461057415e-05,
      "loss": 2.0127,
      "step": 3423
    },
    {
      "epoch": 4.768802228412256,
      "grad_norm": 3.684056043624878,
      "learning_rate": 3.7367822626492326e-05,
      "loss": 1.9572,
      "step": 3424
    },
    {
      "epoch": 4.770194986072424,
      "grad_norm": 3.913855791091919,
      "learning_rate": 3.735787379192723e-05,
      "loss": 1.9114,
      "step": 3425
    },
    {
      "epoch": 4.77158774373259,
      "grad_norm": 3.8981807231903076,
      "learning_rate": 3.7347924957362134e-05,
      "loss": 2.2089,
      "step": 3426
    },
    {
      "epoch": 4.772980501392758,
      "grad_norm": 3.620997905731201,
      "learning_rate": 3.733797612279704e-05,
      "loss": 1.7761,
      "step": 3427
    },
    {
      "epoch": 4.774373259052925,
      "grad_norm": 3.9985365867614746,
      "learning_rate": 3.732802728823195e-05,
      "loss": 1.6269,
      "step": 3428
    },
    {
      "epoch": 4.775766016713092,
      "grad_norm": 3.124957323074341,
      "learning_rate": 3.731807845366685e-05,
      "loss": 1.6921,
      "step": 3429
    },
    {
      "epoch": 4.777158774373259,
      "grad_norm": 3.801938533782959,
      "learning_rate": 3.730812961910176e-05,
      "loss": 2.0213,
      "step": 3430
    },
    {
      "epoch": 4.778551532033426,
      "grad_norm": 3.622676134109497,
      "learning_rate": 3.729818078453666e-05,
      "loss": 1.7529,
      "step": 3431
    },
    {
      "epoch": 4.779944289693593,
      "grad_norm": 4.099602222442627,
      "learning_rate": 3.728823194997157e-05,
      "loss": 1.8388,
      "step": 3432
    },
    {
      "epoch": 4.781337047353761,
      "grad_norm": 3.208228826522827,
      "learning_rate": 3.7278283115406476e-05,
      "loss": 1.6376,
      "step": 3433
    },
    {
      "epoch": 4.782729805013927,
      "grad_norm": 4.3150105476379395,
      "learning_rate": 3.726833428084139e-05,
      "loss": 1.4145,
      "step": 3434
    },
    {
      "epoch": 4.784122562674095,
      "grad_norm": 4.187281608581543,
      "learning_rate": 3.7258385446276284e-05,
      "loss": 2.0179,
      "step": 3435
    },
    {
      "epoch": 4.7855153203342615,
      "grad_norm": 4.075669288635254,
      "learning_rate": 3.7248436611711195e-05,
      "loss": 1.7502,
      "step": 3436
    },
    {
      "epoch": 4.786908077994429,
      "grad_norm": 3.068662166595459,
      "learning_rate": 3.72384877771461e-05,
      "loss": 1.5979,
      "step": 3437
    },
    {
      "epoch": 4.788300835654596,
      "grad_norm": 3.2184665203094482,
      "learning_rate": 3.722853894258101e-05,
      "loss": 1.5834,
      "step": 3438
    },
    {
      "epoch": 4.789693593314763,
      "grad_norm": 3.7012887001037598,
      "learning_rate": 3.721859010801592e-05,
      "loss": 1.6719,
      "step": 3439
    },
    {
      "epoch": 4.791086350974931,
      "grad_norm": 3.8084328174591064,
      "learning_rate": 3.720864127345082e-05,
      "loss": 1.9171,
      "step": 3440
    },
    {
      "epoch": 4.7924791086350975,
      "grad_norm": 3.2045934200286865,
      "learning_rate": 3.719869243888573e-05,
      "loss": 1.5753,
      "step": 3441
    },
    {
      "epoch": 4.793871866295264,
      "grad_norm": 3.7816498279571533,
      "learning_rate": 3.718874360432063e-05,
      "loss": 2.1607,
      "step": 3442
    },
    {
      "epoch": 4.795264623955432,
      "grad_norm": 3.9390110969543457,
      "learning_rate": 3.7178794769755544e-05,
      "loss": 2.2207,
      "step": 3443
    },
    {
      "epoch": 4.796657381615599,
      "grad_norm": 3.879908561706543,
      "learning_rate": 3.716884593519045e-05,
      "loss": 1.7727,
      "step": 3444
    },
    {
      "epoch": 4.798050139275766,
      "grad_norm": 3.260246753692627,
      "learning_rate": 3.715889710062535e-05,
      "loss": 1.7271,
      "step": 3445
    },
    {
      "epoch": 4.7994428969359335,
      "grad_norm": 3.562711238861084,
      "learning_rate": 3.7148948266060256e-05,
      "loss": 1.5258,
      "step": 3446
    },
    {
      "epoch": 4.8008356545961,
      "grad_norm": 4.049978733062744,
      "learning_rate": 3.713899943149517e-05,
      "loss": 2.1744,
      "step": 3447
    },
    {
      "epoch": 4.802228412256268,
      "grad_norm": 3.903008460998535,
      "learning_rate": 3.712905059693007e-05,
      "loss": 1.7587,
      "step": 3448
    },
    {
      "epoch": 4.803621169916434,
      "grad_norm": 3.714756965637207,
      "learning_rate": 3.711910176236498e-05,
      "loss": 1.7595,
      "step": 3449
    },
    {
      "epoch": 4.805013927576602,
      "grad_norm": 3.900696277618408,
      "learning_rate": 3.710915292779988e-05,
      "loss": 1.9418,
      "step": 3450
    },
    {
      "epoch": 4.806406685236769,
      "grad_norm": 4.046336650848389,
      "learning_rate": 3.709920409323479e-05,
      "loss": 2.1615,
      "step": 3451
    },
    {
      "epoch": 4.807799442896936,
      "grad_norm": 2.856611728668213,
      "learning_rate": 3.7089255258669694e-05,
      "loss": 1.3214,
      "step": 3452
    },
    {
      "epoch": 4.809192200557103,
      "grad_norm": 3.4597105979919434,
      "learning_rate": 3.7079306424104605e-05,
      "loss": 1.4322,
      "step": 3453
    },
    {
      "epoch": 4.81058495821727,
      "grad_norm": 3.6220455169677734,
      "learning_rate": 3.706935758953951e-05,
      "loss": 1.9229,
      "step": 3454
    },
    {
      "epoch": 4.811977715877437,
      "grad_norm": 3.5215296745300293,
      "learning_rate": 3.705940875497441e-05,
      "loss": 1.8309,
      "step": 3455
    },
    {
      "epoch": 4.813370473537605,
      "grad_norm": 4.009708404541016,
      "learning_rate": 3.704945992040932e-05,
      "loss": 1.731,
      "step": 3456
    },
    {
      "epoch": 4.814763231197771,
      "grad_norm": 3.590940475463867,
      "learning_rate": 3.703951108584423e-05,
      "loss": 1.8334,
      "step": 3457
    },
    {
      "epoch": 4.816155988857939,
      "grad_norm": 3.6801607608795166,
      "learning_rate": 3.702956225127913e-05,
      "loss": 1.982,
      "step": 3458
    },
    {
      "epoch": 4.8175487465181055,
      "grad_norm": 3.8056976795196533,
      "learning_rate": 3.701961341671404e-05,
      "loss": 1.7785,
      "step": 3459
    },
    {
      "epoch": 4.818941504178273,
      "grad_norm": 3.730794906616211,
      "learning_rate": 3.700966458214894e-05,
      "loss": 1.7429,
      "step": 3460
    },
    {
      "epoch": 4.82033426183844,
      "grad_norm": 3.3455824851989746,
      "learning_rate": 3.699971574758385e-05,
      "loss": 1.6526,
      "step": 3461
    },
    {
      "epoch": 4.821727019498607,
      "grad_norm": 4.02980899810791,
      "learning_rate": 3.6989766913018755e-05,
      "loss": 1.7632,
      "step": 3462
    },
    {
      "epoch": 4.823119777158775,
      "grad_norm": 3.611778497695923,
      "learning_rate": 3.6979818078453666e-05,
      "loss": 1.524,
      "step": 3463
    },
    {
      "epoch": 4.8245125348189415,
      "grad_norm": 3.3080599308013916,
      "learning_rate": 3.696986924388858e-05,
      "loss": 1.7114,
      "step": 3464
    },
    {
      "epoch": 4.825905292479108,
      "grad_norm": 3.5850789546966553,
      "learning_rate": 3.6959920409323474e-05,
      "loss": 1.4328,
      "step": 3465
    },
    {
      "epoch": 4.827298050139276,
      "grad_norm": 3.918975353240967,
      "learning_rate": 3.694997157475838e-05,
      "loss": 1.9371,
      "step": 3466
    },
    {
      "epoch": 4.828690807799443,
      "grad_norm": 4.104087829589844,
      "learning_rate": 3.694002274019329e-05,
      "loss": 1.971,
      "step": 3467
    },
    {
      "epoch": 4.83008356545961,
      "grad_norm": 3.403681993484497,
      "learning_rate": 3.69300739056282e-05,
      "loss": 1.9956,
      "step": 3468
    },
    {
      "epoch": 4.8314763231197775,
      "grad_norm": 3.927757978439331,
      "learning_rate": 3.69201250710631e-05,
      "loss": 1.6046,
      "step": 3469
    },
    {
      "epoch": 4.832869080779944,
      "grad_norm": 4.31494140625,
      "learning_rate": 3.691017623649801e-05,
      "loss": 1.8073,
      "step": 3470
    },
    {
      "epoch": 4.834261838440112,
      "grad_norm": 3.8312408924102783,
      "learning_rate": 3.690022740193291e-05,
      "loss": 1.6918,
      "step": 3471
    },
    {
      "epoch": 4.835654596100278,
      "grad_norm": 3.8084018230438232,
      "learning_rate": 3.689027856736782e-05,
      "loss": 1.9344,
      "step": 3472
    },
    {
      "epoch": 4.837047353760446,
      "grad_norm": 3.9358959197998047,
      "learning_rate": 3.688032973280273e-05,
      "loss": 1.9348,
      "step": 3473
    },
    {
      "epoch": 4.838440111420613,
      "grad_norm": 4.218131065368652,
      "learning_rate": 3.687038089823763e-05,
      "loss": 1.7566,
      "step": 3474
    },
    {
      "epoch": 4.83983286908078,
      "grad_norm": 3.7826144695281982,
      "learning_rate": 3.6860432063672535e-05,
      "loss": 1.8968,
      "step": 3475
    },
    {
      "epoch": 4.841225626740947,
      "grad_norm": 4.4024200439453125,
      "learning_rate": 3.6850483229107446e-05,
      "loss": 1.9906,
      "step": 3476
    },
    {
      "epoch": 4.842618384401114,
      "grad_norm": 3.888853073120117,
      "learning_rate": 3.684053439454235e-05,
      "loss": 1.8098,
      "step": 3477
    },
    {
      "epoch": 4.844011142061281,
      "grad_norm": 3.795907974243164,
      "learning_rate": 3.683058555997726e-05,
      "loss": 2.0344,
      "step": 3478
    },
    {
      "epoch": 4.845403899721449,
      "grad_norm": 4.283677577972412,
      "learning_rate": 3.682063672541216e-05,
      "loss": 1.7764,
      "step": 3479
    },
    {
      "epoch": 4.846796657381615,
      "grad_norm": 3.312293767929077,
      "learning_rate": 3.681068789084707e-05,
      "loss": 1.6458,
      "step": 3480
    },
    {
      "epoch": 4.848189415041783,
      "grad_norm": 3.818984270095825,
      "learning_rate": 3.680073905628197e-05,
      "loss": 1.9209,
      "step": 3481
    },
    {
      "epoch": 4.84958217270195,
      "grad_norm": 3.8890469074249268,
      "learning_rate": 3.6790790221716884e-05,
      "loss": 2.1366,
      "step": 3482
    },
    {
      "epoch": 4.850974930362117,
      "grad_norm": 4.156975269317627,
      "learning_rate": 3.678084138715179e-05,
      "loss": 1.8456,
      "step": 3483
    },
    {
      "epoch": 4.852367688022284,
      "grad_norm": 3.5987191200256348,
      "learning_rate": 3.677089255258669e-05,
      "loss": 1.7846,
      "step": 3484
    },
    {
      "epoch": 4.853760445682451,
      "grad_norm": 4.193032741546631,
      "learning_rate": 3.6760943718021596e-05,
      "loss": 2.1748,
      "step": 3485
    },
    {
      "epoch": 4.855153203342619,
      "grad_norm": 3.4617221355438232,
      "learning_rate": 3.675099488345651e-05,
      "loss": 1.5625,
      "step": 3486
    },
    {
      "epoch": 4.8565459610027855,
      "grad_norm": 3.506861686706543,
      "learning_rate": 3.674104604889141e-05,
      "loss": 1.6646,
      "step": 3487
    },
    {
      "epoch": 4.857938718662953,
      "grad_norm": 3.827099323272705,
      "learning_rate": 3.673109721432632e-05,
      "loss": 2.0211,
      "step": 3488
    },
    {
      "epoch": 4.85933147632312,
      "grad_norm": 4.375415802001953,
      "learning_rate": 3.672114837976122e-05,
      "loss": 2.2855,
      "step": 3489
    },
    {
      "epoch": 4.860724233983287,
      "grad_norm": 3.5745644569396973,
      "learning_rate": 3.671119954519613e-05,
      "loss": 1.8843,
      "step": 3490
    },
    {
      "epoch": 4.862116991643454,
      "grad_norm": 4.019834518432617,
      "learning_rate": 3.6701250710631034e-05,
      "loss": 1.6538,
      "step": 3491
    },
    {
      "epoch": 4.8635097493036215,
      "grad_norm": 4.201918125152588,
      "learning_rate": 3.6691301876065945e-05,
      "loss": 1.6402,
      "step": 3492
    },
    {
      "epoch": 4.864902506963788,
      "grad_norm": 3.5356197357177734,
      "learning_rate": 3.6681353041500856e-05,
      "loss": 1.6881,
      "step": 3493
    },
    {
      "epoch": 4.866295264623956,
      "grad_norm": 3.792555332183838,
      "learning_rate": 3.667140420693575e-05,
      "loss": 2.1385,
      "step": 3494
    },
    {
      "epoch": 4.867688022284122,
      "grad_norm": 3.5409021377563477,
      "learning_rate": 3.666145537237066e-05,
      "loss": 1.6562,
      "step": 3495
    },
    {
      "epoch": 4.86908077994429,
      "grad_norm": 3.8827669620513916,
      "learning_rate": 3.665150653780557e-05,
      "loss": 1.8948,
      "step": 3496
    },
    {
      "epoch": 4.870473537604457,
      "grad_norm": 3.8804070949554443,
      "learning_rate": 3.664155770324048e-05,
      "loss": 1.6538,
      "step": 3497
    },
    {
      "epoch": 4.871866295264624,
      "grad_norm": 3.7477757930755615,
      "learning_rate": 3.663160886867538e-05,
      "loss": 1.733,
      "step": 3498
    },
    {
      "epoch": 4.873259052924791,
      "grad_norm": 3.519540548324585,
      "learning_rate": 3.662166003411029e-05,
      "loss": 2.0008,
      "step": 3499
    },
    {
      "epoch": 4.874651810584958,
      "grad_norm": 3.5823540687561035,
      "learning_rate": 3.661171119954519e-05,
      "loss": 1.645,
      "step": 3500
    },
    {
      "epoch": 4.876044568245125,
      "grad_norm": 4.832265377044678,
      "learning_rate": 3.66017623649801e-05,
      "loss": 1.6547,
      "step": 3501
    },
    {
      "epoch": 4.8774373259052926,
      "grad_norm": 3.9604856967926025,
      "learning_rate": 3.6591813530415006e-05,
      "loss": 2.0138,
      "step": 3502
    },
    {
      "epoch": 4.878830083565459,
      "grad_norm": 3.8182246685028076,
      "learning_rate": 3.658186469584992e-05,
      "loss": 1.6823,
      "step": 3503
    },
    {
      "epoch": 4.880222841225627,
      "grad_norm": 3.1158838272094727,
      "learning_rate": 3.6571915861284814e-05,
      "loss": 1.594,
      "step": 3504
    },
    {
      "epoch": 4.881615598885794,
      "grad_norm": 3.876467227935791,
      "learning_rate": 3.6561967026719725e-05,
      "loss": 1.7696,
      "step": 3505
    },
    {
      "epoch": 4.883008356545961,
      "grad_norm": 3.868966817855835,
      "learning_rate": 3.655201819215463e-05,
      "loss": 1.9037,
      "step": 3506
    },
    {
      "epoch": 4.884401114206128,
      "grad_norm": 3.630047082901001,
      "learning_rate": 3.654206935758954e-05,
      "loss": 1.504,
      "step": 3507
    },
    {
      "epoch": 4.885793871866295,
      "grad_norm": 3.239949941635132,
      "learning_rate": 3.6532120523024444e-05,
      "loss": 1.3695,
      "step": 3508
    },
    {
      "epoch": 4.887186629526463,
      "grad_norm": 4.650155544281006,
      "learning_rate": 3.652217168845935e-05,
      "loss": 2.0668,
      "step": 3509
    },
    {
      "epoch": 4.888579387186629,
      "grad_norm": 3.227562427520752,
      "learning_rate": 3.651222285389425e-05,
      "loss": 1.6167,
      "step": 3510
    },
    {
      "epoch": 4.889972144846797,
      "grad_norm": 3.7339327335357666,
      "learning_rate": 3.650227401932916e-05,
      "loss": 1.631,
      "step": 3511
    },
    {
      "epoch": 4.891364902506964,
      "grad_norm": 3.2770235538482666,
      "learning_rate": 3.649232518476407e-05,
      "loss": 1.6926,
      "step": 3512
    },
    {
      "epoch": 4.892757660167131,
      "grad_norm": 4.106587886810303,
      "learning_rate": 3.648237635019897e-05,
      "loss": 1.7791,
      "step": 3513
    },
    {
      "epoch": 4.894150417827298,
      "grad_norm": 4.150693416595459,
      "learning_rate": 3.6472427515633875e-05,
      "loss": 1.6498,
      "step": 3514
    },
    {
      "epoch": 4.895543175487465,
      "grad_norm": 3.1088902950286865,
      "learning_rate": 3.6462478681068786e-05,
      "loss": 1.641,
      "step": 3515
    },
    {
      "epoch": 4.896935933147632,
      "grad_norm": 3.614098310470581,
      "learning_rate": 3.645252984650369e-05,
      "loss": 1.518,
      "step": 3516
    },
    {
      "epoch": 4.8983286908078,
      "grad_norm": 3.6146419048309326,
      "learning_rate": 3.64425810119386e-05,
      "loss": 1.9188,
      "step": 3517
    },
    {
      "epoch": 4.899721448467966,
      "grad_norm": 3.583493947982788,
      "learning_rate": 3.64326321773735e-05,
      "loss": 1.8812,
      "step": 3518
    },
    {
      "epoch": 4.901114206128134,
      "grad_norm": 4.195201873779297,
      "learning_rate": 3.642268334280841e-05,
      "loss": 2.0226,
      "step": 3519
    },
    {
      "epoch": 4.9025069637883005,
      "grad_norm": 3.681187868118286,
      "learning_rate": 3.641273450824331e-05,
      "loss": 1.8083,
      "step": 3520
    },
    {
      "epoch": 4.903899721448468,
      "grad_norm": 3.536716938018799,
      "learning_rate": 3.6402785673678224e-05,
      "loss": 1.6227,
      "step": 3521
    },
    {
      "epoch": 4.905292479108635,
      "grad_norm": 3.566226005554199,
      "learning_rate": 3.6392836839113135e-05,
      "loss": 1.817,
      "step": 3522
    },
    {
      "epoch": 4.906685236768802,
      "grad_norm": 4.066131591796875,
      "learning_rate": 3.638288800454803e-05,
      "loss": 2.0704,
      "step": 3523
    },
    {
      "epoch": 4.908077994428969,
      "grad_norm": 3.2537622451782227,
      "learning_rate": 3.6372939169982936e-05,
      "loss": 1.7477,
      "step": 3524
    },
    {
      "epoch": 4.9094707520891365,
      "grad_norm": 4.299447059631348,
      "learning_rate": 3.636299033541785e-05,
      "loss": 1.8149,
      "step": 3525
    },
    {
      "epoch": 4.910863509749303,
      "grad_norm": 7.490352153778076,
      "learning_rate": 3.635304150085276e-05,
      "loss": 1.3041,
      "step": 3526
    },
    {
      "epoch": 4.912256267409471,
      "grad_norm": 4.535417079925537,
      "learning_rate": 3.634309266628766e-05,
      "loss": 1.9322,
      "step": 3527
    },
    {
      "epoch": 4.913649025069638,
      "grad_norm": 3.613455295562744,
      "learning_rate": 3.6333143831722566e-05,
      "loss": 1.7951,
      "step": 3528
    },
    {
      "epoch": 4.915041782729805,
      "grad_norm": 3.567254066467285,
      "learning_rate": 3.632319499715747e-05,
      "loss": 1.9571,
      "step": 3529
    },
    {
      "epoch": 4.9164345403899725,
      "grad_norm": 3.3372793197631836,
      "learning_rate": 3.631324616259238e-05,
      "loss": 1.5199,
      "step": 3530
    },
    {
      "epoch": 4.917827298050139,
      "grad_norm": 3.837397336959839,
      "learning_rate": 3.6303297328027285e-05,
      "loss": 1.5254,
      "step": 3531
    },
    {
      "epoch": 4.919220055710307,
      "grad_norm": 5.37261962890625,
      "learning_rate": 3.6293348493462196e-05,
      "loss": 1.8113,
      "step": 3532
    },
    {
      "epoch": 4.920612813370473,
      "grad_norm": 3.5950417518615723,
      "learning_rate": 3.6283399658897093e-05,
      "loss": 1.7721,
      "step": 3533
    },
    {
      "epoch": 4.922005571030641,
      "grad_norm": 3.993096113204956,
      "learning_rate": 3.6273450824332004e-05,
      "loss": 1.9407,
      "step": 3534
    },
    {
      "epoch": 4.923398328690808,
      "grad_norm": 4.111208915710449,
      "learning_rate": 3.626350198976691e-05,
      "loss": 1.4839,
      "step": 3535
    },
    {
      "epoch": 4.924791086350975,
      "grad_norm": 4.141539573669434,
      "learning_rate": 3.625355315520182e-05,
      "loss": 1.8831,
      "step": 3536
    },
    {
      "epoch": 4.926183844011142,
      "grad_norm": 3.235837697982788,
      "learning_rate": 3.624360432063672e-05,
      "loss": 1.625,
      "step": 3537
    },
    {
      "epoch": 4.927576601671309,
      "grad_norm": 3.5801455974578857,
      "learning_rate": 3.623365548607163e-05,
      "loss": 1.6913,
      "step": 3538
    },
    {
      "epoch": 4.928969359331476,
      "grad_norm": 4.944547176361084,
      "learning_rate": 3.622370665150653e-05,
      "loss": 1.6985,
      "step": 3539
    },
    {
      "epoch": 4.930362116991644,
      "grad_norm": 3.601283550262451,
      "learning_rate": 3.621375781694144e-05,
      "loss": 1.6085,
      "step": 3540
    },
    {
      "epoch": 4.93175487465181,
      "grad_norm": 5.916871070861816,
      "learning_rate": 3.6203808982376346e-05,
      "loss": 1.947,
      "step": 3541
    },
    {
      "epoch": 4.933147632311978,
      "grad_norm": 5.327152729034424,
      "learning_rate": 3.619386014781126e-05,
      "loss": 1.8474,
      "step": 3542
    },
    {
      "epoch": 4.9345403899721445,
      "grad_norm": 3.569394588470459,
      "learning_rate": 3.6183911313246155e-05,
      "loss": 1.8715,
      "step": 3543
    },
    {
      "epoch": 4.935933147632312,
      "grad_norm": 3.9451210498809814,
      "learning_rate": 3.6173962478681065e-05,
      "loss": 1.9203,
      "step": 3544
    },
    {
      "epoch": 4.937325905292479,
      "grad_norm": 3.293942928314209,
      "learning_rate": 3.616401364411597e-05,
      "loss": 1.4795,
      "step": 3545
    },
    {
      "epoch": 4.938718662952646,
      "grad_norm": 3.8608896732330322,
      "learning_rate": 3.615406480955088e-05,
      "loss": 1.7539,
      "step": 3546
    },
    {
      "epoch": 4.940111420612814,
      "grad_norm": 3.3696210384368896,
      "learning_rate": 3.614411597498579e-05,
      "loss": 1.5044,
      "step": 3547
    },
    {
      "epoch": 4.9415041782729805,
      "grad_norm": 3.858611583709717,
      "learning_rate": 3.613416714042069e-05,
      "loss": 1.5807,
      "step": 3548
    },
    {
      "epoch": 4.942896935933147,
      "grad_norm": 3.8843953609466553,
      "learning_rate": 3.612421830585559e-05,
      "loss": 2.3736,
      "step": 3549
    },
    {
      "epoch": 4.944289693593315,
      "grad_norm": 3.9137628078460693,
      "learning_rate": 3.61142694712905e-05,
      "loss": 1.7743,
      "step": 3550
    },
    {
      "epoch": 4.945682451253482,
      "grad_norm": 3.500291347503662,
      "learning_rate": 3.6104320636725414e-05,
      "loss": 1.9175,
      "step": 3551
    },
    {
      "epoch": 4.947075208913649,
      "grad_norm": 3.385045289993286,
      "learning_rate": 3.609437180216032e-05,
      "loss": 1.6842,
      "step": 3552
    },
    {
      "epoch": 4.9484679665738165,
      "grad_norm": 6.60980224609375,
      "learning_rate": 3.608442296759522e-05,
      "loss": 1.7951,
      "step": 3553
    },
    {
      "epoch": 4.949860724233983,
      "grad_norm": 3.7822020053863525,
      "learning_rate": 3.6074474133030126e-05,
      "loss": 1.557,
      "step": 3554
    },
    {
      "epoch": 4.951253481894151,
      "grad_norm": 3.5959599018096924,
      "learning_rate": 3.606452529846504e-05,
      "loss": 1.7329,
      "step": 3555
    },
    {
      "epoch": 4.952646239554317,
      "grad_norm": 3.565088987350464,
      "learning_rate": 3.605457646389994e-05,
      "loss": 1.6584,
      "step": 3556
    },
    {
      "epoch": 4.954038997214485,
      "grad_norm": 3.444204807281494,
      "learning_rate": 3.6044627629334845e-05,
      "loss": 1.9458,
      "step": 3557
    },
    {
      "epoch": 4.955431754874652,
      "grad_norm": 3.907775402069092,
      "learning_rate": 3.603467879476975e-05,
      "loss": 1.7874,
      "step": 3558
    },
    {
      "epoch": 4.956824512534819,
      "grad_norm": 3.466712474822998,
      "learning_rate": 3.602472996020466e-05,
      "loss": 1.5166,
      "step": 3559
    },
    {
      "epoch": 4.958217270194986,
      "grad_norm": 3.595690965652466,
      "learning_rate": 3.6014781125639564e-05,
      "loss": 1.9107,
      "step": 3560
    },
    {
      "epoch": 4.959610027855153,
      "grad_norm": 4.689317226409912,
      "learning_rate": 3.6004832291074475e-05,
      "loss": 2.0874,
      "step": 3561
    },
    {
      "epoch": 4.96100278551532,
      "grad_norm": 3.4974448680877686,
      "learning_rate": 3.599488345650937e-05,
      "loss": 1.9071,
      "step": 3562
    },
    {
      "epoch": 4.962395543175488,
      "grad_norm": 4.746645450592041,
      "learning_rate": 3.5984934621944283e-05,
      "loss": 1.6609,
      "step": 3563
    },
    {
      "epoch": 4.963788300835654,
      "grad_norm": 3.7093594074249268,
      "learning_rate": 3.597498578737919e-05,
      "loss": 1.7767,
      "step": 3564
    },
    {
      "epoch": 4.965181058495822,
      "grad_norm": 3.8114004135131836,
      "learning_rate": 3.59650369528141e-05,
      "loss": 2.0213,
      "step": 3565
    },
    {
      "epoch": 4.9665738161559885,
      "grad_norm": 3.6004419326782227,
      "learning_rate": 3.5955088118249e-05,
      "loss": 1.8763,
      "step": 3566
    },
    {
      "epoch": 4.967966573816156,
      "grad_norm": 3.915437698364258,
      "learning_rate": 3.5945139283683907e-05,
      "loss": 1.7432,
      "step": 3567
    },
    {
      "epoch": 4.969359331476323,
      "grad_norm": 3.8346073627471924,
      "learning_rate": 3.593519044911881e-05,
      "loss": 1.9158,
      "step": 3568
    },
    {
      "epoch": 4.97075208913649,
      "grad_norm": 3.7203516960144043,
      "learning_rate": 3.592524161455372e-05,
      "loss": 1.5696,
      "step": 3569
    },
    {
      "epoch": 4.972144846796658,
      "grad_norm": 3.4037482738494873,
      "learning_rate": 3.5915292779988626e-05,
      "loss": 1.7977,
      "step": 3570
    },
    {
      "epoch": 4.9735376044568245,
      "grad_norm": 2.969834566116333,
      "learning_rate": 3.5905343945423536e-05,
      "loss": 1.561,
      "step": 3571
    },
    {
      "epoch": 4.974930362116992,
      "grad_norm": 3.761119842529297,
      "learning_rate": 3.5895395110858434e-05,
      "loss": 1.5814,
      "step": 3572
    },
    {
      "epoch": 4.976323119777159,
      "grad_norm": 4.0623979568481445,
      "learning_rate": 3.5885446276293345e-05,
      "loss": 1.9379,
      "step": 3573
    },
    {
      "epoch": 4.977715877437326,
      "grad_norm": 4.7986626625061035,
      "learning_rate": 3.587549744172825e-05,
      "loss": 1.8651,
      "step": 3574
    },
    {
      "epoch": 4.979108635097493,
      "grad_norm": 3.2597880363464355,
      "learning_rate": 3.586554860716316e-05,
      "loss": 1.6728,
      "step": 3575
    },
    {
      "epoch": 4.9805013927576605,
      "grad_norm": 4.0020527839660645,
      "learning_rate": 3.585559977259807e-05,
      "loss": 1.4982,
      "step": 3576
    },
    {
      "epoch": 4.981894150417827,
      "grad_norm": 3.393110513687134,
      "learning_rate": 3.584565093803297e-05,
      "loss": 1.6846,
      "step": 3577
    },
    {
      "epoch": 4.983286908077995,
      "grad_norm": 3.4916343688964844,
      "learning_rate": 3.583570210346787e-05,
      "loss": 1.6884,
      "step": 3578
    },
    {
      "epoch": 4.984679665738161,
      "grad_norm": 4.145401954650879,
      "learning_rate": 3.582575326890278e-05,
      "loss": 1.6155,
      "step": 3579
    },
    {
      "epoch": 4.986072423398329,
      "grad_norm": 3.450040578842163,
      "learning_rate": 3.581580443433769e-05,
      "loss": 1.5963,
      "step": 3580
    },
    {
      "epoch": 4.987465181058496,
      "grad_norm": 4.401824474334717,
      "learning_rate": 3.58058555997726e-05,
      "loss": 2.0137,
      "step": 3581
    },
    {
      "epoch": 4.988857938718663,
      "grad_norm": 3.62214994430542,
      "learning_rate": 3.57959067652075e-05,
      "loss": 1.8588,
      "step": 3582
    },
    {
      "epoch": 4.99025069637883,
      "grad_norm": 4.877983570098877,
      "learning_rate": 3.5785957930642406e-05,
      "loss": 2.2727,
      "step": 3583
    },
    {
      "epoch": 4.991643454038997,
      "grad_norm": 3.382337808609009,
      "learning_rate": 3.5776009096077316e-05,
      "loss": 1.514,
      "step": 3584
    },
    {
      "epoch": 4.993036211699164,
      "grad_norm": 3.798260450363159,
      "learning_rate": 3.576606026151222e-05,
      "loss": 1.5972,
      "step": 3585
    },
    {
      "epoch": 4.994428969359332,
      "grad_norm": 4.307199478149414,
      "learning_rate": 3.575611142694713e-05,
      "loss": 1.6715,
      "step": 3586
    },
    {
      "epoch": 4.995821727019498,
      "grad_norm": 3.9402637481689453,
      "learning_rate": 3.574616259238203e-05,
      "loss": 1.7135,
      "step": 3587
    },
    {
      "epoch": 4.997214484679666,
      "grad_norm": 3.9421639442443848,
      "learning_rate": 3.573621375781694e-05,
      "loss": 2.3556,
      "step": 3588
    },
    {
      "epoch": 4.998607242339833,
      "grad_norm": 3.5371787548065186,
      "learning_rate": 3.5726264923251844e-05,
      "loss": 1.6727,
      "step": 3589
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.5636303424835205,
      "learning_rate": 3.5716316088686754e-05,
      "loss": 1.4795,
      "step": 3590
    },
    {
      "epoch": 5.0013927576601676,
      "grad_norm": 6.826588153839111,
      "learning_rate": 3.570636725412166e-05,
      "loss": 1.4351,
      "step": 3591
    },
    {
      "epoch": 5.002785515320334,
      "grad_norm": 3.244941234588623,
      "learning_rate": 3.569641841955656e-05,
      "loss": 1.4898,
      "step": 3592
    },
    {
      "epoch": 5.004178272980502,
      "grad_norm": 2.8874032497406006,
      "learning_rate": 3.568646958499147e-05,
      "loss": 1.38,
      "step": 3593
    },
    {
      "epoch": 5.005571030640668,
      "grad_norm": 3.873886823654175,
      "learning_rate": 3.567652075042638e-05,
      "loss": 1.4358,
      "step": 3594
    },
    {
      "epoch": 5.006963788300836,
      "grad_norm": 3.3421318531036377,
      "learning_rate": 3.566657191586128e-05,
      "loss": 1.3531,
      "step": 3595
    },
    {
      "epoch": 5.008356545961003,
      "grad_norm": 3.3321070671081543,
      "learning_rate": 3.565662308129619e-05,
      "loss": 1.4568,
      "step": 3596
    },
    {
      "epoch": 5.00974930362117,
      "grad_norm": 3.352158546447754,
      "learning_rate": 3.564667424673109e-05,
      "loss": 1.5221,
      "step": 3597
    },
    {
      "epoch": 5.011142061281337,
      "grad_norm": 3.3171162605285645,
      "learning_rate": 3.5636725412166e-05,
      "loss": 1.4319,
      "step": 3598
    },
    {
      "epoch": 5.012534818941504,
      "grad_norm": 3.1241772174835205,
      "learning_rate": 3.5626776577600905e-05,
      "loss": 1.6024,
      "step": 3599
    },
    {
      "epoch": 5.013927576601671,
      "grad_norm": 2.815913677215576,
      "learning_rate": 3.5616827743035816e-05,
      "loss": 1.1249,
      "step": 3600
    },
    {
      "epoch": 5.015320334261839,
      "grad_norm": 3.150897741317749,
      "learning_rate": 3.560687890847071e-05,
      "loss": 1.3773,
      "step": 3601
    },
    {
      "epoch": 5.016713091922005,
      "grad_norm": 3.2558064460754395,
      "learning_rate": 3.5596930073905624e-05,
      "loss": 1.5564,
      "step": 3602
    },
    {
      "epoch": 5.018105849582173,
      "grad_norm": 2.789930582046509,
      "learning_rate": 3.558698123934053e-05,
      "loss": 0.979,
      "step": 3603
    },
    {
      "epoch": 5.0194986072423395,
      "grad_norm": 3.3550565242767334,
      "learning_rate": 3.557703240477544e-05,
      "loss": 1.8087,
      "step": 3604
    },
    {
      "epoch": 5.020891364902507,
      "grad_norm": 3.4039390087127686,
      "learning_rate": 3.556708357021035e-05,
      "loss": 1.4341,
      "step": 3605
    },
    {
      "epoch": 5.022284122562674,
      "grad_norm": 4.22948694229126,
      "learning_rate": 3.555713473564525e-05,
      "loss": 1.3164,
      "step": 3606
    },
    {
      "epoch": 5.023676880222841,
      "grad_norm": 3.8464784622192383,
      "learning_rate": 3.554718590108015e-05,
      "loss": 1.23,
      "step": 3607
    },
    {
      "epoch": 5.025069637883008,
      "grad_norm": 3.5483877658843994,
      "learning_rate": 3.553723706651506e-05,
      "loss": 1.2695,
      "step": 3608
    },
    {
      "epoch": 5.0264623955431755,
      "grad_norm": 3.2288761138916016,
      "learning_rate": 3.552728823194997e-05,
      "loss": 1.2609,
      "step": 3609
    },
    {
      "epoch": 5.027855153203342,
      "grad_norm": 3.2547757625579834,
      "learning_rate": 3.551733939738488e-05,
      "loss": 1.5669,
      "step": 3610
    },
    {
      "epoch": 5.02924791086351,
      "grad_norm": 2.923926591873169,
      "learning_rate": 3.550739056281978e-05,
      "loss": 1.4724,
      "step": 3611
    },
    {
      "epoch": 5.030640668523677,
      "grad_norm": 3.3608334064483643,
      "learning_rate": 3.5497441728254685e-05,
      "loss": 1.5402,
      "step": 3612
    },
    {
      "epoch": 5.032033426183844,
      "grad_norm": 3.311286449432373,
      "learning_rate": 3.5487492893689596e-05,
      "loss": 1.3233,
      "step": 3613
    },
    {
      "epoch": 5.0334261838440115,
      "grad_norm": 3.551703691482544,
      "learning_rate": 3.54775440591245e-05,
      "loss": 1.7847,
      "step": 3614
    },
    {
      "epoch": 5.034818941504178,
      "grad_norm": 3.485243797302246,
      "learning_rate": 3.546759522455941e-05,
      "loss": 1.5752,
      "step": 3615
    },
    {
      "epoch": 5.036211699164346,
      "grad_norm": 3.2045066356658936,
      "learning_rate": 3.545764638999431e-05,
      "loss": 1.2544,
      "step": 3616
    },
    {
      "epoch": 5.037604456824512,
      "grad_norm": 3.0100302696228027,
      "learning_rate": 3.544769755542922e-05,
      "loss": 1.2859,
      "step": 3617
    },
    {
      "epoch": 5.03899721448468,
      "grad_norm": 3.7358620166778564,
      "learning_rate": 3.543774872086412e-05,
      "loss": 1.171,
      "step": 3618
    },
    {
      "epoch": 5.040389972144847,
      "grad_norm": 3.3091506958007812,
      "learning_rate": 3.5427799886299034e-05,
      "loss": 1.4943,
      "step": 3619
    },
    {
      "epoch": 5.041782729805014,
      "grad_norm": 3.4011595249176025,
      "learning_rate": 3.541785105173394e-05,
      "loss": 1.6702,
      "step": 3620
    },
    {
      "epoch": 5.043175487465181,
      "grad_norm": 3.6097514629364014,
      "learning_rate": 3.540790221716884e-05,
      "loss": 1.8783,
      "step": 3621
    },
    {
      "epoch": 5.044568245125348,
      "grad_norm": 3.3654654026031494,
      "learning_rate": 3.5397953382603746e-05,
      "loss": 1.6149,
      "step": 3622
    },
    {
      "epoch": 5.045961002785515,
      "grad_norm": 3.2377398014068604,
      "learning_rate": 3.538800454803866e-05,
      "loss": 1.5624,
      "step": 3623
    },
    {
      "epoch": 5.047353760445683,
      "grad_norm": 3.1147098541259766,
      "learning_rate": 3.537805571347356e-05,
      "loss": 1.4554,
      "step": 3624
    },
    {
      "epoch": 5.048746518105849,
      "grad_norm": 3.2079262733459473,
      "learning_rate": 3.536810687890847e-05,
      "loss": 1.7468,
      "step": 3625
    },
    {
      "epoch": 5.050139275766017,
      "grad_norm": 2.7711851596832275,
      "learning_rate": 3.535815804434337e-05,
      "loss": 1.2825,
      "step": 3626
    },
    {
      "epoch": 5.0515320334261835,
      "grad_norm": 4.073577880859375,
      "learning_rate": 3.534820920977828e-05,
      "loss": 1.2833,
      "step": 3627
    },
    {
      "epoch": 5.052924791086351,
      "grad_norm": 3.3940837383270264,
      "learning_rate": 3.5338260375213184e-05,
      "loss": 1.4993,
      "step": 3628
    },
    {
      "epoch": 5.054317548746518,
      "grad_norm": 3.7020013332366943,
      "learning_rate": 3.5328311540648095e-05,
      "loss": 1.8527,
      "step": 3629
    },
    {
      "epoch": 5.055710306406685,
      "grad_norm": 3.2459218502044678,
      "learning_rate": 3.5318362706083e-05,
      "loss": 1.5839,
      "step": 3630
    },
    {
      "epoch": 5.057103064066852,
      "grad_norm": 2.827864646911621,
      "learning_rate": 3.53084138715179e-05,
      "loss": 1.3861,
      "step": 3631
    },
    {
      "epoch": 5.0584958217270195,
      "grad_norm": 3.8657968044281006,
      "learning_rate": 3.529846503695281e-05,
      "loss": 1.8368,
      "step": 3632
    },
    {
      "epoch": 5.059888579387187,
      "grad_norm": 3.375610113143921,
      "learning_rate": 3.528851620238772e-05,
      "loss": 1.713,
      "step": 3633
    },
    {
      "epoch": 5.061281337047354,
      "grad_norm": 3.10119366645813,
      "learning_rate": 3.527856736782263e-05,
      "loss": 1.48,
      "step": 3634
    },
    {
      "epoch": 5.062674094707521,
      "grad_norm": 3.3581926822662354,
      "learning_rate": 3.526861853325753e-05,
      "loss": 1.4784,
      "step": 3635
    },
    {
      "epoch": 5.064066852367688,
      "grad_norm": 3.330660581588745,
      "learning_rate": 3.525866969869243e-05,
      "loss": 1.5715,
      "step": 3636
    },
    {
      "epoch": 5.0654596100278555,
      "grad_norm": 3.0002896785736084,
      "learning_rate": 3.524872086412734e-05,
      "loss": 1.3374,
      "step": 3637
    },
    {
      "epoch": 5.066852367688022,
      "grad_norm": 3.3638551235198975,
      "learning_rate": 3.523877202956225e-05,
      "loss": 1.4083,
      "step": 3638
    },
    {
      "epoch": 5.06824512534819,
      "grad_norm": 3.5368261337280273,
      "learning_rate": 3.5228823194997156e-05,
      "loss": 1.3304,
      "step": 3639
    },
    {
      "epoch": 5.069637883008356,
      "grad_norm": 3.7917847633361816,
      "learning_rate": 3.521887436043207e-05,
      "loss": 1.6021,
      "step": 3640
    },
    {
      "epoch": 5.071030640668524,
      "grad_norm": 3.3980519771575928,
      "learning_rate": 3.5208925525866964e-05,
      "loss": 1.772,
      "step": 3641
    },
    {
      "epoch": 5.072423398328691,
      "grad_norm": 3.2356252670288086,
      "learning_rate": 3.5198976691301875e-05,
      "loss": 1.5132,
      "step": 3642
    },
    {
      "epoch": 5.073816155988858,
      "grad_norm": 3.7738420963287354,
      "learning_rate": 3.518902785673678e-05,
      "loss": 1.4201,
      "step": 3643
    },
    {
      "epoch": 5.075208913649025,
      "grad_norm": 2.7929110527038574,
      "learning_rate": 3.517907902217169e-05,
      "loss": 1.1405,
      "step": 3644
    },
    {
      "epoch": 5.076601671309192,
      "grad_norm": 3.162795066833496,
      "learning_rate": 3.516913018760659e-05,
      "loss": 1.7356,
      "step": 3645
    },
    {
      "epoch": 5.077994428969359,
      "grad_norm": 3.6293771266937256,
      "learning_rate": 3.51591813530415e-05,
      "loss": 1.6064,
      "step": 3646
    },
    {
      "epoch": 5.079387186629527,
      "grad_norm": 3.622387647628784,
      "learning_rate": 3.51492325184764e-05,
      "loss": 1.6365,
      "step": 3647
    },
    {
      "epoch": 5.080779944289693,
      "grad_norm": 3.730894088745117,
      "learning_rate": 3.513928368391131e-05,
      "loss": 1.6652,
      "step": 3648
    },
    {
      "epoch": 5.082172701949861,
      "grad_norm": 2.8982667922973633,
      "learning_rate": 3.512933484934622e-05,
      "loss": 1.1618,
      "step": 3649
    },
    {
      "epoch": 5.0835654596100275,
      "grad_norm": 3.325928211212158,
      "learning_rate": 3.511938601478112e-05,
      "loss": 1.5396,
      "step": 3650
    },
    {
      "epoch": 5.084958217270195,
      "grad_norm": 2.7456600666046143,
      "learning_rate": 3.5109437180216025e-05,
      "loss": 1.1465,
      "step": 3651
    },
    {
      "epoch": 5.086350974930362,
      "grad_norm": 3.5660476684570312,
      "learning_rate": 3.5099488345650936e-05,
      "loss": 1.5811,
      "step": 3652
    },
    {
      "epoch": 5.087743732590529,
      "grad_norm": 3.5288870334625244,
      "learning_rate": 3.508953951108584e-05,
      "loss": 1.787,
      "step": 3653
    },
    {
      "epoch": 5.089136490250697,
      "grad_norm": 3.0032618045806885,
      "learning_rate": 3.507959067652075e-05,
      "loss": 1.3188,
      "step": 3654
    },
    {
      "epoch": 5.0905292479108635,
      "grad_norm": 3.1997663974761963,
      "learning_rate": 3.506964184195565e-05,
      "loss": 1.4057,
      "step": 3655
    },
    {
      "epoch": 5.091922005571031,
      "grad_norm": 3.2036070823669434,
      "learning_rate": 3.505969300739056e-05,
      "loss": 1.425,
      "step": 3656
    },
    {
      "epoch": 5.093314763231198,
      "grad_norm": 3.3824801445007324,
      "learning_rate": 3.504974417282546e-05,
      "loss": 1.6274,
      "step": 3657
    },
    {
      "epoch": 5.094707520891365,
      "grad_norm": 3.0331170558929443,
      "learning_rate": 3.5039795338260374e-05,
      "loss": 1.2897,
      "step": 3658
    },
    {
      "epoch": 5.096100278551532,
      "grad_norm": 3.129055976867676,
      "learning_rate": 3.5029846503695285e-05,
      "loss": 1.5442,
      "step": 3659
    },
    {
      "epoch": 5.0974930362116995,
      "grad_norm": 3.255399227142334,
      "learning_rate": 3.501989766913018e-05,
      "loss": 1.3974,
      "step": 3660
    },
    {
      "epoch": 5.098885793871866,
      "grad_norm": 3.486367702484131,
      "learning_rate": 3.5009948834565086e-05,
      "loss": 1.4974,
      "step": 3661
    },
    {
      "epoch": 5.100278551532034,
      "grad_norm": 3.3133184909820557,
      "learning_rate": 3.5e-05,
      "loss": 1.5463,
      "step": 3662
    },
    {
      "epoch": 5.1016713091922,
      "grad_norm": 3.101465940475464,
      "learning_rate": 3.499005116543491e-05,
      "loss": 1.2295,
      "step": 3663
    },
    {
      "epoch": 5.103064066852368,
      "grad_norm": 2.898435115814209,
      "learning_rate": 3.498010233086981e-05,
      "loss": 1.3083,
      "step": 3664
    },
    {
      "epoch": 5.104456824512535,
      "grad_norm": 7.913095474243164,
      "learning_rate": 3.4970153496304716e-05,
      "loss": 1.5649,
      "step": 3665
    },
    {
      "epoch": 5.105849582172702,
      "grad_norm": 2.8633811473846436,
      "learning_rate": 3.496020466173962e-05,
      "loss": 1.3258,
      "step": 3666
    },
    {
      "epoch": 5.107242339832869,
      "grad_norm": 3.5359859466552734,
      "learning_rate": 3.495025582717453e-05,
      "loss": 1.7369,
      "step": 3667
    },
    {
      "epoch": 5.108635097493036,
      "grad_norm": 3.1591122150421143,
      "learning_rate": 3.4940306992609435e-05,
      "loss": 1.3495,
      "step": 3668
    },
    {
      "epoch": 5.110027855153203,
      "grad_norm": 3.358830690383911,
      "learning_rate": 3.493035815804434e-05,
      "loss": 1.599,
      "step": 3669
    },
    {
      "epoch": 5.111420612813371,
      "grad_norm": 2.988240957260132,
      "learning_rate": 3.492040932347925e-05,
      "loss": 1.4411,
      "step": 3670
    },
    {
      "epoch": 5.112813370473537,
      "grad_norm": 3.552689790725708,
      "learning_rate": 3.4910460488914154e-05,
      "loss": 1.5288,
      "step": 3671
    },
    {
      "epoch": 5.114206128133705,
      "grad_norm": 3.24813175201416,
      "learning_rate": 3.490051165434906e-05,
      "loss": 1.6736,
      "step": 3672
    },
    {
      "epoch": 5.1155988857938715,
      "grad_norm": 3.0939056873321533,
      "learning_rate": 3.489056281978396e-05,
      "loss": 1.2816,
      "step": 3673
    },
    {
      "epoch": 5.116991643454039,
      "grad_norm": 3.16573166847229,
      "learning_rate": 3.488061398521887e-05,
      "loss": 1.435,
      "step": 3674
    },
    {
      "epoch": 5.118384401114207,
      "grad_norm": 3.6670165061950684,
      "learning_rate": 3.487066515065378e-05,
      "loss": 1.4094,
      "step": 3675
    },
    {
      "epoch": 5.119777158774373,
      "grad_norm": 3.0940935611724854,
      "learning_rate": 3.486071631608868e-05,
      "loss": 1.5027,
      "step": 3676
    },
    {
      "epoch": 5.121169916434541,
      "grad_norm": 3.1497867107391357,
      "learning_rate": 3.485076748152359e-05,
      "loss": 1.2258,
      "step": 3677
    },
    {
      "epoch": 5.1225626740947074,
      "grad_norm": 2.927849054336548,
      "learning_rate": 3.4840818646958496e-05,
      "loss": 1.3154,
      "step": 3678
    },
    {
      "epoch": 5.123955431754875,
      "grad_norm": 3.4527838230133057,
      "learning_rate": 3.48308698123934e-05,
      "loss": 1.4148,
      "step": 3679
    },
    {
      "epoch": 5.125348189415042,
      "grad_norm": 6.127671718597412,
      "learning_rate": 3.482092097782831e-05,
      "loss": 1.4014,
      "step": 3680
    },
    {
      "epoch": 5.126740947075209,
      "grad_norm": 3.6558492183685303,
      "learning_rate": 3.4810972143263215e-05,
      "loss": 1.4898,
      "step": 3681
    },
    {
      "epoch": 5.128133704735376,
      "grad_norm": 4.456975936889648,
      "learning_rate": 3.480102330869812e-05,
      "loss": 1.6801,
      "step": 3682
    },
    {
      "epoch": 5.129526462395543,
      "grad_norm": 3.446354866027832,
      "learning_rate": 3.479107447413302e-05,
      "loss": 1.6189,
      "step": 3683
    },
    {
      "epoch": 5.13091922005571,
      "grad_norm": 3.003618001937866,
      "learning_rate": 3.4781125639567934e-05,
      "loss": 1.5376,
      "step": 3684
    },
    {
      "epoch": 5.132311977715878,
      "grad_norm": 3.30557918548584,
      "learning_rate": 3.477117680500284e-05,
      "loss": 1.4387,
      "step": 3685
    },
    {
      "epoch": 5.133704735376044,
      "grad_norm": 3.599032163619995,
      "learning_rate": 3.476122797043774e-05,
      "loss": 1.7406,
      "step": 3686
    },
    {
      "epoch": 5.135097493036212,
      "grad_norm": 3.6804332733154297,
      "learning_rate": 3.475127913587265e-05,
      "loss": 1.8243,
      "step": 3687
    },
    {
      "epoch": 5.1364902506963785,
      "grad_norm": 2.995417356491089,
      "learning_rate": 3.474133030130756e-05,
      "loss": 1.4121,
      "step": 3688
    },
    {
      "epoch": 5.137883008356546,
      "grad_norm": 2.8882551193237305,
      "learning_rate": 3.473138146674246e-05,
      "loss": 1.2631,
      "step": 3689
    },
    {
      "epoch": 5.139275766016713,
      "grad_norm": 3.0136148929595947,
      "learning_rate": 3.472143263217737e-05,
      "loss": 1.5149,
      "step": 3690
    },
    {
      "epoch": 5.14066852367688,
      "grad_norm": 3.514186382293701,
      "learning_rate": 3.4711483797612276e-05,
      "loss": 1.6868,
      "step": 3691
    },
    {
      "epoch": 5.142061281337047,
      "grad_norm": 3.5419411659240723,
      "learning_rate": 3.470153496304719e-05,
      "loss": 1.7281,
      "step": 3692
    },
    {
      "epoch": 5.1434540389972145,
      "grad_norm": 3.458965301513672,
      "learning_rate": 3.469158612848209e-05,
      "loss": 1.5011,
      "step": 3693
    },
    {
      "epoch": 5.144846796657381,
      "grad_norm": 3.9583911895751953,
      "learning_rate": 3.4681637293916995e-05,
      "loss": 1.5089,
      "step": 3694
    },
    {
      "epoch": 5.146239554317549,
      "grad_norm": 3.239198684692383,
      "learning_rate": 3.46716884593519e-05,
      "loss": 1.5014,
      "step": 3695
    },
    {
      "epoch": 5.147632311977716,
      "grad_norm": 3.087223768234253,
      "learning_rate": 3.466173962478681e-05,
      "loss": 1.5708,
      "step": 3696
    },
    {
      "epoch": 5.149025069637883,
      "grad_norm": 3.3014509677886963,
      "learning_rate": 3.4651790790221714e-05,
      "loss": 1.5735,
      "step": 3697
    },
    {
      "epoch": 5.1504178272980505,
      "grad_norm": 3.4644100666046143,
      "learning_rate": 3.464184195565662e-05,
      "loss": 1.5335,
      "step": 3698
    },
    {
      "epoch": 5.151810584958217,
      "grad_norm": 3.331061601638794,
      "learning_rate": 3.463189312109153e-05,
      "loss": 1.4699,
      "step": 3699
    },
    {
      "epoch": 5.153203342618385,
      "grad_norm": 3.240514039993286,
      "learning_rate": 3.462194428652643e-05,
      "loss": 1.2046,
      "step": 3700
    },
    {
      "epoch": 5.154596100278551,
      "grad_norm": 3.537564277648926,
      "learning_rate": 3.461199545196134e-05,
      "loss": 1.4906,
      "step": 3701
    },
    {
      "epoch": 5.155988857938719,
      "grad_norm": 4.372491836547852,
      "learning_rate": 3.460204661739625e-05,
      "loss": 1.5823,
      "step": 3702
    },
    {
      "epoch": 5.157381615598886,
      "grad_norm": 3.1347551345825195,
      "learning_rate": 3.459209778283115e-05,
      "loss": 1.3192,
      "step": 3703
    },
    {
      "epoch": 5.158774373259053,
      "grad_norm": 3.348072052001953,
      "learning_rate": 3.4582148948266056e-05,
      "loss": 1.478,
      "step": 3704
    },
    {
      "epoch": 5.16016713091922,
      "grad_norm": 3.973855972290039,
      "learning_rate": 3.457220011370096e-05,
      "loss": 1.7168,
      "step": 3705
    },
    {
      "epoch": 5.161559888579387,
      "grad_norm": 3.481712579727173,
      "learning_rate": 3.456225127913587e-05,
      "loss": 1.5055,
      "step": 3706
    },
    {
      "epoch": 5.162952646239554,
      "grad_norm": 3.2055375576019287,
      "learning_rate": 3.4552302444570775e-05,
      "loss": 1.3556,
      "step": 3707
    },
    {
      "epoch": 5.164345403899722,
      "grad_norm": 2.8089704513549805,
      "learning_rate": 3.454235361000568e-05,
      "loss": 1.3841,
      "step": 3708
    },
    {
      "epoch": 5.165738161559888,
      "grad_norm": 2.9740664958953857,
      "learning_rate": 3.453240477544059e-05,
      "loss": 1.5267,
      "step": 3709
    },
    {
      "epoch": 5.167130919220056,
      "grad_norm": 3.2898945808410645,
      "learning_rate": 3.4522455940875494e-05,
      "loss": 1.2577,
      "step": 3710
    },
    {
      "epoch": 5.1685236768802225,
      "grad_norm": 3.3988595008850098,
      "learning_rate": 3.45125071063104e-05,
      "loss": 1.3032,
      "step": 3711
    },
    {
      "epoch": 5.16991643454039,
      "grad_norm": 3.444390058517456,
      "learning_rate": 3.450255827174531e-05,
      "loss": 1.4137,
      "step": 3712
    },
    {
      "epoch": 5.171309192200557,
      "grad_norm": 2.9359185695648193,
      "learning_rate": 3.449260943718021e-05,
      "loss": 1.3204,
      "step": 3713
    },
    {
      "epoch": 5.172701949860724,
      "grad_norm": 3.4491429328918457,
      "learning_rate": 3.448266060261512e-05,
      "loss": 1.3318,
      "step": 3714
    },
    {
      "epoch": 5.174094707520891,
      "grad_norm": 3.563459873199463,
      "learning_rate": 3.447271176805002e-05,
      "loss": 1.3859,
      "step": 3715
    },
    {
      "epoch": 5.1754874651810585,
      "grad_norm": 4.120718479156494,
      "learning_rate": 3.446276293348493e-05,
      "loss": 1.708,
      "step": 3716
    },
    {
      "epoch": 5.176880222841225,
      "grad_norm": 3.0903916358947754,
      "learning_rate": 3.4452814098919836e-05,
      "loss": 1.303,
      "step": 3717
    },
    {
      "epoch": 5.178272980501393,
      "grad_norm": 3.113990068435669,
      "learning_rate": 3.444286526435474e-05,
      "loss": 1.3673,
      "step": 3718
    },
    {
      "epoch": 5.17966573816156,
      "grad_norm": 3.122307538986206,
      "learning_rate": 3.443291642978965e-05,
      "loss": 1.3382,
      "step": 3719
    },
    {
      "epoch": 5.181058495821727,
      "grad_norm": 3.141376495361328,
      "learning_rate": 3.4422967595224555e-05,
      "loss": 1.4865,
      "step": 3720
    },
    {
      "epoch": 5.1824512534818945,
      "grad_norm": 2.90157151222229,
      "learning_rate": 3.4413018760659466e-05,
      "loss": 1.4591,
      "step": 3721
    },
    {
      "epoch": 5.183844011142061,
      "grad_norm": 3.2187490463256836,
      "learning_rate": 3.440306992609437e-05,
      "loss": 1.4385,
      "step": 3722
    },
    {
      "epoch": 5.185236768802229,
      "grad_norm": 3.39043927192688,
      "learning_rate": 3.4393121091529274e-05,
      "loss": 1.7379,
      "step": 3723
    },
    {
      "epoch": 5.186629526462395,
      "grad_norm": 3.1450371742248535,
      "learning_rate": 3.4383172256964185e-05,
      "loss": 1.5507,
      "step": 3724
    },
    {
      "epoch": 5.188022284122563,
      "grad_norm": 3.4779839515686035,
      "learning_rate": 3.437322342239909e-05,
      "loss": 1.3501,
      "step": 3725
    },
    {
      "epoch": 5.18941504178273,
      "grad_norm": 3.0941572189331055,
      "learning_rate": 3.436327458783399e-05,
      "loss": 1.2161,
      "step": 3726
    },
    {
      "epoch": 5.190807799442897,
      "grad_norm": 3.643517017364502,
      "learning_rate": 3.43533257532689e-05,
      "loss": 1.3892,
      "step": 3727
    },
    {
      "epoch": 5.192200557103064,
      "grad_norm": 3.092085123062134,
      "learning_rate": 3.434337691870381e-05,
      "loss": 1.4778,
      "step": 3728
    },
    {
      "epoch": 5.193593314763231,
      "grad_norm": 3.0536091327667236,
      "learning_rate": 3.433342808413871e-05,
      "loss": 1.3603,
      "step": 3729
    },
    {
      "epoch": 5.194986072423398,
      "grad_norm": 3.3252575397491455,
      "learning_rate": 3.4323479249573616e-05,
      "loss": 1.593,
      "step": 3730
    },
    {
      "epoch": 5.196378830083566,
      "grad_norm": 3.5904736518859863,
      "learning_rate": 3.431353041500853e-05,
      "loss": 1.6221,
      "step": 3731
    },
    {
      "epoch": 5.197771587743732,
      "grad_norm": 3.3816256523132324,
      "learning_rate": 3.430358158044343e-05,
      "loss": 1.4817,
      "step": 3732
    },
    {
      "epoch": 5.1991643454039,
      "grad_norm": 3.674961566925049,
      "learning_rate": 3.4293632745878335e-05,
      "loss": 1.7566,
      "step": 3733
    },
    {
      "epoch": 5.2005571030640665,
      "grad_norm": 3.5699350833892822,
      "learning_rate": 3.4283683911313246e-05,
      "loss": 1.6321,
      "step": 3734
    },
    {
      "epoch": 5.201949860724234,
      "grad_norm": 3.3462531566619873,
      "learning_rate": 3.427373507674815e-05,
      "loss": 1.432,
      "step": 3735
    },
    {
      "epoch": 5.203342618384401,
      "grad_norm": 3.26177978515625,
      "learning_rate": 3.4263786242183054e-05,
      "loss": 1.3974,
      "step": 3736
    },
    {
      "epoch": 5.204735376044568,
      "grad_norm": 3.213848829269409,
      "learning_rate": 3.425383740761796e-05,
      "loss": 1.2837,
      "step": 3737
    },
    {
      "epoch": 5.206128133704736,
      "grad_norm": 3.2751214504241943,
      "learning_rate": 3.424388857305287e-05,
      "loss": 1.3041,
      "step": 3738
    },
    {
      "epoch": 5.2075208913649025,
      "grad_norm": 3.244657516479492,
      "learning_rate": 3.423393973848777e-05,
      "loss": 1.5692,
      "step": 3739
    },
    {
      "epoch": 5.20891364902507,
      "grad_norm": 3.5669984817504883,
      "learning_rate": 3.422399090392268e-05,
      "loss": 1.4378,
      "step": 3740
    },
    {
      "epoch": 5.210306406685237,
      "grad_norm": 2.9351229667663574,
      "learning_rate": 3.421404206935759e-05,
      "loss": 1.198,
      "step": 3741
    },
    {
      "epoch": 5.211699164345404,
      "grad_norm": 3.4855310916900635,
      "learning_rate": 3.420409323479249e-05,
      "loss": 1.5302,
      "step": 3742
    },
    {
      "epoch": 5.213091922005571,
      "grad_norm": 2.974017381668091,
      "learning_rate": 3.4194144400227396e-05,
      "loss": 1.2527,
      "step": 3743
    },
    {
      "epoch": 5.2144846796657385,
      "grad_norm": 3.190744400024414,
      "learning_rate": 3.41841955656623e-05,
      "loss": 1.4676,
      "step": 3744
    },
    {
      "epoch": 5.215877437325905,
      "grad_norm": 3.7585272789001465,
      "learning_rate": 3.417424673109721e-05,
      "loss": 1.9378,
      "step": 3745
    },
    {
      "epoch": 5.217270194986073,
      "grad_norm": 3.2659990787506104,
      "learning_rate": 3.416429789653212e-05,
      "loss": 1.4307,
      "step": 3746
    },
    {
      "epoch": 5.218662952646239,
      "grad_norm": 3.387321949005127,
      "learning_rate": 3.4154349061967026e-05,
      "loss": 1.4444,
      "step": 3747
    },
    {
      "epoch": 5.220055710306407,
      "grad_norm": 3.2577359676361084,
      "learning_rate": 3.414440022740193e-05,
      "loss": 1.5049,
      "step": 3748
    },
    {
      "epoch": 5.221448467966574,
      "grad_norm": 3.1732935905456543,
      "learning_rate": 3.4134451392836834e-05,
      "loss": 1.5884,
      "step": 3749
    },
    {
      "epoch": 5.222841225626741,
      "grad_norm": 3.5600271224975586,
      "learning_rate": 3.4124502558271745e-05,
      "loss": 1.8239,
      "step": 3750
    },
    {
      "epoch": 5.224233983286908,
      "grad_norm": 3.043459415435791,
      "learning_rate": 3.411455372370665e-05,
      "loss": 1.3098,
      "step": 3751
    },
    {
      "epoch": 5.225626740947075,
      "grad_norm": 3.24894380569458,
      "learning_rate": 3.4104604889141553e-05,
      "loss": 1.1873,
      "step": 3752
    },
    {
      "epoch": 5.227019498607242,
      "grad_norm": 3.9069149494171143,
      "learning_rate": 3.4094656054576464e-05,
      "loss": 1.3383,
      "step": 3753
    },
    {
      "epoch": 5.22841225626741,
      "grad_norm": 2.6927483081817627,
      "learning_rate": 3.408470722001137e-05,
      "loss": 1.0927,
      "step": 3754
    },
    {
      "epoch": 5.229805013927576,
      "grad_norm": 3.414612293243408,
      "learning_rate": 3.407475838544627e-05,
      "loss": 1.3463,
      "step": 3755
    },
    {
      "epoch": 5.231197771587744,
      "grad_norm": 3.3081040382385254,
      "learning_rate": 3.406480955088118e-05,
      "loss": 1.4498,
      "step": 3756
    },
    {
      "epoch": 5.2325905292479105,
      "grad_norm": 4.235991954803467,
      "learning_rate": 3.405486071631609e-05,
      "loss": 1.7435,
      "step": 3757
    },
    {
      "epoch": 5.233983286908078,
      "grad_norm": 3.453193426132202,
      "learning_rate": 3.404491188175099e-05,
      "loss": 1.3693,
      "step": 3758
    },
    {
      "epoch": 5.235376044568245,
      "grad_norm": 3.6752874851226807,
      "learning_rate": 3.4034963047185895e-05,
      "loss": 1.448,
      "step": 3759
    },
    {
      "epoch": 5.236768802228412,
      "grad_norm": 3.6105453968048096,
      "learning_rate": 3.4025014212620806e-05,
      "loss": 1.6106,
      "step": 3760
    },
    {
      "epoch": 5.23816155988858,
      "grad_norm": 2.91959547996521,
      "learning_rate": 3.401506537805571e-05,
      "loss": 1.3171,
      "step": 3761
    },
    {
      "epoch": 5.2395543175487465,
      "grad_norm": 3.491471290588379,
      "learning_rate": 3.4005116543490614e-05,
      "loss": 1.4736,
      "step": 3762
    },
    {
      "epoch": 5.240947075208914,
      "grad_norm": 3.8632736206054688,
      "learning_rate": 3.3995167708925525e-05,
      "loss": 1.4344,
      "step": 3763
    },
    {
      "epoch": 5.242339832869081,
      "grad_norm": 3.767866849899292,
      "learning_rate": 3.398521887436043e-05,
      "loss": 1.2747,
      "step": 3764
    },
    {
      "epoch": 5.243732590529248,
      "grad_norm": 3.0411815643310547,
      "learning_rate": 3.3975270039795333e-05,
      "loss": 1.3078,
      "step": 3765
    },
    {
      "epoch": 5.245125348189415,
      "grad_norm": 3.113668918609619,
      "learning_rate": 3.396532120523024e-05,
      "loss": 1.4455,
      "step": 3766
    },
    {
      "epoch": 5.2465181058495824,
      "grad_norm": 3.476451873779297,
      "learning_rate": 3.395537237066515e-05,
      "loss": 1.712,
      "step": 3767
    },
    {
      "epoch": 5.247910863509749,
      "grad_norm": 3.588526964187622,
      "learning_rate": 3.394542353610005e-05,
      "loss": 1.5368,
      "step": 3768
    },
    {
      "epoch": 5.249303621169917,
      "grad_norm": 2.9576282501220703,
      "learning_rate": 3.3935474701534957e-05,
      "loss": 1.2277,
      "step": 3769
    },
    {
      "epoch": 5.250696378830083,
      "grad_norm": 3.746317148208618,
      "learning_rate": 3.392552586696987e-05,
      "loss": 1.6772,
      "step": 3770
    },
    {
      "epoch": 5.252089136490251,
      "grad_norm": 3.6188101768493652,
      "learning_rate": 3.391557703240477e-05,
      "loss": 1.8135,
      "step": 3771
    },
    {
      "epoch": 5.2534818941504176,
      "grad_norm": 2.8830478191375732,
      "learning_rate": 3.3905628197839676e-05,
      "loss": 1.3099,
      "step": 3772
    },
    {
      "epoch": 5.254874651810585,
      "grad_norm": 3.50219988822937,
      "learning_rate": 3.3895679363274586e-05,
      "loss": 1.4418,
      "step": 3773
    },
    {
      "epoch": 5.256267409470752,
      "grad_norm": 3.100625991821289,
      "learning_rate": 3.388573052870949e-05,
      "loss": 1.0908,
      "step": 3774
    },
    {
      "epoch": 5.257660167130919,
      "grad_norm": 2.9331486225128174,
      "learning_rate": 3.38757816941444e-05,
      "loss": 1.0758,
      "step": 3775
    },
    {
      "epoch": 5.259052924791086,
      "grad_norm": 3.0599770545959473,
      "learning_rate": 3.3865832859579305e-05,
      "loss": 1.3869,
      "step": 3776
    },
    {
      "epoch": 5.2604456824512535,
      "grad_norm": 2.9075305461883545,
      "learning_rate": 3.385588402501421e-05,
      "loss": 1.12,
      "step": 3777
    },
    {
      "epoch": 5.26183844011142,
      "grad_norm": 3.2149698734283447,
      "learning_rate": 3.384593519044912e-05,
      "loss": 1.352,
      "step": 3778
    },
    {
      "epoch": 5.263231197771588,
      "grad_norm": 3.346226453781128,
      "learning_rate": 3.3835986355884024e-05,
      "loss": 1.1678,
      "step": 3779
    },
    {
      "epoch": 5.264623955431755,
      "grad_norm": 3.9807395935058594,
      "learning_rate": 3.382603752131893e-05,
      "loss": 1.9116,
      "step": 3780
    },
    {
      "epoch": 5.266016713091922,
      "grad_norm": 3.3949508666992188,
      "learning_rate": 3.381608868675383e-05,
      "loss": 1.587,
      "step": 3781
    },
    {
      "epoch": 5.2674094707520895,
      "grad_norm": 3.343817710876465,
      "learning_rate": 3.3806139852188743e-05,
      "loss": 1.3577,
      "step": 3782
    },
    {
      "epoch": 5.268802228412256,
      "grad_norm": 3.4596922397613525,
      "learning_rate": 3.379619101762365e-05,
      "loss": 1.2473,
      "step": 3783
    },
    {
      "epoch": 5.270194986072424,
      "grad_norm": 2.865809679031372,
      "learning_rate": 3.378624218305855e-05,
      "loss": 1.4484,
      "step": 3784
    },
    {
      "epoch": 5.27158774373259,
      "grad_norm": 3.792099952697754,
      "learning_rate": 3.377629334849346e-05,
      "loss": 1.6435,
      "step": 3785
    },
    {
      "epoch": 5.272980501392758,
      "grad_norm": 4.077797889709473,
      "learning_rate": 3.3766344513928367e-05,
      "loss": 1.4314,
      "step": 3786
    },
    {
      "epoch": 5.274373259052925,
      "grad_norm": 3.7077529430389404,
      "learning_rate": 3.375639567936327e-05,
      "loss": 1.4233,
      "step": 3787
    },
    {
      "epoch": 5.275766016713092,
      "grad_norm": 3.0881690979003906,
      "learning_rate": 3.3746446844798175e-05,
      "loss": 1.2292,
      "step": 3788
    },
    {
      "epoch": 5.277158774373259,
      "grad_norm": 4.164643287658691,
      "learning_rate": 3.3736498010233086e-05,
      "loss": 2.0144,
      "step": 3789
    },
    {
      "epoch": 5.278551532033426,
      "grad_norm": 3.2468597888946533,
      "learning_rate": 3.372654917566799e-05,
      "loss": 1.4749,
      "step": 3790
    },
    {
      "epoch": 5.279944289693593,
      "grad_norm": 3.39668607711792,
      "learning_rate": 3.3716600341102894e-05,
      "loss": 1.3774,
      "step": 3791
    },
    {
      "epoch": 5.281337047353761,
      "grad_norm": 3.3434839248657227,
      "learning_rate": 3.3706651506537805e-05,
      "loss": 1.5875,
      "step": 3792
    },
    {
      "epoch": 5.282729805013927,
      "grad_norm": 3.2673895359039307,
      "learning_rate": 3.369670267197271e-05,
      "loss": 1.5603,
      "step": 3793
    },
    {
      "epoch": 5.284122562674095,
      "grad_norm": 3.286590099334717,
      "learning_rate": 3.368675383740761e-05,
      "loss": 1.3379,
      "step": 3794
    },
    {
      "epoch": 5.2855153203342615,
      "grad_norm": 3.299992084503174,
      "learning_rate": 3.3676805002842524e-05,
      "loss": 1.5589,
      "step": 3795
    },
    {
      "epoch": 5.286908077994429,
      "grad_norm": 3.6170554161071777,
      "learning_rate": 3.366685616827743e-05,
      "loss": 1.6821,
      "step": 3796
    },
    {
      "epoch": 5.288300835654596,
      "grad_norm": 3.3409528732299805,
      "learning_rate": 3.365690733371233e-05,
      "loss": 1.5481,
      "step": 3797
    },
    {
      "epoch": 5.289693593314763,
      "grad_norm": 3.209533929824829,
      "learning_rate": 3.3646958499147236e-05,
      "loss": 1.3921,
      "step": 3798
    },
    {
      "epoch": 5.29108635097493,
      "grad_norm": 3.2695765495300293,
      "learning_rate": 3.3637009664582147e-05,
      "loss": 1.2969,
      "step": 3799
    },
    {
      "epoch": 5.2924791086350975,
      "grad_norm": 3.567899703979492,
      "learning_rate": 3.362706083001706e-05,
      "loss": 1.5348,
      "step": 3800
    },
    {
      "epoch": 5.293871866295264,
      "grad_norm": 3.4814798831939697,
      "learning_rate": 3.3617111995451955e-05,
      "loss": 1.228,
      "step": 3801
    },
    {
      "epoch": 5.295264623955432,
      "grad_norm": 3.332674503326416,
      "learning_rate": 3.3607163160886866e-05,
      "loss": 1.2396,
      "step": 3802
    },
    {
      "epoch": 5.296657381615599,
      "grad_norm": 4.195977687835693,
      "learning_rate": 3.359721432632177e-05,
      "loss": 1.5116,
      "step": 3803
    },
    {
      "epoch": 5.298050139275766,
      "grad_norm": 3.3662610054016113,
      "learning_rate": 3.358726549175668e-05,
      "loss": 1.5008,
      "step": 3804
    },
    {
      "epoch": 5.2994428969359335,
      "grad_norm": 3.4025464057922363,
      "learning_rate": 3.3577316657191585e-05,
      "loss": 1.6831,
      "step": 3805
    },
    {
      "epoch": 5.3008356545961,
      "grad_norm": 3.5710318088531494,
      "learning_rate": 3.356736782262649e-05,
      "loss": 1.577,
      "step": 3806
    },
    {
      "epoch": 5.302228412256268,
      "grad_norm": 3.1233208179473877,
      "learning_rate": 3.35574189880614e-05,
      "loss": 1.4833,
      "step": 3807
    },
    {
      "epoch": 5.303621169916434,
      "grad_norm": 3.377131462097168,
      "learning_rate": 3.3547470153496304e-05,
      "loss": 1.6813,
      "step": 3808
    },
    {
      "epoch": 5.305013927576602,
      "grad_norm": 3.482731342315674,
      "learning_rate": 3.353752131893121e-05,
      "loss": 1.5225,
      "step": 3809
    },
    {
      "epoch": 5.306406685236769,
      "grad_norm": 3.1043663024902344,
      "learning_rate": 3.352757248436611e-05,
      "loss": 1.2276,
      "step": 3810
    },
    {
      "epoch": 5.307799442896936,
      "grad_norm": 3.382906913757324,
      "learning_rate": 3.351762364980102e-05,
      "loss": 1.3289,
      "step": 3811
    },
    {
      "epoch": 5.309192200557103,
      "grad_norm": 3.4567394256591797,
      "learning_rate": 3.350767481523593e-05,
      "loss": 1.4484,
      "step": 3812
    },
    {
      "epoch": 5.31058495821727,
      "grad_norm": 3.7682783603668213,
      "learning_rate": 3.349772598067083e-05,
      "loss": 1.9119,
      "step": 3813
    },
    {
      "epoch": 5.311977715877437,
      "grad_norm": 3.658247470855713,
      "learning_rate": 3.348777714610574e-05,
      "loss": 1.6523,
      "step": 3814
    },
    {
      "epoch": 5.313370473537605,
      "grad_norm": 2.793503522872925,
      "learning_rate": 3.3477828311540646e-05,
      "loss": 1.2379,
      "step": 3815
    },
    {
      "epoch": 5.314763231197771,
      "grad_norm": 3.6581826210021973,
      "learning_rate": 3.346787947697555e-05,
      "loss": 1.6143,
      "step": 3816
    },
    {
      "epoch": 5.316155988857939,
      "grad_norm": 3.794476270675659,
      "learning_rate": 3.345793064241046e-05,
      "loss": 1.5009,
      "step": 3817
    },
    {
      "epoch": 5.3175487465181055,
      "grad_norm": 3.8433616161346436,
      "learning_rate": 3.3447981807845365e-05,
      "loss": 1.4117,
      "step": 3818
    },
    {
      "epoch": 5.318941504178273,
      "grad_norm": 2.8389317989349365,
      "learning_rate": 3.343803297328027e-05,
      "loss": 1.2281,
      "step": 3819
    },
    {
      "epoch": 5.32033426183844,
      "grad_norm": 3.3493552207946777,
      "learning_rate": 3.342808413871517e-05,
      "loss": 1.4341,
      "step": 3820
    },
    {
      "epoch": 5.321727019498607,
      "grad_norm": 4.652235507965088,
      "learning_rate": 3.3418135304150084e-05,
      "loss": 1.3654,
      "step": 3821
    },
    {
      "epoch": 5.323119777158775,
      "grad_norm": 3.134294271469116,
      "learning_rate": 3.340818646958499e-05,
      "loss": 1.5002,
      "step": 3822
    },
    {
      "epoch": 5.3245125348189415,
      "grad_norm": 3.2703158855438232,
      "learning_rate": 3.339823763501989e-05,
      "loss": 1.4319,
      "step": 3823
    },
    {
      "epoch": 5.325905292479109,
      "grad_norm": 3.316800117492676,
      "learning_rate": 3.33882888004548e-05,
      "loss": 1.4161,
      "step": 3824
    },
    {
      "epoch": 5.327298050139276,
      "grad_norm": 2.9278154373168945,
      "learning_rate": 3.337833996588971e-05,
      "loss": 1.2484,
      "step": 3825
    },
    {
      "epoch": 5.328690807799443,
      "grad_norm": 3.4309632778167725,
      "learning_rate": 3.336839113132461e-05,
      "loss": 1.4719,
      "step": 3826
    },
    {
      "epoch": 5.33008356545961,
      "grad_norm": 3.2639284133911133,
      "learning_rate": 3.3358442296759515e-05,
      "loss": 1.474,
      "step": 3827
    },
    {
      "epoch": 5.3314763231197775,
      "grad_norm": 3.263230323791504,
      "learning_rate": 3.3348493462194426e-05,
      "loss": 1.5239,
      "step": 3828
    },
    {
      "epoch": 5.332869080779944,
      "grad_norm": 3.6524100303649902,
      "learning_rate": 3.3338544627629337e-05,
      "loss": 1.4968,
      "step": 3829
    },
    {
      "epoch": 5.334261838440112,
      "grad_norm": 3.077557325363159,
      "learning_rate": 3.3328595793064234e-05,
      "loss": 1.0446,
      "step": 3830
    },
    {
      "epoch": 5.335654596100278,
      "grad_norm": 3.468564987182617,
      "learning_rate": 3.3318646958499145e-05,
      "loss": 1.5119,
      "step": 3831
    },
    {
      "epoch": 5.337047353760446,
      "grad_norm": 3.4484708309173584,
      "learning_rate": 3.330869812393405e-05,
      "loss": 1.5385,
      "step": 3832
    },
    {
      "epoch": 5.338440111420613,
      "grad_norm": 3.177473306655884,
      "learning_rate": 3.329874928936896e-05,
      "loss": 1.1245,
      "step": 3833
    },
    {
      "epoch": 5.33983286908078,
      "grad_norm": 3.467501163482666,
      "learning_rate": 3.3288800454803864e-05,
      "loss": 1.576,
      "step": 3834
    },
    {
      "epoch": 5.341225626740947,
      "grad_norm": 3.00364089012146,
      "learning_rate": 3.327885162023877e-05,
      "loss": 1.2415,
      "step": 3835
    },
    {
      "epoch": 5.342618384401114,
      "grad_norm": 4.731228351593018,
      "learning_rate": 3.326890278567368e-05,
      "loss": 1.3672,
      "step": 3836
    },
    {
      "epoch": 5.344011142061281,
      "grad_norm": 3.201103925704956,
      "learning_rate": 3.325895395110858e-05,
      "loss": 1.4716,
      "step": 3837
    },
    {
      "epoch": 5.345403899721449,
      "grad_norm": 3.115396022796631,
      "learning_rate": 3.324900511654349e-05,
      "loss": 1.2102,
      "step": 3838
    },
    {
      "epoch": 5.346796657381615,
      "grad_norm": 3.1591529846191406,
      "learning_rate": 3.32390562819784e-05,
      "loss": 1.2851,
      "step": 3839
    },
    {
      "epoch": 5.348189415041783,
      "grad_norm": 2.949540138244629,
      "learning_rate": 3.32291074474133e-05,
      "loss": 1.2902,
      "step": 3840
    },
    {
      "epoch": 5.3495821727019495,
      "grad_norm": 3.423433780670166,
      "learning_rate": 3.3219158612848206e-05,
      "loss": 1.4414,
      "step": 3841
    },
    {
      "epoch": 5.350974930362117,
      "grad_norm": 3.4719269275665283,
      "learning_rate": 3.320920977828311e-05,
      "loss": 1.5675,
      "step": 3842
    },
    {
      "epoch": 5.352367688022284,
      "grad_norm": 3.1271326541900635,
      "learning_rate": 3.319926094371802e-05,
      "loss": 1.1859,
      "step": 3843
    },
    {
      "epoch": 5.353760445682451,
      "grad_norm": 3.346649646759033,
      "learning_rate": 3.3189312109152925e-05,
      "loss": 1.2755,
      "step": 3844
    },
    {
      "epoch": 5.355153203342619,
      "grad_norm": 3.6410646438598633,
      "learning_rate": 3.317936327458783e-05,
      "loss": 1.6861,
      "step": 3845
    },
    {
      "epoch": 5.3565459610027855,
      "grad_norm": 3.6382384300231934,
      "learning_rate": 3.316941444002274e-05,
      "loss": 1.5115,
      "step": 3846
    },
    {
      "epoch": 5.357938718662953,
      "grad_norm": 3.3190386295318604,
      "learning_rate": 3.3159465605457644e-05,
      "loss": 1.5371,
      "step": 3847
    },
    {
      "epoch": 5.35933147632312,
      "grad_norm": 5.246037006378174,
      "learning_rate": 3.314951677089255e-05,
      "loss": 1.5903,
      "step": 3848
    },
    {
      "epoch": 5.360724233983287,
      "grad_norm": 5.078573226928711,
      "learning_rate": 3.313956793632745e-05,
      "loss": 1.3486,
      "step": 3849
    },
    {
      "epoch": 5.362116991643454,
      "grad_norm": 2.915461778640747,
      "learning_rate": 3.312961910176236e-05,
      "loss": 1.3303,
      "step": 3850
    },
    {
      "epoch": 5.3635097493036215,
      "grad_norm": 4.221773147583008,
      "learning_rate": 3.311967026719727e-05,
      "loss": 1.562,
      "step": 3851
    },
    {
      "epoch": 5.364902506963788,
      "grad_norm": 4.714282035827637,
      "learning_rate": 3.310972143263217e-05,
      "loss": 1.506,
      "step": 3852
    },
    {
      "epoch": 5.366295264623956,
      "grad_norm": 3.0521111488342285,
      "learning_rate": 3.309977259806708e-05,
      "loss": 1.1443,
      "step": 3853
    },
    {
      "epoch": 5.367688022284122,
      "grad_norm": 3.915872097015381,
      "learning_rate": 3.3089823763501986e-05,
      "loss": 1.6592,
      "step": 3854
    },
    {
      "epoch": 5.36908077994429,
      "grad_norm": 3.1428794860839844,
      "learning_rate": 3.307987492893689e-05,
      "loss": 1.3548,
      "step": 3855
    },
    {
      "epoch": 5.370473537604457,
      "grad_norm": 3.683709144592285,
      "learning_rate": 3.30699260943718e-05,
      "loss": 1.6792,
      "step": 3856
    },
    {
      "epoch": 5.371866295264624,
      "grad_norm": 3.304654359817505,
      "learning_rate": 3.3059977259806705e-05,
      "loss": 1.4264,
      "step": 3857
    },
    {
      "epoch": 5.373259052924791,
      "grad_norm": 3.9296340942382812,
      "learning_rate": 3.3050028425241616e-05,
      "loss": 1.7694,
      "step": 3858
    },
    {
      "epoch": 5.374651810584958,
      "grad_norm": 3.66880202293396,
      "learning_rate": 3.304007959067652e-05,
      "loss": 1.9336,
      "step": 3859
    },
    {
      "epoch": 5.376044568245125,
      "grad_norm": 3.4877336025238037,
      "learning_rate": 3.3030130756111424e-05,
      "loss": 1.248,
      "step": 3860
    },
    {
      "epoch": 5.3774373259052926,
      "grad_norm": 3.447391986846924,
      "learning_rate": 3.3020181921546335e-05,
      "loss": 1.5482,
      "step": 3861
    },
    {
      "epoch": 5.378830083565459,
      "grad_norm": 6.742697715759277,
      "learning_rate": 3.301023308698124e-05,
      "loss": 1.5847,
      "step": 3862
    },
    {
      "epoch": 5.380222841225627,
      "grad_norm": 3.357435464859009,
      "learning_rate": 3.300028425241614e-05,
      "loss": 1.3434,
      "step": 3863
    },
    {
      "epoch": 5.381615598885794,
      "grad_norm": 3.4757871627807617,
      "learning_rate": 3.299033541785105e-05,
      "loss": 1.4059,
      "step": 3864
    },
    {
      "epoch": 5.383008356545961,
      "grad_norm": 4.1926350593566895,
      "learning_rate": 3.298038658328596e-05,
      "loss": 1.5506,
      "step": 3865
    },
    {
      "epoch": 5.3844011142061285,
      "grad_norm": 3.4357733726501465,
      "learning_rate": 3.297043774872086e-05,
      "loss": 1.6557,
      "step": 3866
    },
    {
      "epoch": 5.385793871866295,
      "grad_norm": 3.7321584224700928,
      "learning_rate": 3.2960488914155766e-05,
      "loss": 1.612,
      "step": 3867
    },
    {
      "epoch": 5.387186629526463,
      "grad_norm": 3.0687263011932373,
      "learning_rate": 3.295054007959068e-05,
      "loss": 1.4702,
      "step": 3868
    },
    {
      "epoch": 5.388579387186629,
      "grad_norm": 3.526613473892212,
      "learning_rate": 3.294059124502558e-05,
      "loss": 1.6784,
      "step": 3869
    },
    {
      "epoch": 5.389972144846797,
      "grad_norm": 3.673717737197876,
      "learning_rate": 3.2930642410460485e-05,
      "loss": 1.3199,
      "step": 3870
    },
    {
      "epoch": 5.391364902506964,
      "grad_norm": 3.002877712249756,
      "learning_rate": 3.292069357589539e-05,
      "loss": 1.2214,
      "step": 3871
    },
    {
      "epoch": 5.392757660167131,
      "grad_norm": 3.2002980709075928,
      "learning_rate": 3.29107447413303e-05,
      "loss": 1.4956,
      "step": 3872
    },
    {
      "epoch": 5.394150417827298,
      "grad_norm": 2.985261917114258,
      "learning_rate": 3.2900795906765204e-05,
      "loss": 1.3173,
      "step": 3873
    },
    {
      "epoch": 5.395543175487465,
      "grad_norm": 3.5086417198181152,
      "learning_rate": 3.289084707220011e-05,
      "loss": 1.3256,
      "step": 3874
    },
    {
      "epoch": 5.396935933147632,
      "grad_norm": 2.9871435165405273,
      "learning_rate": 3.288089823763502e-05,
      "loss": 1.4463,
      "step": 3875
    },
    {
      "epoch": 5.3983286908078,
      "grad_norm": 3.4604439735412598,
      "learning_rate": 3.287094940306992e-05,
      "loss": 1.4626,
      "step": 3876
    },
    {
      "epoch": 5.399721448467966,
      "grad_norm": 3.3530375957489014,
      "learning_rate": 3.286100056850483e-05,
      "loss": 1.3948,
      "step": 3877
    },
    {
      "epoch": 5.401114206128134,
      "grad_norm": 4.094565391540527,
      "learning_rate": 3.285105173393974e-05,
      "loss": 1.4901,
      "step": 3878
    },
    {
      "epoch": 5.4025069637883005,
      "grad_norm": 3.299420118331909,
      "learning_rate": 3.284110289937464e-05,
      "loss": 1.4508,
      "step": 3879
    },
    {
      "epoch": 5.403899721448468,
      "grad_norm": 3.154717445373535,
      "learning_rate": 3.2831154064809546e-05,
      "loss": 1.1577,
      "step": 3880
    },
    {
      "epoch": 5.405292479108635,
      "grad_norm": 3.361546277999878,
      "learning_rate": 3.282120523024445e-05,
      "loss": 1.3684,
      "step": 3881
    },
    {
      "epoch": 5.406685236768802,
      "grad_norm": 4.2829909324646,
      "learning_rate": 3.281125639567936e-05,
      "loss": 1.5438,
      "step": 3882
    },
    {
      "epoch": 5.408077994428969,
      "grad_norm": 3.4301254749298096,
      "learning_rate": 3.2801307561114265e-05,
      "loss": 1.3747,
      "step": 3883
    },
    {
      "epoch": 5.4094707520891365,
      "grad_norm": 3.082698106765747,
      "learning_rate": 3.279135872654917e-05,
      "loss": 1.3142,
      "step": 3884
    },
    {
      "epoch": 5.410863509749303,
      "grad_norm": 3.568859815597534,
      "learning_rate": 3.278140989198408e-05,
      "loss": 1.6341,
      "step": 3885
    },
    {
      "epoch": 5.412256267409471,
      "grad_norm": 2.8952035903930664,
      "learning_rate": 3.2771461057418984e-05,
      "loss": 1.2079,
      "step": 3886
    },
    {
      "epoch": 5.413649025069638,
      "grad_norm": 3.228323221206665,
      "learning_rate": 3.2761512222853895e-05,
      "loss": 1.4066,
      "step": 3887
    },
    {
      "epoch": 5.415041782729805,
      "grad_norm": 3.5163016319274902,
      "learning_rate": 3.27515633882888e-05,
      "loss": 1.5318,
      "step": 3888
    },
    {
      "epoch": 5.4164345403899725,
      "grad_norm": 3.3017849922180176,
      "learning_rate": 3.27416145537237e-05,
      "loss": 1.2807,
      "step": 3889
    },
    {
      "epoch": 5.417827298050139,
      "grad_norm": 3.5008184909820557,
      "learning_rate": 3.2731665719158614e-05,
      "loss": 1.4937,
      "step": 3890
    },
    {
      "epoch": 5.419220055710307,
      "grad_norm": 3.5234081745147705,
      "learning_rate": 3.272171688459352e-05,
      "loss": 1.4323,
      "step": 3891
    },
    {
      "epoch": 5.420612813370473,
      "grad_norm": 3.7362451553344727,
      "learning_rate": 3.271176805002842e-05,
      "loss": 1.4941,
      "step": 3892
    },
    {
      "epoch": 5.422005571030641,
      "grad_norm": 2.8877837657928467,
      "learning_rate": 3.2701819215463326e-05,
      "loss": 1.3242,
      "step": 3893
    },
    {
      "epoch": 5.423398328690808,
      "grad_norm": 3.115877866744995,
      "learning_rate": 3.269187038089824e-05,
      "loss": 1.3975,
      "step": 3894
    },
    {
      "epoch": 5.424791086350975,
      "grad_norm": 3.7254276275634766,
      "learning_rate": 3.268192154633314e-05,
      "loss": 1.394,
      "step": 3895
    },
    {
      "epoch": 5.426183844011142,
      "grad_norm": 3.117959499359131,
      "learning_rate": 3.2671972711768045e-05,
      "loss": 1.3566,
      "step": 3896
    },
    {
      "epoch": 5.427576601671309,
      "grad_norm": 3.875413417816162,
      "learning_rate": 3.2662023877202956e-05,
      "loss": 1.3033,
      "step": 3897
    },
    {
      "epoch": 5.428969359331476,
      "grad_norm": 3.9885122776031494,
      "learning_rate": 3.265207504263786e-05,
      "loss": 1.6364,
      "step": 3898
    },
    {
      "epoch": 5.430362116991644,
      "grad_norm": 2.8498659133911133,
      "learning_rate": 3.2642126208072764e-05,
      "loss": 1.3442,
      "step": 3899
    },
    {
      "epoch": 5.43175487465181,
      "grad_norm": 3.3412253856658936,
      "learning_rate": 3.2632177373507675e-05,
      "loss": 1.3209,
      "step": 3900
    },
    {
      "epoch": 5.433147632311978,
      "grad_norm": 3.333075523376465,
      "learning_rate": 3.262222853894258e-05,
      "loss": 1.5815,
      "step": 3901
    },
    {
      "epoch": 5.4345403899721445,
      "grad_norm": 3.547705888748169,
      "learning_rate": 3.261227970437748e-05,
      "loss": 1.6916,
      "step": 3902
    },
    {
      "epoch": 5.435933147632312,
      "grad_norm": 3.714667320251465,
      "learning_rate": 3.260233086981239e-05,
      "loss": 1.5171,
      "step": 3903
    },
    {
      "epoch": 5.437325905292479,
      "grad_norm": 3.384195566177368,
      "learning_rate": 3.25923820352473e-05,
      "loss": 1.4183,
      "step": 3904
    },
    {
      "epoch": 5.438718662952646,
      "grad_norm": 3.1017160415649414,
      "learning_rate": 3.25824332006822e-05,
      "loss": 1.4028,
      "step": 3905
    },
    {
      "epoch": 5.440111420612814,
      "grad_norm": 4.297664165496826,
      "learning_rate": 3.2572484366117106e-05,
      "loss": 1.5649,
      "step": 3906
    },
    {
      "epoch": 5.4415041782729805,
      "grad_norm": 3.9159302711486816,
      "learning_rate": 3.256253553155202e-05,
      "loss": 1.4022,
      "step": 3907
    },
    {
      "epoch": 5.442896935933147,
      "grad_norm": 3.391209840774536,
      "learning_rate": 3.255258669698692e-05,
      "loss": 1.3308,
      "step": 3908
    },
    {
      "epoch": 5.444289693593315,
      "grad_norm": 3.9618213176727295,
      "learning_rate": 3.2542637862421825e-05,
      "loss": 1.5808,
      "step": 3909
    },
    {
      "epoch": 5.445682451253482,
      "grad_norm": 3.6104896068573,
      "learning_rate": 3.253268902785673e-05,
      "loss": 1.4468,
      "step": 3910
    },
    {
      "epoch": 5.447075208913649,
      "grad_norm": 3.209728240966797,
      "learning_rate": 3.252274019329164e-05,
      "loss": 1.3776,
      "step": 3911
    },
    {
      "epoch": 5.4484679665738165,
      "grad_norm": 3.180717706680298,
      "learning_rate": 3.251279135872655e-05,
      "loss": 1.2867,
      "step": 3912
    },
    {
      "epoch": 5.449860724233983,
      "grad_norm": 3.3016552925109863,
      "learning_rate": 3.250284252416145e-05,
      "loss": 1.3566,
      "step": 3913
    },
    {
      "epoch": 5.451253481894151,
      "grad_norm": 3.617339611053467,
      "learning_rate": 3.249289368959636e-05,
      "loss": 1.6453,
      "step": 3914
    },
    {
      "epoch": 5.452646239554317,
      "grad_norm": 3.0904672145843506,
      "learning_rate": 3.248294485503126e-05,
      "loss": 1.3079,
      "step": 3915
    },
    {
      "epoch": 5.454038997214485,
      "grad_norm": 3.719731330871582,
      "learning_rate": 3.2472996020466174e-05,
      "loss": 1.4557,
      "step": 3916
    },
    {
      "epoch": 5.455431754874652,
      "grad_norm": 3.3843271732330322,
      "learning_rate": 3.246304718590108e-05,
      "loss": 1.4903,
      "step": 3917
    },
    {
      "epoch": 5.456824512534819,
      "grad_norm": 3.0128655433654785,
      "learning_rate": 3.245309835133598e-05,
      "loss": 1.191,
      "step": 3918
    },
    {
      "epoch": 5.458217270194986,
      "grad_norm": 3.1585681438446045,
      "learning_rate": 3.244314951677089e-05,
      "loss": 1.4083,
      "step": 3919
    },
    {
      "epoch": 5.459610027855153,
      "grad_norm": 3.3376874923706055,
      "learning_rate": 3.24332006822058e-05,
      "loss": 1.6112,
      "step": 3920
    },
    {
      "epoch": 5.46100278551532,
      "grad_norm": 3.072366237640381,
      "learning_rate": 3.24232518476407e-05,
      "loss": 1.1329,
      "step": 3921
    },
    {
      "epoch": 5.462395543175488,
      "grad_norm": 3.20548415184021,
      "learning_rate": 3.241330301307561e-05,
      "loss": 1.4211,
      "step": 3922
    },
    {
      "epoch": 5.463788300835654,
      "grad_norm": 3.1418893337249756,
      "learning_rate": 3.2403354178510516e-05,
      "loss": 1.3566,
      "step": 3923
    },
    {
      "epoch": 5.465181058495822,
      "grad_norm": 3.166402578353882,
      "learning_rate": 3.239340534394542e-05,
      "loss": 1.1901,
      "step": 3924
    },
    {
      "epoch": 5.4665738161559885,
      "grad_norm": 3.5038914680480957,
      "learning_rate": 3.2383456509380324e-05,
      "loss": 1.5809,
      "step": 3925
    },
    {
      "epoch": 5.467966573816156,
      "grad_norm": 3.311577081680298,
      "learning_rate": 3.2373507674815235e-05,
      "loss": 1.4073,
      "step": 3926
    },
    {
      "epoch": 5.469359331476323,
      "grad_norm": 3.3907663822174072,
      "learning_rate": 3.236355884025014e-05,
      "loss": 1.2873,
      "step": 3927
    },
    {
      "epoch": 5.47075208913649,
      "grad_norm": 3.972849130630493,
      "learning_rate": 3.235361000568504e-05,
      "loss": 1.4817,
      "step": 3928
    },
    {
      "epoch": 5.472144846796658,
      "grad_norm": 3.3276312351226807,
      "learning_rate": 3.2343661171119954e-05,
      "loss": 1.284,
      "step": 3929
    },
    {
      "epoch": 5.4735376044568245,
      "grad_norm": 2.837625026702881,
      "learning_rate": 3.233371233655486e-05,
      "loss": 1.099,
      "step": 3930
    },
    {
      "epoch": 5.474930362116992,
      "grad_norm": 3.2056424617767334,
      "learning_rate": 3.232376350198976e-05,
      "loss": 1.3397,
      "step": 3931
    },
    {
      "epoch": 5.476323119777159,
      "grad_norm": 3.7157018184661865,
      "learning_rate": 3.2313814667424666e-05,
      "loss": 1.6531,
      "step": 3932
    },
    {
      "epoch": 5.477715877437326,
      "grad_norm": 3.524932861328125,
      "learning_rate": 3.230386583285958e-05,
      "loss": 1.2025,
      "step": 3933
    },
    {
      "epoch": 5.479108635097493,
      "grad_norm": 3.128558874130249,
      "learning_rate": 3.229391699829448e-05,
      "loss": 1.3906,
      "step": 3934
    },
    {
      "epoch": 5.4805013927576605,
      "grad_norm": 2.9211344718933105,
      "learning_rate": 3.2283968163729385e-05,
      "loss": 1.2087,
      "step": 3935
    },
    {
      "epoch": 5.481894150417827,
      "grad_norm": 3.499192714691162,
      "learning_rate": 3.2274019329164296e-05,
      "loss": 1.3862,
      "step": 3936
    },
    {
      "epoch": 5.483286908077995,
      "grad_norm": 2.7200067043304443,
      "learning_rate": 3.22640704945992e-05,
      "loss": 1.1445,
      "step": 3937
    },
    {
      "epoch": 5.484679665738161,
      "grad_norm": 3.847712755203247,
      "learning_rate": 3.2254121660034104e-05,
      "loss": 1.3631,
      "step": 3938
    },
    {
      "epoch": 5.486072423398329,
      "grad_norm": 4.017714977264404,
      "learning_rate": 3.2244172825469015e-05,
      "loss": 1.6958,
      "step": 3939
    },
    {
      "epoch": 5.487465181058496,
      "grad_norm": 3.3647825717926025,
      "learning_rate": 3.223422399090392e-05,
      "loss": 1.376,
      "step": 3940
    },
    {
      "epoch": 5.488857938718663,
      "grad_norm": 2.96148943901062,
      "learning_rate": 3.222427515633883e-05,
      "loss": 1.1843,
      "step": 3941
    },
    {
      "epoch": 5.49025069637883,
      "grad_norm": 3.516737222671509,
      "learning_rate": 3.221432632177373e-05,
      "loss": 1.6118,
      "step": 3942
    },
    {
      "epoch": 5.491643454038997,
      "grad_norm": 2.888239622116089,
      "learning_rate": 3.220437748720864e-05,
      "loss": 1.2634,
      "step": 3943
    },
    {
      "epoch": 5.493036211699164,
      "grad_norm": 3.1161155700683594,
      "learning_rate": 3.219442865264355e-05,
      "loss": 1.2417,
      "step": 3944
    },
    {
      "epoch": 5.494428969359332,
      "grad_norm": 3.035766124725342,
      "learning_rate": 3.218447981807845e-05,
      "loss": 1.2204,
      "step": 3945
    },
    {
      "epoch": 5.495821727019498,
      "grad_norm": 3.7109148502349854,
      "learning_rate": 3.217453098351336e-05,
      "loss": 1.5756,
      "step": 3946
    },
    {
      "epoch": 5.497214484679666,
      "grad_norm": 3.837930679321289,
      "learning_rate": 3.216458214894826e-05,
      "loss": 1.187,
      "step": 3947
    },
    {
      "epoch": 5.498607242339833,
      "grad_norm": 3.503160238265991,
      "learning_rate": 3.215463331438317e-05,
      "loss": 1.3288,
      "step": 3948
    },
    {
      "epoch": 5.5,
      "grad_norm": 2.700408697128296,
      "learning_rate": 3.2144684479818076e-05,
      "loss": 1.1584,
      "step": 3949
    },
    {
      "epoch": 5.501392757660167,
      "grad_norm": 3.317624092102051,
      "learning_rate": 3.213473564525298e-05,
      "loss": 1.3599,
      "step": 3950
    },
    {
      "epoch": 5.502785515320334,
      "grad_norm": 3.801151990890503,
      "learning_rate": 3.212478681068789e-05,
      "loss": 1.4942,
      "step": 3951
    },
    {
      "epoch": 5.504178272980502,
      "grad_norm": 3.7417783737182617,
      "learning_rate": 3.2114837976122795e-05,
      "loss": 1.1773,
      "step": 3952
    },
    {
      "epoch": 5.505571030640668,
      "grad_norm": 3.4485790729522705,
      "learning_rate": 3.21048891415577e-05,
      "loss": 1.4223,
      "step": 3953
    },
    {
      "epoch": 5.506963788300836,
      "grad_norm": 3.1726791858673096,
      "learning_rate": 3.2094940306992603e-05,
      "loss": 1.2985,
      "step": 3954
    },
    {
      "epoch": 5.508356545961003,
      "grad_norm": 3.7761385440826416,
      "learning_rate": 3.2084991472427514e-05,
      "loss": 1.3546,
      "step": 3955
    },
    {
      "epoch": 5.50974930362117,
      "grad_norm": 3.064495086669922,
      "learning_rate": 3.207504263786242e-05,
      "loss": 1.3108,
      "step": 3956
    },
    {
      "epoch": 5.511142061281337,
      "grad_norm": 3.1682968139648438,
      "learning_rate": 3.206509380329732e-05,
      "loss": 1.2865,
      "step": 3957
    },
    {
      "epoch": 5.512534818941504,
      "grad_norm": 2.9550204277038574,
      "learning_rate": 3.205514496873223e-05,
      "loss": 1.1755,
      "step": 3958
    },
    {
      "epoch": 5.513927576601671,
      "grad_norm": 3.430180788040161,
      "learning_rate": 3.204519613416714e-05,
      "loss": 1.3638,
      "step": 3959
    },
    {
      "epoch": 5.515320334261839,
      "grad_norm": 3.227559804916382,
      "learning_rate": 3.203524729960204e-05,
      "loss": 1.3122,
      "step": 3960
    },
    {
      "epoch": 5.516713091922005,
      "grad_norm": 2.9942450523376465,
      "learning_rate": 3.202529846503695e-05,
      "loss": 1.2739,
      "step": 3961
    },
    {
      "epoch": 5.518105849582173,
      "grad_norm": 4.473596096038818,
      "learning_rate": 3.2015349630471856e-05,
      "loss": 1.5011,
      "step": 3962
    },
    {
      "epoch": 5.5194986072423395,
      "grad_norm": 3.246814250946045,
      "learning_rate": 3.200540079590676e-05,
      "loss": 1.4086,
      "step": 3963
    },
    {
      "epoch": 5.520891364902507,
      "grad_norm": 3.2652289867401123,
      "learning_rate": 3.1995451961341665e-05,
      "loss": 1.4842,
      "step": 3964
    },
    {
      "epoch": 5.522284122562674,
      "grad_norm": 3.1682024002075195,
      "learning_rate": 3.1985503126776575e-05,
      "loss": 1.3278,
      "step": 3965
    },
    {
      "epoch": 5.523676880222841,
      "grad_norm": 3.0474061965942383,
      "learning_rate": 3.197555429221148e-05,
      "loss": 1.3106,
      "step": 3966
    },
    {
      "epoch": 5.525069637883008,
      "grad_norm": 3.323626756668091,
      "learning_rate": 3.1965605457646384e-05,
      "loss": 1.3913,
      "step": 3967
    },
    {
      "epoch": 5.5264623955431755,
      "grad_norm": 3.210984230041504,
      "learning_rate": 3.1955656623081294e-05,
      "loss": 1.2502,
      "step": 3968
    },
    {
      "epoch": 5.527855153203342,
      "grad_norm": 4.018118858337402,
      "learning_rate": 3.19457077885162e-05,
      "loss": 1.1894,
      "step": 3969
    },
    {
      "epoch": 5.52924791086351,
      "grad_norm": 2.9094223976135254,
      "learning_rate": 3.193575895395111e-05,
      "loss": 1.1678,
      "step": 3970
    },
    {
      "epoch": 5.530640668523677,
      "grad_norm": 3.7920138835906982,
      "learning_rate": 3.192581011938601e-05,
      "loss": 1.4131,
      "step": 3971
    },
    {
      "epoch": 5.532033426183844,
      "grad_norm": 3.238229513168335,
      "learning_rate": 3.191586128482092e-05,
      "loss": 1.4148,
      "step": 3972
    },
    {
      "epoch": 5.5334261838440115,
      "grad_norm": 3.8313090801239014,
      "learning_rate": 3.190591245025583e-05,
      "loss": 1.6562,
      "step": 3973
    },
    {
      "epoch": 5.534818941504178,
      "grad_norm": 5.136152744293213,
      "learning_rate": 3.189596361569073e-05,
      "loss": 1.4177,
      "step": 3974
    },
    {
      "epoch": 5.536211699164346,
      "grad_norm": 3.5499415397644043,
      "learning_rate": 3.1886014781125636e-05,
      "loss": 1.449,
      "step": 3975
    },
    {
      "epoch": 5.537604456824512,
      "grad_norm": 2.9771597385406494,
      "learning_rate": 3.187606594656054e-05,
      "loss": 1.1568,
      "step": 3976
    },
    {
      "epoch": 5.53899721448468,
      "grad_norm": 3.252581834793091,
      "learning_rate": 3.186611711199545e-05,
      "loss": 1.2848,
      "step": 3977
    },
    {
      "epoch": 5.540389972144847,
      "grad_norm": 3.4198732376098633,
      "learning_rate": 3.1856168277430355e-05,
      "loss": 1.3266,
      "step": 3978
    },
    {
      "epoch": 5.541782729805014,
      "grad_norm": 3.03037691116333,
      "learning_rate": 3.184621944286526e-05,
      "loss": 1.2643,
      "step": 3979
    },
    {
      "epoch": 5.543175487465181,
      "grad_norm": 3.268641233444214,
      "learning_rate": 3.183627060830017e-05,
      "loss": 1.468,
      "step": 3980
    },
    {
      "epoch": 5.544568245125348,
      "grad_norm": 4.733526229858398,
      "learning_rate": 3.1826321773735074e-05,
      "loss": 1.5814,
      "step": 3981
    },
    {
      "epoch": 5.545961002785515,
      "grad_norm": 4.59187650680542,
      "learning_rate": 3.181637293916998e-05,
      "loss": 1.6145,
      "step": 3982
    },
    {
      "epoch": 5.547353760445683,
      "grad_norm": 3.706252098083496,
      "learning_rate": 3.180642410460489e-05,
      "loss": 1.4103,
      "step": 3983
    },
    {
      "epoch": 5.548746518105849,
      "grad_norm": 2.993542194366455,
      "learning_rate": 3.1796475270039793e-05,
      "loss": 1.1197,
      "step": 3984
    },
    {
      "epoch": 5.550139275766017,
      "grad_norm": 3.199744939804077,
      "learning_rate": 3.17865264354747e-05,
      "loss": 1.5324,
      "step": 3985
    },
    {
      "epoch": 5.5515320334261835,
      "grad_norm": 3.2078278064727783,
      "learning_rate": 3.17765776009096e-05,
      "loss": 1.3773,
      "step": 3986
    },
    {
      "epoch": 5.552924791086351,
      "grad_norm": 3.8803422451019287,
      "learning_rate": 3.176662876634451e-05,
      "loss": 1.3681,
      "step": 3987
    },
    {
      "epoch": 5.554317548746518,
      "grad_norm": 3.2783961296081543,
      "learning_rate": 3.1756679931779417e-05,
      "loss": 1.4319,
      "step": 3988
    },
    {
      "epoch": 5.555710306406685,
      "grad_norm": 6.760887145996094,
      "learning_rate": 3.174673109721432e-05,
      "loss": 1.2848,
      "step": 3989
    },
    {
      "epoch": 5.557103064066853,
      "grad_norm": 3.0715172290802,
      "learning_rate": 3.173678226264923e-05,
      "loss": 1.3646,
      "step": 3990
    },
    {
      "epoch": 5.5584958217270195,
      "grad_norm": 3.3107216358184814,
      "learning_rate": 3.1726833428084136e-05,
      "loss": 1.5385,
      "step": 3991
    },
    {
      "epoch": 5.559888579387186,
      "grad_norm": 3.1300973892211914,
      "learning_rate": 3.171688459351904e-05,
      "loss": 1.3358,
      "step": 3992
    },
    {
      "epoch": 5.561281337047354,
      "grad_norm": 3.567563056945801,
      "learning_rate": 3.170693575895395e-05,
      "loss": 1.3929,
      "step": 3993
    },
    {
      "epoch": 5.562674094707521,
      "grad_norm": 3.478569269180298,
      "learning_rate": 3.1696986924388855e-05,
      "loss": 1.4469,
      "step": 3994
    },
    {
      "epoch": 5.564066852367688,
      "grad_norm": 3.3098061084747314,
      "learning_rate": 3.168703808982376e-05,
      "loss": 1.2011,
      "step": 3995
    },
    {
      "epoch": 5.5654596100278555,
      "grad_norm": 3.060587167739868,
      "learning_rate": 3.167708925525866e-05,
      "loss": 1.3309,
      "step": 3996
    },
    {
      "epoch": 5.566852367688022,
      "grad_norm": 2.8655707836151123,
      "learning_rate": 3.1667140420693574e-05,
      "loss": 1.2528,
      "step": 3997
    },
    {
      "epoch": 5.56824512534819,
      "grad_norm": 3.060734748840332,
      "learning_rate": 3.165719158612848e-05,
      "loss": 1.5167,
      "step": 3998
    },
    {
      "epoch": 5.569637883008356,
      "grad_norm": 3.5213658809661865,
      "learning_rate": 3.164724275156339e-05,
      "loss": 1.504,
      "step": 3999
    },
    {
      "epoch": 5.571030640668524,
      "grad_norm": 2.8732447624206543,
      "learning_rate": 3.163729391699829e-05,
      "loss": 1.4204,
      "step": 4000
    },
    {
      "epoch": 5.572423398328691,
      "grad_norm": 3.171318769454956,
      "learning_rate": 3.16273450824332e-05,
      "loss": 1.2582,
      "step": 4001
    },
    {
      "epoch": 5.573816155988858,
      "grad_norm": 3.1689140796661377,
      "learning_rate": 3.161739624786811e-05,
      "loss": 1.2342,
      "step": 4002
    },
    {
      "epoch": 5.575208913649025,
      "grad_norm": 3.2862606048583984,
      "learning_rate": 3.160744741330301e-05,
      "loss": 1.2617,
      "step": 4003
    },
    {
      "epoch": 5.576601671309192,
      "grad_norm": 3.740560293197632,
      "learning_rate": 3.1597498578737916e-05,
      "loss": 1.598,
      "step": 4004
    },
    {
      "epoch": 5.577994428969359,
      "grad_norm": 3.5847156047821045,
      "learning_rate": 3.1587549744172826e-05,
      "loss": 1.3149,
      "step": 4005
    },
    {
      "epoch": 5.579387186629527,
      "grad_norm": 3.189554452896118,
      "learning_rate": 3.157760090960773e-05,
      "loss": 1.2423,
      "step": 4006
    },
    {
      "epoch": 5.580779944289693,
      "grad_norm": 3.277650833129883,
      "learning_rate": 3.1567652075042635e-05,
      "loss": 1.3428,
      "step": 4007
    },
    {
      "epoch": 5.582172701949861,
      "grad_norm": 3.879124879837036,
      "learning_rate": 3.155770324047754e-05,
      "loss": 1.3948,
      "step": 4008
    },
    {
      "epoch": 5.5835654596100275,
      "grad_norm": 3.133305072784424,
      "learning_rate": 3.154775440591245e-05,
      "loss": 1.4726,
      "step": 4009
    },
    {
      "epoch": 5.584958217270195,
      "grad_norm": 3.1936330795288086,
      "learning_rate": 3.1537805571347354e-05,
      "loss": 1.0911,
      "step": 4010
    },
    {
      "epoch": 5.586350974930362,
      "grad_norm": 3.7738914489746094,
      "learning_rate": 3.152785673678226e-05,
      "loss": 1.5611,
      "step": 4011
    },
    {
      "epoch": 5.587743732590529,
      "grad_norm": 3.346017837524414,
      "learning_rate": 3.151790790221717e-05,
      "loss": 1.4063,
      "step": 4012
    },
    {
      "epoch": 5.589136490250697,
      "grad_norm": 3.697584629058838,
      "learning_rate": 3.150795906765207e-05,
      "loss": 1.4792,
      "step": 4013
    },
    {
      "epoch": 5.5905292479108635,
      "grad_norm": 3.020110845565796,
      "learning_rate": 3.149801023308698e-05,
      "loss": 1.2814,
      "step": 4014
    },
    {
      "epoch": 5.591922005571031,
      "grad_norm": 4.786131381988525,
      "learning_rate": 3.148806139852189e-05,
      "loss": 1.3081,
      "step": 4015
    },
    {
      "epoch": 5.593314763231198,
      "grad_norm": 3.489091396331787,
      "learning_rate": 3.147811256395679e-05,
      "loss": 1.3555,
      "step": 4016
    },
    {
      "epoch": 5.594707520891365,
      "grad_norm": 3.1483490467071533,
      "learning_rate": 3.1468163729391696e-05,
      "loss": 1.4145,
      "step": 4017
    },
    {
      "epoch": 5.596100278551532,
      "grad_norm": 3.646615505218506,
      "learning_rate": 3.14582148948266e-05,
      "loss": 1.5155,
      "step": 4018
    },
    {
      "epoch": 5.5974930362116995,
      "grad_norm": 3.1701669692993164,
      "learning_rate": 3.144826606026151e-05,
      "loss": 1.3475,
      "step": 4019
    },
    {
      "epoch": 5.598885793871866,
      "grad_norm": 3.67806339263916,
      "learning_rate": 3.1438317225696415e-05,
      "loss": 1.5856,
      "step": 4020
    },
    {
      "epoch": 5.600278551532034,
      "grad_norm": 3.745802402496338,
      "learning_rate": 3.142836839113132e-05,
      "loss": 1.7255,
      "step": 4021
    },
    {
      "epoch": 5.6016713091922,
      "grad_norm": 3.7711427211761475,
      "learning_rate": 3.141841955656623e-05,
      "loss": 1.5741,
      "step": 4022
    },
    {
      "epoch": 5.603064066852368,
      "grad_norm": 3.1372427940368652,
      "learning_rate": 3.1408470722001134e-05,
      "loss": 1.2582,
      "step": 4023
    },
    {
      "epoch": 5.604456824512535,
      "grad_norm": 2.933021306991577,
      "learning_rate": 3.139852188743604e-05,
      "loss": 1.2301,
      "step": 4024
    },
    {
      "epoch": 5.605849582172702,
      "grad_norm": 2.75142240524292,
      "learning_rate": 3.138857305287094e-05,
      "loss": 1.0087,
      "step": 4025
    },
    {
      "epoch": 5.607242339832869,
      "grad_norm": 3.52231502532959,
      "learning_rate": 3.137862421830585e-05,
      "loss": 1.513,
      "step": 4026
    },
    {
      "epoch": 5.608635097493036,
      "grad_norm": 3.888263702392578,
      "learning_rate": 3.1368675383740764e-05,
      "loss": 1.6725,
      "step": 4027
    },
    {
      "epoch": 5.610027855153203,
      "grad_norm": 3.135993480682373,
      "learning_rate": 3.135872654917567e-05,
      "loss": 1.3214,
      "step": 4028
    },
    {
      "epoch": 5.611420612813371,
      "grad_norm": 2.9921624660491943,
      "learning_rate": 3.134877771461057e-05,
      "loss": 1.1343,
      "step": 4029
    },
    {
      "epoch": 5.612813370473537,
      "grad_norm": 3.49458646774292,
      "learning_rate": 3.1338828880045476e-05,
      "loss": 1.3498,
      "step": 4030
    },
    {
      "epoch": 5.614206128133705,
      "grad_norm": 3.1544137001037598,
      "learning_rate": 3.132888004548039e-05,
      "loss": 1.1794,
      "step": 4031
    },
    {
      "epoch": 5.615598885793872,
      "grad_norm": 3.238394260406494,
      "learning_rate": 3.131893121091529e-05,
      "loss": 1.3715,
      "step": 4032
    },
    {
      "epoch": 5.616991643454039,
      "grad_norm": 3.5697274208068848,
      "learning_rate": 3.1308982376350195e-05,
      "loss": 1.457,
      "step": 4033
    },
    {
      "epoch": 5.618384401114206,
      "grad_norm": 3.0798115730285645,
      "learning_rate": 3.1299033541785106e-05,
      "loss": 1.5299,
      "step": 4034
    },
    {
      "epoch": 5.619777158774373,
      "grad_norm": 2.8842337131500244,
      "learning_rate": 3.128908470722001e-05,
      "loss": 1.234,
      "step": 4035
    },
    {
      "epoch": 5.621169916434541,
      "grad_norm": 3.178698778152466,
      "learning_rate": 3.1279135872654914e-05,
      "loss": 1.3407,
      "step": 4036
    },
    {
      "epoch": 5.6225626740947074,
      "grad_norm": 4.927010536193848,
      "learning_rate": 3.1269187038089825e-05,
      "loss": 1.2835,
      "step": 4037
    },
    {
      "epoch": 5.623955431754875,
      "grad_norm": 3.533519983291626,
      "learning_rate": 3.125923820352473e-05,
      "loss": 1.1547,
      "step": 4038
    },
    {
      "epoch": 5.625348189415042,
      "grad_norm": 3.076141595840454,
      "learning_rate": 3.124928936895963e-05,
      "loss": 1.3105,
      "step": 4039
    },
    {
      "epoch": 5.626740947075209,
      "grad_norm": 3.7596917152404785,
      "learning_rate": 3.123934053439454e-05,
      "loss": 1.3619,
      "step": 4040
    },
    {
      "epoch": 5.628133704735376,
      "grad_norm": 3.091172456741333,
      "learning_rate": 3.122939169982945e-05,
      "loss": 1.2091,
      "step": 4041
    },
    {
      "epoch": 5.629526462395543,
      "grad_norm": 3.2155330181121826,
      "learning_rate": 3.121944286526435e-05,
      "loss": 1.2206,
      "step": 4042
    },
    {
      "epoch": 5.63091922005571,
      "grad_norm": 3.5321922302246094,
      "learning_rate": 3.1209494030699256e-05,
      "loss": 1.4917,
      "step": 4043
    },
    {
      "epoch": 5.632311977715878,
      "grad_norm": 3.7770276069641113,
      "learning_rate": 3.119954519613417e-05,
      "loss": 1.5045,
      "step": 4044
    },
    {
      "epoch": 5.633704735376044,
      "grad_norm": 3.619699239730835,
      "learning_rate": 3.118959636156907e-05,
      "loss": 1.2959,
      "step": 4045
    },
    {
      "epoch": 5.635097493036212,
      "grad_norm": 3.2381839752197266,
      "learning_rate": 3.1179647527003975e-05,
      "loss": 1.3476,
      "step": 4046
    },
    {
      "epoch": 5.6364902506963785,
      "grad_norm": 3.1673707962036133,
      "learning_rate": 3.116969869243888e-05,
      "loss": 1.3016,
      "step": 4047
    },
    {
      "epoch": 5.637883008356546,
      "grad_norm": 3.2762868404388428,
      "learning_rate": 3.115974985787379e-05,
      "loss": 1.331,
      "step": 4048
    },
    {
      "epoch": 5.639275766016713,
      "grad_norm": 3.825324773788452,
      "learning_rate": 3.1149801023308694e-05,
      "loss": 1.5748,
      "step": 4049
    },
    {
      "epoch": 5.64066852367688,
      "grad_norm": 3.2996816635131836,
      "learning_rate": 3.11398521887436e-05,
      "loss": 1.4018,
      "step": 4050
    },
    {
      "epoch": 5.642061281337047,
      "grad_norm": 3.464852809906006,
      "learning_rate": 3.112990335417851e-05,
      "loss": 1.4006,
      "step": 4051
    },
    {
      "epoch": 5.6434540389972145,
      "grad_norm": 3.3543944358825684,
      "learning_rate": 3.111995451961341e-05,
      "loss": 1.3048,
      "step": 4052
    },
    {
      "epoch": 5.644846796657381,
      "grad_norm": 3.9715123176574707,
      "learning_rate": 3.1110005685048324e-05,
      "loss": 1.381,
      "step": 4053
    },
    {
      "epoch": 5.646239554317549,
      "grad_norm": 3.423572063446045,
      "learning_rate": 3.110005685048323e-05,
      "loss": 1.4331,
      "step": 4054
    },
    {
      "epoch": 5.647632311977716,
      "grad_norm": 3.0576534271240234,
      "learning_rate": 3.109010801591813e-05,
      "loss": 1.4171,
      "step": 4055
    },
    {
      "epoch": 5.649025069637883,
      "grad_norm": 3.4302568435668945,
      "learning_rate": 3.108015918135304e-05,
      "loss": 1.4861,
      "step": 4056
    },
    {
      "epoch": 5.65041782729805,
      "grad_norm": 3.7648096084594727,
      "learning_rate": 3.107021034678795e-05,
      "loss": 1.5186,
      "step": 4057
    },
    {
      "epoch": 5.651810584958217,
      "grad_norm": 3.2167465686798096,
      "learning_rate": 3.106026151222285e-05,
      "loss": 1.5403,
      "step": 4058
    },
    {
      "epoch": 5.653203342618385,
      "grad_norm": 3.5104517936706543,
      "learning_rate": 3.1050312677657755e-05,
      "loss": 1.2002,
      "step": 4059
    },
    {
      "epoch": 5.654596100278551,
      "grad_norm": 4.983837127685547,
      "learning_rate": 3.1040363843092666e-05,
      "loss": 1.343,
      "step": 4060
    },
    {
      "epoch": 5.655988857938719,
      "grad_norm": 3.564985752105713,
      "learning_rate": 3.103041500852757e-05,
      "loss": 1.4958,
      "step": 4061
    },
    {
      "epoch": 5.657381615598886,
      "grad_norm": 3.7939937114715576,
      "learning_rate": 3.1020466173962474e-05,
      "loss": 1.4437,
      "step": 4062
    },
    {
      "epoch": 5.658774373259053,
      "grad_norm": 3.5555875301361084,
      "learning_rate": 3.1010517339397385e-05,
      "loss": 1.2997,
      "step": 4063
    },
    {
      "epoch": 5.66016713091922,
      "grad_norm": 2.7710838317871094,
      "learning_rate": 3.100056850483229e-05,
      "loss": 1.008,
      "step": 4064
    },
    {
      "epoch": 5.661559888579387,
      "grad_norm": 3.532623767852783,
      "learning_rate": 3.099061967026719e-05,
      "loss": 1.6255,
      "step": 4065
    },
    {
      "epoch": 5.662952646239554,
      "grad_norm": 3.8716318607330322,
      "learning_rate": 3.0980670835702104e-05,
      "loss": 1.3818,
      "step": 4066
    },
    {
      "epoch": 5.664345403899722,
      "grad_norm": 3.149153470993042,
      "learning_rate": 3.097072200113701e-05,
      "loss": 1.3997,
      "step": 4067
    },
    {
      "epoch": 5.665738161559888,
      "grad_norm": 3.526543378829956,
      "learning_rate": 3.096077316657191e-05,
      "loss": 1.3464,
      "step": 4068
    },
    {
      "epoch": 5.667130919220056,
      "grad_norm": 3.077786922454834,
      "learning_rate": 3.0950824332006816e-05,
      "loss": 1.1928,
      "step": 4069
    },
    {
      "epoch": 5.6685236768802225,
      "grad_norm": 3.1415812969207764,
      "learning_rate": 3.094087549744173e-05,
      "loss": 1.3901,
      "step": 4070
    },
    {
      "epoch": 5.66991643454039,
      "grad_norm": 3.5665364265441895,
      "learning_rate": 3.093092666287663e-05,
      "loss": 1.1578,
      "step": 4071
    },
    {
      "epoch": 5.671309192200557,
      "grad_norm": 3.6948559284210205,
      "learning_rate": 3.0920977828311535e-05,
      "loss": 1.5388,
      "step": 4072
    },
    {
      "epoch": 5.672701949860724,
      "grad_norm": 3.619840621948242,
      "learning_rate": 3.0911028993746446e-05,
      "loss": 1.5306,
      "step": 4073
    },
    {
      "epoch": 5.674094707520892,
      "grad_norm": 3.6319541931152344,
      "learning_rate": 3.090108015918135e-05,
      "loss": 1.6165,
      "step": 4074
    },
    {
      "epoch": 5.6754874651810585,
      "grad_norm": 3.2584638595581055,
      "learning_rate": 3.0891131324616254e-05,
      "loss": 1.2343,
      "step": 4075
    },
    {
      "epoch": 5.676880222841225,
      "grad_norm": 3.5607893466949463,
      "learning_rate": 3.0881182490051165e-05,
      "loss": 1.5375,
      "step": 4076
    },
    {
      "epoch": 5.678272980501393,
      "grad_norm": 3.6344685554504395,
      "learning_rate": 3.087123365548607e-05,
      "loss": 1.468,
      "step": 4077
    },
    {
      "epoch": 5.67966573816156,
      "grad_norm": 3.4850354194641113,
      "learning_rate": 3.086128482092097e-05,
      "loss": 1.3957,
      "step": 4078
    },
    {
      "epoch": 5.681058495821727,
      "grad_norm": 3.4105865955352783,
      "learning_rate": 3.085133598635588e-05,
      "loss": 1.3103,
      "step": 4079
    },
    {
      "epoch": 5.6824512534818945,
      "grad_norm": 3.7823452949523926,
      "learning_rate": 3.084138715179079e-05,
      "loss": 1.519,
      "step": 4080
    },
    {
      "epoch": 5.683844011142061,
      "grad_norm": 4.052648544311523,
      "learning_rate": 3.083143831722569e-05,
      "loss": 1.4329,
      "step": 4081
    },
    {
      "epoch": 5.685236768802229,
      "grad_norm": 3.6204118728637695,
      "learning_rate": 3.08214894826606e-05,
      "loss": 1.4498,
      "step": 4082
    },
    {
      "epoch": 5.686629526462395,
      "grad_norm": 3.150477170944214,
      "learning_rate": 3.081154064809551e-05,
      "loss": 1.51,
      "step": 4083
    },
    {
      "epoch": 5.688022284122563,
      "grad_norm": 3.188532590866089,
      "learning_rate": 3.080159181353041e-05,
      "loss": 1.5072,
      "step": 4084
    },
    {
      "epoch": 5.68941504178273,
      "grad_norm": 3.531343936920166,
      "learning_rate": 3.079164297896532e-05,
      "loss": 1.3479,
      "step": 4085
    },
    {
      "epoch": 5.690807799442897,
      "grad_norm": 3.5683412551879883,
      "learning_rate": 3.0781694144400226e-05,
      "loss": 1.8486,
      "step": 4086
    },
    {
      "epoch": 5.692200557103064,
      "grad_norm": 2.9277305603027344,
      "learning_rate": 3.077174530983513e-05,
      "loss": 1.0993,
      "step": 4087
    },
    {
      "epoch": 5.693593314763231,
      "grad_norm": 3.2109014987945557,
      "learning_rate": 3.076179647527004e-05,
      "loss": 1.3071,
      "step": 4088
    },
    {
      "epoch": 5.694986072423398,
      "grad_norm": 3.4162259101867676,
      "learning_rate": 3.0751847640704945e-05,
      "loss": 1.4459,
      "step": 4089
    },
    {
      "epoch": 5.696378830083566,
      "grad_norm": 3.262427806854248,
      "learning_rate": 3.074189880613985e-05,
      "loss": 1.271,
      "step": 4090
    },
    {
      "epoch": 5.697771587743732,
      "grad_norm": 3.9636881351470947,
      "learning_rate": 3.073194997157475e-05,
      "loss": 1.1394,
      "step": 4091
    },
    {
      "epoch": 5.6991643454039,
      "grad_norm": 4.1263885498046875,
      "learning_rate": 3.0722001137009664e-05,
      "loss": 1.2762,
      "step": 4092
    },
    {
      "epoch": 5.7005571030640665,
      "grad_norm": 3.5942258834838867,
      "learning_rate": 3.071205230244457e-05,
      "loss": 1.5294,
      "step": 4093
    },
    {
      "epoch": 5.701949860724234,
      "grad_norm": 3.5829882621765137,
      "learning_rate": 3.070210346787947e-05,
      "loss": 1.6246,
      "step": 4094
    },
    {
      "epoch": 5.703342618384401,
      "grad_norm": 3.2303061485290527,
      "learning_rate": 3.069215463331438e-05,
      "loss": 1.2655,
      "step": 4095
    },
    {
      "epoch": 5.704735376044568,
      "grad_norm": 2.990727186203003,
      "learning_rate": 3.068220579874929e-05,
      "loss": 1.2716,
      "step": 4096
    },
    {
      "epoch": 5.706128133704736,
      "grad_norm": 3.8842310905456543,
      "learning_rate": 3.067225696418419e-05,
      "loss": 1.6645,
      "step": 4097
    },
    {
      "epoch": 5.7075208913649025,
      "grad_norm": 3.260706901550293,
      "learning_rate": 3.06623081296191e-05,
      "loss": 1.3646,
      "step": 4098
    },
    {
      "epoch": 5.708913649025069,
      "grad_norm": 3.34687876701355,
      "learning_rate": 3.0652359295054006e-05,
      "loss": 1.4088,
      "step": 4099
    },
    {
      "epoch": 5.710306406685237,
      "grad_norm": 3.53373384475708,
      "learning_rate": 3.064241046048891e-05,
      "loss": 1.4936,
      "step": 4100
    },
    {
      "epoch": 5.711699164345404,
      "grad_norm": 3.292091131210327,
      "learning_rate": 3.0632461625923814e-05,
      "loss": 1.2043,
      "step": 4101
    },
    {
      "epoch": 5.713091922005571,
      "grad_norm": 3.288079023361206,
      "learning_rate": 3.0622512791358725e-05,
      "loss": 1.5383,
      "step": 4102
    },
    {
      "epoch": 5.7144846796657385,
      "grad_norm": 3.857797861099243,
      "learning_rate": 3.061256395679363e-05,
      "loss": 1.3433,
      "step": 4103
    },
    {
      "epoch": 5.715877437325905,
      "grad_norm": 3.4742066860198975,
      "learning_rate": 3.060261512222853e-05,
      "loss": 1.4904,
      "step": 4104
    },
    {
      "epoch": 5.717270194986073,
      "grad_norm": 2.6940255165100098,
      "learning_rate": 3.0592666287663444e-05,
      "loss": 1.1487,
      "step": 4105
    },
    {
      "epoch": 5.718662952646239,
      "grad_norm": 3.0581483840942383,
      "learning_rate": 3.058271745309835e-05,
      "loss": 1.2375,
      "step": 4106
    },
    {
      "epoch": 5.720055710306407,
      "grad_norm": 3.8022048473358154,
      "learning_rate": 3.057276861853325e-05,
      "loss": 1.5468,
      "step": 4107
    },
    {
      "epoch": 5.721448467966574,
      "grad_norm": 3.0157785415649414,
      "learning_rate": 3.0562819783968156e-05,
      "loss": 1.2329,
      "step": 4108
    },
    {
      "epoch": 5.722841225626741,
      "grad_norm": 2.8184762001037598,
      "learning_rate": 3.055287094940307e-05,
      "loss": 1.1472,
      "step": 4109
    },
    {
      "epoch": 5.724233983286908,
      "grad_norm": 3.1936144828796387,
      "learning_rate": 3.054292211483798e-05,
      "loss": 1.2509,
      "step": 4110
    },
    {
      "epoch": 5.725626740947075,
      "grad_norm": 4.00202751159668,
      "learning_rate": 3.053297328027288e-05,
      "loss": 1.6249,
      "step": 4111
    },
    {
      "epoch": 5.727019498607242,
      "grad_norm": 3.5175881385803223,
      "learning_rate": 3.0523024445707786e-05,
      "loss": 1.2482,
      "step": 4112
    },
    {
      "epoch": 5.72841225626741,
      "grad_norm": 3.3421785831451416,
      "learning_rate": 3.051307561114269e-05,
      "loss": 1.1038,
      "step": 4113
    },
    {
      "epoch": 5.729805013927576,
      "grad_norm": 3.4558494091033936,
      "learning_rate": 3.0503126776577598e-05,
      "loss": 1.3373,
      "step": 4114
    },
    {
      "epoch": 5.731197771587744,
      "grad_norm": 3.02663254737854,
      "learning_rate": 3.0493177942012505e-05,
      "loss": 1.5091,
      "step": 4115
    },
    {
      "epoch": 5.732590529247911,
      "grad_norm": 2.962890386581421,
      "learning_rate": 3.048322910744741e-05,
      "loss": 1.1872,
      "step": 4116
    },
    {
      "epoch": 5.733983286908078,
      "grad_norm": 3.380704879760742,
      "learning_rate": 3.0473280272882317e-05,
      "loss": 1.2305,
      "step": 4117
    },
    {
      "epoch": 5.735376044568245,
      "grad_norm": 3.2142016887664795,
      "learning_rate": 3.046333143831722e-05,
      "loss": 1.3784,
      "step": 4118
    },
    {
      "epoch": 5.736768802228412,
      "grad_norm": 2.860426187515259,
      "learning_rate": 3.0453382603752128e-05,
      "loss": 1.179,
      "step": 4119
    },
    {
      "epoch": 5.73816155988858,
      "grad_norm": 3.4320616722106934,
      "learning_rate": 3.0443433769187036e-05,
      "loss": 1.4381,
      "step": 4120
    },
    {
      "epoch": 5.7395543175487465,
      "grad_norm": 3.2273716926574707,
      "learning_rate": 3.043348493462194e-05,
      "loss": 1.2319,
      "step": 4121
    },
    {
      "epoch": 5.740947075208914,
      "grad_norm": 3.51316499710083,
      "learning_rate": 3.042353610005685e-05,
      "loss": 1.6669,
      "step": 4122
    },
    {
      "epoch": 5.742339832869081,
      "grad_norm": 2.818206787109375,
      "learning_rate": 3.041358726549175e-05,
      "loss": 1.1421,
      "step": 4123
    },
    {
      "epoch": 5.743732590529248,
      "grad_norm": 3.244997262954712,
      "learning_rate": 3.0403638430926662e-05,
      "loss": 1.2826,
      "step": 4124
    },
    {
      "epoch": 5.745125348189415,
      "grad_norm": 3.031724452972412,
      "learning_rate": 3.0393689596361566e-05,
      "loss": 1.2507,
      "step": 4125
    },
    {
      "epoch": 5.7465181058495824,
      "grad_norm": 3.2619550228118896,
      "learning_rate": 3.0383740761796474e-05,
      "loss": 1.2841,
      "step": 4126
    },
    {
      "epoch": 5.747910863509749,
      "grad_norm": 3.3931632041931152,
      "learning_rate": 3.037379192723138e-05,
      "loss": 1.3306,
      "step": 4127
    },
    {
      "epoch": 5.749303621169917,
      "grad_norm": 3.3280770778656006,
      "learning_rate": 3.0363843092666285e-05,
      "loss": 1.4295,
      "step": 4128
    },
    {
      "epoch": 5.750696378830083,
      "grad_norm": 3.037776470184326,
      "learning_rate": 3.0353894258101193e-05,
      "loss": 1.2161,
      "step": 4129
    },
    {
      "epoch": 5.752089136490251,
      "grad_norm": 3.152137517929077,
      "learning_rate": 3.0343945423536097e-05,
      "loss": 1.4588,
      "step": 4130
    },
    {
      "epoch": 5.7534818941504176,
      "grad_norm": 3.9945080280303955,
      "learning_rate": 3.0333996588971004e-05,
      "loss": 1.4879,
      "step": 4131
    },
    {
      "epoch": 5.754874651810585,
      "grad_norm": 3.6053826808929443,
      "learning_rate": 3.032404775440591e-05,
      "loss": 1.4856,
      "step": 4132
    },
    {
      "epoch": 5.756267409470752,
      "grad_norm": 3.221766710281372,
      "learning_rate": 3.0314098919840816e-05,
      "loss": 1.3328,
      "step": 4133
    },
    {
      "epoch": 5.757660167130919,
      "grad_norm": 3.8348989486694336,
      "learning_rate": 3.0304150085275723e-05,
      "loss": 1.493,
      "step": 4134
    },
    {
      "epoch": 5.759052924791086,
      "grad_norm": 3.57279372215271,
      "learning_rate": 3.0294201250710627e-05,
      "loss": 1.4078,
      "step": 4135
    },
    {
      "epoch": 5.7604456824512535,
      "grad_norm": 3.2637948989868164,
      "learning_rate": 3.0284252416145535e-05,
      "loss": 1.2166,
      "step": 4136
    },
    {
      "epoch": 5.76183844011142,
      "grad_norm": 3.51096773147583,
      "learning_rate": 3.0274303581580442e-05,
      "loss": 1.5175,
      "step": 4137
    },
    {
      "epoch": 5.763231197771588,
      "grad_norm": 3.0070669651031494,
      "learning_rate": 3.0264354747015346e-05,
      "loss": 1.0949,
      "step": 4138
    },
    {
      "epoch": 5.764623955431755,
      "grad_norm": 3.6944706439971924,
      "learning_rate": 3.0254405912450254e-05,
      "loss": 1.457,
      "step": 4139
    },
    {
      "epoch": 5.766016713091922,
      "grad_norm": 3.577955484390259,
      "learning_rate": 3.0244457077885158e-05,
      "loss": 1.4939,
      "step": 4140
    },
    {
      "epoch": 5.767409470752089,
      "grad_norm": 3.0088469982147217,
      "learning_rate": 3.0234508243320065e-05,
      "loss": 1.3536,
      "step": 4141
    },
    {
      "epoch": 5.768802228412256,
      "grad_norm": 3.2172975540161133,
      "learning_rate": 3.0224559408754973e-05,
      "loss": 1.3453,
      "step": 4142
    },
    {
      "epoch": 5.770194986072424,
      "grad_norm": 2.8926684856414795,
      "learning_rate": 3.0214610574189877e-05,
      "loss": 1.2709,
      "step": 4143
    },
    {
      "epoch": 5.77158774373259,
      "grad_norm": 2.8702945709228516,
      "learning_rate": 3.0204661739624784e-05,
      "loss": 1.0931,
      "step": 4144
    },
    {
      "epoch": 5.772980501392758,
      "grad_norm": 3.240285634994507,
      "learning_rate": 3.019471290505969e-05,
      "loss": 0.9439,
      "step": 4145
    },
    {
      "epoch": 5.774373259052925,
      "grad_norm": 3.271115779876709,
      "learning_rate": 3.0184764070494596e-05,
      "loss": 1.4925,
      "step": 4146
    },
    {
      "epoch": 5.775766016713092,
      "grad_norm": 3.020358085632324,
      "learning_rate": 3.01748152359295e-05,
      "loss": 1.2876,
      "step": 4147
    },
    {
      "epoch": 5.777158774373259,
      "grad_norm": 3.0515573024749756,
      "learning_rate": 3.0164866401364407e-05,
      "loss": 1.0941,
      "step": 4148
    },
    {
      "epoch": 5.778551532033426,
      "grad_norm": 3.369999408721924,
      "learning_rate": 3.0154917566799315e-05,
      "loss": 1.2354,
      "step": 4149
    },
    {
      "epoch": 5.779944289693593,
      "grad_norm": 4.339197158813477,
      "learning_rate": 3.014496873223422e-05,
      "loss": 1.9558,
      "step": 4150
    },
    {
      "epoch": 5.781337047353761,
      "grad_norm": 3.7317991256713867,
      "learning_rate": 3.013501989766913e-05,
      "loss": 1.2342,
      "step": 4151
    },
    {
      "epoch": 5.782729805013927,
      "grad_norm": 3.094419002532959,
      "learning_rate": 3.012507106310403e-05,
      "loss": 1.1601,
      "step": 4152
    },
    {
      "epoch": 5.784122562674095,
      "grad_norm": 2.936570644378662,
      "learning_rate": 3.011512222853894e-05,
      "loss": 1.1971,
      "step": 4153
    },
    {
      "epoch": 5.7855153203342615,
      "grad_norm": 3.0709304809570312,
      "learning_rate": 3.010517339397385e-05,
      "loss": 1.1642,
      "step": 4154
    },
    {
      "epoch": 5.786908077994429,
      "grad_norm": 2.7466046810150146,
      "learning_rate": 3.0095224559408753e-05,
      "loss": 0.943,
      "step": 4155
    },
    {
      "epoch": 5.788300835654596,
      "grad_norm": 3.3312766551971436,
      "learning_rate": 3.008527572484366e-05,
      "loss": 1.4297,
      "step": 4156
    },
    {
      "epoch": 5.789693593314763,
      "grad_norm": 3.617126703262329,
      "learning_rate": 3.0075326890278564e-05,
      "loss": 1.3232,
      "step": 4157
    },
    {
      "epoch": 5.791086350974931,
      "grad_norm": 2.7909839153289795,
      "learning_rate": 3.0065378055713472e-05,
      "loss": 0.8445,
      "step": 4158
    },
    {
      "epoch": 5.7924791086350975,
      "grad_norm": 3.9032790660858154,
      "learning_rate": 3.005542922114838e-05,
      "loss": 1.6201,
      "step": 4159
    },
    {
      "epoch": 5.793871866295264,
      "grad_norm": 3.076613426208496,
      "learning_rate": 3.0045480386583283e-05,
      "loss": 1.3427,
      "step": 4160
    },
    {
      "epoch": 5.795264623955432,
      "grad_norm": 4.433011531829834,
      "learning_rate": 3.003553155201819e-05,
      "loss": 1.3001,
      "step": 4161
    },
    {
      "epoch": 5.796657381615599,
      "grad_norm": 3.7605056762695312,
      "learning_rate": 3.0025582717453095e-05,
      "loss": 1.3189,
      "step": 4162
    },
    {
      "epoch": 5.798050139275766,
      "grad_norm": 3.1744890213012695,
      "learning_rate": 3.0015633882888002e-05,
      "loss": 1.4072,
      "step": 4163
    },
    {
      "epoch": 5.7994428969359335,
      "grad_norm": 3.4143013954162598,
      "learning_rate": 3.000568504832291e-05,
      "loss": 1.3463,
      "step": 4164
    },
    {
      "epoch": 5.8008356545961,
      "grad_norm": 3.0669949054718018,
      "learning_rate": 2.9995736213757814e-05,
      "loss": 1.227,
      "step": 4165
    },
    {
      "epoch": 5.802228412256268,
      "grad_norm": 3.3583242893218994,
      "learning_rate": 2.998578737919272e-05,
      "loss": 1.4553,
      "step": 4166
    },
    {
      "epoch": 5.803621169916434,
      "grad_norm": 3.753074884414673,
      "learning_rate": 2.9975838544627625e-05,
      "loss": 1.5123,
      "step": 4167
    },
    {
      "epoch": 5.805013927576602,
      "grad_norm": 3.2437632083892822,
      "learning_rate": 2.9965889710062533e-05,
      "loss": 1.0924,
      "step": 4168
    },
    {
      "epoch": 5.806406685236769,
      "grad_norm": 3.4535083770751953,
      "learning_rate": 2.9955940875497437e-05,
      "loss": 1.3092,
      "step": 4169
    },
    {
      "epoch": 5.807799442896936,
      "grad_norm": 3.2413816452026367,
      "learning_rate": 2.9945992040932344e-05,
      "loss": 1.4041,
      "step": 4170
    },
    {
      "epoch": 5.809192200557103,
      "grad_norm": 3.1956446170806885,
      "learning_rate": 2.9936043206367252e-05,
      "loss": 1.3137,
      "step": 4171
    },
    {
      "epoch": 5.81058495821727,
      "grad_norm": 3.5328941345214844,
      "learning_rate": 2.9926094371802156e-05,
      "loss": 1.5614,
      "step": 4172
    },
    {
      "epoch": 5.811977715877437,
      "grad_norm": 3.4689059257507324,
      "learning_rate": 2.9916145537237063e-05,
      "loss": 1.4592,
      "step": 4173
    },
    {
      "epoch": 5.813370473537605,
      "grad_norm": 3.411120891571045,
      "learning_rate": 2.9906196702671968e-05,
      "loss": 1.3544,
      "step": 4174
    },
    {
      "epoch": 5.814763231197771,
      "grad_norm": 3.5169613361358643,
      "learning_rate": 2.9896247868106875e-05,
      "loss": 1.541,
      "step": 4175
    },
    {
      "epoch": 5.816155988857939,
      "grad_norm": 2.96293568611145,
      "learning_rate": 2.9886299033541782e-05,
      "loss": 1.0733,
      "step": 4176
    },
    {
      "epoch": 5.8175487465181055,
      "grad_norm": 3.3015525341033936,
      "learning_rate": 2.9876350198976687e-05,
      "loss": 1.5115,
      "step": 4177
    },
    {
      "epoch": 5.818941504178273,
      "grad_norm": 2.850428581237793,
      "learning_rate": 2.9866401364411597e-05,
      "loss": 1.2325,
      "step": 4178
    },
    {
      "epoch": 5.82033426183844,
      "grad_norm": 3.6030232906341553,
      "learning_rate": 2.9856452529846498e-05,
      "loss": 1.4975,
      "step": 4179
    },
    {
      "epoch": 5.821727019498607,
      "grad_norm": 3.1357882022857666,
      "learning_rate": 2.984650369528141e-05,
      "loss": 1.4008,
      "step": 4180
    },
    {
      "epoch": 5.823119777158775,
      "grad_norm": 3.1677186489105225,
      "learning_rate": 2.9836554860716316e-05,
      "loss": 1.2208,
      "step": 4181
    },
    {
      "epoch": 5.8245125348189415,
      "grad_norm": 3.15153431892395,
      "learning_rate": 2.982660602615122e-05,
      "loss": 1.3147,
      "step": 4182
    },
    {
      "epoch": 5.825905292479108,
      "grad_norm": 3.3322243690490723,
      "learning_rate": 2.9816657191586128e-05,
      "loss": 1.3051,
      "step": 4183
    },
    {
      "epoch": 5.827298050139276,
      "grad_norm": 3.433753252029419,
      "learning_rate": 2.9806708357021032e-05,
      "loss": 1.4682,
      "step": 4184
    },
    {
      "epoch": 5.828690807799443,
      "grad_norm": 3.517406940460205,
      "learning_rate": 2.979675952245594e-05,
      "loss": 1.383,
      "step": 4185
    },
    {
      "epoch": 5.83008356545961,
      "grad_norm": 10.21381664276123,
      "learning_rate": 2.9786810687890847e-05,
      "loss": 1.2204,
      "step": 4186
    },
    {
      "epoch": 5.8314763231197775,
      "grad_norm": 3.349642515182495,
      "learning_rate": 2.977686185332575e-05,
      "loss": 1.3719,
      "step": 4187
    },
    {
      "epoch": 5.832869080779944,
      "grad_norm": 3.384669303894043,
      "learning_rate": 2.976691301876066e-05,
      "loss": 1.42,
      "step": 4188
    },
    {
      "epoch": 5.834261838440112,
      "grad_norm": 3.919182300567627,
      "learning_rate": 2.9756964184195563e-05,
      "loss": 1.5257,
      "step": 4189
    },
    {
      "epoch": 5.835654596100278,
      "grad_norm": 3.144423484802246,
      "learning_rate": 2.974701534963047e-05,
      "loss": 1.4346,
      "step": 4190
    },
    {
      "epoch": 5.837047353760446,
      "grad_norm": 3.360140800476074,
      "learning_rate": 2.9737066515065374e-05,
      "loss": 1.628,
      "step": 4191
    },
    {
      "epoch": 5.838440111420613,
      "grad_norm": 3.2371537685394287,
      "learning_rate": 2.972711768050028e-05,
      "loss": 1.3053,
      "step": 4192
    },
    {
      "epoch": 5.83983286908078,
      "grad_norm": 3.167996883392334,
      "learning_rate": 2.971716884593519e-05,
      "loss": 1.3235,
      "step": 4193
    },
    {
      "epoch": 5.841225626740947,
      "grad_norm": 3.457688570022583,
      "learning_rate": 2.9707220011370093e-05,
      "loss": 1.3575,
      "step": 4194
    },
    {
      "epoch": 5.842618384401114,
      "grad_norm": 3.499300956726074,
      "learning_rate": 2.9697271176805e-05,
      "loss": 1.407,
      "step": 4195
    },
    {
      "epoch": 5.844011142061281,
      "grad_norm": 3.7530620098114014,
      "learning_rate": 2.9687322342239905e-05,
      "loss": 1.3062,
      "step": 4196
    },
    {
      "epoch": 5.845403899721449,
      "grad_norm": 3.6542885303497314,
      "learning_rate": 2.9677373507674812e-05,
      "loss": 1.4013,
      "step": 4197
    },
    {
      "epoch": 5.846796657381615,
      "grad_norm": 3.788815975189209,
      "learning_rate": 2.966742467310972e-05,
      "loss": 1.7402,
      "step": 4198
    },
    {
      "epoch": 5.848189415041783,
      "grad_norm": 3.198843479156494,
      "learning_rate": 2.9657475838544624e-05,
      "loss": 1.1983,
      "step": 4199
    },
    {
      "epoch": 5.84958217270195,
      "grad_norm": 3.5123395919799805,
      "learning_rate": 2.964752700397953e-05,
      "loss": 1.3479,
      "step": 4200
    },
    {
      "epoch": 5.850974930362117,
      "grad_norm": 3.186678409576416,
      "learning_rate": 2.9637578169414435e-05,
      "loss": 1.3358,
      "step": 4201
    },
    {
      "epoch": 5.852367688022284,
      "grad_norm": 3.2986679077148438,
      "learning_rate": 2.9627629334849343e-05,
      "loss": 1.4147,
      "step": 4202
    },
    {
      "epoch": 5.853760445682451,
      "grad_norm": 3.591884136199951,
      "learning_rate": 2.961768050028425e-05,
      "loss": 1.252,
      "step": 4203
    },
    {
      "epoch": 5.855153203342619,
      "grad_norm": 4.0336503982543945,
      "learning_rate": 2.9607731665719154e-05,
      "loss": 1.545,
      "step": 4204
    },
    {
      "epoch": 5.8565459610027855,
      "grad_norm": 3.571032762527466,
      "learning_rate": 2.959778283115406e-05,
      "loss": 1.4103,
      "step": 4205
    },
    {
      "epoch": 5.857938718662953,
      "grad_norm": 3.2887074947357178,
      "learning_rate": 2.9587833996588966e-05,
      "loss": 1.3755,
      "step": 4206
    },
    {
      "epoch": 5.85933147632312,
      "grad_norm": 2.8116984367370605,
      "learning_rate": 2.9577885162023877e-05,
      "loss": 1.0999,
      "step": 4207
    },
    {
      "epoch": 5.860724233983287,
      "grad_norm": 3.0161566734313965,
      "learning_rate": 2.9567936327458784e-05,
      "loss": 1.2861,
      "step": 4208
    },
    {
      "epoch": 5.862116991643454,
      "grad_norm": 3.3580868244171143,
      "learning_rate": 2.9557987492893688e-05,
      "loss": 1.3462,
      "step": 4209
    },
    {
      "epoch": 5.8635097493036215,
      "grad_norm": 3.395779609680176,
      "learning_rate": 2.9548038658328596e-05,
      "loss": 1.3673,
      "step": 4210
    },
    {
      "epoch": 5.864902506963788,
      "grad_norm": 3.057870388031006,
      "learning_rate": 2.95380898237635e-05,
      "loss": 1.0038,
      "step": 4211
    },
    {
      "epoch": 5.866295264623956,
      "grad_norm": 2.9318737983703613,
      "learning_rate": 2.9528140989198407e-05,
      "loss": 1.415,
      "step": 4212
    },
    {
      "epoch": 5.867688022284122,
      "grad_norm": 3.789201021194458,
      "learning_rate": 2.951819215463331e-05,
      "loss": 1.2342,
      "step": 4213
    },
    {
      "epoch": 5.86908077994429,
      "grad_norm": 3.4682655334472656,
      "learning_rate": 2.950824332006822e-05,
      "loss": 1.3848,
      "step": 4214
    },
    {
      "epoch": 5.870473537604457,
      "grad_norm": 3.241466522216797,
      "learning_rate": 2.9498294485503126e-05,
      "loss": 1.6755,
      "step": 4215
    },
    {
      "epoch": 5.871866295264624,
      "grad_norm": 3.4985568523406982,
      "learning_rate": 2.948834565093803e-05,
      "loss": 1.2251,
      "step": 4216
    },
    {
      "epoch": 5.873259052924791,
      "grad_norm": 3.2599987983703613,
      "learning_rate": 2.9478396816372938e-05,
      "loss": 1.308,
      "step": 4217
    },
    {
      "epoch": 5.874651810584958,
      "grad_norm": 3.268479108810425,
      "learning_rate": 2.946844798180784e-05,
      "loss": 1.2174,
      "step": 4218
    },
    {
      "epoch": 5.876044568245125,
      "grad_norm": 3.1882810592651367,
      "learning_rate": 2.945849914724275e-05,
      "loss": 1.2337,
      "step": 4219
    },
    {
      "epoch": 5.8774373259052926,
      "grad_norm": 3.6085448265075684,
      "learning_rate": 2.9448550312677657e-05,
      "loss": 1.7037,
      "step": 4220
    },
    {
      "epoch": 5.878830083565459,
      "grad_norm": 3.001582384109497,
      "learning_rate": 2.943860147811256e-05,
      "loss": 1.2232,
      "step": 4221
    },
    {
      "epoch": 5.880222841225627,
      "grad_norm": 3.2147443294525146,
      "learning_rate": 2.9428652643547468e-05,
      "loss": 1.2098,
      "step": 4222
    },
    {
      "epoch": 5.881615598885794,
      "grad_norm": 3.6638669967651367,
      "learning_rate": 2.9418703808982372e-05,
      "loss": 1.5788,
      "step": 4223
    },
    {
      "epoch": 5.883008356545961,
      "grad_norm": 2.899076461791992,
      "learning_rate": 2.940875497441728e-05,
      "loss": 0.9433,
      "step": 4224
    },
    {
      "epoch": 5.884401114206128,
      "grad_norm": 3.244516372680664,
      "learning_rate": 2.9398806139852187e-05,
      "loss": 1.4991,
      "step": 4225
    },
    {
      "epoch": 5.885793871866295,
      "grad_norm": 3.1038339138031006,
      "learning_rate": 2.938885730528709e-05,
      "loss": 1.0099,
      "step": 4226
    },
    {
      "epoch": 5.887186629526463,
      "grad_norm": 2.9299960136413574,
      "learning_rate": 2.9378908470722e-05,
      "loss": 1.1283,
      "step": 4227
    },
    {
      "epoch": 5.888579387186629,
      "grad_norm": 3.4644925594329834,
      "learning_rate": 2.9368959636156903e-05,
      "loss": 1.4124,
      "step": 4228
    },
    {
      "epoch": 5.889972144846797,
      "grad_norm": 3.1208677291870117,
      "learning_rate": 2.935901080159181e-05,
      "loss": 1.5089,
      "step": 4229
    },
    {
      "epoch": 5.891364902506964,
      "grad_norm": 3.68741774559021,
      "learning_rate": 2.9349061967026718e-05,
      "loss": 1.3013,
      "step": 4230
    },
    {
      "epoch": 5.892757660167131,
      "grad_norm": 3.068450450897217,
      "learning_rate": 2.9339113132461622e-05,
      "loss": 1.0985,
      "step": 4231
    },
    {
      "epoch": 5.894150417827298,
      "grad_norm": 3.0982754230499268,
      "learning_rate": 2.932916429789653e-05,
      "loss": 1.1662,
      "step": 4232
    },
    {
      "epoch": 5.895543175487465,
      "grad_norm": 3.9106130599975586,
      "learning_rate": 2.9319215463331433e-05,
      "loss": 1.6737,
      "step": 4233
    },
    {
      "epoch": 5.896935933147632,
      "grad_norm": 3.0923335552215576,
      "learning_rate": 2.9309266628766344e-05,
      "loss": 1.0318,
      "step": 4234
    },
    {
      "epoch": 5.8983286908078,
      "grad_norm": 3.1148061752319336,
      "learning_rate": 2.9299317794201245e-05,
      "loss": 1.074,
      "step": 4235
    },
    {
      "epoch": 5.899721448467966,
      "grad_norm": 3.4015417098999023,
      "learning_rate": 2.9289368959636156e-05,
      "loss": 1.54,
      "step": 4236
    },
    {
      "epoch": 5.901114206128134,
      "grad_norm": 3.440244436264038,
      "learning_rate": 2.9279420125071063e-05,
      "loss": 1.2495,
      "step": 4237
    },
    {
      "epoch": 5.9025069637883005,
      "grad_norm": 3.1867411136627197,
      "learning_rate": 2.9269471290505967e-05,
      "loss": 1.2098,
      "step": 4238
    },
    {
      "epoch": 5.903899721448468,
      "grad_norm": 3.6312007904052734,
      "learning_rate": 2.9259522455940875e-05,
      "loss": 1.6599,
      "step": 4239
    },
    {
      "epoch": 5.905292479108635,
      "grad_norm": 3.264538288116455,
      "learning_rate": 2.924957362137578e-05,
      "loss": 1.1512,
      "step": 4240
    },
    {
      "epoch": 5.906685236768802,
      "grad_norm": 3.1987524032592773,
      "learning_rate": 2.9239624786810686e-05,
      "loss": 1.3559,
      "step": 4241
    },
    {
      "epoch": 5.908077994428969,
      "grad_norm": 3.2913811206817627,
      "learning_rate": 2.9229675952245594e-05,
      "loss": 1.3152,
      "step": 4242
    },
    {
      "epoch": 5.9094707520891365,
      "grad_norm": 3.5033583641052246,
      "learning_rate": 2.9219727117680498e-05,
      "loss": 1.3457,
      "step": 4243
    },
    {
      "epoch": 5.910863509749303,
      "grad_norm": 3.714859962463379,
      "learning_rate": 2.9209778283115405e-05,
      "loss": 1.5776,
      "step": 4244
    },
    {
      "epoch": 5.912256267409471,
      "grad_norm": 3.511657476425171,
      "learning_rate": 2.919982944855031e-05,
      "loss": 1.523,
      "step": 4245
    },
    {
      "epoch": 5.913649025069638,
      "grad_norm": 3.459751605987549,
      "learning_rate": 2.9189880613985217e-05,
      "loss": 1.2272,
      "step": 4246
    },
    {
      "epoch": 5.915041782729805,
      "grad_norm": 3.478177070617676,
      "learning_rate": 2.9179931779420124e-05,
      "loss": 1.5362,
      "step": 4247
    },
    {
      "epoch": 5.9164345403899725,
      "grad_norm": 3.2239317893981934,
      "learning_rate": 2.9169982944855028e-05,
      "loss": 1.0754,
      "step": 4248
    },
    {
      "epoch": 5.917827298050139,
      "grad_norm": 3.0095248222351074,
      "learning_rate": 2.9160034110289936e-05,
      "loss": 1.3243,
      "step": 4249
    },
    {
      "epoch": 5.919220055710307,
      "grad_norm": 3.2675821781158447,
      "learning_rate": 2.915008527572484e-05,
      "loss": 1.048,
      "step": 4250
    },
    {
      "epoch": 5.920612813370473,
      "grad_norm": 3.6328365802764893,
      "learning_rate": 2.9140136441159747e-05,
      "loss": 1.4779,
      "step": 4251
    },
    {
      "epoch": 5.922005571030641,
      "grad_norm": 3.0589301586151123,
      "learning_rate": 2.913018760659465e-05,
      "loss": 1.3803,
      "step": 4252
    },
    {
      "epoch": 5.923398328690808,
      "grad_norm": 3.189467430114746,
      "learning_rate": 2.912023877202956e-05,
      "loss": 1.2764,
      "step": 4253
    },
    {
      "epoch": 5.924791086350975,
      "grad_norm": 3.7781779766082764,
      "learning_rate": 2.9110289937464466e-05,
      "loss": 1.3555,
      "step": 4254
    },
    {
      "epoch": 5.926183844011142,
      "grad_norm": 3.6133739948272705,
      "learning_rate": 2.910034110289937e-05,
      "loss": 1.3811,
      "step": 4255
    },
    {
      "epoch": 5.927576601671309,
      "grad_norm": 3.5860114097595215,
      "learning_rate": 2.9090392268334278e-05,
      "loss": 1.5179,
      "step": 4256
    },
    {
      "epoch": 5.928969359331476,
      "grad_norm": 4.614382743835449,
      "learning_rate": 2.9080443433769182e-05,
      "loss": 1.2738,
      "step": 4257
    },
    {
      "epoch": 5.930362116991644,
      "grad_norm": 8.058302879333496,
      "learning_rate": 2.907049459920409e-05,
      "loss": 1.409,
      "step": 4258
    },
    {
      "epoch": 5.93175487465181,
      "grad_norm": 4.036209583282471,
      "learning_rate": 2.9060545764638997e-05,
      "loss": 1.3674,
      "step": 4259
    },
    {
      "epoch": 5.933147632311978,
      "grad_norm": 3.4181134700775146,
      "learning_rate": 2.90505969300739e-05,
      "loss": 1.1795,
      "step": 4260
    },
    {
      "epoch": 5.9345403899721445,
      "grad_norm": 3.0663325786590576,
      "learning_rate": 2.904064809550881e-05,
      "loss": 1.066,
      "step": 4261
    },
    {
      "epoch": 5.935933147632312,
      "grad_norm": 3.1197872161865234,
      "learning_rate": 2.9030699260943712e-05,
      "loss": 1.2864,
      "step": 4262
    },
    {
      "epoch": 5.937325905292479,
      "grad_norm": 3.1056711673736572,
      "learning_rate": 2.9020750426378623e-05,
      "loss": 1.031,
      "step": 4263
    },
    {
      "epoch": 5.938718662952646,
      "grad_norm": 4.707518100738525,
      "learning_rate": 2.901080159181353e-05,
      "loss": 1.2698,
      "step": 4264
    },
    {
      "epoch": 5.940111420612814,
      "grad_norm": 3.6590993404388428,
      "learning_rate": 2.9000852757248435e-05,
      "loss": 1.4503,
      "step": 4265
    },
    {
      "epoch": 5.9415041782729805,
      "grad_norm": 3.107004165649414,
      "learning_rate": 2.8990903922683342e-05,
      "loss": 1.1497,
      "step": 4266
    },
    {
      "epoch": 5.942896935933147,
      "grad_norm": 3.4972264766693115,
      "learning_rate": 2.8980955088118246e-05,
      "loss": 1.4542,
      "step": 4267
    },
    {
      "epoch": 5.944289693593315,
      "grad_norm": 3.4089934825897217,
      "learning_rate": 2.8971006253553154e-05,
      "loss": 1.3729,
      "step": 4268
    },
    {
      "epoch": 5.945682451253482,
      "grad_norm": 3.4364445209503174,
      "learning_rate": 2.896105741898806e-05,
      "loss": 1.6305,
      "step": 4269
    },
    {
      "epoch": 5.947075208913649,
      "grad_norm": 2.60617995262146,
      "learning_rate": 2.8951108584422965e-05,
      "loss": 1.1789,
      "step": 4270
    },
    {
      "epoch": 5.9484679665738165,
      "grad_norm": 3.87069034576416,
      "learning_rate": 2.8941159749857873e-05,
      "loss": 1.3996,
      "step": 4271
    },
    {
      "epoch": 5.949860724233983,
      "grad_norm": 2.914503335952759,
      "learning_rate": 2.8931210915292777e-05,
      "loss": 1.1826,
      "step": 4272
    },
    {
      "epoch": 5.951253481894151,
      "grad_norm": 3.2550365924835205,
      "learning_rate": 2.8921262080727684e-05,
      "loss": 1.5075,
      "step": 4273
    },
    {
      "epoch": 5.952646239554317,
      "grad_norm": 3.322737693786621,
      "learning_rate": 2.891131324616259e-05,
      "loss": 1.4445,
      "step": 4274
    },
    {
      "epoch": 5.954038997214485,
      "grad_norm": 3.8737401962280273,
      "learning_rate": 2.8901364411597496e-05,
      "loss": 1.2271,
      "step": 4275
    },
    {
      "epoch": 5.955431754874652,
      "grad_norm": 2.9896321296691895,
      "learning_rate": 2.8891415577032403e-05,
      "loss": 1.2047,
      "step": 4276
    },
    {
      "epoch": 5.956824512534819,
      "grad_norm": 3.766829490661621,
      "learning_rate": 2.8881466742467307e-05,
      "loss": 1.1086,
      "step": 4277
    },
    {
      "epoch": 5.958217270194986,
      "grad_norm": 3.792320728302002,
      "learning_rate": 2.8871517907902215e-05,
      "loss": 1.1187,
      "step": 4278
    },
    {
      "epoch": 5.959610027855153,
      "grad_norm": 3.7886745929718018,
      "learning_rate": 2.886156907333712e-05,
      "loss": 1.1448,
      "step": 4279
    },
    {
      "epoch": 5.96100278551532,
      "grad_norm": 3.2813446521759033,
      "learning_rate": 2.8851620238772026e-05,
      "loss": 1.2934,
      "step": 4280
    },
    {
      "epoch": 5.962395543175488,
      "grad_norm": 3.670713186264038,
      "learning_rate": 2.8841671404206934e-05,
      "loss": 1.504,
      "step": 4281
    },
    {
      "epoch": 5.963788300835654,
      "grad_norm": 3.573119640350342,
      "learning_rate": 2.8831722569641838e-05,
      "loss": 1.1831,
      "step": 4282
    },
    {
      "epoch": 5.965181058495822,
      "grad_norm": 3.8057022094726562,
      "learning_rate": 2.8821773735076745e-05,
      "loss": 1.5948,
      "step": 4283
    },
    {
      "epoch": 5.9665738161559885,
      "grad_norm": 3.496189594268799,
      "learning_rate": 2.881182490051165e-05,
      "loss": 1.3005,
      "step": 4284
    },
    {
      "epoch": 5.967966573816156,
      "grad_norm": 3.821584463119507,
      "learning_rate": 2.8801876065946557e-05,
      "loss": 1.4228,
      "step": 4285
    },
    {
      "epoch": 5.969359331476323,
      "grad_norm": 3.7349889278411865,
      "learning_rate": 2.8791927231381464e-05,
      "loss": 1.4551,
      "step": 4286
    },
    {
      "epoch": 5.97075208913649,
      "grad_norm": 3.0262346267700195,
      "learning_rate": 2.878197839681637e-05,
      "loss": 1.0351,
      "step": 4287
    },
    {
      "epoch": 5.972144846796658,
      "grad_norm": 3.5886991024017334,
      "learning_rate": 2.8772029562251276e-05,
      "loss": 1.5945,
      "step": 4288
    },
    {
      "epoch": 5.9735376044568245,
      "grad_norm": 4.177419185638428,
      "learning_rate": 2.876208072768618e-05,
      "loss": 1.4383,
      "step": 4289
    },
    {
      "epoch": 5.974930362116992,
      "grad_norm": 3.3451030254364014,
      "learning_rate": 2.875213189312109e-05,
      "loss": 1.138,
      "step": 4290
    },
    {
      "epoch": 5.976323119777159,
      "grad_norm": 3.395125389099121,
      "learning_rate": 2.8742183058556e-05,
      "loss": 1.4852,
      "step": 4291
    },
    {
      "epoch": 5.977715877437326,
      "grad_norm": 3.767486095428467,
      "learning_rate": 2.8732234223990902e-05,
      "loss": 1.5145,
      "step": 4292
    },
    {
      "epoch": 5.979108635097493,
      "grad_norm": 3.584868907928467,
      "learning_rate": 2.872228538942581e-05,
      "loss": 1.5485,
      "step": 4293
    },
    {
      "epoch": 5.9805013927576605,
      "grad_norm": 3.1592671871185303,
      "learning_rate": 2.8712336554860714e-05,
      "loss": 1.3211,
      "step": 4294
    },
    {
      "epoch": 5.981894150417827,
      "grad_norm": 7.4499831199646,
      "learning_rate": 2.870238772029562e-05,
      "loss": 1.4428,
      "step": 4295
    },
    {
      "epoch": 5.983286908077995,
      "grad_norm": 3.348339796066284,
      "learning_rate": 2.8692438885730526e-05,
      "loss": 1.1395,
      "step": 4296
    },
    {
      "epoch": 5.984679665738161,
      "grad_norm": 3.1295952796936035,
      "learning_rate": 2.8682490051165433e-05,
      "loss": 1.4351,
      "step": 4297
    },
    {
      "epoch": 5.986072423398329,
      "grad_norm": 3.1701250076293945,
      "learning_rate": 2.867254121660034e-05,
      "loss": 1.37,
      "step": 4298
    },
    {
      "epoch": 5.987465181058496,
      "grad_norm": 3.4694182872772217,
      "learning_rate": 2.8662592382035245e-05,
      "loss": 1.3412,
      "step": 4299
    },
    {
      "epoch": 5.988857938718663,
      "grad_norm": 3.265918254852295,
      "learning_rate": 2.8652643547470152e-05,
      "loss": 1.3245,
      "step": 4300
    },
    {
      "epoch": 5.99025069637883,
      "grad_norm": 3.54642653465271,
      "learning_rate": 2.8642694712905056e-05,
      "loss": 1.3947,
      "step": 4301
    },
    {
      "epoch": 5.991643454038997,
      "grad_norm": 3.4953207969665527,
      "learning_rate": 2.8632745878339964e-05,
      "loss": 1.2307,
      "step": 4302
    },
    {
      "epoch": 5.993036211699164,
      "grad_norm": 3.104365825653076,
      "learning_rate": 2.862279704377487e-05,
      "loss": 1.2262,
      "step": 4303
    },
    {
      "epoch": 5.994428969359332,
      "grad_norm": 3.410261869430542,
      "learning_rate": 2.8612848209209775e-05,
      "loss": 1.2567,
      "step": 4304
    },
    {
      "epoch": 5.995821727019498,
      "grad_norm": 3.7561678886413574,
      "learning_rate": 2.8602899374644683e-05,
      "loss": 1.5103,
      "step": 4305
    },
    {
      "epoch": 5.997214484679666,
      "grad_norm": 2.7139320373535156,
      "learning_rate": 2.8592950540079587e-05,
      "loss": 1.003,
      "step": 4306
    },
    {
      "epoch": 5.998607242339833,
      "grad_norm": 3.683227300643921,
      "learning_rate": 2.8583001705514494e-05,
      "loss": 1.171,
      "step": 4307
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.929830312728882,
      "learning_rate": 2.85730528709494e-05,
      "loss": 1.0885,
      "step": 4308
    },
    {
      "epoch": 6.0013927576601676,
      "grad_norm": 3.2844910621643066,
      "learning_rate": 2.8563104036384306e-05,
      "loss": 1.0426,
      "step": 4309
    },
    {
      "epoch": 6.002785515320334,
      "grad_norm": 3.4652326107025146,
      "learning_rate": 2.8553155201819213e-05,
      "loss": 1.3491,
      "step": 4310
    },
    {
      "epoch": 6.004178272980502,
      "grad_norm": 3.4769561290740967,
      "learning_rate": 2.8543206367254117e-05,
      "loss": 1.3252,
      "step": 4311
    },
    {
      "epoch": 6.005571030640668,
      "grad_norm": 3.0269827842712402,
      "learning_rate": 2.8533257532689025e-05,
      "loss": 1.2031,
      "step": 4312
    },
    {
      "epoch": 6.006963788300836,
      "grad_norm": 2.7160186767578125,
      "learning_rate": 2.8523308698123932e-05,
      "loss": 0.8197,
      "step": 4313
    },
    {
      "epoch": 6.008356545961003,
      "grad_norm": 3.1668927669525146,
      "learning_rate": 2.8513359863558836e-05,
      "loss": 1.1986,
      "step": 4314
    },
    {
      "epoch": 6.00974930362117,
      "grad_norm": 3.0711967945098877,
      "learning_rate": 2.8503411028993744e-05,
      "loss": 1.206,
      "step": 4315
    },
    {
      "epoch": 6.011142061281337,
      "grad_norm": 3.2880496978759766,
      "learning_rate": 2.8493462194428648e-05,
      "loss": 1.4118,
      "step": 4316
    },
    {
      "epoch": 6.012534818941504,
      "grad_norm": 2.922769069671631,
      "learning_rate": 2.8483513359863555e-05,
      "loss": 1.1758,
      "step": 4317
    },
    {
      "epoch": 6.013927576601671,
      "grad_norm": 3.00602650642395,
      "learning_rate": 2.847356452529846e-05,
      "loss": 1.2837,
      "step": 4318
    },
    {
      "epoch": 6.015320334261839,
      "grad_norm": 3.077406644821167,
      "learning_rate": 2.846361569073337e-05,
      "loss": 1.383,
      "step": 4319
    },
    {
      "epoch": 6.016713091922005,
      "grad_norm": 3.1378235816955566,
      "learning_rate": 2.8453666856168278e-05,
      "loss": 0.9669,
      "step": 4320
    },
    {
      "epoch": 6.018105849582173,
      "grad_norm": 2.8575522899627686,
      "learning_rate": 2.844371802160318e-05,
      "loss": 1.0698,
      "step": 4321
    },
    {
      "epoch": 6.0194986072423395,
      "grad_norm": 2.893104314804077,
      "learning_rate": 2.843376918703809e-05,
      "loss": 1.2204,
      "step": 4322
    },
    {
      "epoch": 6.020891364902507,
      "grad_norm": 3.258876085281372,
      "learning_rate": 2.8423820352472993e-05,
      "loss": 1.1175,
      "step": 4323
    },
    {
      "epoch": 6.022284122562674,
      "grad_norm": 2.6355395317077637,
      "learning_rate": 2.84138715179079e-05,
      "loss": 0.7807,
      "step": 4324
    },
    {
      "epoch": 6.023676880222841,
      "grad_norm": 2.9973530769348145,
      "learning_rate": 2.8403922683342808e-05,
      "loss": 1.0241,
      "step": 4325
    },
    {
      "epoch": 6.025069637883008,
      "grad_norm": 3.0352323055267334,
      "learning_rate": 2.8393973848777712e-05,
      "loss": 1.1441,
      "step": 4326
    },
    {
      "epoch": 6.0264623955431755,
      "grad_norm": 3.1082520484924316,
      "learning_rate": 2.838402501421262e-05,
      "loss": 1.0703,
      "step": 4327
    },
    {
      "epoch": 6.027855153203342,
      "grad_norm": 3.024894952774048,
      "learning_rate": 2.8374076179647524e-05,
      "loss": 1.0597,
      "step": 4328
    },
    {
      "epoch": 6.02924791086351,
      "grad_norm": 3.3348255157470703,
      "learning_rate": 2.836412734508243e-05,
      "loss": 1.3704,
      "step": 4329
    },
    {
      "epoch": 6.030640668523677,
      "grad_norm": 2.8505239486694336,
      "learning_rate": 2.835417851051734e-05,
      "loss": 0.9118,
      "step": 4330
    },
    {
      "epoch": 6.032033426183844,
      "grad_norm": 2.8822379112243652,
      "learning_rate": 2.8344229675952243e-05,
      "loss": 1.0973,
      "step": 4331
    },
    {
      "epoch": 6.0334261838440115,
      "grad_norm": 3.1755788326263428,
      "learning_rate": 2.833428084138715e-05,
      "loss": 0.9507,
      "step": 4332
    },
    {
      "epoch": 6.034818941504178,
      "grad_norm": 2.5899782180786133,
      "learning_rate": 2.8324332006822054e-05,
      "loss": 0.9368,
      "step": 4333
    },
    {
      "epoch": 6.036211699164346,
      "grad_norm": 2.606720209121704,
      "learning_rate": 2.8314383172256962e-05,
      "loss": 0.9887,
      "step": 4334
    },
    {
      "epoch": 6.037604456824512,
      "grad_norm": 2.9312844276428223,
      "learning_rate": 2.830443433769187e-05,
      "loss": 1.0675,
      "step": 4335
    },
    {
      "epoch": 6.03899721448468,
      "grad_norm": 4.057625770568848,
      "learning_rate": 2.8294485503126773e-05,
      "loss": 1.2126,
      "step": 4336
    },
    {
      "epoch": 6.040389972144847,
      "grad_norm": 2.791701316833496,
      "learning_rate": 2.828453666856168e-05,
      "loss": 1.1019,
      "step": 4337
    },
    {
      "epoch": 6.041782729805014,
      "grad_norm": 2.6868085861206055,
      "learning_rate": 2.8274587833996585e-05,
      "loss": 0.8865,
      "step": 4338
    },
    {
      "epoch": 6.043175487465181,
      "grad_norm": 3.106109142303467,
      "learning_rate": 2.8264638999431492e-05,
      "loss": 1.1152,
      "step": 4339
    },
    {
      "epoch": 6.044568245125348,
      "grad_norm": 3.098327875137329,
      "learning_rate": 2.8254690164866396e-05,
      "loss": 1.3081,
      "step": 4340
    },
    {
      "epoch": 6.045961002785515,
      "grad_norm": 2.9800162315368652,
      "learning_rate": 2.8244741330301304e-05,
      "loss": 1.0212,
      "step": 4341
    },
    {
      "epoch": 6.047353760445683,
      "grad_norm": 2.679067850112915,
      "learning_rate": 2.823479249573621e-05,
      "loss": 0.9206,
      "step": 4342
    },
    {
      "epoch": 6.048746518105849,
      "grad_norm": 2.7627615928649902,
      "learning_rate": 2.8224843661171115e-05,
      "loss": 0.9192,
      "step": 4343
    },
    {
      "epoch": 6.050139275766017,
      "grad_norm": 2.181574821472168,
      "learning_rate": 2.8214894826606023e-05,
      "loss": 0.7656,
      "step": 4344
    },
    {
      "epoch": 6.0515320334261835,
      "grad_norm": 3.5677804946899414,
      "learning_rate": 2.8204945992040927e-05,
      "loss": 0.9876,
      "step": 4345
    },
    {
      "epoch": 6.052924791086351,
      "grad_norm": 3.1361148357391357,
      "learning_rate": 2.8194997157475834e-05,
      "loss": 1.1568,
      "step": 4346
    },
    {
      "epoch": 6.054317548746518,
      "grad_norm": 2.8180770874023438,
      "learning_rate": 2.8185048322910745e-05,
      "loss": 0.9822,
      "step": 4347
    },
    {
      "epoch": 6.055710306406685,
      "grad_norm": 2.7147934436798096,
      "learning_rate": 2.817509948834565e-05,
      "loss": 0.9644,
      "step": 4348
    },
    {
      "epoch": 6.057103064066852,
      "grad_norm": 3.5782716274261475,
      "learning_rate": 2.8165150653780557e-05,
      "loss": 1.0967,
      "step": 4349
    },
    {
      "epoch": 6.0584958217270195,
      "grad_norm": 3.360884189605713,
      "learning_rate": 2.815520181921546e-05,
      "loss": 0.8963,
      "step": 4350
    },
    {
      "epoch": 6.059888579387187,
      "grad_norm": 2.692962169647217,
      "learning_rate": 2.8145252984650368e-05,
      "loss": 1.2067,
      "step": 4351
    },
    {
      "epoch": 6.061281337047354,
      "grad_norm": 3.3448727130889893,
      "learning_rate": 2.8135304150085276e-05,
      "loss": 1.1568,
      "step": 4352
    },
    {
      "epoch": 6.062674094707521,
      "grad_norm": 3.0642430782318115,
      "learning_rate": 2.812535531552018e-05,
      "loss": 0.9382,
      "step": 4353
    },
    {
      "epoch": 6.064066852367688,
      "grad_norm": 3.427797555923462,
      "learning_rate": 2.8115406480955087e-05,
      "loss": 1.0851,
      "step": 4354
    },
    {
      "epoch": 6.0654596100278555,
      "grad_norm": 3.3021860122680664,
      "learning_rate": 2.810545764638999e-05,
      "loss": 0.9075,
      "step": 4355
    },
    {
      "epoch": 6.066852367688022,
      "grad_norm": 3.595940113067627,
      "learning_rate": 2.80955088118249e-05,
      "loss": 1.2991,
      "step": 4356
    },
    {
      "epoch": 6.06824512534819,
      "grad_norm": 3.1379806995391846,
      "learning_rate": 2.8085559977259806e-05,
      "loss": 1.1563,
      "step": 4357
    },
    {
      "epoch": 6.069637883008356,
      "grad_norm": 3.217236280441284,
      "learning_rate": 2.807561114269471e-05,
      "loss": 1.3122,
      "step": 4358
    },
    {
      "epoch": 6.071030640668524,
      "grad_norm": 3.2345099449157715,
      "learning_rate": 2.8065662308129618e-05,
      "loss": 1.1902,
      "step": 4359
    },
    {
      "epoch": 6.072423398328691,
      "grad_norm": 2.8844897747039795,
      "learning_rate": 2.8055713473564522e-05,
      "loss": 1.1736,
      "step": 4360
    },
    {
      "epoch": 6.073816155988858,
      "grad_norm": 2.722585439682007,
      "learning_rate": 2.804576463899943e-05,
      "loss": 1.079,
      "step": 4361
    },
    {
      "epoch": 6.075208913649025,
      "grad_norm": 3.402970790863037,
      "learning_rate": 2.8035815804434333e-05,
      "loss": 1.3633,
      "step": 4362
    },
    {
      "epoch": 6.076601671309192,
      "grad_norm": 3.323962450027466,
      "learning_rate": 2.802586696986924e-05,
      "loss": 1.2016,
      "step": 4363
    },
    {
      "epoch": 6.077994428969359,
      "grad_norm": 2.6824100017547607,
      "learning_rate": 2.801591813530415e-05,
      "loss": 1.1126,
      "step": 4364
    },
    {
      "epoch": 6.079387186629527,
      "grad_norm": 2.8601081371307373,
      "learning_rate": 2.8005969300739052e-05,
      "loss": 1.1906,
      "step": 4365
    },
    {
      "epoch": 6.080779944289693,
      "grad_norm": 3.135958194732666,
      "learning_rate": 2.799602046617396e-05,
      "loss": 1.0786,
      "step": 4366
    },
    {
      "epoch": 6.082172701949861,
      "grad_norm": 2.9230966567993164,
      "learning_rate": 2.7986071631608864e-05,
      "loss": 1.1039,
      "step": 4367
    },
    {
      "epoch": 6.0835654596100275,
      "grad_norm": 3.0857913494110107,
      "learning_rate": 2.797612279704377e-05,
      "loss": 1.0456,
      "step": 4368
    },
    {
      "epoch": 6.084958217270195,
      "grad_norm": 3.4706621170043945,
      "learning_rate": 2.796617396247868e-05,
      "loss": 1.2033,
      "step": 4369
    },
    {
      "epoch": 6.086350974930362,
      "grad_norm": 2.7866897583007812,
      "learning_rate": 2.7956225127913583e-05,
      "loss": 0.7676,
      "step": 4370
    },
    {
      "epoch": 6.087743732590529,
      "grad_norm": 3.1143743991851807,
      "learning_rate": 2.794627629334849e-05,
      "loss": 1.2894,
      "step": 4371
    },
    {
      "epoch": 6.089136490250697,
      "grad_norm": 2.9497530460357666,
      "learning_rate": 2.7936327458783394e-05,
      "loss": 0.8985,
      "step": 4372
    },
    {
      "epoch": 6.0905292479108635,
      "grad_norm": 2.946164131164551,
      "learning_rate": 2.7926378624218302e-05,
      "loss": 1.0038,
      "step": 4373
    },
    {
      "epoch": 6.091922005571031,
      "grad_norm": 3.4043126106262207,
      "learning_rate": 2.7916429789653213e-05,
      "loss": 1.2819,
      "step": 4374
    },
    {
      "epoch": 6.093314763231198,
      "grad_norm": 3.185086965560913,
      "learning_rate": 2.7906480955088117e-05,
      "loss": 1.3676,
      "step": 4375
    },
    {
      "epoch": 6.094707520891365,
      "grad_norm": 3.4986045360565186,
      "learning_rate": 2.7896532120523024e-05,
      "loss": 1.0832,
      "step": 4376
    },
    {
      "epoch": 6.096100278551532,
      "grad_norm": 4.043941497802734,
      "learning_rate": 2.788658328595793e-05,
      "loss": 1.059,
      "step": 4377
    },
    {
      "epoch": 6.0974930362116995,
      "grad_norm": 3.022905111312866,
      "learning_rate": 2.7876634451392836e-05,
      "loss": 1.1137,
      "step": 4378
    },
    {
      "epoch": 6.098885793871866,
      "grad_norm": 2.5002658367156982,
      "learning_rate": 2.7866685616827743e-05,
      "loss": 0.9002,
      "step": 4379
    },
    {
      "epoch": 6.100278551532034,
      "grad_norm": 3.5448226928710938,
      "learning_rate": 2.7856736782262647e-05,
      "loss": 1.2442,
      "step": 4380
    },
    {
      "epoch": 6.1016713091922,
      "grad_norm": 2.954012870788574,
      "learning_rate": 2.7846787947697555e-05,
      "loss": 0.9172,
      "step": 4381
    },
    {
      "epoch": 6.103064066852368,
      "grad_norm": 3.0760223865509033,
      "learning_rate": 2.783683911313246e-05,
      "loss": 1.0496,
      "step": 4382
    },
    {
      "epoch": 6.104456824512535,
      "grad_norm": 2.6588072776794434,
      "learning_rate": 2.7826890278567366e-05,
      "loss": 1.0475,
      "step": 4383
    },
    {
      "epoch": 6.105849582172702,
      "grad_norm": 3.1436309814453125,
      "learning_rate": 2.781694144400227e-05,
      "loss": 1.0528,
      "step": 4384
    },
    {
      "epoch": 6.107242339832869,
      "grad_norm": 2.879953622817993,
      "learning_rate": 2.7806992609437178e-05,
      "loss": 1.0093,
      "step": 4385
    },
    {
      "epoch": 6.108635097493036,
      "grad_norm": 3.093233346939087,
      "learning_rate": 2.7797043774872085e-05,
      "loss": 1.0947,
      "step": 4386
    },
    {
      "epoch": 6.110027855153203,
      "grad_norm": 2.9168739318847656,
      "learning_rate": 2.778709494030699e-05,
      "loss": 1.1076,
      "step": 4387
    },
    {
      "epoch": 6.111420612813371,
      "grad_norm": 3.2096879482269287,
      "learning_rate": 2.7777146105741897e-05,
      "loss": 1.0626,
      "step": 4388
    },
    {
      "epoch": 6.112813370473537,
      "grad_norm": 2.9247891902923584,
      "learning_rate": 2.77671972711768e-05,
      "loss": 1.0168,
      "step": 4389
    },
    {
      "epoch": 6.114206128133705,
      "grad_norm": 3.6016478538513184,
      "learning_rate": 2.775724843661171e-05,
      "loss": 1.3709,
      "step": 4390
    },
    {
      "epoch": 6.1155988857938715,
      "grad_norm": 2.836681604385376,
      "learning_rate": 2.7747299602046616e-05,
      "loss": 1.0959,
      "step": 4391
    },
    {
      "epoch": 6.116991643454039,
      "grad_norm": 2.8651485443115234,
      "learning_rate": 2.773735076748152e-05,
      "loss": 1.1398,
      "step": 4392
    },
    {
      "epoch": 6.118384401114207,
      "grad_norm": 2.827662706375122,
      "learning_rate": 2.7727401932916427e-05,
      "loss": 1.0757,
      "step": 4393
    },
    {
      "epoch": 6.119777158774373,
      "grad_norm": 2.425075054168701,
      "learning_rate": 2.771745309835133e-05,
      "loss": 0.7563,
      "step": 4394
    },
    {
      "epoch": 6.121169916434541,
      "grad_norm": 2.821561098098755,
      "learning_rate": 2.770750426378624e-05,
      "loss": 0.8801,
      "step": 4395
    },
    {
      "epoch": 6.1225626740947074,
      "grad_norm": 2.993879795074463,
      "learning_rate": 2.7697555429221146e-05,
      "loss": 0.8296,
      "step": 4396
    },
    {
      "epoch": 6.123955431754875,
      "grad_norm": 3.3326897621154785,
      "learning_rate": 2.768760659465605e-05,
      "loss": 1.1056,
      "step": 4397
    },
    {
      "epoch": 6.125348189415042,
      "grad_norm": 4.613895416259766,
      "learning_rate": 2.7677657760090958e-05,
      "loss": 1.1644,
      "step": 4398
    },
    {
      "epoch": 6.126740947075209,
      "grad_norm": 3.4767098426818848,
      "learning_rate": 2.7667708925525862e-05,
      "loss": 1.2945,
      "step": 4399
    },
    {
      "epoch": 6.128133704735376,
      "grad_norm": 3.333099842071533,
      "learning_rate": 2.765776009096077e-05,
      "loss": 1.1153,
      "step": 4400
    },
    {
      "epoch": 6.129526462395543,
      "grad_norm": 2.639409065246582,
      "learning_rate": 2.764781125639568e-05,
      "loss": 0.9969,
      "step": 4401
    },
    {
      "epoch": 6.13091922005571,
      "grad_norm": 2.9064671993255615,
      "learning_rate": 2.763786242183058e-05,
      "loss": 0.9964,
      "step": 4402
    },
    {
      "epoch": 6.132311977715878,
      "grad_norm": 2.537750482559204,
      "learning_rate": 2.7627913587265492e-05,
      "loss": 0.913,
      "step": 4403
    },
    {
      "epoch": 6.133704735376044,
      "grad_norm": 2.829019546508789,
      "learning_rate": 2.7617964752700396e-05,
      "loss": 0.9105,
      "step": 4404
    },
    {
      "epoch": 6.135097493036212,
      "grad_norm": 3.537351369857788,
      "learning_rate": 2.7608015918135303e-05,
      "loss": 1.0077,
      "step": 4405
    },
    {
      "epoch": 6.1364902506963785,
      "grad_norm": 2.815483570098877,
      "learning_rate": 2.7598067083570208e-05,
      "loss": 1.0808,
      "step": 4406
    },
    {
      "epoch": 6.137883008356546,
      "grad_norm": 3.0557570457458496,
      "learning_rate": 2.7588118249005115e-05,
      "loss": 0.9736,
      "step": 4407
    },
    {
      "epoch": 6.139275766016713,
      "grad_norm": 2.9972572326660156,
      "learning_rate": 2.7578169414440022e-05,
      "loss": 1.1775,
      "step": 4408
    },
    {
      "epoch": 6.14066852367688,
      "grad_norm": 3.338998794555664,
      "learning_rate": 2.7568220579874927e-05,
      "loss": 1.2179,
      "step": 4409
    },
    {
      "epoch": 6.142061281337047,
      "grad_norm": 3.2950048446655273,
      "learning_rate": 2.7558271745309834e-05,
      "loss": 1.2744,
      "step": 4410
    },
    {
      "epoch": 6.1434540389972145,
      "grad_norm": 3.7593259811401367,
      "learning_rate": 2.7548322910744738e-05,
      "loss": 1.0514,
      "step": 4411
    },
    {
      "epoch": 6.144846796657381,
      "grad_norm": 3.059964656829834,
      "learning_rate": 2.7538374076179646e-05,
      "loss": 1.1883,
      "step": 4412
    },
    {
      "epoch": 6.146239554317549,
      "grad_norm": 2.710592269897461,
      "learning_rate": 2.7528425241614553e-05,
      "loss": 0.8026,
      "step": 4413
    },
    {
      "epoch": 6.147632311977716,
      "grad_norm": 3.130387544631958,
      "learning_rate": 2.7518476407049457e-05,
      "loss": 1.2624,
      "step": 4414
    },
    {
      "epoch": 6.149025069637883,
      "grad_norm": 3.216870069503784,
      "learning_rate": 2.7508527572484365e-05,
      "loss": 1.0916,
      "step": 4415
    },
    {
      "epoch": 6.1504178272980505,
      "grad_norm": 2.8923614025115967,
      "learning_rate": 2.749857873791927e-05,
      "loss": 1.0776,
      "step": 4416
    },
    {
      "epoch": 6.151810584958217,
      "grad_norm": 2.855976104736328,
      "learning_rate": 2.7488629903354176e-05,
      "loss": 1.0076,
      "step": 4417
    },
    {
      "epoch": 6.153203342618385,
      "grad_norm": 3.043898105621338,
      "learning_rate": 2.7478681068789084e-05,
      "loss": 0.9476,
      "step": 4418
    },
    {
      "epoch": 6.154596100278551,
      "grad_norm": 3.2427806854248047,
      "learning_rate": 2.7468732234223988e-05,
      "loss": 1.1407,
      "step": 4419
    },
    {
      "epoch": 6.155988857938719,
      "grad_norm": 2.4030675888061523,
      "learning_rate": 2.7458783399658895e-05,
      "loss": 0.8496,
      "step": 4420
    },
    {
      "epoch": 6.157381615598886,
      "grad_norm": 2.9312615394592285,
      "learning_rate": 2.74488345650938e-05,
      "loss": 1.0472,
      "step": 4421
    },
    {
      "epoch": 6.158774373259053,
      "grad_norm": 2.7273006439208984,
      "learning_rate": 2.7438885730528707e-05,
      "loss": 1.0319,
      "step": 4422
    },
    {
      "epoch": 6.16016713091922,
      "grad_norm": 2.653557062149048,
      "learning_rate": 2.7428936895963614e-05,
      "loss": 1.0142,
      "step": 4423
    },
    {
      "epoch": 6.161559888579387,
      "grad_norm": 2.500417947769165,
      "learning_rate": 2.7418988061398518e-05,
      "loss": 0.7894,
      "step": 4424
    },
    {
      "epoch": 6.162952646239554,
      "grad_norm": 3.8580479621887207,
      "learning_rate": 2.7409039226833426e-05,
      "loss": 0.9945,
      "step": 4425
    },
    {
      "epoch": 6.164345403899722,
      "grad_norm": 2.5263960361480713,
      "learning_rate": 2.739909039226833e-05,
      "loss": 0.7399,
      "step": 4426
    },
    {
      "epoch": 6.165738161559888,
      "grad_norm": 3.226283073425293,
      "learning_rate": 2.7389141557703237e-05,
      "loss": 1.0198,
      "step": 4427
    },
    {
      "epoch": 6.167130919220056,
      "grad_norm": 3.1259124279022217,
      "learning_rate": 2.737919272313814e-05,
      "loss": 1.0195,
      "step": 4428
    },
    {
      "epoch": 6.1685236768802225,
      "grad_norm": 2.9556758403778076,
      "learning_rate": 2.736924388857305e-05,
      "loss": 0.9375,
      "step": 4429
    },
    {
      "epoch": 6.16991643454039,
      "grad_norm": 3.263813018798828,
      "learning_rate": 2.735929505400796e-05,
      "loss": 1.273,
      "step": 4430
    },
    {
      "epoch": 6.171309192200557,
      "grad_norm": 2.9776573181152344,
      "learning_rate": 2.7349346219442864e-05,
      "loss": 1.2021,
      "step": 4431
    },
    {
      "epoch": 6.172701949860724,
      "grad_norm": 2.862586498260498,
      "learning_rate": 2.733939738487777e-05,
      "loss": 1.0884,
      "step": 4432
    },
    {
      "epoch": 6.174094707520891,
      "grad_norm": 3.386265993118286,
      "learning_rate": 2.7329448550312675e-05,
      "loss": 1.2135,
      "step": 4433
    },
    {
      "epoch": 6.1754874651810585,
      "grad_norm": 3.206834316253662,
      "learning_rate": 2.7319499715747583e-05,
      "loss": 1.1536,
      "step": 4434
    },
    {
      "epoch": 6.176880222841225,
      "grad_norm": 3.6165032386779785,
      "learning_rate": 2.730955088118249e-05,
      "loss": 1.3512,
      "step": 4435
    },
    {
      "epoch": 6.178272980501393,
      "grad_norm": 2.75239634513855,
      "learning_rate": 2.7299602046617394e-05,
      "loss": 1.1119,
      "step": 4436
    },
    {
      "epoch": 6.17966573816156,
      "grad_norm": 3.360097646713257,
      "learning_rate": 2.72896532120523e-05,
      "loss": 1.3204,
      "step": 4437
    },
    {
      "epoch": 6.181058495821727,
      "grad_norm": 3.0297152996063232,
      "learning_rate": 2.7279704377487206e-05,
      "loss": 0.9583,
      "step": 4438
    },
    {
      "epoch": 6.1824512534818945,
      "grad_norm": 3.2654929161071777,
      "learning_rate": 2.7269755542922113e-05,
      "loss": 1.0884,
      "step": 4439
    },
    {
      "epoch": 6.183844011142061,
      "grad_norm": 3.2654433250427246,
      "learning_rate": 2.725980670835702e-05,
      "loss": 1.0021,
      "step": 4440
    },
    {
      "epoch": 6.185236768802229,
      "grad_norm": 2.9692389965057373,
      "learning_rate": 2.7249857873791925e-05,
      "loss": 1.0119,
      "step": 4441
    },
    {
      "epoch": 6.186629526462395,
      "grad_norm": 3.0016331672668457,
      "learning_rate": 2.7239909039226832e-05,
      "loss": 1.0429,
      "step": 4442
    },
    {
      "epoch": 6.188022284122563,
      "grad_norm": 3.1427507400512695,
      "learning_rate": 2.7229960204661736e-05,
      "loss": 1.121,
      "step": 4443
    },
    {
      "epoch": 6.18941504178273,
      "grad_norm": 3.3997702598571777,
      "learning_rate": 2.7220011370096644e-05,
      "loss": 1.3677,
      "step": 4444
    },
    {
      "epoch": 6.190807799442897,
      "grad_norm": 2.856048583984375,
      "learning_rate": 2.7210062535531548e-05,
      "loss": 1.1527,
      "step": 4445
    },
    {
      "epoch": 6.192200557103064,
      "grad_norm": 2.6541314125061035,
      "learning_rate": 2.7200113700966455e-05,
      "loss": 0.9218,
      "step": 4446
    },
    {
      "epoch": 6.193593314763231,
      "grad_norm": 3.25169038772583,
      "learning_rate": 2.7190164866401363e-05,
      "loss": 1.3243,
      "step": 4447
    },
    {
      "epoch": 6.194986072423398,
      "grad_norm": 2.727038860321045,
      "learning_rate": 2.7180216031836267e-05,
      "loss": 1.0709,
      "step": 4448
    },
    {
      "epoch": 6.196378830083566,
      "grad_norm": 3.0074374675750732,
      "learning_rate": 2.7170267197271174e-05,
      "loss": 1.136,
      "step": 4449
    },
    {
      "epoch": 6.197771587743732,
      "grad_norm": 3.121025800704956,
      "learning_rate": 2.716031836270608e-05,
      "loss": 1.2403,
      "step": 4450
    },
    {
      "epoch": 6.1991643454039,
      "grad_norm": 3.5447516441345215,
      "learning_rate": 2.7150369528140986e-05,
      "loss": 1.0727,
      "step": 4451
    },
    {
      "epoch": 6.2005571030640665,
      "grad_norm": 3.221599578857422,
      "learning_rate": 2.7140420693575893e-05,
      "loss": 1.1793,
      "step": 4452
    },
    {
      "epoch": 6.201949860724234,
      "grad_norm": 2.852677345275879,
      "learning_rate": 2.7130471859010797e-05,
      "loss": 1.1591,
      "step": 4453
    },
    {
      "epoch": 6.203342618384401,
      "grad_norm": 2.823099136352539,
      "learning_rate": 2.7120523024445705e-05,
      "loss": 0.9864,
      "step": 4454
    },
    {
      "epoch": 6.204735376044568,
      "grad_norm": 2.9134204387664795,
      "learning_rate": 2.711057418988061e-05,
      "loss": 1.1051,
      "step": 4455
    },
    {
      "epoch": 6.206128133704736,
      "grad_norm": 3.437868356704712,
      "learning_rate": 2.7100625355315516e-05,
      "loss": 1.1092,
      "step": 4456
    },
    {
      "epoch": 6.2075208913649025,
      "grad_norm": 3.328800916671753,
      "learning_rate": 2.7090676520750427e-05,
      "loss": 1.1272,
      "step": 4457
    },
    {
      "epoch": 6.20891364902507,
      "grad_norm": 3.6444592475891113,
      "learning_rate": 2.7080727686185328e-05,
      "loss": 1.4687,
      "step": 4458
    },
    {
      "epoch": 6.210306406685237,
      "grad_norm": 3.0198705196380615,
      "learning_rate": 2.707077885162024e-05,
      "loss": 1.0773,
      "step": 4459
    },
    {
      "epoch": 6.211699164345404,
      "grad_norm": 3.073758363723755,
      "learning_rate": 2.7060830017055143e-05,
      "loss": 1.2888,
      "step": 4460
    },
    {
      "epoch": 6.213091922005571,
      "grad_norm": 2.8396661281585693,
      "learning_rate": 2.705088118249005e-05,
      "loss": 1.0346,
      "step": 4461
    },
    {
      "epoch": 6.2144846796657385,
      "grad_norm": 2.5152876377105713,
      "learning_rate": 2.7040932347924958e-05,
      "loss": 0.8936,
      "step": 4462
    },
    {
      "epoch": 6.215877437325905,
      "grad_norm": 2.855208158493042,
      "learning_rate": 2.7030983513359862e-05,
      "loss": 0.9975,
      "step": 4463
    },
    {
      "epoch": 6.217270194986073,
      "grad_norm": 3.0980777740478516,
      "learning_rate": 2.702103467879477e-05,
      "loss": 1.1262,
      "step": 4464
    },
    {
      "epoch": 6.218662952646239,
      "grad_norm": 3.4908106327056885,
      "learning_rate": 2.7011085844229673e-05,
      "loss": 0.8788,
      "step": 4465
    },
    {
      "epoch": 6.220055710306407,
      "grad_norm": 2.9197728633880615,
      "learning_rate": 2.700113700966458e-05,
      "loss": 1.0582,
      "step": 4466
    },
    {
      "epoch": 6.221448467966574,
      "grad_norm": 2.9431533813476562,
      "learning_rate": 2.6991188175099485e-05,
      "loss": 1.3294,
      "step": 4467
    },
    {
      "epoch": 6.222841225626741,
      "grad_norm": 3.098060369491577,
      "learning_rate": 2.6981239340534392e-05,
      "loss": 0.9908,
      "step": 4468
    },
    {
      "epoch": 6.224233983286908,
      "grad_norm": 2.678053855895996,
      "learning_rate": 2.69712905059693e-05,
      "loss": 0.9772,
      "step": 4469
    },
    {
      "epoch": 6.225626740947075,
      "grad_norm": 4.149014949798584,
      "learning_rate": 2.6961341671404204e-05,
      "loss": 0.9701,
      "step": 4470
    },
    {
      "epoch": 6.227019498607242,
      "grad_norm": 3.3159608840942383,
      "learning_rate": 2.695139283683911e-05,
      "loss": 1.1843,
      "step": 4471
    },
    {
      "epoch": 6.22841225626741,
      "grad_norm": 2.8942387104034424,
      "learning_rate": 2.6941444002274015e-05,
      "loss": 0.9166,
      "step": 4472
    },
    {
      "epoch": 6.229805013927576,
      "grad_norm": 3.1490697860717773,
      "learning_rate": 2.6931495167708923e-05,
      "loss": 1.087,
      "step": 4473
    },
    {
      "epoch": 6.231197771587744,
      "grad_norm": 3.0416064262390137,
      "learning_rate": 2.692154633314383e-05,
      "loss": 1.069,
      "step": 4474
    },
    {
      "epoch": 6.2325905292479105,
      "grad_norm": 3.295464277267456,
      "learning_rate": 2.6911597498578734e-05,
      "loss": 1.0931,
      "step": 4475
    },
    {
      "epoch": 6.233983286908078,
      "grad_norm": 2.9210784435272217,
      "learning_rate": 2.6901648664013642e-05,
      "loss": 0.8531,
      "step": 4476
    },
    {
      "epoch": 6.235376044568245,
      "grad_norm": 3.215038537979126,
      "learning_rate": 2.6891699829448546e-05,
      "loss": 1.1697,
      "step": 4477
    },
    {
      "epoch": 6.236768802228412,
      "grad_norm": 2.8260319232940674,
      "learning_rate": 2.6881750994883453e-05,
      "loss": 1.1174,
      "step": 4478
    },
    {
      "epoch": 6.23816155988858,
      "grad_norm": 2.42219614982605,
      "learning_rate": 2.687180216031836e-05,
      "loss": 0.8495,
      "step": 4479
    },
    {
      "epoch": 6.2395543175487465,
      "grad_norm": 3.514204740524292,
      "learning_rate": 2.6861853325753265e-05,
      "loss": 1.2002,
      "step": 4480
    },
    {
      "epoch": 6.240947075208914,
      "grad_norm": 3.183506488800049,
      "learning_rate": 2.6851904491188172e-05,
      "loss": 1.0287,
      "step": 4481
    },
    {
      "epoch": 6.242339832869081,
      "grad_norm": 3.0259408950805664,
      "learning_rate": 2.6841955656623076e-05,
      "loss": 0.9779,
      "step": 4482
    },
    {
      "epoch": 6.243732590529248,
      "grad_norm": 3.318598747253418,
      "learning_rate": 2.6832006822057984e-05,
      "loss": 1.0195,
      "step": 4483
    },
    {
      "epoch": 6.245125348189415,
      "grad_norm": 2.9120140075683594,
      "learning_rate": 2.6822057987492895e-05,
      "loss": 1.1234,
      "step": 4484
    },
    {
      "epoch": 6.2465181058495824,
      "grad_norm": 3.182194709777832,
      "learning_rate": 2.6812109152927795e-05,
      "loss": 1.2401,
      "step": 4485
    },
    {
      "epoch": 6.247910863509749,
      "grad_norm": 2.682565450668335,
      "learning_rate": 2.6802160318362706e-05,
      "loss": 1.0734,
      "step": 4486
    },
    {
      "epoch": 6.249303621169917,
      "grad_norm": 3.1601169109344482,
      "learning_rate": 2.679221148379761e-05,
      "loss": 1.203,
      "step": 4487
    },
    {
      "epoch": 6.250696378830083,
      "grad_norm": 2.9164812564849854,
      "learning_rate": 2.6782262649232518e-05,
      "loss": 1.2156,
      "step": 4488
    },
    {
      "epoch": 6.252089136490251,
      "grad_norm": 2.9238650798797607,
      "learning_rate": 2.6772313814667422e-05,
      "loss": 1.0867,
      "step": 4489
    },
    {
      "epoch": 6.2534818941504176,
      "grad_norm": 2.9365594387054443,
      "learning_rate": 2.676236498010233e-05,
      "loss": 1.1165,
      "step": 4490
    },
    {
      "epoch": 6.254874651810585,
      "grad_norm": 2.6445999145507812,
      "learning_rate": 2.6752416145537237e-05,
      "loss": 1.0766,
      "step": 4491
    },
    {
      "epoch": 6.256267409470752,
      "grad_norm": 2.6067068576812744,
      "learning_rate": 2.674246731097214e-05,
      "loss": 1.0061,
      "step": 4492
    },
    {
      "epoch": 6.257660167130919,
      "grad_norm": 3.738860607147217,
      "learning_rate": 2.673251847640705e-05,
      "loss": 1.0554,
      "step": 4493
    },
    {
      "epoch": 6.259052924791086,
      "grad_norm": 3.5218794345855713,
      "learning_rate": 2.6722569641841952e-05,
      "loss": 1.1866,
      "step": 4494
    },
    {
      "epoch": 6.2604456824512535,
      "grad_norm": 2.8027846813201904,
      "learning_rate": 2.671262080727686e-05,
      "loss": 0.9919,
      "step": 4495
    },
    {
      "epoch": 6.26183844011142,
      "grad_norm": 2.9861958026885986,
      "learning_rate": 2.6702671972711767e-05,
      "loss": 1.1291,
      "step": 4496
    },
    {
      "epoch": 6.263231197771588,
      "grad_norm": 3.143234968185425,
      "learning_rate": 2.669272313814667e-05,
      "loss": 0.8558,
      "step": 4497
    },
    {
      "epoch": 6.264623955431755,
      "grad_norm": 2.6168835163116455,
      "learning_rate": 2.668277430358158e-05,
      "loss": 0.944,
      "step": 4498
    },
    {
      "epoch": 6.266016713091922,
      "grad_norm": 3.6969757080078125,
      "learning_rate": 2.6672825469016483e-05,
      "loss": 1.1644,
      "step": 4499
    },
    {
      "epoch": 6.2674094707520895,
      "grad_norm": 2.8025553226470947,
      "learning_rate": 2.666287663445139e-05,
      "loss": 1.009,
      "step": 4500
    },
    {
      "epoch": 6.268802228412256,
      "grad_norm": 2.835324764251709,
      "learning_rate": 2.6652927799886298e-05,
      "loss": 0.9676,
      "step": 4501
    },
    {
      "epoch": 6.270194986072424,
      "grad_norm": 3.0212371349334717,
      "learning_rate": 2.6642978965321202e-05,
      "loss": 1.0412,
      "step": 4502
    },
    {
      "epoch": 6.27158774373259,
      "grad_norm": 3.110295295715332,
      "learning_rate": 2.663303013075611e-05,
      "loss": 1.0001,
      "step": 4503
    },
    {
      "epoch": 6.272980501392758,
      "grad_norm": 3.0676543712615967,
      "learning_rate": 2.6623081296191014e-05,
      "loss": 0.9887,
      "step": 4504
    },
    {
      "epoch": 6.274373259052925,
      "grad_norm": 3.4942009449005127,
      "learning_rate": 2.661313246162592e-05,
      "loss": 1.2879,
      "step": 4505
    },
    {
      "epoch": 6.275766016713092,
      "grad_norm": 3.999800682067871,
      "learning_rate": 2.660318362706083e-05,
      "loss": 1.1081,
      "step": 4506
    },
    {
      "epoch": 6.277158774373259,
      "grad_norm": 6.410895824432373,
      "learning_rate": 2.6593234792495733e-05,
      "loss": 1.2086,
      "step": 4507
    },
    {
      "epoch": 6.278551532033426,
      "grad_norm": 2.805187463760376,
      "learning_rate": 2.658328595793064e-05,
      "loss": 1.0742,
      "step": 4508
    },
    {
      "epoch": 6.279944289693593,
      "grad_norm": 3.0566322803497314,
      "learning_rate": 2.6573337123365544e-05,
      "loss": 1.0853,
      "step": 4509
    },
    {
      "epoch": 6.281337047353761,
      "grad_norm": 2.8116209506988525,
      "learning_rate": 2.656338828880045e-05,
      "loss": 0.8955,
      "step": 4510
    },
    {
      "epoch": 6.282729805013927,
      "grad_norm": 2.5915589332580566,
      "learning_rate": 2.6553439454235356e-05,
      "loss": 0.7699,
      "step": 4511
    },
    {
      "epoch": 6.284122562674095,
      "grad_norm": 3.1562087535858154,
      "learning_rate": 2.6543490619670263e-05,
      "loss": 0.9947,
      "step": 4512
    },
    {
      "epoch": 6.2855153203342615,
      "grad_norm": 2.7000367641448975,
      "learning_rate": 2.6533541785105174e-05,
      "loss": 0.9771,
      "step": 4513
    },
    {
      "epoch": 6.286908077994429,
      "grad_norm": 2.8173441886901855,
      "learning_rate": 2.6523592950540075e-05,
      "loss": 1.0611,
      "step": 4514
    },
    {
      "epoch": 6.288300835654596,
      "grad_norm": 2.9903464317321777,
      "learning_rate": 2.6513644115974986e-05,
      "loss": 1.0874,
      "step": 4515
    },
    {
      "epoch": 6.289693593314763,
      "grad_norm": 3.273923397064209,
      "learning_rate": 2.650369528140989e-05,
      "loss": 1.0667,
      "step": 4516
    },
    {
      "epoch": 6.29108635097493,
      "grad_norm": 3.0306897163391113,
      "learning_rate": 2.6493746446844797e-05,
      "loss": 1.0308,
      "step": 4517
    },
    {
      "epoch": 6.2924791086350975,
      "grad_norm": 2.948537588119507,
      "learning_rate": 2.6483797612279705e-05,
      "loss": 0.9645,
      "step": 4518
    },
    {
      "epoch": 6.293871866295264,
      "grad_norm": 2.477851390838623,
      "learning_rate": 2.647384877771461e-05,
      "loss": 0.7948,
      "step": 4519
    },
    {
      "epoch": 6.295264623955432,
      "grad_norm": 2.6905126571655273,
      "learning_rate": 2.6463899943149516e-05,
      "loss": 0.9911,
      "step": 4520
    },
    {
      "epoch": 6.296657381615599,
      "grad_norm": 2.6656227111816406,
      "learning_rate": 2.645395110858442e-05,
      "loss": 0.9001,
      "step": 4521
    },
    {
      "epoch": 6.298050139275766,
      "grad_norm": 2.6179158687591553,
      "learning_rate": 2.6444002274019328e-05,
      "loss": 0.8578,
      "step": 4522
    },
    {
      "epoch": 6.2994428969359335,
      "grad_norm": 2.883927822113037,
      "learning_rate": 2.6434053439454235e-05,
      "loss": 1.0552,
      "step": 4523
    },
    {
      "epoch": 6.3008356545961,
      "grad_norm": 2.9616405963897705,
      "learning_rate": 2.642410460488914e-05,
      "loss": 1.0918,
      "step": 4524
    },
    {
      "epoch": 6.302228412256268,
      "grad_norm": 2.888624906539917,
      "learning_rate": 2.6414155770324047e-05,
      "loss": 1.2251,
      "step": 4525
    },
    {
      "epoch": 6.303621169916434,
      "grad_norm": 3.91120982170105,
      "learning_rate": 2.640420693575895e-05,
      "loss": 1.2118,
      "step": 4526
    },
    {
      "epoch": 6.305013927576602,
      "grad_norm": 2.8514249324798584,
      "learning_rate": 2.6394258101193858e-05,
      "loss": 0.9345,
      "step": 4527
    },
    {
      "epoch": 6.306406685236769,
      "grad_norm": 2.232351303100586,
      "learning_rate": 2.6384309266628766e-05,
      "loss": 0.7428,
      "step": 4528
    },
    {
      "epoch": 6.307799442896936,
      "grad_norm": 2.5607776641845703,
      "learning_rate": 2.637436043206367e-05,
      "loss": 0.7218,
      "step": 4529
    },
    {
      "epoch": 6.309192200557103,
      "grad_norm": 2.8364899158477783,
      "learning_rate": 2.6364411597498577e-05,
      "loss": 1.019,
      "step": 4530
    },
    {
      "epoch": 6.31058495821727,
      "grad_norm": 2.9029994010925293,
      "learning_rate": 2.635446276293348e-05,
      "loss": 1.183,
      "step": 4531
    },
    {
      "epoch": 6.311977715877437,
      "grad_norm": 3.020650863647461,
      "learning_rate": 2.634451392836839e-05,
      "loss": 1.0243,
      "step": 4532
    },
    {
      "epoch": 6.313370473537605,
      "grad_norm": 2.6905677318573,
      "learning_rate": 2.6334565093803293e-05,
      "loss": 0.8836,
      "step": 4533
    },
    {
      "epoch": 6.314763231197771,
      "grad_norm": 2.8317041397094727,
      "learning_rate": 2.63246162592382e-05,
      "loss": 0.9563,
      "step": 4534
    },
    {
      "epoch": 6.316155988857939,
      "grad_norm": 3.1123011112213135,
      "learning_rate": 2.6314667424673108e-05,
      "loss": 1.1499,
      "step": 4535
    },
    {
      "epoch": 6.3175487465181055,
      "grad_norm": 3.1473801136016846,
      "learning_rate": 2.6304718590108012e-05,
      "loss": 1.1845,
      "step": 4536
    },
    {
      "epoch": 6.318941504178273,
      "grad_norm": 2.918656349182129,
      "learning_rate": 2.629476975554292e-05,
      "loss": 1.2055,
      "step": 4537
    },
    {
      "epoch": 6.32033426183844,
      "grad_norm": 3.118304491043091,
      "learning_rate": 2.6284820920977823e-05,
      "loss": 1.3297,
      "step": 4538
    },
    {
      "epoch": 6.321727019498607,
      "grad_norm": 2.6895864009857178,
      "learning_rate": 2.627487208641273e-05,
      "loss": 1.184,
      "step": 4539
    },
    {
      "epoch": 6.323119777158775,
      "grad_norm": 2.6843533515930176,
      "learning_rate": 2.626492325184764e-05,
      "loss": 1.1,
      "step": 4540
    },
    {
      "epoch": 6.3245125348189415,
      "grad_norm": 3.0688438415527344,
      "learning_rate": 2.6254974417282542e-05,
      "loss": 1.1488,
      "step": 4541
    },
    {
      "epoch": 6.325905292479109,
      "grad_norm": 2.361597776412964,
      "learning_rate": 2.6245025582717453e-05,
      "loss": 0.8017,
      "step": 4542
    },
    {
      "epoch": 6.327298050139276,
      "grad_norm": 3.0336103439331055,
      "learning_rate": 2.6235076748152354e-05,
      "loss": 1.1591,
      "step": 4543
    },
    {
      "epoch": 6.328690807799443,
      "grad_norm": 2.713315010070801,
      "learning_rate": 2.6225127913587265e-05,
      "loss": 0.9068,
      "step": 4544
    },
    {
      "epoch": 6.33008356545961,
      "grad_norm": 2.8144984245300293,
      "learning_rate": 2.6215179079022172e-05,
      "loss": 1.0332,
      "step": 4545
    },
    {
      "epoch": 6.3314763231197775,
      "grad_norm": 3.7023849487304688,
      "learning_rate": 2.6205230244457076e-05,
      "loss": 1.1772,
      "step": 4546
    },
    {
      "epoch": 6.332869080779944,
      "grad_norm": 3.373838186264038,
      "learning_rate": 2.6195281409891984e-05,
      "loss": 1.203,
      "step": 4547
    },
    {
      "epoch": 6.334261838440112,
      "grad_norm": 2.8458609580993652,
      "learning_rate": 2.6185332575326888e-05,
      "loss": 1.1772,
      "step": 4548
    },
    {
      "epoch": 6.335654596100278,
      "grad_norm": 2.9250094890594482,
      "learning_rate": 2.6175383740761795e-05,
      "loss": 1.2326,
      "step": 4549
    },
    {
      "epoch": 6.337047353760446,
      "grad_norm": 2.676616907119751,
      "learning_rate": 2.6165434906196703e-05,
      "loss": 1.0522,
      "step": 4550
    },
    {
      "epoch": 6.338440111420613,
      "grad_norm": 3.5170352458953857,
      "learning_rate": 2.6155486071631607e-05,
      "loss": 0.9828,
      "step": 4551
    },
    {
      "epoch": 6.33983286908078,
      "grad_norm": 3.3684849739074707,
      "learning_rate": 2.6145537237066514e-05,
      "loss": 1.0967,
      "step": 4552
    },
    {
      "epoch": 6.341225626740947,
      "grad_norm": 2.9907102584838867,
      "learning_rate": 2.6135588402501418e-05,
      "loss": 1.0293,
      "step": 4553
    },
    {
      "epoch": 6.342618384401114,
      "grad_norm": 2.501798629760742,
      "learning_rate": 2.6125639567936326e-05,
      "loss": 0.9224,
      "step": 4554
    },
    {
      "epoch": 6.344011142061281,
      "grad_norm": 3.068803310394287,
      "learning_rate": 2.611569073337123e-05,
      "loss": 1.229,
      "step": 4555
    },
    {
      "epoch": 6.345403899721449,
      "grad_norm": 3.6298787593841553,
      "learning_rate": 2.6105741898806137e-05,
      "loss": 1.2226,
      "step": 4556
    },
    {
      "epoch": 6.346796657381615,
      "grad_norm": 4.111496448516846,
      "learning_rate": 2.6095793064241045e-05,
      "loss": 1.3388,
      "step": 4557
    },
    {
      "epoch": 6.348189415041783,
      "grad_norm": 2.6730756759643555,
      "learning_rate": 2.608584422967595e-05,
      "loss": 0.9996,
      "step": 4558
    },
    {
      "epoch": 6.3495821727019495,
      "grad_norm": 2.889697313308716,
      "learning_rate": 2.6075895395110856e-05,
      "loss": 1.0544,
      "step": 4559
    },
    {
      "epoch": 6.350974930362117,
      "grad_norm": 2.5355618000030518,
      "learning_rate": 2.606594656054576e-05,
      "loss": 0.9156,
      "step": 4560
    },
    {
      "epoch": 6.352367688022284,
      "grad_norm": 2.8138792514801025,
      "learning_rate": 2.6055997725980668e-05,
      "loss": 0.9829,
      "step": 4561
    },
    {
      "epoch": 6.353760445682451,
      "grad_norm": 3.2617111206054688,
      "learning_rate": 2.6046048891415575e-05,
      "loss": 1.1128,
      "step": 4562
    },
    {
      "epoch": 6.355153203342619,
      "grad_norm": 2.78334641456604,
      "learning_rate": 2.603610005685048e-05,
      "loss": 0.9547,
      "step": 4563
    },
    {
      "epoch": 6.3565459610027855,
      "grad_norm": 3.299154043197632,
      "learning_rate": 2.6026151222285387e-05,
      "loss": 1.1187,
      "step": 4564
    },
    {
      "epoch": 6.357938718662953,
      "grad_norm": 3.1545679569244385,
      "learning_rate": 2.601620238772029e-05,
      "loss": 0.9695,
      "step": 4565
    },
    {
      "epoch": 6.35933147632312,
      "grad_norm": 3.2818355560302734,
      "learning_rate": 2.60062535531552e-05,
      "loss": 1.1084,
      "step": 4566
    },
    {
      "epoch": 6.360724233983287,
      "grad_norm": 2.984417676925659,
      "learning_rate": 2.5996304718590106e-05,
      "loss": 0.9893,
      "step": 4567
    },
    {
      "epoch": 6.362116991643454,
      "grad_norm": 3.286147117614746,
      "learning_rate": 2.598635588402501e-05,
      "loss": 1.3968,
      "step": 4568
    },
    {
      "epoch": 6.3635097493036215,
      "grad_norm": 3.4263768196105957,
      "learning_rate": 2.597640704945992e-05,
      "loss": 1.2713,
      "step": 4569
    },
    {
      "epoch": 6.364902506963788,
      "grad_norm": 2.975414514541626,
      "learning_rate": 2.596645821489482e-05,
      "loss": 1.2003,
      "step": 4570
    },
    {
      "epoch": 6.366295264623956,
      "grad_norm": 2.4351494312286377,
      "learning_rate": 2.5956509380329732e-05,
      "loss": 0.9211,
      "step": 4571
    },
    {
      "epoch": 6.367688022284122,
      "grad_norm": 3.6765012741088867,
      "learning_rate": 2.594656054576464e-05,
      "loss": 1.0735,
      "step": 4572
    },
    {
      "epoch": 6.36908077994429,
      "grad_norm": 3.1842734813690186,
      "learning_rate": 2.5936611711199544e-05,
      "loss": 1.1938,
      "step": 4573
    },
    {
      "epoch": 6.370473537604457,
      "grad_norm": 3.16499400138855,
      "learning_rate": 2.592666287663445e-05,
      "loss": 1.3683,
      "step": 4574
    },
    {
      "epoch": 6.371866295264624,
      "grad_norm": 2.929141044616699,
      "learning_rate": 2.5916714042069355e-05,
      "loss": 1.0272,
      "step": 4575
    },
    {
      "epoch": 6.373259052924791,
      "grad_norm": 3.245147228240967,
      "learning_rate": 2.5906765207504263e-05,
      "loss": 1.1937,
      "step": 4576
    },
    {
      "epoch": 6.374651810584958,
      "grad_norm": 3.472449541091919,
      "learning_rate": 2.5896816372939167e-05,
      "loss": 1.3552,
      "step": 4577
    },
    {
      "epoch": 6.376044568245125,
      "grad_norm": 2.93967866897583,
      "learning_rate": 2.5886867538374074e-05,
      "loss": 0.9408,
      "step": 4578
    },
    {
      "epoch": 6.3774373259052926,
      "grad_norm": 3.1434576511383057,
      "learning_rate": 2.5876918703808982e-05,
      "loss": 1.1183,
      "step": 4579
    },
    {
      "epoch": 6.378830083565459,
      "grad_norm": 2.9085288047790527,
      "learning_rate": 2.5866969869243886e-05,
      "loss": 0.9078,
      "step": 4580
    },
    {
      "epoch": 6.380222841225627,
      "grad_norm": 4.131046772003174,
      "learning_rate": 2.5857021034678793e-05,
      "loss": 1.1996,
      "step": 4581
    },
    {
      "epoch": 6.381615598885794,
      "grad_norm": 3.206854820251465,
      "learning_rate": 2.5847072200113697e-05,
      "loss": 1.1907,
      "step": 4582
    },
    {
      "epoch": 6.383008356545961,
      "grad_norm": 3.003754138946533,
      "learning_rate": 2.5837123365548605e-05,
      "loss": 1.1902,
      "step": 4583
    },
    {
      "epoch": 6.3844011142061285,
      "grad_norm": 3.2510647773742676,
      "learning_rate": 2.5827174530983512e-05,
      "loss": 1.2185,
      "step": 4584
    },
    {
      "epoch": 6.385793871866295,
      "grad_norm": 3.215944528579712,
      "learning_rate": 2.5817225696418416e-05,
      "loss": 1.0377,
      "step": 4585
    },
    {
      "epoch": 6.387186629526463,
      "grad_norm": 3.019026756286621,
      "learning_rate": 2.5807276861853324e-05,
      "loss": 1.1211,
      "step": 4586
    },
    {
      "epoch": 6.388579387186629,
      "grad_norm": 2.9317314624786377,
      "learning_rate": 2.5797328027288228e-05,
      "loss": 1.0638,
      "step": 4587
    },
    {
      "epoch": 6.389972144846797,
      "grad_norm": 2.9449985027313232,
      "learning_rate": 2.5787379192723135e-05,
      "loss": 0.9736,
      "step": 4588
    },
    {
      "epoch": 6.391364902506964,
      "grad_norm": 3.33158540725708,
      "learning_rate": 2.5777430358158043e-05,
      "loss": 1.0651,
      "step": 4589
    },
    {
      "epoch": 6.392757660167131,
      "grad_norm": 2.5780160427093506,
      "learning_rate": 2.5767481523592947e-05,
      "loss": 1.0129,
      "step": 4590
    },
    {
      "epoch": 6.394150417827298,
      "grad_norm": 3.0476839542388916,
      "learning_rate": 2.5757532689027854e-05,
      "loss": 1.0014,
      "step": 4591
    },
    {
      "epoch": 6.395543175487465,
      "grad_norm": 3.832833766937256,
      "learning_rate": 2.574758385446276e-05,
      "loss": 0.9095,
      "step": 4592
    },
    {
      "epoch": 6.396935933147632,
      "grad_norm": 2.8570642471313477,
      "learning_rate": 2.5737635019897666e-05,
      "loss": 0.968,
      "step": 4593
    },
    {
      "epoch": 6.3983286908078,
      "grad_norm": 3.2362117767333984,
      "learning_rate": 2.5727686185332573e-05,
      "loss": 1.0836,
      "step": 4594
    },
    {
      "epoch": 6.399721448467966,
      "grad_norm": 2.5439629554748535,
      "learning_rate": 2.5717737350767478e-05,
      "loss": 0.9426,
      "step": 4595
    },
    {
      "epoch": 6.401114206128134,
      "grad_norm": 3.088709592819214,
      "learning_rate": 2.5707788516202385e-05,
      "loss": 1.1695,
      "step": 4596
    },
    {
      "epoch": 6.4025069637883005,
      "grad_norm": 2.907419443130493,
      "learning_rate": 2.569783968163729e-05,
      "loss": 1.0151,
      "step": 4597
    },
    {
      "epoch": 6.403899721448468,
      "grad_norm": 3.443770408630371,
      "learning_rate": 2.56878908470722e-05,
      "loss": 1.1531,
      "step": 4598
    },
    {
      "epoch": 6.405292479108635,
      "grad_norm": 3.1635823249816895,
      "learning_rate": 2.56779420125071e-05,
      "loss": 1.0283,
      "step": 4599
    },
    {
      "epoch": 6.406685236768802,
      "grad_norm": 2.810631513595581,
      "learning_rate": 2.566799317794201e-05,
      "loss": 1.0196,
      "step": 4600
    },
    {
      "epoch": 6.408077994428969,
      "grad_norm": 2.9825165271759033,
      "learning_rate": 2.565804434337692e-05,
      "loss": 1.3187,
      "step": 4601
    },
    {
      "epoch": 6.4094707520891365,
      "grad_norm": 4.131698131561279,
      "learning_rate": 2.5648095508811823e-05,
      "loss": 1.0415,
      "step": 4602
    },
    {
      "epoch": 6.410863509749303,
      "grad_norm": 2.7010226249694824,
      "learning_rate": 2.563814667424673e-05,
      "loss": 0.9784,
      "step": 4603
    },
    {
      "epoch": 6.412256267409471,
      "grad_norm": 4.202706813812256,
      "learning_rate": 2.5628197839681635e-05,
      "loss": 0.9339,
      "step": 4604
    },
    {
      "epoch": 6.413649025069638,
      "grad_norm": 3.086254119873047,
      "learning_rate": 2.5618249005116542e-05,
      "loss": 0.8264,
      "step": 4605
    },
    {
      "epoch": 6.415041782729805,
      "grad_norm": 2.9942097663879395,
      "learning_rate": 2.560830017055145e-05,
      "loss": 1.0963,
      "step": 4606
    },
    {
      "epoch": 6.4164345403899725,
      "grad_norm": 3.093679428100586,
      "learning_rate": 2.5598351335986354e-05,
      "loss": 1.1648,
      "step": 4607
    },
    {
      "epoch": 6.417827298050139,
      "grad_norm": 2.9956912994384766,
      "learning_rate": 2.558840250142126e-05,
      "loss": 1.0576,
      "step": 4608
    },
    {
      "epoch": 6.419220055710307,
      "grad_norm": 3.0251107215881348,
      "learning_rate": 2.5578453666856165e-05,
      "loss": 0.9207,
      "step": 4609
    },
    {
      "epoch": 6.420612813370473,
      "grad_norm": 2.97255539894104,
      "learning_rate": 2.5568504832291073e-05,
      "loss": 1.0619,
      "step": 4610
    },
    {
      "epoch": 6.422005571030641,
      "grad_norm": 2.7506778240203857,
      "learning_rate": 2.555855599772598e-05,
      "loss": 1.0987,
      "step": 4611
    },
    {
      "epoch": 6.423398328690808,
      "grad_norm": 2.845851182937622,
      "learning_rate": 2.5548607163160884e-05,
      "loss": 0.9654,
      "step": 4612
    },
    {
      "epoch": 6.424791086350975,
      "grad_norm": 2.7043139934539795,
      "learning_rate": 2.553865832859579e-05,
      "loss": 0.962,
      "step": 4613
    },
    {
      "epoch": 6.426183844011142,
      "grad_norm": 3.166102170944214,
      "learning_rate": 2.5528709494030696e-05,
      "loss": 1.2582,
      "step": 4614
    },
    {
      "epoch": 6.427576601671309,
      "grad_norm": 2.819932460784912,
      "learning_rate": 2.5518760659465603e-05,
      "loss": 1.0251,
      "step": 4615
    },
    {
      "epoch": 6.428969359331476,
      "grad_norm": 4.832479476928711,
      "learning_rate": 2.550881182490051e-05,
      "loss": 0.9833,
      "step": 4616
    },
    {
      "epoch": 6.430362116991644,
      "grad_norm": 2.6273465156555176,
      "learning_rate": 2.5498862990335415e-05,
      "loss": 0.8972,
      "step": 4617
    },
    {
      "epoch": 6.43175487465181,
      "grad_norm": 3.3082454204559326,
      "learning_rate": 2.5488914155770322e-05,
      "loss": 1.0965,
      "step": 4618
    },
    {
      "epoch": 6.433147632311978,
      "grad_norm": 3.153756856918335,
      "learning_rate": 2.5478965321205226e-05,
      "loss": 1.1271,
      "step": 4619
    },
    {
      "epoch": 6.4345403899721445,
      "grad_norm": 3.0597243309020996,
      "learning_rate": 2.5469016486640134e-05,
      "loss": 1.1421,
      "step": 4620
    },
    {
      "epoch": 6.435933147632312,
      "grad_norm": 2.59921932220459,
      "learning_rate": 2.5459067652075038e-05,
      "loss": 0.892,
      "step": 4621
    },
    {
      "epoch": 6.437325905292479,
      "grad_norm": 2.940541982650757,
      "learning_rate": 2.5449118817509945e-05,
      "loss": 1.1599,
      "step": 4622
    },
    {
      "epoch": 6.438718662952646,
      "grad_norm": 2.700190782546997,
      "learning_rate": 2.5439169982944853e-05,
      "loss": 0.8381,
      "step": 4623
    },
    {
      "epoch": 6.440111420612814,
      "grad_norm": 2.8738365173339844,
      "learning_rate": 2.5429221148379757e-05,
      "loss": 1.0608,
      "step": 4624
    },
    {
      "epoch": 6.4415041782729805,
      "grad_norm": 3.1238081455230713,
      "learning_rate": 2.5419272313814668e-05,
      "loss": 1.0864,
      "step": 4625
    },
    {
      "epoch": 6.442896935933147,
      "grad_norm": 2.8366260528564453,
      "learning_rate": 2.5409323479249568e-05,
      "loss": 0.9429,
      "step": 4626
    },
    {
      "epoch": 6.444289693593315,
      "grad_norm": 3.097949981689453,
      "learning_rate": 2.539937464468448e-05,
      "loss": 1.0946,
      "step": 4627
    },
    {
      "epoch": 6.445682451253482,
      "grad_norm": 4.2688422203063965,
      "learning_rate": 2.5389425810119387e-05,
      "loss": 1.0365,
      "step": 4628
    },
    {
      "epoch": 6.447075208913649,
      "grad_norm": 2.8227672576904297,
      "learning_rate": 2.537947697555429e-05,
      "loss": 1.1122,
      "step": 4629
    },
    {
      "epoch": 6.4484679665738165,
      "grad_norm": 3.191445827484131,
      "learning_rate": 2.5369528140989198e-05,
      "loss": 1.275,
      "step": 4630
    },
    {
      "epoch": 6.449860724233983,
      "grad_norm": 3.0683882236480713,
      "learning_rate": 2.5359579306424102e-05,
      "loss": 1.0693,
      "step": 4631
    },
    {
      "epoch": 6.451253481894151,
      "grad_norm": 2.7502219676971436,
      "learning_rate": 2.534963047185901e-05,
      "loss": 0.9373,
      "step": 4632
    },
    {
      "epoch": 6.452646239554317,
      "grad_norm": 2.8942861557006836,
      "learning_rate": 2.5339681637293917e-05,
      "loss": 1.1393,
      "step": 4633
    },
    {
      "epoch": 6.454038997214485,
      "grad_norm": 2.999723196029663,
      "learning_rate": 2.532973280272882e-05,
      "loss": 0.8987,
      "step": 4634
    },
    {
      "epoch": 6.455431754874652,
      "grad_norm": 2.8345420360565186,
      "learning_rate": 2.531978396816373e-05,
      "loss": 1.0373,
      "step": 4635
    },
    {
      "epoch": 6.456824512534819,
      "grad_norm": 2.9698808193206787,
      "learning_rate": 2.5309835133598633e-05,
      "loss": 1.2927,
      "step": 4636
    },
    {
      "epoch": 6.458217270194986,
      "grad_norm": 2.84393048286438,
      "learning_rate": 2.529988629903354e-05,
      "loss": 0.9664,
      "step": 4637
    },
    {
      "epoch": 6.459610027855153,
      "grad_norm": 2.92636775970459,
      "learning_rate": 2.5289937464468448e-05,
      "loss": 1.1069,
      "step": 4638
    },
    {
      "epoch": 6.46100278551532,
      "grad_norm": 3.5068624019622803,
      "learning_rate": 2.527998862990335e-05,
      "loss": 0.959,
      "step": 4639
    },
    {
      "epoch": 6.462395543175488,
      "grad_norm": 2.868551015853882,
      "learning_rate": 2.527003979533826e-05,
      "loss": 1.1338,
      "step": 4640
    },
    {
      "epoch": 6.463788300835654,
      "grad_norm": 3.7234344482421875,
      "learning_rate": 2.5260090960773163e-05,
      "loss": 0.8658,
      "step": 4641
    },
    {
      "epoch": 6.465181058495822,
      "grad_norm": 3.3896803855895996,
      "learning_rate": 2.525014212620807e-05,
      "loss": 1.1506,
      "step": 4642
    },
    {
      "epoch": 6.4665738161559885,
      "grad_norm": 2.8069541454315186,
      "learning_rate": 2.5240193291642975e-05,
      "loss": 1.0095,
      "step": 4643
    },
    {
      "epoch": 6.467966573816156,
      "grad_norm": 2.879835605621338,
      "learning_rate": 2.5230244457077882e-05,
      "loss": 1.1378,
      "step": 4644
    },
    {
      "epoch": 6.469359331476323,
      "grad_norm": 3.1890411376953125,
      "learning_rate": 2.522029562251279e-05,
      "loss": 1.3044,
      "step": 4645
    },
    {
      "epoch": 6.47075208913649,
      "grad_norm": 3.6902568340301514,
      "learning_rate": 2.5210346787947694e-05,
      "loss": 1.1575,
      "step": 4646
    },
    {
      "epoch": 6.472144846796658,
      "grad_norm": 3.353255033493042,
      "learning_rate": 2.52003979533826e-05,
      "loss": 1.1234,
      "step": 4647
    },
    {
      "epoch": 6.4735376044568245,
      "grad_norm": 3.339797258377075,
      "learning_rate": 2.5190449118817505e-05,
      "loss": 1.3185,
      "step": 4648
    },
    {
      "epoch": 6.474930362116992,
      "grad_norm": 2.759061813354492,
      "learning_rate": 2.5180500284252413e-05,
      "loss": 0.8372,
      "step": 4649
    },
    {
      "epoch": 6.476323119777159,
      "grad_norm": 3.0016627311706543,
      "learning_rate": 2.517055144968732e-05,
      "loss": 0.9608,
      "step": 4650
    },
    {
      "epoch": 6.477715877437326,
      "grad_norm": 3.210092782974243,
      "learning_rate": 2.5160602615122224e-05,
      "loss": 1.1231,
      "step": 4651
    },
    {
      "epoch": 6.479108635097493,
      "grad_norm": 2.6207337379455566,
      "learning_rate": 2.5150653780557132e-05,
      "loss": 0.8677,
      "step": 4652
    },
    {
      "epoch": 6.4805013927576605,
      "grad_norm": 2.916144847869873,
      "learning_rate": 2.5140704945992036e-05,
      "loss": 1.0909,
      "step": 4653
    },
    {
      "epoch": 6.481894150417827,
      "grad_norm": 3.0932586193084717,
      "learning_rate": 2.5130756111426947e-05,
      "loss": 1.2029,
      "step": 4654
    },
    {
      "epoch": 6.483286908077995,
      "grad_norm": 2.834848403930664,
      "learning_rate": 2.5120807276861854e-05,
      "loss": 1.0181,
      "step": 4655
    },
    {
      "epoch": 6.484679665738161,
      "grad_norm": 3.0963003635406494,
      "learning_rate": 2.5110858442296758e-05,
      "loss": 0.9916,
      "step": 4656
    },
    {
      "epoch": 6.486072423398329,
      "grad_norm": 2.952533721923828,
      "learning_rate": 2.5100909607731666e-05,
      "loss": 0.9978,
      "step": 4657
    },
    {
      "epoch": 6.487465181058496,
      "grad_norm": 3.1474313735961914,
      "learning_rate": 2.509096077316657e-05,
      "loss": 1.1436,
      "step": 4658
    },
    {
      "epoch": 6.488857938718663,
      "grad_norm": 2.6717944145202637,
      "learning_rate": 2.5081011938601477e-05,
      "loss": 0.8815,
      "step": 4659
    },
    {
      "epoch": 6.49025069637883,
      "grad_norm": 3.0935699939727783,
      "learning_rate": 2.507106310403638e-05,
      "loss": 1.1013,
      "step": 4660
    },
    {
      "epoch": 6.491643454038997,
      "grad_norm": 3.12080454826355,
      "learning_rate": 2.506111426947129e-05,
      "loss": 1.098,
      "step": 4661
    },
    {
      "epoch": 6.493036211699164,
      "grad_norm": 3.0284221172332764,
      "learning_rate": 2.5051165434906196e-05,
      "loss": 1.1351,
      "step": 4662
    },
    {
      "epoch": 6.494428969359332,
      "grad_norm": 3.2609972953796387,
      "learning_rate": 2.50412166003411e-05,
      "loss": 1.2471,
      "step": 4663
    },
    {
      "epoch": 6.495821727019498,
      "grad_norm": 3.3732669353485107,
      "learning_rate": 2.5031267765776008e-05,
      "loss": 0.9781,
      "step": 4664
    },
    {
      "epoch": 6.497214484679666,
      "grad_norm": 3.2345945835113525,
      "learning_rate": 2.5021318931210912e-05,
      "loss": 1.2421,
      "step": 4665
    },
    {
      "epoch": 6.498607242339833,
      "grad_norm": 3.080406904220581,
      "learning_rate": 2.501137009664582e-05,
      "loss": 1.0013,
      "step": 4666
    },
    {
      "epoch": 6.5,
      "grad_norm": 4.212028980255127,
      "learning_rate": 2.5001421262080727e-05,
      "loss": 0.8842,
      "step": 4667
    },
    {
      "epoch": 6.501392757660167,
      "grad_norm": 2.7986233234405518,
      "learning_rate": 2.499147242751563e-05,
      "loss": 1.0137,
      "step": 4668
    },
    {
      "epoch": 6.502785515320334,
      "grad_norm": 3.28351092338562,
      "learning_rate": 2.4981523592950538e-05,
      "loss": 1.0776,
      "step": 4669
    },
    {
      "epoch": 6.504178272980502,
      "grad_norm": 2.7406668663024902,
      "learning_rate": 2.4971574758385442e-05,
      "loss": 1.0218,
      "step": 4670
    },
    {
      "epoch": 6.505571030640668,
      "grad_norm": 2.6849937438964844,
      "learning_rate": 2.496162592382035e-05,
      "loss": 0.8841,
      "step": 4671
    },
    {
      "epoch": 6.506963788300836,
      "grad_norm": 3.0540127754211426,
      "learning_rate": 2.4951677089255257e-05,
      "loss": 0.9514,
      "step": 4672
    },
    {
      "epoch": 6.508356545961003,
      "grad_norm": 3.3202834129333496,
      "learning_rate": 2.494172825469016e-05,
      "loss": 1.1729,
      "step": 4673
    },
    {
      "epoch": 6.50974930362117,
      "grad_norm": 3.6286327838897705,
      "learning_rate": 2.493177942012507e-05,
      "loss": 0.9764,
      "step": 4674
    },
    {
      "epoch": 6.511142061281337,
      "grad_norm": 3.012465238571167,
      "learning_rate": 2.4921830585559973e-05,
      "loss": 1.1726,
      "step": 4675
    },
    {
      "epoch": 6.512534818941504,
      "grad_norm": 3.3141849040985107,
      "learning_rate": 2.491188175099488e-05,
      "loss": 1.1859,
      "step": 4676
    },
    {
      "epoch": 6.513927576601671,
      "grad_norm": 3.096423625946045,
      "learning_rate": 2.4901932916429788e-05,
      "loss": 1.1932,
      "step": 4677
    },
    {
      "epoch": 6.515320334261839,
      "grad_norm": 2.589930772781372,
      "learning_rate": 2.4891984081864692e-05,
      "loss": 0.7744,
      "step": 4678
    },
    {
      "epoch": 6.516713091922005,
      "grad_norm": 2.926628351211548,
      "learning_rate": 2.48820352472996e-05,
      "loss": 0.8675,
      "step": 4679
    },
    {
      "epoch": 6.518105849582173,
      "grad_norm": 3.745406150817871,
      "learning_rate": 2.4872086412734503e-05,
      "loss": 1.1075,
      "step": 4680
    },
    {
      "epoch": 6.5194986072423395,
      "grad_norm": 3.591724157333374,
      "learning_rate": 2.4862137578169414e-05,
      "loss": 0.9095,
      "step": 4681
    },
    {
      "epoch": 6.520891364902507,
      "grad_norm": 2.5653629302978516,
      "learning_rate": 2.4852188743604315e-05,
      "loss": 0.9503,
      "step": 4682
    },
    {
      "epoch": 6.522284122562674,
      "grad_norm": 3.108179807662964,
      "learning_rate": 2.4842239909039226e-05,
      "loss": 1.1386,
      "step": 4683
    },
    {
      "epoch": 6.523676880222841,
      "grad_norm": 3.205650806427002,
      "learning_rate": 2.4832291074474133e-05,
      "loss": 1.0266,
      "step": 4684
    },
    {
      "epoch": 6.525069637883008,
      "grad_norm": 2.7524988651275635,
      "learning_rate": 2.4822342239909037e-05,
      "loss": 1.0027,
      "step": 4685
    },
    {
      "epoch": 6.5264623955431755,
      "grad_norm": 2.7650883197784424,
      "learning_rate": 2.4812393405343945e-05,
      "loss": 0.9588,
      "step": 4686
    },
    {
      "epoch": 6.527855153203342,
      "grad_norm": 3.339064598083496,
      "learning_rate": 2.480244457077885e-05,
      "loss": 1.1757,
      "step": 4687
    },
    {
      "epoch": 6.52924791086351,
      "grad_norm": 3.030686140060425,
      "learning_rate": 2.4792495736213756e-05,
      "loss": 0.9491,
      "step": 4688
    },
    {
      "epoch": 6.530640668523677,
      "grad_norm": 3.2274553775787354,
      "learning_rate": 2.4782546901648664e-05,
      "loss": 1.2159,
      "step": 4689
    },
    {
      "epoch": 6.532033426183844,
      "grad_norm": 3.967367172241211,
      "learning_rate": 2.4772598067083568e-05,
      "loss": 1.3414,
      "step": 4690
    },
    {
      "epoch": 6.5334261838440115,
      "grad_norm": 3.752794027328491,
      "learning_rate": 2.4762649232518475e-05,
      "loss": 1.1098,
      "step": 4691
    },
    {
      "epoch": 6.534818941504178,
      "grad_norm": 3.29585337638855,
      "learning_rate": 2.475270039795338e-05,
      "loss": 1.1735,
      "step": 4692
    },
    {
      "epoch": 6.536211699164346,
      "grad_norm": 3.18642520904541,
      "learning_rate": 2.4742751563388287e-05,
      "loss": 1.3173,
      "step": 4693
    },
    {
      "epoch": 6.537604456824512,
      "grad_norm": 2.4995405673980713,
      "learning_rate": 2.4732802728823194e-05,
      "loss": 0.8807,
      "step": 4694
    },
    {
      "epoch": 6.53899721448468,
      "grad_norm": 2.5437867641448975,
      "learning_rate": 2.47228538942581e-05,
      "loss": 0.9895,
      "step": 4695
    },
    {
      "epoch": 6.540389972144847,
      "grad_norm": 3.1137325763702393,
      "learning_rate": 2.4712905059693006e-05,
      "loss": 1.2948,
      "step": 4696
    },
    {
      "epoch": 6.541782729805014,
      "grad_norm": 3.344264507293701,
      "learning_rate": 2.470295622512791e-05,
      "loss": 1.3989,
      "step": 4697
    },
    {
      "epoch": 6.543175487465181,
      "grad_norm": 3.345322370529175,
      "learning_rate": 2.4693007390562817e-05,
      "loss": 1.2011,
      "step": 4698
    },
    {
      "epoch": 6.544568245125348,
      "grad_norm": 2.8151438236236572,
      "learning_rate": 2.4683058555997725e-05,
      "loss": 0.9592,
      "step": 4699
    },
    {
      "epoch": 6.545961002785515,
      "grad_norm": 2.728134870529175,
      "learning_rate": 2.467310972143263e-05,
      "loss": 0.9775,
      "step": 4700
    },
    {
      "epoch": 6.547353760445683,
      "grad_norm": 2.9509196281433105,
      "learning_rate": 2.4663160886867536e-05,
      "loss": 1.1995,
      "step": 4701
    },
    {
      "epoch": 6.548746518105849,
      "grad_norm": 3.4511666297912598,
      "learning_rate": 2.465321205230244e-05,
      "loss": 1.1744,
      "step": 4702
    },
    {
      "epoch": 6.550139275766017,
      "grad_norm": 2.9215309619903564,
      "learning_rate": 2.4643263217737348e-05,
      "loss": 0.9128,
      "step": 4703
    },
    {
      "epoch": 6.5515320334261835,
      "grad_norm": 2.8722798824310303,
      "learning_rate": 2.4633314383172252e-05,
      "loss": 0.9149,
      "step": 4704
    },
    {
      "epoch": 6.552924791086351,
      "grad_norm": 2.5877463817596436,
      "learning_rate": 2.462336554860716e-05,
      "loss": 0.8475,
      "step": 4705
    },
    {
      "epoch": 6.554317548746518,
      "grad_norm": 3.049581527709961,
      "learning_rate": 2.4613416714042067e-05,
      "loss": 0.9758,
      "step": 4706
    },
    {
      "epoch": 6.555710306406685,
      "grad_norm": 4.076055526733398,
      "learning_rate": 2.460346787947697e-05,
      "loss": 0.9216,
      "step": 4707
    },
    {
      "epoch": 6.557103064066853,
      "grad_norm": 2.760948657989502,
      "learning_rate": 2.459351904491188e-05,
      "loss": 0.8467,
      "step": 4708
    },
    {
      "epoch": 6.5584958217270195,
      "grad_norm": 2.908273458480835,
      "learning_rate": 2.4583570210346783e-05,
      "loss": 0.9611,
      "step": 4709
    },
    {
      "epoch": 6.559888579387186,
      "grad_norm": 2.923372268676758,
      "learning_rate": 2.4573621375781693e-05,
      "loss": 1.0828,
      "step": 4710
    },
    {
      "epoch": 6.561281337047354,
      "grad_norm": 3.5005199909210205,
      "learning_rate": 2.45636725412166e-05,
      "loss": 1.0053,
      "step": 4711
    },
    {
      "epoch": 6.562674094707521,
      "grad_norm": 3.210413694381714,
      "learning_rate": 2.4553723706651505e-05,
      "loss": 1.1403,
      "step": 4712
    },
    {
      "epoch": 6.564066852367688,
      "grad_norm": 3.584632635116577,
      "learning_rate": 2.4543774872086412e-05,
      "loss": 1.1124,
      "step": 4713
    },
    {
      "epoch": 6.5654596100278555,
      "grad_norm": 3.279512882232666,
      "learning_rate": 2.4533826037521317e-05,
      "loss": 1.1136,
      "step": 4714
    },
    {
      "epoch": 6.566852367688022,
      "grad_norm": 3.1212546825408936,
      "learning_rate": 2.4523877202956224e-05,
      "loss": 0.9316,
      "step": 4715
    },
    {
      "epoch": 6.56824512534819,
      "grad_norm": 2.8250529766082764,
      "learning_rate": 2.451392836839113e-05,
      "loss": 1.0379,
      "step": 4716
    },
    {
      "epoch": 6.569637883008356,
      "grad_norm": 3.100564479827881,
      "learning_rate": 2.4503979533826036e-05,
      "loss": 1.0759,
      "step": 4717
    },
    {
      "epoch": 6.571030640668524,
      "grad_norm": 3.492800235748291,
      "learning_rate": 2.4494030699260943e-05,
      "loss": 1.0498,
      "step": 4718
    },
    {
      "epoch": 6.572423398328691,
      "grad_norm": 2.630836009979248,
      "learning_rate": 2.4484081864695847e-05,
      "loss": 0.8225,
      "step": 4719
    },
    {
      "epoch": 6.573816155988858,
      "grad_norm": 2.584005117416382,
      "learning_rate": 2.4474133030130755e-05,
      "loss": 1.018,
      "step": 4720
    },
    {
      "epoch": 6.575208913649025,
      "grad_norm": 2.8945672512054443,
      "learning_rate": 2.4464184195565662e-05,
      "loss": 0.9315,
      "step": 4721
    },
    {
      "epoch": 6.576601671309192,
      "grad_norm": 3.119586229324341,
      "learning_rate": 2.4454235361000566e-05,
      "loss": 1.2142,
      "step": 4722
    },
    {
      "epoch": 6.577994428969359,
      "grad_norm": 3.0555262565612793,
      "learning_rate": 2.4444286526435474e-05,
      "loss": 0.9332,
      "step": 4723
    },
    {
      "epoch": 6.579387186629527,
      "grad_norm": 2.706542730331421,
      "learning_rate": 2.4434337691870378e-05,
      "loss": 0.9654,
      "step": 4724
    },
    {
      "epoch": 6.580779944289693,
      "grad_norm": 3.118386745452881,
      "learning_rate": 2.4424388857305285e-05,
      "loss": 1.4078,
      "step": 4725
    },
    {
      "epoch": 6.582172701949861,
      "grad_norm": 3.0359275341033936,
      "learning_rate": 2.441444002274019e-05,
      "loss": 1.1302,
      "step": 4726
    },
    {
      "epoch": 6.5835654596100275,
      "grad_norm": 2.9190561771392822,
      "learning_rate": 2.4404491188175097e-05,
      "loss": 1.0922,
      "step": 4727
    },
    {
      "epoch": 6.584958217270195,
      "grad_norm": 2.814833641052246,
      "learning_rate": 2.4394542353610004e-05,
      "loss": 0.8571,
      "step": 4728
    },
    {
      "epoch": 6.586350974930362,
      "grad_norm": 2.996880292892456,
      "learning_rate": 2.4384593519044908e-05,
      "loss": 0.8809,
      "step": 4729
    },
    {
      "epoch": 6.587743732590529,
      "grad_norm": 3.106786012649536,
      "learning_rate": 2.4374644684479816e-05,
      "loss": 1.0673,
      "step": 4730
    },
    {
      "epoch": 6.589136490250697,
      "grad_norm": 2.694575309753418,
      "learning_rate": 2.436469584991472e-05,
      "loss": 1.0702,
      "step": 4731
    },
    {
      "epoch": 6.5905292479108635,
      "grad_norm": 2.9143412113189697,
      "learning_rate": 2.4354747015349627e-05,
      "loss": 0.9471,
      "step": 4732
    },
    {
      "epoch": 6.591922005571031,
      "grad_norm": 2.377490282058716,
      "learning_rate": 2.4344798180784535e-05,
      "loss": 0.6882,
      "step": 4733
    },
    {
      "epoch": 6.593314763231198,
      "grad_norm": 2.679741144180298,
      "learning_rate": 2.433484934621944e-05,
      "loss": 0.8205,
      "step": 4734
    },
    {
      "epoch": 6.594707520891365,
      "grad_norm": 2.5864362716674805,
      "learning_rate": 2.4324900511654346e-05,
      "loss": 1.0964,
      "step": 4735
    },
    {
      "epoch": 6.596100278551532,
      "grad_norm": 2.768282175064087,
      "learning_rate": 2.431495167708925e-05,
      "loss": 0.8348,
      "step": 4736
    },
    {
      "epoch": 6.5974930362116995,
      "grad_norm": 3.049933910369873,
      "learning_rate": 2.430500284252416e-05,
      "loss": 1.1628,
      "step": 4737
    },
    {
      "epoch": 6.598885793871866,
      "grad_norm": 3.35349440574646,
      "learning_rate": 2.429505400795907e-05,
      "loss": 1.2172,
      "step": 4738
    },
    {
      "epoch": 6.600278551532034,
      "grad_norm": 3.1047682762145996,
      "learning_rate": 2.4285105173393973e-05,
      "loss": 0.9607,
      "step": 4739
    },
    {
      "epoch": 6.6016713091922,
      "grad_norm": 2.599682331085205,
      "learning_rate": 2.427515633882888e-05,
      "loss": 0.8289,
      "step": 4740
    },
    {
      "epoch": 6.603064066852368,
      "grad_norm": 2.819077491760254,
      "learning_rate": 2.4265207504263784e-05,
      "loss": 1.2117,
      "step": 4741
    },
    {
      "epoch": 6.604456824512535,
      "grad_norm": 3.0849640369415283,
      "learning_rate": 2.425525866969869e-05,
      "loss": 1.0484,
      "step": 4742
    },
    {
      "epoch": 6.605849582172702,
      "grad_norm": 3.0797269344329834,
      "learning_rate": 2.42453098351336e-05,
      "loss": 0.9468,
      "step": 4743
    },
    {
      "epoch": 6.607242339832869,
      "grad_norm": 2.7996320724487305,
      "learning_rate": 2.4235361000568503e-05,
      "loss": 0.8944,
      "step": 4744
    },
    {
      "epoch": 6.608635097493036,
      "grad_norm": 2.971978187561035,
      "learning_rate": 2.422541216600341e-05,
      "loss": 1.0863,
      "step": 4745
    },
    {
      "epoch": 6.610027855153203,
      "grad_norm": 2.5103025436401367,
      "learning_rate": 2.4215463331438315e-05,
      "loss": 0.7514,
      "step": 4746
    },
    {
      "epoch": 6.611420612813371,
      "grad_norm": 3.3744678497314453,
      "learning_rate": 2.4205514496873222e-05,
      "loss": 0.7911,
      "step": 4747
    },
    {
      "epoch": 6.612813370473537,
      "grad_norm": 2.7324295043945312,
      "learning_rate": 2.4195565662308126e-05,
      "loss": 0.8992,
      "step": 4748
    },
    {
      "epoch": 6.614206128133705,
      "grad_norm": 2.771432876586914,
      "learning_rate": 2.4185616827743034e-05,
      "loss": 0.9506,
      "step": 4749
    },
    {
      "epoch": 6.615598885793872,
      "grad_norm": 3.6363251209259033,
      "learning_rate": 2.417566799317794e-05,
      "loss": 1.2449,
      "step": 4750
    },
    {
      "epoch": 6.616991643454039,
      "grad_norm": 3.2286977767944336,
      "learning_rate": 2.4165719158612845e-05,
      "loss": 1.2933,
      "step": 4751
    },
    {
      "epoch": 6.618384401114206,
      "grad_norm": 2.9055263996124268,
      "learning_rate": 2.4155770324047753e-05,
      "loss": 1.1623,
      "step": 4752
    },
    {
      "epoch": 6.619777158774373,
      "grad_norm": 3.2123911380767822,
      "learning_rate": 2.4145821489482657e-05,
      "loss": 1.0442,
      "step": 4753
    },
    {
      "epoch": 6.621169916434541,
      "grad_norm": 3.097463607788086,
      "learning_rate": 2.4135872654917564e-05,
      "loss": 0.9573,
      "step": 4754
    },
    {
      "epoch": 6.6225626740947074,
      "grad_norm": 2.822230815887451,
      "learning_rate": 2.4125923820352472e-05,
      "loss": 1.026,
      "step": 4755
    },
    {
      "epoch": 6.623955431754875,
      "grad_norm": 3.2758986949920654,
      "learning_rate": 2.4115974985787376e-05,
      "loss": 1.0508,
      "step": 4756
    },
    {
      "epoch": 6.625348189415042,
      "grad_norm": 2.6393706798553467,
      "learning_rate": 2.4106026151222283e-05,
      "loss": 0.8872,
      "step": 4757
    },
    {
      "epoch": 6.626740947075209,
      "grad_norm": 2.6567206382751465,
      "learning_rate": 2.4096077316657187e-05,
      "loss": 0.9524,
      "step": 4758
    },
    {
      "epoch": 6.628133704735376,
      "grad_norm": 3.7393648624420166,
      "learning_rate": 2.4086128482092095e-05,
      "loss": 1.002,
      "step": 4759
    },
    {
      "epoch": 6.629526462395543,
      "grad_norm": 2.7445218563079834,
      "learning_rate": 2.4076179647527002e-05,
      "loss": 0.9604,
      "step": 4760
    },
    {
      "epoch": 6.63091922005571,
      "grad_norm": 2.7881252765655518,
      "learning_rate": 2.4066230812961906e-05,
      "loss": 0.8748,
      "step": 4761
    },
    {
      "epoch": 6.632311977715878,
      "grad_norm": 2.8022923469543457,
      "learning_rate": 2.4056281978396814e-05,
      "loss": 0.9567,
      "step": 4762
    },
    {
      "epoch": 6.633704735376044,
      "grad_norm": 2.6584715843200684,
      "learning_rate": 2.4046333143831718e-05,
      "loss": 0.8187,
      "step": 4763
    },
    {
      "epoch": 6.635097493036212,
      "grad_norm": 2.696336507797241,
      "learning_rate": 2.4036384309266625e-05,
      "loss": 0.9962,
      "step": 4764
    },
    {
      "epoch": 6.6364902506963785,
      "grad_norm": 2.634849786758423,
      "learning_rate": 2.4026435474701536e-05,
      "loss": 0.9339,
      "step": 4765
    },
    {
      "epoch": 6.637883008356546,
      "grad_norm": 2.7696452140808105,
      "learning_rate": 2.401648664013644e-05,
      "loss": 0.9384,
      "step": 4766
    },
    {
      "epoch": 6.639275766016713,
      "grad_norm": 2.92993426322937,
      "learning_rate": 2.4006537805571348e-05,
      "loss": 1.0852,
      "step": 4767
    },
    {
      "epoch": 6.64066852367688,
      "grad_norm": 3.3232686519622803,
      "learning_rate": 2.3996588971006252e-05,
      "loss": 1.1916,
      "step": 4768
    },
    {
      "epoch": 6.642061281337047,
      "grad_norm": 4.471713066101074,
      "learning_rate": 2.398664013644116e-05,
      "loss": 1.1487,
      "step": 4769
    },
    {
      "epoch": 6.6434540389972145,
      "grad_norm": 2.7921526432037354,
      "learning_rate": 2.3976691301876063e-05,
      "loss": 0.9126,
      "step": 4770
    },
    {
      "epoch": 6.644846796657381,
      "grad_norm": 3.0503218173980713,
      "learning_rate": 2.396674246731097e-05,
      "loss": 1.0404,
      "step": 4771
    },
    {
      "epoch": 6.646239554317549,
      "grad_norm": 2.566943407058716,
      "learning_rate": 2.3956793632745878e-05,
      "loss": 0.8334,
      "step": 4772
    },
    {
      "epoch": 6.647632311977716,
      "grad_norm": 3.2688379287719727,
      "learning_rate": 2.3946844798180782e-05,
      "loss": 0.947,
      "step": 4773
    },
    {
      "epoch": 6.649025069637883,
      "grad_norm": 3.023160457611084,
      "learning_rate": 2.393689596361569e-05,
      "loss": 1.06,
      "step": 4774
    },
    {
      "epoch": 6.65041782729805,
      "grad_norm": 3.023071527481079,
      "learning_rate": 2.3926947129050594e-05,
      "loss": 1.0209,
      "step": 4775
    },
    {
      "epoch": 6.651810584958217,
      "grad_norm": 2.9008774757385254,
      "learning_rate": 2.39169982944855e-05,
      "loss": 1.1054,
      "step": 4776
    },
    {
      "epoch": 6.653203342618385,
      "grad_norm": 2.7631795406341553,
      "learning_rate": 2.390704945992041e-05,
      "loss": 0.9356,
      "step": 4777
    },
    {
      "epoch": 6.654596100278551,
      "grad_norm": 2.6900882720947266,
      "learning_rate": 2.3897100625355313e-05,
      "loss": 0.8931,
      "step": 4778
    },
    {
      "epoch": 6.655988857938719,
      "grad_norm": 2.8253777027130127,
      "learning_rate": 2.388715179079022e-05,
      "loss": 1.0883,
      "step": 4779
    },
    {
      "epoch": 6.657381615598886,
      "grad_norm": 2.7304461002349854,
      "learning_rate": 2.3877202956225124e-05,
      "loss": 1.068,
      "step": 4780
    },
    {
      "epoch": 6.658774373259053,
      "grad_norm": 3.2457969188690186,
      "learning_rate": 2.3867254121660032e-05,
      "loss": 1.0144,
      "step": 4781
    },
    {
      "epoch": 6.66016713091922,
      "grad_norm": 3.1068785190582275,
      "learning_rate": 2.385730528709494e-05,
      "loss": 0.962,
      "step": 4782
    },
    {
      "epoch": 6.661559888579387,
      "grad_norm": 3.2889511585235596,
      "learning_rate": 2.3847356452529843e-05,
      "loss": 1.0686,
      "step": 4783
    },
    {
      "epoch": 6.662952646239554,
      "grad_norm": 3.2533276081085205,
      "learning_rate": 2.383740761796475e-05,
      "loss": 0.9554,
      "step": 4784
    },
    {
      "epoch": 6.664345403899722,
      "grad_norm": 3.1829142570495605,
      "learning_rate": 2.3827458783399655e-05,
      "loss": 1.0605,
      "step": 4785
    },
    {
      "epoch": 6.665738161559888,
      "grad_norm": 2.736853837966919,
      "learning_rate": 2.3817509948834562e-05,
      "loss": 0.9768,
      "step": 4786
    },
    {
      "epoch": 6.667130919220056,
      "grad_norm": 3.2278525829315186,
      "learning_rate": 2.380756111426947e-05,
      "loss": 1.0798,
      "step": 4787
    },
    {
      "epoch": 6.6685236768802225,
      "grad_norm": 3.094557762145996,
      "learning_rate": 2.3797612279704374e-05,
      "loss": 1.0931,
      "step": 4788
    },
    {
      "epoch": 6.66991643454039,
      "grad_norm": 2.9677906036376953,
      "learning_rate": 2.378766344513928e-05,
      "loss": 1.1247,
      "step": 4789
    },
    {
      "epoch": 6.671309192200557,
      "grad_norm": 2.5754425525665283,
      "learning_rate": 2.3777714610574185e-05,
      "loss": 0.7597,
      "step": 4790
    },
    {
      "epoch": 6.672701949860724,
      "grad_norm": 2.9235668182373047,
      "learning_rate": 2.3767765776009093e-05,
      "loss": 1.137,
      "step": 4791
    },
    {
      "epoch": 6.674094707520892,
      "grad_norm": 3.0413553714752197,
      "learning_rate": 2.3757816941443997e-05,
      "loss": 1.1092,
      "step": 4792
    },
    {
      "epoch": 6.6754874651810585,
      "grad_norm": 2.9375197887420654,
      "learning_rate": 2.3747868106878904e-05,
      "loss": 1.0091,
      "step": 4793
    },
    {
      "epoch": 6.676880222841225,
      "grad_norm": 2.6702022552490234,
      "learning_rate": 2.3737919272313815e-05,
      "loss": 0.9394,
      "step": 4794
    },
    {
      "epoch": 6.678272980501393,
      "grad_norm": 7.281650066375732,
      "learning_rate": 2.372797043774872e-05,
      "loss": 1.1282,
      "step": 4795
    },
    {
      "epoch": 6.67966573816156,
      "grad_norm": 2.7565412521362305,
      "learning_rate": 2.3718021603183627e-05,
      "loss": 0.9102,
      "step": 4796
    },
    {
      "epoch": 6.681058495821727,
      "grad_norm": 3.12174391746521,
      "learning_rate": 2.370807276861853e-05,
      "loss": 1.1007,
      "step": 4797
    },
    {
      "epoch": 6.6824512534818945,
      "grad_norm": 3.0235435962677,
      "learning_rate": 2.369812393405344e-05,
      "loss": 1.1611,
      "step": 4798
    },
    {
      "epoch": 6.683844011142061,
      "grad_norm": 2.963348388671875,
      "learning_rate": 2.3688175099488346e-05,
      "loss": 0.9288,
      "step": 4799
    },
    {
      "epoch": 6.685236768802229,
      "grad_norm": 3.0740180015563965,
      "learning_rate": 2.367822626492325e-05,
      "loss": 0.9463,
      "step": 4800
    },
    {
      "epoch": 6.686629526462395,
      "grad_norm": 3.2976651191711426,
      "learning_rate": 2.3668277430358157e-05,
      "loss": 1.284,
      "step": 4801
    },
    {
      "epoch": 6.688022284122563,
      "grad_norm": 3.295809745788574,
      "learning_rate": 2.365832859579306e-05,
      "loss": 1.0205,
      "step": 4802
    },
    {
      "epoch": 6.68941504178273,
      "grad_norm": 3.338949680328369,
      "learning_rate": 2.364837976122797e-05,
      "loss": 1.3288,
      "step": 4803
    },
    {
      "epoch": 6.690807799442897,
      "grad_norm": 3.4728505611419678,
      "learning_rate": 2.3638430926662876e-05,
      "loss": 0.9714,
      "step": 4804
    },
    {
      "epoch": 6.692200557103064,
      "grad_norm": 2.947736978530884,
      "learning_rate": 2.362848209209778e-05,
      "loss": 1.1931,
      "step": 4805
    },
    {
      "epoch": 6.693593314763231,
      "grad_norm": 2.6117374897003174,
      "learning_rate": 2.3618533257532688e-05,
      "loss": 0.8665,
      "step": 4806
    },
    {
      "epoch": 6.694986072423398,
      "grad_norm": 2.668964385986328,
      "learning_rate": 2.3608584422967592e-05,
      "loss": 0.8744,
      "step": 4807
    },
    {
      "epoch": 6.696378830083566,
      "grad_norm": 2.5390703678131104,
      "learning_rate": 2.35986355884025e-05,
      "loss": 1.0152,
      "step": 4808
    },
    {
      "epoch": 6.697771587743732,
      "grad_norm": 2.81025767326355,
      "learning_rate": 2.3588686753837407e-05,
      "loss": 0.9227,
      "step": 4809
    },
    {
      "epoch": 6.6991643454039,
      "grad_norm": 3.149915933609009,
      "learning_rate": 2.357873791927231e-05,
      "loss": 1.142,
      "step": 4810
    },
    {
      "epoch": 6.7005571030640665,
      "grad_norm": 3.086437702178955,
      "learning_rate": 2.356878908470722e-05,
      "loss": 1.2156,
      "step": 4811
    },
    {
      "epoch": 6.701949860724234,
      "grad_norm": 3.5764501094818115,
      "learning_rate": 2.3558840250142123e-05,
      "loss": 1.164,
      "step": 4812
    },
    {
      "epoch": 6.703342618384401,
      "grad_norm": 6.518548011779785,
      "learning_rate": 2.354889141557703e-05,
      "loss": 0.9738,
      "step": 4813
    },
    {
      "epoch": 6.704735376044568,
      "grad_norm": 2.80240797996521,
      "learning_rate": 2.3538942581011934e-05,
      "loss": 0.9904,
      "step": 4814
    },
    {
      "epoch": 6.706128133704736,
      "grad_norm": 3.277841091156006,
      "learning_rate": 2.352899374644684e-05,
      "loss": 0.9248,
      "step": 4815
    },
    {
      "epoch": 6.7075208913649025,
      "grad_norm": 3.215149164199829,
      "learning_rate": 2.351904491188175e-05,
      "loss": 0.9506,
      "step": 4816
    },
    {
      "epoch": 6.708913649025069,
      "grad_norm": 3.2901055812835693,
      "learning_rate": 2.3509096077316653e-05,
      "loss": 1.2405,
      "step": 4817
    },
    {
      "epoch": 6.710306406685237,
      "grad_norm": 2.942967653274536,
      "learning_rate": 2.349914724275156e-05,
      "loss": 1.1618,
      "step": 4818
    },
    {
      "epoch": 6.711699164345404,
      "grad_norm": 2.7844669818878174,
      "learning_rate": 2.3489198408186465e-05,
      "loss": 1.0405,
      "step": 4819
    },
    {
      "epoch": 6.713091922005571,
      "grad_norm": 3.1708924770355225,
      "learning_rate": 2.3479249573621372e-05,
      "loss": 1.3228,
      "step": 4820
    },
    {
      "epoch": 6.7144846796657385,
      "grad_norm": 3.003021240234375,
      "learning_rate": 2.3469300739056283e-05,
      "loss": 1.0084,
      "step": 4821
    },
    {
      "epoch": 6.715877437325905,
      "grad_norm": 3.2701704502105713,
      "learning_rate": 2.3459351904491187e-05,
      "loss": 1.1496,
      "step": 4822
    },
    {
      "epoch": 6.717270194986073,
      "grad_norm": 3.280702590942383,
      "learning_rate": 2.3449403069926094e-05,
      "loss": 0.8639,
      "step": 4823
    },
    {
      "epoch": 6.718662952646239,
      "grad_norm": 2.408778667449951,
      "learning_rate": 2.3439454235361e-05,
      "loss": 0.7839,
      "step": 4824
    },
    {
      "epoch": 6.720055710306407,
      "grad_norm": 3.651390314102173,
      "learning_rate": 2.3429505400795906e-05,
      "loss": 1.1629,
      "step": 4825
    },
    {
      "epoch": 6.721448467966574,
      "grad_norm": 3.0692644119262695,
      "learning_rate": 2.3419556566230813e-05,
      "loss": 1.2663,
      "step": 4826
    },
    {
      "epoch": 6.722841225626741,
      "grad_norm": 2.9470434188842773,
      "learning_rate": 2.3409607731665718e-05,
      "loss": 1.1087,
      "step": 4827
    },
    {
      "epoch": 6.724233983286908,
      "grad_norm": 2.759246826171875,
      "learning_rate": 2.3399658897100625e-05,
      "loss": 0.9773,
      "step": 4828
    },
    {
      "epoch": 6.725626740947075,
      "grad_norm": 2.8075499534606934,
      "learning_rate": 2.338971006253553e-05,
      "loss": 0.749,
      "step": 4829
    },
    {
      "epoch": 6.727019498607242,
      "grad_norm": 2.875091791152954,
      "learning_rate": 2.3379761227970437e-05,
      "loss": 1.0342,
      "step": 4830
    },
    {
      "epoch": 6.72841225626741,
      "grad_norm": 3.200989246368408,
      "learning_rate": 2.3369812393405344e-05,
      "loss": 1.0565,
      "step": 4831
    },
    {
      "epoch": 6.729805013927576,
      "grad_norm": 2.9280784130096436,
      "learning_rate": 2.3359863558840248e-05,
      "loss": 0.9753,
      "step": 4832
    },
    {
      "epoch": 6.731197771587744,
      "grad_norm": 2.8760924339294434,
      "learning_rate": 2.3349914724275156e-05,
      "loss": 1.0051,
      "step": 4833
    },
    {
      "epoch": 6.732590529247911,
      "grad_norm": 2.6996631622314453,
      "learning_rate": 2.333996588971006e-05,
      "loss": 0.9777,
      "step": 4834
    },
    {
      "epoch": 6.733983286908078,
      "grad_norm": 3.017238140106201,
      "learning_rate": 2.3330017055144967e-05,
      "loss": 0.8874,
      "step": 4835
    },
    {
      "epoch": 6.735376044568245,
      "grad_norm": 3.1263272762298584,
      "learning_rate": 2.332006822057987e-05,
      "loss": 0.9936,
      "step": 4836
    },
    {
      "epoch": 6.736768802228412,
      "grad_norm": 3.133610248565674,
      "learning_rate": 2.331011938601478e-05,
      "loss": 0.9494,
      "step": 4837
    },
    {
      "epoch": 6.73816155988858,
      "grad_norm": 2.646540403366089,
      "learning_rate": 2.3300170551449686e-05,
      "loss": 0.856,
      "step": 4838
    },
    {
      "epoch": 6.7395543175487465,
      "grad_norm": 3.041557550430298,
      "learning_rate": 2.329022171688459e-05,
      "loss": 0.9367,
      "step": 4839
    },
    {
      "epoch": 6.740947075208914,
      "grad_norm": 2.9367825984954834,
      "learning_rate": 2.3280272882319498e-05,
      "loss": 1.0595,
      "step": 4840
    },
    {
      "epoch": 6.742339832869081,
      "grad_norm": 3.084501266479492,
      "learning_rate": 2.3270324047754402e-05,
      "loss": 0.8773,
      "step": 4841
    },
    {
      "epoch": 6.743732590529248,
      "grad_norm": 3.071359634399414,
      "learning_rate": 2.326037521318931e-05,
      "loss": 1.049,
      "step": 4842
    },
    {
      "epoch": 6.745125348189415,
      "grad_norm": 3.3054821491241455,
      "learning_rate": 2.3250426378624217e-05,
      "loss": 1.1725,
      "step": 4843
    },
    {
      "epoch": 6.7465181058495824,
      "grad_norm": 2.5150833129882812,
      "learning_rate": 2.324047754405912e-05,
      "loss": 0.7595,
      "step": 4844
    },
    {
      "epoch": 6.747910863509749,
      "grad_norm": 2.453789472579956,
      "learning_rate": 2.3230528709494028e-05,
      "loss": 0.891,
      "step": 4845
    },
    {
      "epoch": 6.749303621169917,
      "grad_norm": 2.9088315963745117,
      "learning_rate": 2.3220579874928932e-05,
      "loss": 1.0837,
      "step": 4846
    },
    {
      "epoch": 6.750696378830083,
      "grad_norm": 2.64329195022583,
      "learning_rate": 2.321063104036384e-05,
      "loss": 0.8332,
      "step": 4847
    },
    {
      "epoch": 6.752089136490251,
      "grad_norm": 2.5478029251098633,
      "learning_rate": 2.320068220579875e-05,
      "loss": 0.9777,
      "step": 4848
    },
    {
      "epoch": 6.7534818941504176,
      "grad_norm": 2.7847177982330322,
      "learning_rate": 2.319073337123365e-05,
      "loss": 0.9929,
      "step": 4849
    },
    {
      "epoch": 6.754874651810585,
      "grad_norm": 2.8384597301483154,
      "learning_rate": 2.3180784536668562e-05,
      "loss": 1.0823,
      "step": 4850
    },
    {
      "epoch": 6.756267409470752,
      "grad_norm": 3.376582145690918,
      "learning_rate": 2.3170835702103466e-05,
      "loss": 1.1255,
      "step": 4851
    },
    {
      "epoch": 6.757660167130919,
      "grad_norm": 2.6226806640625,
      "learning_rate": 2.3160886867538374e-05,
      "loss": 0.8217,
      "step": 4852
    },
    {
      "epoch": 6.759052924791086,
      "grad_norm": 2.9121294021606445,
      "learning_rate": 2.3150938032973278e-05,
      "loss": 1.0547,
      "step": 4853
    },
    {
      "epoch": 6.7604456824512535,
      "grad_norm": 3.0356545448303223,
      "learning_rate": 2.3140989198408185e-05,
      "loss": 1.0487,
      "step": 4854
    },
    {
      "epoch": 6.76183844011142,
      "grad_norm": 3.278174638748169,
      "learning_rate": 2.3131040363843093e-05,
      "loss": 0.8931,
      "step": 4855
    },
    {
      "epoch": 6.763231197771588,
      "grad_norm": 3.302827835083008,
      "learning_rate": 2.3121091529277997e-05,
      "loss": 1.1367,
      "step": 4856
    },
    {
      "epoch": 6.764623955431755,
      "grad_norm": 2.7777512073516846,
      "learning_rate": 2.3111142694712904e-05,
      "loss": 0.9247,
      "step": 4857
    },
    {
      "epoch": 6.766016713091922,
      "grad_norm": 2.9615275859832764,
      "learning_rate": 2.3101193860147808e-05,
      "loss": 0.9697,
      "step": 4858
    },
    {
      "epoch": 6.767409470752089,
      "grad_norm": 3.0469748973846436,
      "learning_rate": 2.3091245025582716e-05,
      "loss": 1.2354,
      "step": 4859
    },
    {
      "epoch": 6.768802228412256,
      "grad_norm": 3.0362789630889893,
      "learning_rate": 2.3081296191017623e-05,
      "loss": 0.9566,
      "step": 4860
    },
    {
      "epoch": 6.770194986072424,
      "grad_norm": 2.6087095737457275,
      "learning_rate": 2.3071347356452527e-05,
      "loss": 1.0061,
      "step": 4861
    },
    {
      "epoch": 6.77158774373259,
      "grad_norm": 3.168748140335083,
      "learning_rate": 2.3061398521887435e-05,
      "loss": 0.8952,
      "step": 4862
    },
    {
      "epoch": 6.772980501392758,
      "grad_norm": 2.2951464653015137,
      "learning_rate": 2.305144968732234e-05,
      "loss": 0.7511,
      "step": 4863
    },
    {
      "epoch": 6.774373259052925,
      "grad_norm": 2.996347188949585,
      "learning_rate": 2.3041500852757246e-05,
      "loss": 0.8698,
      "step": 4864
    },
    {
      "epoch": 6.775766016713092,
      "grad_norm": 2.909731149673462,
      "learning_rate": 2.3031552018192154e-05,
      "loss": 1.0732,
      "step": 4865
    },
    {
      "epoch": 6.777158774373259,
      "grad_norm": 3.417818546295166,
      "learning_rate": 2.3021603183627058e-05,
      "loss": 1.0244,
      "step": 4866
    },
    {
      "epoch": 6.778551532033426,
      "grad_norm": 2.939164400100708,
      "learning_rate": 2.3011654349061965e-05,
      "loss": 0.6911,
      "step": 4867
    },
    {
      "epoch": 6.779944289693593,
      "grad_norm": 2.8852765560150146,
      "learning_rate": 2.300170551449687e-05,
      "loss": 1.0562,
      "step": 4868
    },
    {
      "epoch": 6.781337047353761,
      "grad_norm": 2.835871934890747,
      "learning_rate": 2.2991756679931777e-05,
      "loss": 0.9956,
      "step": 4869
    },
    {
      "epoch": 6.782729805013927,
      "grad_norm": 3.3156003952026367,
      "learning_rate": 2.2981807845366684e-05,
      "loss": 1.0879,
      "step": 4870
    },
    {
      "epoch": 6.784122562674095,
      "grad_norm": 3.052037000656128,
      "learning_rate": 2.297185901080159e-05,
      "loss": 1.0994,
      "step": 4871
    },
    {
      "epoch": 6.7855153203342615,
      "grad_norm": 2.884131908416748,
      "learning_rate": 2.2961910176236496e-05,
      "loss": 1.028,
      "step": 4872
    },
    {
      "epoch": 6.786908077994429,
      "grad_norm": 2.5954511165618896,
      "learning_rate": 2.29519613416714e-05,
      "loss": 0.8262,
      "step": 4873
    },
    {
      "epoch": 6.788300835654596,
      "grad_norm": 3.0728766918182373,
      "learning_rate": 2.2942012507106307e-05,
      "loss": 1.1055,
      "step": 4874
    },
    {
      "epoch": 6.789693593314763,
      "grad_norm": 2.8305482864379883,
      "learning_rate": 2.293206367254121e-05,
      "loss": 0.9566,
      "step": 4875
    },
    {
      "epoch": 6.791086350974931,
      "grad_norm": 2.665647029876709,
      "learning_rate": 2.292211483797612e-05,
      "loss": 0.8036,
      "step": 4876
    },
    {
      "epoch": 6.7924791086350975,
      "grad_norm": 2.9441335201263428,
      "learning_rate": 2.291216600341103e-05,
      "loss": 1.0342,
      "step": 4877
    },
    {
      "epoch": 6.793871866295264,
      "grad_norm": 4.162250995635986,
      "learning_rate": 2.2902217168845934e-05,
      "loss": 0.7745,
      "step": 4878
    },
    {
      "epoch": 6.795264623955432,
      "grad_norm": 2.99715518951416,
      "learning_rate": 2.289226833428084e-05,
      "loss": 1.1928,
      "step": 4879
    },
    {
      "epoch": 6.796657381615599,
      "grad_norm": 3.211400270462036,
      "learning_rate": 2.2882319499715745e-05,
      "loss": 0.9964,
      "step": 4880
    },
    {
      "epoch": 6.798050139275766,
      "grad_norm": 2.7228331565856934,
      "learning_rate": 2.2872370665150653e-05,
      "loss": 0.9169,
      "step": 4881
    },
    {
      "epoch": 6.7994428969359335,
      "grad_norm": 2.626319646835327,
      "learning_rate": 2.286242183058556e-05,
      "loss": 0.8734,
      "step": 4882
    },
    {
      "epoch": 6.8008356545961,
      "grad_norm": 2.9297358989715576,
      "learning_rate": 2.2852472996020464e-05,
      "loss": 1.199,
      "step": 4883
    },
    {
      "epoch": 6.802228412256268,
      "grad_norm": 2.9165637493133545,
      "learning_rate": 2.2842524161455372e-05,
      "loss": 1.0327,
      "step": 4884
    },
    {
      "epoch": 6.803621169916434,
      "grad_norm": 2.8690686225891113,
      "learning_rate": 2.2832575326890276e-05,
      "loss": 0.913,
      "step": 4885
    },
    {
      "epoch": 6.805013927576602,
      "grad_norm": 4.286510467529297,
      "learning_rate": 2.2822626492325183e-05,
      "loss": 0.9175,
      "step": 4886
    },
    {
      "epoch": 6.806406685236769,
      "grad_norm": 2.6812071800231934,
      "learning_rate": 2.281267765776009e-05,
      "loss": 0.8382,
      "step": 4887
    },
    {
      "epoch": 6.807799442896936,
      "grad_norm": 3.066721200942993,
      "learning_rate": 2.2802728823194995e-05,
      "loss": 0.9701,
      "step": 4888
    },
    {
      "epoch": 6.809192200557103,
      "grad_norm": 3.317391872406006,
      "learning_rate": 2.2792779988629902e-05,
      "loss": 1.1758,
      "step": 4889
    },
    {
      "epoch": 6.81058495821727,
      "grad_norm": 2.943735122680664,
      "learning_rate": 2.2782831154064806e-05,
      "loss": 0.9858,
      "step": 4890
    },
    {
      "epoch": 6.811977715877437,
      "grad_norm": 3.0627267360687256,
      "learning_rate": 2.2772882319499714e-05,
      "loss": 1.1379,
      "step": 4891
    },
    {
      "epoch": 6.813370473537605,
      "grad_norm": 2.906496524810791,
      "learning_rate": 2.276293348493462e-05,
      "loss": 0.9958,
      "step": 4892
    },
    {
      "epoch": 6.814763231197771,
      "grad_norm": 3.279766798019409,
      "learning_rate": 2.2752984650369525e-05,
      "loss": 1.0487,
      "step": 4893
    },
    {
      "epoch": 6.816155988857939,
      "grad_norm": 2.6565163135528564,
      "learning_rate": 2.2743035815804433e-05,
      "loss": 0.8614,
      "step": 4894
    },
    {
      "epoch": 6.8175487465181055,
      "grad_norm": 2.6851377487182617,
      "learning_rate": 2.2733086981239337e-05,
      "loss": 0.8974,
      "step": 4895
    },
    {
      "epoch": 6.818941504178273,
      "grad_norm": 2.879284143447876,
      "learning_rate": 2.2723138146674244e-05,
      "loss": 0.899,
      "step": 4896
    },
    {
      "epoch": 6.82033426183844,
      "grad_norm": 3.6307334899902344,
      "learning_rate": 2.271318931210915e-05,
      "loss": 1.2624,
      "step": 4897
    },
    {
      "epoch": 6.821727019498607,
      "grad_norm": 3.167030096054077,
      "learning_rate": 2.2703240477544056e-05,
      "loss": 1.0152,
      "step": 4898
    },
    {
      "epoch": 6.823119777158775,
      "grad_norm": 2.485337018966675,
      "learning_rate": 2.2693291642978963e-05,
      "loss": 0.8044,
      "step": 4899
    },
    {
      "epoch": 6.8245125348189415,
      "grad_norm": 3.0829617977142334,
      "learning_rate": 2.2683342808413868e-05,
      "loss": 1.2321,
      "step": 4900
    },
    {
      "epoch": 6.825905292479108,
      "grad_norm": 3.10870361328125,
      "learning_rate": 2.2673393973848775e-05,
      "loss": 0.907,
      "step": 4901
    },
    {
      "epoch": 6.827298050139276,
      "grad_norm": 3.1099343299865723,
      "learning_rate": 2.266344513928368e-05,
      "loss": 0.9595,
      "step": 4902
    },
    {
      "epoch": 6.828690807799443,
      "grad_norm": 3.0900535583496094,
      "learning_rate": 2.2653496304718587e-05,
      "loss": 1.0796,
      "step": 4903
    },
    {
      "epoch": 6.83008356545961,
      "grad_norm": 3.012258529663086,
      "learning_rate": 2.2643547470153497e-05,
      "loss": 0.9799,
      "step": 4904
    },
    {
      "epoch": 6.8314763231197775,
      "grad_norm": 2.9189305305480957,
      "learning_rate": 2.2633598635588398e-05,
      "loss": 0.8298,
      "step": 4905
    },
    {
      "epoch": 6.832869080779944,
      "grad_norm": 3.2075531482696533,
      "learning_rate": 2.262364980102331e-05,
      "loss": 1.2077,
      "step": 4906
    },
    {
      "epoch": 6.834261838440112,
      "grad_norm": 3.070300340652466,
      "learning_rate": 2.2613700966458213e-05,
      "loss": 1.1166,
      "step": 4907
    },
    {
      "epoch": 6.835654596100278,
      "grad_norm": 2.730436325073242,
      "learning_rate": 2.260375213189312e-05,
      "loss": 1.137,
      "step": 4908
    },
    {
      "epoch": 6.837047353760446,
      "grad_norm": 2.8196377754211426,
      "learning_rate": 2.2593803297328028e-05,
      "loss": 0.8568,
      "step": 4909
    },
    {
      "epoch": 6.838440111420613,
      "grad_norm": 3.2209866046905518,
      "learning_rate": 2.2583854462762932e-05,
      "loss": 1.1469,
      "step": 4910
    },
    {
      "epoch": 6.83983286908078,
      "grad_norm": 3.142430067062378,
      "learning_rate": 2.257390562819784e-05,
      "loss": 1.1244,
      "step": 4911
    },
    {
      "epoch": 6.841225626740947,
      "grad_norm": 2.8419435024261475,
      "learning_rate": 2.2563956793632744e-05,
      "loss": 0.8251,
      "step": 4912
    },
    {
      "epoch": 6.842618384401114,
      "grad_norm": 3.740586519241333,
      "learning_rate": 2.255400795906765e-05,
      "loss": 1.0554,
      "step": 4913
    },
    {
      "epoch": 6.844011142061281,
      "grad_norm": 2.52078914642334,
      "learning_rate": 2.254405912450256e-05,
      "loss": 0.8627,
      "step": 4914
    },
    {
      "epoch": 6.845403899721449,
      "grad_norm": 3.096212387084961,
      "learning_rate": 2.2534110289937463e-05,
      "loss": 1.0125,
      "step": 4915
    },
    {
      "epoch": 6.846796657381615,
      "grad_norm": 3.0929019451141357,
      "learning_rate": 2.252416145537237e-05,
      "loss": 1.1388,
      "step": 4916
    },
    {
      "epoch": 6.848189415041783,
      "grad_norm": 2.9732069969177246,
      "learning_rate": 2.2514212620807274e-05,
      "loss": 1.0101,
      "step": 4917
    },
    {
      "epoch": 6.84958217270195,
      "grad_norm": 3.1545639038085938,
      "learning_rate": 2.250426378624218e-05,
      "loss": 1.0123,
      "step": 4918
    },
    {
      "epoch": 6.850974930362117,
      "grad_norm": 3.0801069736480713,
      "learning_rate": 2.2494314951677086e-05,
      "loss": 1.1101,
      "step": 4919
    },
    {
      "epoch": 6.852367688022284,
      "grad_norm": 3.8419761657714844,
      "learning_rate": 2.2484366117111993e-05,
      "loss": 0.8489,
      "step": 4920
    },
    {
      "epoch": 6.853760445682451,
      "grad_norm": 3.294038772583008,
      "learning_rate": 2.24744172825469e-05,
      "loss": 1.1795,
      "step": 4921
    },
    {
      "epoch": 6.855153203342619,
      "grad_norm": 3.377178430557251,
      "learning_rate": 2.2464468447981805e-05,
      "loss": 1.3105,
      "step": 4922
    },
    {
      "epoch": 6.8565459610027855,
      "grad_norm": 3.4253134727478027,
      "learning_rate": 2.2454519613416712e-05,
      "loss": 1.1638,
      "step": 4923
    },
    {
      "epoch": 6.857938718662953,
      "grad_norm": 2.7638607025146484,
      "learning_rate": 2.2444570778851616e-05,
      "loss": 0.9349,
      "step": 4924
    },
    {
      "epoch": 6.85933147632312,
      "grad_norm": 2.787493944168091,
      "learning_rate": 2.2434621944286524e-05,
      "loss": 0.9409,
      "step": 4925
    },
    {
      "epoch": 6.860724233983287,
      "grad_norm": 2.876359224319458,
      "learning_rate": 2.242467310972143e-05,
      "loss": 0.852,
      "step": 4926
    },
    {
      "epoch": 6.862116991643454,
      "grad_norm": 3.78804874420166,
      "learning_rate": 2.2414724275156335e-05,
      "loss": 1.2609,
      "step": 4927
    },
    {
      "epoch": 6.8635097493036215,
      "grad_norm": 2.9731123447418213,
      "learning_rate": 2.2404775440591243e-05,
      "loss": 1.0511,
      "step": 4928
    },
    {
      "epoch": 6.864902506963788,
      "grad_norm": 2.677267074584961,
      "learning_rate": 2.2394826606026147e-05,
      "loss": 0.9392,
      "step": 4929
    },
    {
      "epoch": 6.866295264623956,
      "grad_norm": 2.678102970123291,
      "learning_rate": 2.2384877771461054e-05,
      "loss": 0.9781,
      "step": 4930
    },
    {
      "epoch": 6.867688022284122,
      "grad_norm": 3.74009108543396,
      "learning_rate": 2.2374928936895965e-05,
      "loss": 0.9932,
      "step": 4931
    },
    {
      "epoch": 6.86908077994429,
      "grad_norm": 2.4893105030059814,
      "learning_rate": 2.2364980102330866e-05,
      "loss": 0.7554,
      "step": 4932
    },
    {
      "epoch": 6.870473537604457,
      "grad_norm": 2.7979495525360107,
      "learning_rate": 2.2355031267765777e-05,
      "loss": 0.819,
      "step": 4933
    },
    {
      "epoch": 6.871866295264624,
      "grad_norm": 2.7396624088287354,
      "learning_rate": 2.234508243320068e-05,
      "loss": 0.8175,
      "step": 4934
    },
    {
      "epoch": 6.873259052924791,
      "grad_norm": 3.2493722438812256,
      "learning_rate": 2.2335133598635588e-05,
      "loss": 0.9786,
      "step": 4935
    },
    {
      "epoch": 6.874651810584958,
      "grad_norm": 2.8006935119628906,
      "learning_rate": 2.2325184764070496e-05,
      "loss": 0.8901,
      "step": 4936
    },
    {
      "epoch": 6.876044568245125,
      "grad_norm": 2.727621078491211,
      "learning_rate": 2.23152359295054e-05,
      "loss": 0.9357,
      "step": 4937
    },
    {
      "epoch": 6.8774373259052926,
      "grad_norm": 3.2232580184936523,
      "learning_rate": 2.2305287094940307e-05,
      "loss": 1.0972,
      "step": 4938
    },
    {
      "epoch": 6.878830083565459,
      "grad_norm": 3.3753669261932373,
      "learning_rate": 2.229533826037521e-05,
      "loss": 0.9813,
      "step": 4939
    },
    {
      "epoch": 6.880222841225627,
      "grad_norm": 3.5995097160339355,
      "learning_rate": 2.228538942581012e-05,
      "loss": 1.5259,
      "step": 4940
    },
    {
      "epoch": 6.881615598885794,
      "grad_norm": 2.841562032699585,
      "learning_rate": 2.2275440591245023e-05,
      "loss": 1.1019,
      "step": 4941
    },
    {
      "epoch": 6.883008356545961,
      "grad_norm": 4.627342224121094,
      "learning_rate": 2.226549175667993e-05,
      "loss": 0.9604,
      "step": 4942
    },
    {
      "epoch": 6.884401114206128,
      "grad_norm": 3.2343509197235107,
      "learning_rate": 2.2255542922114838e-05,
      "loss": 1.168,
      "step": 4943
    },
    {
      "epoch": 6.885793871866295,
      "grad_norm": 3.2918992042541504,
      "learning_rate": 2.224559408754974e-05,
      "loss": 1.2578,
      "step": 4944
    },
    {
      "epoch": 6.887186629526463,
      "grad_norm": 2.8083975315093994,
      "learning_rate": 2.223564525298465e-05,
      "loss": 0.9497,
      "step": 4945
    },
    {
      "epoch": 6.888579387186629,
      "grad_norm": 2.8898448944091797,
      "learning_rate": 2.2225696418419553e-05,
      "loss": 0.9155,
      "step": 4946
    },
    {
      "epoch": 6.889972144846797,
      "grad_norm": 3.120607376098633,
      "learning_rate": 2.221574758385446e-05,
      "loss": 1.1165,
      "step": 4947
    },
    {
      "epoch": 6.891364902506964,
      "grad_norm": 2.907076835632324,
      "learning_rate": 2.2205798749289368e-05,
      "loss": 0.7774,
      "step": 4948
    },
    {
      "epoch": 6.892757660167131,
      "grad_norm": 2.8598861694335938,
      "learning_rate": 2.2195849914724272e-05,
      "loss": 0.8315,
      "step": 4949
    },
    {
      "epoch": 6.894150417827298,
      "grad_norm": 2.8945460319519043,
      "learning_rate": 2.218590108015918e-05,
      "loss": 0.9248,
      "step": 4950
    },
    {
      "epoch": 6.895543175487465,
      "grad_norm": 3.612288236618042,
      "learning_rate": 2.2175952245594084e-05,
      "loss": 1.3396,
      "step": 4951
    },
    {
      "epoch": 6.896935933147632,
      "grad_norm": 2.9821317195892334,
      "learning_rate": 2.216600341102899e-05,
      "loss": 0.8911,
      "step": 4952
    },
    {
      "epoch": 6.8983286908078,
      "grad_norm": 3.016906261444092,
      "learning_rate": 2.21560545764639e-05,
      "loss": 1.109,
      "step": 4953
    },
    {
      "epoch": 6.899721448467966,
      "grad_norm": 3.6679413318634033,
      "learning_rate": 2.2146105741898803e-05,
      "loss": 0.9454,
      "step": 4954
    },
    {
      "epoch": 6.901114206128134,
      "grad_norm": 2.7348663806915283,
      "learning_rate": 2.213615690733371e-05,
      "loss": 1.0042,
      "step": 4955
    },
    {
      "epoch": 6.9025069637883005,
      "grad_norm": 2.812051296234131,
      "learning_rate": 2.2126208072768614e-05,
      "loss": 0.9994,
      "step": 4956
    },
    {
      "epoch": 6.903899721448468,
      "grad_norm": 3.1831793785095215,
      "learning_rate": 2.2116259238203522e-05,
      "loss": 1.2802,
      "step": 4957
    },
    {
      "epoch": 6.905292479108635,
      "grad_norm": 4.422842025756836,
      "learning_rate": 2.210631040363843e-05,
      "loss": 0.8655,
      "step": 4958
    },
    {
      "epoch": 6.906685236768802,
      "grad_norm": 3.1010208129882812,
      "learning_rate": 2.2096361569073333e-05,
      "loss": 1.0346,
      "step": 4959
    },
    {
      "epoch": 6.908077994428969,
      "grad_norm": 2.766726493835449,
      "learning_rate": 2.2086412734508244e-05,
      "loss": 1.0624,
      "step": 4960
    },
    {
      "epoch": 6.9094707520891365,
      "grad_norm": 3.0663726329803467,
      "learning_rate": 2.2076463899943145e-05,
      "loss": 0.9082,
      "step": 4961
    },
    {
      "epoch": 6.910863509749303,
      "grad_norm": 3.0582809448242188,
      "learning_rate": 2.2066515065378056e-05,
      "loss": 1.2008,
      "step": 4962
    },
    {
      "epoch": 6.912256267409471,
      "grad_norm": 2.92802095413208,
      "learning_rate": 2.205656623081296e-05,
      "loss": 1.0997,
      "step": 4963
    },
    {
      "epoch": 6.913649025069638,
      "grad_norm": 3.3929343223571777,
      "learning_rate": 2.2046617396247867e-05,
      "loss": 1.1293,
      "step": 4964
    },
    {
      "epoch": 6.915041782729805,
      "grad_norm": 2.8015310764312744,
      "learning_rate": 2.2036668561682775e-05,
      "loss": 0.9593,
      "step": 4965
    },
    {
      "epoch": 6.9164345403899725,
      "grad_norm": 2.6430611610412598,
      "learning_rate": 2.202671972711768e-05,
      "loss": 0.9387,
      "step": 4966
    },
    {
      "epoch": 6.917827298050139,
      "grad_norm": 3.121490478515625,
      "learning_rate": 2.2016770892552586e-05,
      "loss": 0.8176,
      "step": 4967
    },
    {
      "epoch": 6.919220055710307,
      "grad_norm": 2.7189342975616455,
      "learning_rate": 2.200682205798749e-05,
      "loss": 0.731,
      "step": 4968
    },
    {
      "epoch": 6.920612813370473,
      "grad_norm": 2.969587564468384,
      "learning_rate": 2.1996873223422398e-05,
      "loss": 1.0109,
      "step": 4969
    },
    {
      "epoch": 6.922005571030641,
      "grad_norm": 2.9785587787628174,
      "learning_rate": 2.1986924388857305e-05,
      "loss": 1.0273,
      "step": 4970
    },
    {
      "epoch": 6.923398328690808,
      "grad_norm": 3.127487897872925,
      "learning_rate": 2.197697555429221e-05,
      "loss": 1.1461,
      "step": 4971
    },
    {
      "epoch": 6.924791086350975,
      "grad_norm": 3.0932915210723877,
      "learning_rate": 2.1967026719727117e-05,
      "loss": 0.9521,
      "step": 4972
    },
    {
      "epoch": 6.926183844011142,
      "grad_norm": 3.6915338039398193,
      "learning_rate": 2.195707788516202e-05,
      "loss": 1.2715,
      "step": 4973
    },
    {
      "epoch": 6.927576601671309,
      "grad_norm": 2.8909175395965576,
      "learning_rate": 2.1947129050596928e-05,
      "loss": 1.2073,
      "step": 4974
    },
    {
      "epoch": 6.928969359331476,
      "grad_norm": 2.9219141006469727,
      "learning_rate": 2.1937180216031836e-05,
      "loss": 0.8063,
      "step": 4975
    },
    {
      "epoch": 6.930362116991644,
      "grad_norm": 4.024686336517334,
      "learning_rate": 2.192723138146674e-05,
      "loss": 0.8786,
      "step": 4976
    },
    {
      "epoch": 6.93175487465181,
      "grad_norm": 2.914431571960449,
      "learning_rate": 2.1917282546901647e-05,
      "loss": 0.7909,
      "step": 4977
    },
    {
      "epoch": 6.933147632311978,
      "grad_norm": 3.4128596782684326,
      "learning_rate": 2.190733371233655e-05,
      "loss": 1.1857,
      "step": 4978
    },
    {
      "epoch": 6.9345403899721445,
      "grad_norm": 2.566725254058838,
      "learning_rate": 2.189738487777146e-05,
      "loss": 0.8833,
      "step": 4979
    },
    {
      "epoch": 6.935933147632312,
      "grad_norm": 3.3989527225494385,
      "learning_rate": 2.1887436043206366e-05,
      "loss": 1.0414,
      "step": 4980
    },
    {
      "epoch": 6.937325905292479,
      "grad_norm": 3.1205458641052246,
      "learning_rate": 2.187748720864127e-05,
      "loss": 0.8835,
      "step": 4981
    },
    {
      "epoch": 6.938718662952646,
      "grad_norm": 3.2083778381347656,
      "learning_rate": 2.1867538374076178e-05,
      "loss": 1.2242,
      "step": 4982
    },
    {
      "epoch": 6.940111420612814,
      "grad_norm": 3.1415822505950928,
      "learning_rate": 2.1857589539511082e-05,
      "loss": 0.8324,
      "step": 4983
    },
    {
      "epoch": 6.9415041782729805,
      "grad_norm": 3.6850414276123047,
      "learning_rate": 2.184764070494599e-05,
      "loss": 1.0104,
      "step": 4984
    },
    {
      "epoch": 6.942896935933147,
      "grad_norm": 2.9321935176849365,
      "learning_rate": 2.1837691870380893e-05,
      "loss": 0.8893,
      "step": 4985
    },
    {
      "epoch": 6.944289693593315,
      "grad_norm": 3.8329665660858154,
      "learning_rate": 2.18277430358158e-05,
      "loss": 1.1802,
      "step": 4986
    },
    {
      "epoch": 6.945682451253482,
      "grad_norm": 3.4439475536346436,
      "learning_rate": 2.1817794201250712e-05,
      "loss": 0.8457,
      "step": 4987
    },
    {
      "epoch": 6.947075208913649,
      "grad_norm": 3.002546548843384,
      "learning_rate": 2.1807845366685612e-05,
      "loss": 1.1196,
      "step": 4988
    },
    {
      "epoch": 6.9484679665738165,
      "grad_norm": 3.2283008098602295,
      "learning_rate": 2.1797896532120523e-05,
      "loss": 1.1945,
      "step": 4989
    },
    {
      "epoch": 6.949860724233983,
      "grad_norm": 2.5133228302001953,
      "learning_rate": 2.1787947697555427e-05,
      "loss": 0.7714,
      "step": 4990
    },
    {
      "epoch": 6.951253481894151,
      "grad_norm": 3.184483528137207,
      "learning_rate": 2.1777998862990335e-05,
      "loss": 0.8774,
      "step": 4991
    },
    {
      "epoch": 6.952646239554317,
      "grad_norm": 2.887610912322998,
      "learning_rate": 2.1768050028425242e-05,
      "loss": 0.8744,
      "step": 4992
    },
    {
      "epoch": 6.954038997214485,
      "grad_norm": 2.69132661819458,
      "learning_rate": 2.1758101193860146e-05,
      "loss": 0.8415,
      "step": 4993
    },
    {
      "epoch": 6.955431754874652,
      "grad_norm": 3.511922597885132,
      "learning_rate": 2.1748152359295054e-05,
      "loss": 0.8937,
      "step": 4994
    },
    {
      "epoch": 6.956824512534819,
      "grad_norm": 3.0658130645751953,
      "learning_rate": 2.1738203524729958e-05,
      "loss": 0.8997,
      "step": 4995
    },
    {
      "epoch": 6.958217270194986,
      "grad_norm": 2.714297294616699,
      "learning_rate": 2.1728254690164865e-05,
      "loss": 1.1088,
      "step": 4996
    },
    {
      "epoch": 6.959610027855153,
      "grad_norm": 2.500307083129883,
      "learning_rate": 2.1718305855599773e-05,
      "loss": 0.9374,
      "step": 4997
    },
    {
      "epoch": 6.96100278551532,
      "grad_norm": 2.758538007736206,
      "learning_rate": 2.1708357021034677e-05,
      "loss": 0.9487,
      "step": 4998
    },
    {
      "epoch": 6.962395543175488,
      "grad_norm": 3.0099618434906006,
      "learning_rate": 2.1698408186469584e-05,
      "loss": 1.0268,
      "step": 4999
    },
    {
      "epoch": 6.963788300835654,
      "grad_norm": 3.660125970840454,
      "learning_rate": 2.168845935190449e-05,
      "loss": 1.0164,
      "step": 5000
    },
    {
      "epoch": 6.965181058495822,
      "grad_norm": 3.014194965362549,
      "learning_rate": 2.1678510517339396e-05,
      "loss": 1.1265,
      "step": 5001
    },
    {
      "epoch": 6.9665738161559885,
      "grad_norm": 2.936569929122925,
      "learning_rate": 2.1668561682774303e-05,
      "loss": 0.9369,
      "step": 5002
    },
    {
      "epoch": 6.967966573816156,
      "grad_norm": 3.5617330074310303,
      "learning_rate": 2.1658612848209207e-05,
      "loss": 1.2657,
      "step": 5003
    },
    {
      "epoch": 6.969359331476323,
      "grad_norm": 3.6318094730377197,
      "learning_rate": 2.1648664013644115e-05,
      "loss": 1.1003,
      "step": 5004
    },
    {
      "epoch": 6.97075208913649,
      "grad_norm": 2.7930498123168945,
      "learning_rate": 2.163871517907902e-05,
      "loss": 0.9573,
      "step": 5005
    },
    {
      "epoch": 6.972144846796658,
      "grad_norm": 3.4401938915252686,
      "learning_rate": 2.1628766344513926e-05,
      "loss": 1.0708,
      "step": 5006
    },
    {
      "epoch": 6.9735376044568245,
      "grad_norm": 2.823951005935669,
      "learning_rate": 2.161881750994883e-05,
      "loss": 1.1048,
      "step": 5007
    },
    {
      "epoch": 6.974930362116992,
      "grad_norm": 2.9734506607055664,
      "learning_rate": 2.1608868675383738e-05,
      "loss": 1.0475,
      "step": 5008
    },
    {
      "epoch": 6.976323119777159,
      "grad_norm": 2.6455373764038086,
      "learning_rate": 2.1598919840818645e-05,
      "loss": 0.8868,
      "step": 5009
    },
    {
      "epoch": 6.977715877437326,
      "grad_norm": 3.0616531372070312,
      "learning_rate": 2.158897100625355e-05,
      "loss": 0.8475,
      "step": 5010
    },
    {
      "epoch": 6.979108635097493,
      "grad_norm": 2.7626075744628906,
      "learning_rate": 2.1579022171688457e-05,
      "loss": 0.9221,
      "step": 5011
    },
    {
      "epoch": 6.9805013927576605,
      "grad_norm": 3.4198381900787354,
      "learning_rate": 2.156907333712336e-05,
      "loss": 1.2821,
      "step": 5012
    },
    {
      "epoch": 6.981894150417827,
      "grad_norm": 2.8367843627929688,
      "learning_rate": 2.155912450255827e-05,
      "loss": 0.9523,
      "step": 5013
    },
    {
      "epoch": 6.983286908077995,
      "grad_norm": 3.0300216674804688,
      "learning_rate": 2.1549175667993176e-05,
      "loss": 0.9777,
      "step": 5014
    },
    {
      "epoch": 6.984679665738161,
      "grad_norm": 3.0950534343719482,
      "learning_rate": 2.153922683342808e-05,
      "loss": 1.0669,
      "step": 5015
    },
    {
      "epoch": 6.986072423398329,
      "grad_norm": 3.0042805671691895,
      "learning_rate": 2.152927799886299e-05,
      "loss": 0.8434,
      "step": 5016
    },
    {
      "epoch": 6.987465181058496,
      "grad_norm": 2.593696355819702,
      "learning_rate": 2.151932916429789e-05,
      "loss": 0.7874,
      "step": 5017
    },
    {
      "epoch": 6.988857938718663,
      "grad_norm": 2.6960651874542236,
      "learning_rate": 2.1509380329732802e-05,
      "loss": 0.9296,
      "step": 5018
    },
    {
      "epoch": 6.99025069637883,
      "grad_norm": 3.2014400959014893,
      "learning_rate": 2.149943149516771e-05,
      "loss": 0.9696,
      "step": 5019
    },
    {
      "epoch": 6.991643454038997,
      "grad_norm": 2.7455062866210938,
      "learning_rate": 2.1489482660602614e-05,
      "loss": 0.8764,
      "step": 5020
    },
    {
      "epoch": 6.993036211699164,
      "grad_norm": 3.0886213779449463,
      "learning_rate": 2.147953382603752e-05,
      "loss": 0.8537,
      "step": 5021
    },
    {
      "epoch": 6.994428969359332,
      "grad_norm": 2.88210391998291,
      "learning_rate": 2.1469584991472426e-05,
      "loss": 0.875,
      "step": 5022
    },
    {
      "epoch": 6.995821727019498,
      "grad_norm": 3.03379487991333,
      "learning_rate": 2.1459636156907333e-05,
      "loss": 0.9391,
      "step": 5023
    },
    {
      "epoch": 6.997214484679666,
      "grad_norm": 2.986440420150757,
      "learning_rate": 2.144968732234224e-05,
      "loss": 1.0542,
      "step": 5024
    },
    {
      "epoch": 6.998607242339833,
      "grad_norm": 3.2636523246765137,
      "learning_rate": 2.1439738487777145e-05,
      "loss": 0.9983,
      "step": 5025
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.2439088821411133,
      "learning_rate": 2.1429789653212052e-05,
      "loss": 1.0126,
      "step": 5026
    },
    {
      "epoch": 7.0013927576601676,
      "grad_norm": 2.588900566101074,
      "learning_rate": 2.1419840818646956e-05,
      "loss": 0.7241,
      "step": 5027
    },
    {
      "epoch": 7.002785515320334,
      "grad_norm": 2.518993616104126,
      "learning_rate": 2.1409891984081864e-05,
      "loss": 0.8023,
      "step": 5028
    },
    {
      "epoch": 7.004178272980502,
      "grad_norm": 2.8074703216552734,
      "learning_rate": 2.1399943149516768e-05,
      "loss": 0.983,
      "step": 5029
    },
    {
      "epoch": 7.005571030640668,
      "grad_norm": 2.823148012161255,
      "learning_rate": 2.1389994314951675e-05,
      "loss": 1.0526,
      "step": 5030
    },
    {
      "epoch": 7.006963788300836,
      "grad_norm": 2.761317014694214,
      "learning_rate": 2.1380045480386583e-05,
      "loss": 0.8701,
      "step": 5031
    },
    {
      "epoch": 7.008356545961003,
      "grad_norm": 2.7132506370544434,
      "learning_rate": 2.1370096645821487e-05,
      "loss": 0.7854,
      "step": 5032
    },
    {
      "epoch": 7.00974930362117,
      "grad_norm": 2.4156360626220703,
      "learning_rate": 2.1360147811256394e-05,
      "loss": 0.7814,
      "step": 5033
    },
    {
      "epoch": 7.011142061281337,
      "grad_norm": 2.9176406860351562,
      "learning_rate": 2.1350198976691298e-05,
      "loss": 0.8162,
      "step": 5034
    },
    {
      "epoch": 7.012534818941504,
      "grad_norm": 2.6062848567962646,
      "learning_rate": 2.1340250142126206e-05,
      "loss": 0.8725,
      "step": 5035
    },
    {
      "epoch": 7.013927576601671,
      "grad_norm": 2.6114494800567627,
      "learning_rate": 2.1330301307561113e-05,
      "loss": 0.9053,
      "step": 5036
    },
    {
      "epoch": 7.015320334261839,
      "grad_norm": 2.632460117340088,
      "learning_rate": 2.1320352472996017e-05,
      "loss": 0.7621,
      "step": 5037
    },
    {
      "epoch": 7.016713091922005,
      "grad_norm": 2.561506509780884,
      "learning_rate": 2.1310403638430925e-05,
      "loss": 0.8962,
      "step": 5038
    },
    {
      "epoch": 7.018105849582173,
      "grad_norm": 3.0773773193359375,
      "learning_rate": 2.130045480386583e-05,
      "loss": 0.9322,
      "step": 5039
    },
    {
      "epoch": 7.0194986072423395,
      "grad_norm": 2.61908221244812,
      "learning_rate": 2.1290505969300736e-05,
      "loss": 0.942,
      "step": 5040
    },
    {
      "epoch": 7.020891364902507,
      "grad_norm": 2.6667587757110596,
      "learning_rate": 2.1280557134735644e-05,
      "loss": 0.8027,
      "step": 5041
    },
    {
      "epoch": 7.022284122562674,
      "grad_norm": 2.5058159828186035,
      "learning_rate": 2.1270608300170548e-05,
      "loss": 0.7292,
      "step": 5042
    },
    {
      "epoch": 7.023676880222841,
      "grad_norm": 2.6387643814086914,
      "learning_rate": 2.126065946560546e-05,
      "loss": 0.7403,
      "step": 5043
    },
    {
      "epoch": 7.025069637883008,
      "grad_norm": 2.9340646266937256,
      "learning_rate": 2.125071063104036e-05,
      "loss": 1.1487,
      "step": 5044
    },
    {
      "epoch": 7.0264623955431755,
      "grad_norm": 3.019251823425293,
      "learning_rate": 2.124076179647527e-05,
      "loss": 0.8989,
      "step": 5045
    },
    {
      "epoch": 7.027855153203342,
      "grad_norm": 2.4151124954223633,
      "learning_rate": 2.123081296191017e-05,
      "loss": 0.8461,
      "step": 5046
    },
    {
      "epoch": 7.02924791086351,
      "grad_norm": 2.8883981704711914,
      "learning_rate": 2.122086412734508e-05,
      "loss": 0.9723,
      "step": 5047
    },
    {
      "epoch": 7.030640668523677,
      "grad_norm": 2.4514729976654053,
      "learning_rate": 2.121091529277999e-05,
      "loss": 0.7727,
      "step": 5048
    },
    {
      "epoch": 7.032033426183844,
      "grad_norm": 2.7968037128448486,
      "learning_rate": 2.1200966458214893e-05,
      "loss": 0.8198,
      "step": 5049
    },
    {
      "epoch": 7.0334261838440115,
      "grad_norm": 4.603396892547607,
      "learning_rate": 2.11910176236498e-05,
      "loss": 0.9419,
      "step": 5050
    },
    {
      "epoch": 7.034818941504178,
      "grad_norm": 2.6633098125457764,
      "learning_rate": 2.1181068789084705e-05,
      "loss": 0.8245,
      "step": 5051
    },
    {
      "epoch": 7.036211699164346,
      "grad_norm": 3.1830623149871826,
      "learning_rate": 2.1171119954519612e-05,
      "loss": 1.1548,
      "step": 5052
    },
    {
      "epoch": 7.037604456824512,
      "grad_norm": 2.8647260665893555,
      "learning_rate": 2.116117111995452e-05,
      "loss": 0.8362,
      "step": 5053
    },
    {
      "epoch": 7.03899721448468,
      "grad_norm": 2.6832895278930664,
      "learning_rate": 2.1151222285389424e-05,
      "loss": 0.9162,
      "step": 5054
    },
    {
      "epoch": 7.040389972144847,
      "grad_norm": 2.7217798233032227,
      "learning_rate": 2.114127345082433e-05,
      "loss": 0.8067,
      "step": 5055
    },
    {
      "epoch": 7.041782729805014,
      "grad_norm": 2.254660129547119,
      "learning_rate": 2.1131324616259235e-05,
      "loss": 0.6615,
      "step": 5056
    },
    {
      "epoch": 7.043175487465181,
      "grad_norm": 2.911212205886841,
      "learning_rate": 2.1121375781694143e-05,
      "loss": 0.9648,
      "step": 5057
    },
    {
      "epoch": 7.044568245125348,
      "grad_norm": 2.566112756729126,
      "learning_rate": 2.111142694712905e-05,
      "loss": 0.6842,
      "step": 5058
    },
    {
      "epoch": 7.045961002785515,
      "grad_norm": 2.482128143310547,
      "learning_rate": 2.1101478112563954e-05,
      "loss": 0.7122,
      "step": 5059
    },
    {
      "epoch": 7.047353760445683,
      "grad_norm": 2.571878671646118,
      "learning_rate": 2.1091529277998862e-05,
      "loss": 0.7384,
      "step": 5060
    },
    {
      "epoch": 7.048746518105849,
      "grad_norm": 2.926588773727417,
      "learning_rate": 2.1081580443433766e-05,
      "loss": 1.0739,
      "step": 5061
    },
    {
      "epoch": 7.050139275766017,
      "grad_norm": 2.393764019012451,
      "learning_rate": 2.1071631608868673e-05,
      "loss": 0.7013,
      "step": 5062
    },
    {
      "epoch": 7.0515320334261835,
      "grad_norm": 2.4707753658294678,
      "learning_rate": 2.106168277430358e-05,
      "loss": 0.7492,
      "step": 5063
    },
    {
      "epoch": 7.052924791086351,
      "grad_norm": 2.3997011184692383,
      "learning_rate": 2.1051733939738485e-05,
      "loss": 0.822,
      "step": 5064
    },
    {
      "epoch": 7.054317548746518,
      "grad_norm": 2.5442023277282715,
      "learning_rate": 2.1041785105173392e-05,
      "loss": 0.7408,
      "step": 5065
    },
    {
      "epoch": 7.055710306406685,
      "grad_norm": 2.7726078033447266,
      "learning_rate": 2.1031836270608296e-05,
      "loss": 0.9467,
      "step": 5066
    },
    {
      "epoch": 7.057103064066852,
      "grad_norm": 2.851839542388916,
      "learning_rate": 2.1021887436043204e-05,
      "loss": 1.0702,
      "step": 5067
    },
    {
      "epoch": 7.0584958217270195,
      "grad_norm": 3.1124398708343506,
      "learning_rate": 2.1011938601478108e-05,
      "loss": 0.9318,
      "step": 5068
    },
    {
      "epoch": 7.059888579387187,
      "grad_norm": 2.5048420429229736,
      "learning_rate": 2.1001989766913015e-05,
      "loss": 0.844,
      "step": 5069
    },
    {
      "epoch": 7.061281337047354,
      "grad_norm": 2.6897008419036865,
      "learning_rate": 2.0992040932347923e-05,
      "loss": 0.8532,
      "step": 5070
    },
    {
      "epoch": 7.062674094707521,
      "grad_norm": 2.5082085132598877,
      "learning_rate": 2.0982092097782827e-05,
      "loss": 0.7295,
      "step": 5071
    },
    {
      "epoch": 7.064066852367688,
      "grad_norm": 2.720109224319458,
      "learning_rate": 2.0972143263217738e-05,
      "loss": 0.7822,
      "step": 5072
    },
    {
      "epoch": 7.0654596100278555,
      "grad_norm": 2.456808567047119,
      "learning_rate": 2.096219442865264e-05,
      "loss": 0.76,
      "step": 5073
    },
    {
      "epoch": 7.066852367688022,
      "grad_norm": 2.958230495452881,
      "learning_rate": 2.095224559408755e-05,
      "loss": 0.799,
      "step": 5074
    },
    {
      "epoch": 7.06824512534819,
      "grad_norm": 2.6701319217681885,
      "learning_rate": 2.0942296759522457e-05,
      "loss": 0.9165,
      "step": 5075
    },
    {
      "epoch": 7.069637883008356,
      "grad_norm": 2.6504392623901367,
      "learning_rate": 2.093234792495736e-05,
      "loss": 0.7042,
      "step": 5076
    },
    {
      "epoch": 7.071030640668524,
      "grad_norm": 2.80842924118042,
      "learning_rate": 2.0922399090392268e-05,
      "loss": 0.8763,
      "step": 5077
    },
    {
      "epoch": 7.072423398328691,
      "grad_norm": 4.207406997680664,
      "learning_rate": 2.0912450255827172e-05,
      "loss": 0.9335,
      "step": 5078
    },
    {
      "epoch": 7.073816155988858,
      "grad_norm": 2.4932899475097656,
      "learning_rate": 2.090250142126208e-05,
      "loss": 0.7264,
      "step": 5079
    },
    {
      "epoch": 7.075208913649025,
      "grad_norm": 3.238790988922119,
      "learning_rate": 2.0892552586696987e-05,
      "loss": 0.9106,
      "step": 5080
    },
    {
      "epoch": 7.076601671309192,
      "grad_norm": 3.1530511379241943,
      "learning_rate": 2.088260375213189e-05,
      "loss": 0.9639,
      "step": 5081
    },
    {
      "epoch": 7.077994428969359,
      "grad_norm": 3.010418176651001,
      "learning_rate": 2.08726549175668e-05,
      "loss": 0.8949,
      "step": 5082
    },
    {
      "epoch": 7.079387186629527,
      "grad_norm": 2.6341300010681152,
      "learning_rate": 2.0862706083001703e-05,
      "loss": 0.6999,
      "step": 5083
    },
    {
      "epoch": 7.080779944289693,
      "grad_norm": 2.839822292327881,
      "learning_rate": 2.085275724843661e-05,
      "loss": 0.8473,
      "step": 5084
    },
    {
      "epoch": 7.082172701949861,
      "grad_norm": 2.499387264251709,
      "learning_rate": 2.0842808413871518e-05,
      "loss": 0.8796,
      "step": 5085
    },
    {
      "epoch": 7.0835654596100275,
      "grad_norm": 3.1768767833709717,
      "learning_rate": 2.0832859579306422e-05,
      "loss": 0.8339,
      "step": 5086
    },
    {
      "epoch": 7.084958217270195,
      "grad_norm": 2.5548417568206787,
      "learning_rate": 2.082291074474133e-05,
      "loss": 0.8024,
      "step": 5087
    },
    {
      "epoch": 7.086350974930362,
      "grad_norm": 2.81540846824646,
      "learning_rate": 2.0812961910176233e-05,
      "loss": 0.8978,
      "step": 5088
    },
    {
      "epoch": 7.087743732590529,
      "grad_norm": 2.507047176361084,
      "learning_rate": 2.080301307561114e-05,
      "loss": 0.8813,
      "step": 5089
    },
    {
      "epoch": 7.089136490250697,
      "grad_norm": 2.84696888923645,
      "learning_rate": 2.0793064241046045e-05,
      "loss": 0.6542,
      "step": 5090
    },
    {
      "epoch": 7.0905292479108635,
      "grad_norm": 2.5703725814819336,
      "learning_rate": 2.0783115406480952e-05,
      "loss": 0.8221,
      "step": 5091
    },
    {
      "epoch": 7.091922005571031,
      "grad_norm": 2.9748427867889404,
      "learning_rate": 2.077316657191586e-05,
      "loss": 0.7694,
      "step": 5092
    },
    {
      "epoch": 7.093314763231198,
      "grad_norm": 2.238007068634033,
      "learning_rate": 2.0763217737350764e-05,
      "loss": 0.6962,
      "step": 5093
    },
    {
      "epoch": 7.094707520891365,
      "grad_norm": 2.6781530380249023,
      "learning_rate": 2.075326890278567e-05,
      "loss": 0.8532,
      "step": 5094
    },
    {
      "epoch": 7.096100278551532,
      "grad_norm": 2.589348554611206,
      "learning_rate": 2.0743320068220575e-05,
      "loss": 0.7188,
      "step": 5095
    },
    {
      "epoch": 7.0974930362116995,
      "grad_norm": 2.998473644256592,
      "learning_rate": 2.0733371233655483e-05,
      "loss": 0.9301,
      "step": 5096
    },
    {
      "epoch": 7.098885793871866,
      "grad_norm": 2.7394397258758545,
      "learning_rate": 2.072342239909039e-05,
      "loss": 0.8027,
      "step": 5097
    },
    {
      "epoch": 7.100278551532034,
      "grad_norm": 2.572772979736328,
      "learning_rate": 2.0713473564525294e-05,
      "loss": 0.7165,
      "step": 5098
    },
    {
      "epoch": 7.1016713091922,
      "grad_norm": 2.7826836109161377,
      "learning_rate": 2.0703524729960202e-05,
      "loss": 0.9358,
      "step": 5099
    },
    {
      "epoch": 7.103064066852368,
      "grad_norm": 3.2955169677734375,
      "learning_rate": 2.0693575895395106e-05,
      "loss": 0.8409,
      "step": 5100
    },
    {
      "epoch": 7.104456824512535,
      "grad_norm": 2.598841905593872,
      "learning_rate": 2.0683627060830017e-05,
      "loss": 0.8871,
      "step": 5101
    },
    {
      "epoch": 7.105849582172702,
      "grad_norm": 2.574275255203247,
      "learning_rate": 2.0673678226264924e-05,
      "loss": 0.6962,
      "step": 5102
    },
    {
      "epoch": 7.107242339832869,
      "grad_norm": 2.3531739711761475,
      "learning_rate": 2.066372939169983e-05,
      "loss": 0.6944,
      "step": 5103
    },
    {
      "epoch": 7.108635097493036,
      "grad_norm": 2.294454574584961,
      "learning_rate": 2.0653780557134736e-05,
      "loss": 0.7277,
      "step": 5104
    },
    {
      "epoch": 7.110027855153203,
      "grad_norm": 3.7097771167755127,
      "learning_rate": 2.064383172256964e-05,
      "loss": 0.8847,
      "step": 5105
    },
    {
      "epoch": 7.111420612813371,
      "grad_norm": 2.7075202465057373,
      "learning_rate": 2.0633882888004547e-05,
      "loss": 0.8689,
      "step": 5106
    },
    {
      "epoch": 7.112813370473537,
      "grad_norm": 2.550398111343384,
      "learning_rate": 2.0623934053439455e-05,
      "loss": 0.6873,
      "step": 5107
    },
    {
      "epoch": 7.114206128133705,
      "grad_norm": 2.507601499557495,
      "learning_rate": 2.061398521887436e-05,
      "loss": 0.949,
      "step": 5108
    },
    {
      "epoch": 7.1155988857938715,
      "grad_norm": 2.350555181503296,
      "learning_rate": 2.0604036384309266e-05,
      "loss": 0.7499,
      "step": 5109
    },
    {
      "epoch": 7.116991643454039,
      "grad_norm": 2.904186487197876,
      "learning_rate": 2.059408754974417e-05,
      "loss": 0.9101,
      "step": 5110
    },
    {
      "epoch": 7.118384401114207,
      "grad_norm": 2.7214269638061523,
      "learning_rate": 2.0584138715179078e-05,
      "loss": 0.9063,
      "step": 5111
    },
    {
      "epoch": 7.119777158774373,
      "grad_norm": 2.905555486679077,
      "learning_rate": 2.0574189880613982e-05,
      "loss": 0.8439,
      "step": 5112
    },
    {
      "epoch": 7.121169916434541,
      "grad_norm": 2.5376923084259033,
      "learning_rate": 2.056424104604889e-05,
      "loss": 0.7845,
      "step": 5113
    },
    {
      "epoch": 7.1225626740947074,
      "grad_norm": 2.6675326824188232,
      "learning_rate": 2.0554292211483797e-05,
      "loss": 0.9121,
      "step": 5114
    },
    {
      "epoch": 7.123955431754875,
      "grad_norm": 3.5976321697235107,
      "learning_rate": 2.05443433769187e-05,
      "loss": 0.972,
      "step": 5115
    },
    {
      "epoch": 7.125348189415042,
      "grad_norm": 2.5115559101104736,
      "learning_rate": 2.053439454235361e-05,
      "loss": 0.7596,
      "step": 5116
    },
    {
      "epoch": 7.126740947075209,
      "grad_norm": 2.221853733062744,
      "learning_rate": 2.0524445707788513e-05,
      "loss": 0.6731,
      "step": 5117
    },
    {
      "epoch": 7.128133704735376,
      "grad_norm": 2.7109086513519287,
      "learning_rate": 2.051449687322342e-05,
      "loss": 0.9109,
      "step": 5118
    },
    {
      "epoch": 7.129526462395543,
      "grad_norm": 2.7682974338531494,
      "learning_rate": 2.0504548038658327e-05,
      "loss": 0.8008,
      "step": 5119
    },
    {
      "epoch": 7.13091922005571,
      "grad_norm": 2.805457830429077,
      "learning_rate": 2.049459920409323e-05,
      "loss": 0.8912,
      "step": 5120
    },
    {
      "epoch": 7.132311977715878,
      "grad_norm": 3.161466360092163,
      "learning_rate": 2.048465036952814e-05,
      "loss": 0.9236,
      "step": 5121
    },
    {
      "epoch": 7.133704735376044,
      "grad_norm": 2.6468088626861572,
      "learning_rate": 2.0474701534963043e-05,
      "loss": 0.8531,
      "step": 5122
    },
    {
      "epoch": 7.135097493036212,
      "grad_norm": 2.6612088680267334,
      "learning_rate": 2.046475270039795e-05,
      "loss": 0.8459,
      "step": 5123
    },
    {
      "epoch": 7.1364902506963785,
      "grad_norm": 2.7777202129364014,
      "learning_rate": 2.0454803865832858e-05,
      "loss": 0.9561,
      "step": 5124
    },
    {
      "epoch": 7.137883008356546,
      "grad_norm": 3.0060179233551025,
      "learning_rate": 2.0444855031267762e-05,
      "loss": 0.9452,
      "step": 5125
    },
    {
      "epoch": 7.139275766016713,
      "grad_norm": 3.0659730434417725,
      "learning_rate": 2.043490619670267e-05,
      "loss": 0.8891,
      "step": 5126
    },
    {
      "epoch": 7.14066852367688,
      "grad_norm": 2.5169434547424316,
      "learning_rate": 2.0424957362137574e-05,
      "loss": 0.7024,
      "step": 5127
    },
    {
      "epoch": 7.142061281337047,
      "grad_norm": 2.351672887802124,
      "learning_rate": 2.0415008527572484e-05,
      "loss": 0.6795,
      "step": 5128
    },
    {
      "epoch": 7.1434540389972145,
      "grad_norm": 2.4826416969299316,
      "learning_rate": 2.0405059693007392e-05,
      "loss": 0.745,
      "step": 5129
    },
    {
      "epoch": 7.144846796657381,
      "grad_norm": 2.7235655784606934,
      "learning_rate": 2.0395110858442296e-05,
      "loss": 0.8468,
      "step": 5130
    },
    {
      "epoch": 7.146239554317549,
      "grad_norm": 2.200775623321533,
      "learning_rate": 2.0385162023877203e-05,
      "loss": 0.7456,
      "step": 5131
    },
    {
      "epoch": 7.147632311977716,
      "grad_norm": 2.8050622940063477,
      "learning_rate": 2.0375213189312108e-05,
      "loss": 0.6287,
      "step": 5132
    },
    {
      "epoch": 7.149025069637883,
      "grad_norm": 2.5859289169311523,
      "learning_rate": 2.0365264354747015e-05,
      "loss": 0.8106,
      "step": 5133
    },
    {
      "epoch": 7.1504178272980505,
      "grad_norm": 3.303654909133911,
      "learning_rate": 2.035531552018192e-05,
      "loss": 0.9672,
      "step": 5134
    },
    {
      "epoch": 7.151810584958217,
      "grad_norm": 3.0739753246307373,
      "learning_rate": 2.0345366685616827e-05,
      "loss": 0.8583,
      "step": 5135
    },
    {
      "epoch": 7.153203342618385,
      "grad_norm": 2.913184404373169,
      "learning_rate": 2.0335417851051734e-05,
      "loss": 0.9952,
      "step": 5136
    },
    {
      "epoch": 7.154596100278551,
      "grad_norm": 2.4681808948516846,
      "learning_rate": 2.0325469016486638e-05,
      "loss": 0.8304,
      "step": 5137
    },
    {
      "epoch": 7.155988857938719,
      "grad_norm": 2.6157567501068115,
      "learning_rate": 2.0315520181921546e-05,
      "loss": 0.7232,
      "step": 5138
    },
    {
      "epoch": 7.157381615598886,
      "grad_norm": 2.5770859718322754,
      "learning_rate": 2.030557134735645e-05,
      "loss": 0.8426,
      "step": 5139
    },
    {
      "epoch": 7.158774373259053,
      "grad_norm": 2.7727253437042236,
      "learning_rate": 2.0295622512791357e-05,
      "loss": 0.8057,
      "step": 5140
    },
    {
      "epoch": 7.16016713091922,
      "grad_norm": 2.745166778564453,
      "learning_rate": 2.0285673678226265e-05,
      "loss": 0.9122,
      "step": 5141
    },
    {
      "epoch": 7.161559888579387,
      "grad_norm": 2.6689202785491943,
      "learning_rate": 2.027572484366117e-05,
      "loss": 0.8866,
      "step": 5142
    },
    {
      "epoch": 7.162952646239554,
      "grad_norm": 2.7326817512512207,
      "learning_rate": 2.0265776009096076e-05,
      "loss": 0.9073,
      "step": 5143
    },
    {
      "epoch": 7.164345403899722,
      "grad_norm": 2.2113630771636963,
      "learning_rate": 2.025582717453098e-05,
      "loss": 0.8135,
      "step": 5144
    },
    {
      "epoch": 7.165738161559888,
      "grad_norm": 4.216705322265625,
      "learning_rate": 2.0245878339965888e-05,
      "loss": 0.8057,
      "step": 5145
    },
    {
      "epoch": 7.167130919220056,
      "grad_norm": 2.7352399826049805,
      "learning_rate": 2.0235929505400795e-05,
      "loss": 0.8108,
      "step": 5146
    },
    {
      "epoch": 7.1685236768802225,
      "grad_norm": 2.788586139678955,
      "learning_rate": 2.02259806708357e-05,
      "loss": 0.6843,
      "step": 5147
    },
    {
      "epoch": 7.16991643454039,
      "grad_norm": 2.4862241744995117,
      "learning_rate": 2.0216031836270607e-05,
      "loss": 0.826,
      "step": 5148
    },
    {
      "epoch": 7.171309192200557,
      "grad_norm": 3.241105556488037,
      "learning_rate": 2.020608300170551e-05,
      "loss": 1.0344,
      "step": 5149
    },
    {
      "epoch": 7.172701949860724,
      "grad_norm": 2.471928834915161,
      "learning_rate": 2.0196134167140418e-05,
      "loss": 0.6736,
      "step": 5150
    },
    {
      "epoch": 7.174094707520891,
      "grad_norm": 2.711073875427246,
      "learning_rate": 2.0186185332575326e-05,
      "loss": 0.911,
      "step": 5151
    },
    {
      "epoch": 7.1754874651810585,
      "grad_norm": 2.5336215496063232,
      "learning_rate": 2.017623649801023e-05,
      "loss": 0.936,
      "step": 5152
    },
    {
      "epoch": 7.176880222841225,
      "grad_norm": 2.245718479156494,
      "learning_rate": 2.0166287663445137e-05,
      "loss": 0.632,
      "step": 5153
    },
    {
      "epoch": 7.178272980501393,
      "grad_norm": 2.371717929840088,
      "learning_rate": 2.015633882888004e-05,
      "loss": 0.7806,
      "step": 5154
    },
    {
      "epoch": 7.17966573816156,
      "grad_norm": 2.528045415878296,
      "learning_rate": 2.014638999431495e-05,
      "loss": 0.8388,
      "step": 5155
    },
    {
      "epoch": 7.181058495821727,
      "grad_norm": 2.6796059608459473,
      "learning_rate": 2.0136441159749853e-05,
      "loss": 0.8448,
      "step": 5156
    },
    {
      "epoch": 7.1824512534818945,
      "grad_norm": 2.8471596240997314,
      "learning_rate": 2.0126492325184764e-05,
      "loss": 0.8743,
      "step": 5157
    },
    {
      "epoch": 7.183844011142061,
      "grad_norm": 2.8560431003570557,
      "learning_rate": 2.011654349061967e-05,
      "loss": 0.9575,
      "step": 5158
    },
    {
      "epoch": 7.185236768802229,
      "grad_norm": 3.3372654914855957,
      "learning_rate": 2.0106594656054575e-05,
      "loss": 1.3444,
      "step": 5159
    },
    {
      "epoch": 7.186629526462395,
      "grad_norm": 2.640535593032837,
      "learning_rate": 2.0096645821489483e-05,
      "loss": 0.8092,
      "step": 5160
    },
    {
      "epoch": 7.188022284122563,
      "grad_norm": 2.883291482925415,
      "learning_rate": 2.0086696986924387e-05,
      "loss": 0.7063,
      "step": 5161
    },
    {
      "epoch": 7.18941504178273,
      "grad_norm": 2.960609197616577,
      "learning_rate": 2.0076748152359294e-05,
      "loss": 1.1612,
      "step": 5162
    },
    {
      "epoch": 7.190807799442897,
      "grad_norm": 2.7009124755859375,
      "learning_rate": 2.00667993177942e-05,
      "loss": 1.0279,
      "step": 5163
    },
    {
      "epoch": 7.192200557103064,
      "grad_norm": 2.5741305351257324,
      "learning_rate": 2.0056850483229106e-05,
      "loss": 0.7547,
      "step": 5164
    },
    {
      "epoch": 7.193593314763231,
      "grad_norm": 2.402419090270996,
      "learning_rate": 2.0046901648664013e-05,
      "loss": 0.7029,
      "step": 5165
    },
    {
      "epoch": 7.194986072423398,
      "grad_norm": 2.8439228534698486,
      "learning_rate": 2.0036952814098917e-05,
      "loss": 0.7626,
      "step": 5166
    },
    {
      "epoch": 7.196378830083566,
      "grad_norm": 2.8873085975646973,
      "learning_rate": 2.0027003979533825e-05,
      "loss": 0.9064,
      "step": 5167
    },
    {
      "epoch": 7.197771587743732,
      "grad_norm": 2.371703624725342,
      "learning_rate": 2.0017055144968732e-05,
      "loss": 0.5794,
      "step": 5168
    },
    {
      "epoch": 7.1991643454039,
      "grad_norm": 2.3308355808258057,
      "learning_rate": 2.0007106310403636e-05,
      "loss": 0.6031,
      "step": 5169
    },
    {
      "epoch": 7.2005571030640665,
      "grad_norm": 2.594644784927368,
      "learning_rate": 1.9997157475838544e-05,
      "loss": 0.7663,
      "step": 5170
    },
    {
      "epoch": 7.201949860724234,
      "grad_norm": 2.617103338241577,
      "learning_rate": 1.9987208641273448e-05,
      "loss": 0.801,
      "step": 5171
    },
    {
      "epoch": 7.203342618384401,
      "grad_norm": 3.003443717956543,
      "learning_rate": 1.9977259806708355e-05,
      "loss": 0.9119,
      "step": 5172
    },
    {
      "epoch": 7.204735376044568,
      "grad_norm": 2.8007192611694336,
      "learning_rate": 1.9967310972143263e-05,
      "loss": 0.9762,
      "step": 5173
    },
    {
      "epoch": 7.206128133704736,
      "grad_norm": 2.3412129878997803,
      "learning_rate": 1.9957362137578167e-05,
      "loss": 0.6466,
      "step": 5174
    },
    {
      "epoch": 7.2075208913649025,
      "grad_norm": 3.0760936737060547,
      "learning_rate": 1.9947413303013074e-05,
      "loss": 1.1183,
      "step": 5175
    },
    {
      "epoch": 7.20891364902507,
      "grad_norm": 2.990391969680786,
      "learning_rate": 1.993746446844798e-05,
      "loss": 0.7956,
      "step": 5176
    },
    {
      "epoch": 7.210306406685237,
      "grad_norm": 2.6078553199768066,
      "learning_rate": 1.9927515633882886e-05,
      "loss": 0.742,
      "step": 5177
    },
    {
      "epoch": 7.211699164345404,
      "grad_norm": 3.205244541168213,
      "learning_rate": 1.991756679931779e-05,
      "loss": 0.8327,
      "step": 5178
    },
    {
      "epoch": 7.213091922005571,
      "grad_norm": 3.2693397998809814,
      "learning_rate": 1.9907617964752697e-05,
      "loss": 0.8681,
      "step": 5179
    },
    {
      "epoch": 7.2144846796657385,
      "grad_norm": 3.0199766159057617,
      "learning_rate": 1.9897669130187605e-05,
      "loss": 1.0342,
      "step": 5180
    },
    {
      "epoch": 7.215877437325905,
      "grad_norm": 2.707958221435547,
      "learning_rate": 1.988772029562251e-05,
      "loss": 0.8669,
      "step": 5181
    },
    {
      "epoch": 7.217270194986073,
      "grad_norm": 2.774451971054077,
      "learning_rate": 1.9877771461057416e-05,
      "loss": 0.8854,
      "step": 5182
    },
    {
      "epoch": 7.218662952646239,
      "grad_norm": 2.6234207153320312,
      "learning_rate": 1.986782262649232e-05,
      "loss": 0.6751,
      "step": 5183
    },
    {
      "epoch": 7.220055710306407,
      "grad_norm": 2.521958827972412,
      "learning_rate": 1.985787379192723e-05,
      "loss": 0.7473,
      "step": 5184
    },
    {
      "epoch": 7.221448467966574,
      "grad_norm": 2.656883716583252,
      "learning_rate": 1.984792495736214e-05,
      "loss": 0.8238,
      "step": 5185
    },
    {
      "epoch": 7.222841225626741,
      "grad_norm": 2.4660890102386475,
      "learning_rate": 1.9837976122797043e-05,
      "loss": 0.7723,
      "step": 5186
    },
    {
      "epoch": 7.224233983286908,
      "grad_norm": 2.383047342300415,
      "learning_rate": 1.982802728823195e-05,
      "loss": 0.6429,
      "step": 5187
    },
    {
      "epoch": 7.225626740947075,
      "grad_norm": 2.556492805480957,
      "learning_rate": 1.9818078453666854e-05,
      "loss": 0.7697,
      "step": 5188
    },
    {
      "epoch": 7.227019498607242,
      "grad_norm": 4.027998924255371,
      "learning_rate": 1.9808129619101762e-05,
      "loss": 0.9606,
      "step": 5189
    },
    {
      "epoch": 7.22841225626741,
      "grad_norm": 2.40372896194458,
      "learning_rate": 1.979818078453667e-05,
      "loss": 0.717,
      "step": 5190
    },
    {
      "epoch": 7.229805013927576,
      "grad_norm": 2.6457722187042236,
      "learning_rate": 1.9788231949971573e-05,
      "loss": 0.8272,
      "step": 5191
    },
    {
      "epoch": 7.231197771587744,
      "grad_norm": 2.5490403175354004,
      "learning_rate": 1.977828311540648e-05,
      "loss": 0.824,
      "step": 5192
    },
    {
      "epoch": 7.2325905292479105,
      "grad_norm": 2.8429832458496094,
      "learning_rate": 1.9768334280841385e-05,
      "loss": 0.8409,
      "step": 5193
    },
    {
      "epoch": 7.233983286908078,
      "grad_norm": 2.7641210556030273,
      "learning_rate": 1.9758385446276292e-05,
      "loss": 0.8567,
      "step": 5194
    },
    {
      "epoch": 7.235376044568245,
      "grad_norm": 2.108842134475708,
      "learning_rate": 1.97484366117112e-05,
      "loss": 0.6063,
      "step": 5195
    },
    {
      "epoch": 7.236768802228412,
      "grad_norm": 2.6179306507110596,
      "learning_rate": 1.9738487777146104e-05,
      "loss": 0.672,
      "step": 5196
    },
    {
      "epoch": 7.23816155988858,
      "grad_norm": 2.7160799503326416,
      "learning_rate": 1.972853894258101e-05,
      "loss": 0.8693,
      "step": 5197
    },
    {
      "epoch": 7.2395543175487465,
      "grad_norm": 2.631033420562744,
      "learning_rate": 1.9718590108015915e-05,
      "loss": 0.7287,
      "step": 5198
    },
    {
      "epoch": 7.240947075208914,
      "grad_norm": 2.4316320419311523,
      "learning_rate": 1.9708641273450823e-05,
      "loss": 0.7627,
      "step": 5199
    },
    {
      "epoch": 7.242339832869081,
      "grad_norm": 2.6915841102600098,
      "learning_rate": 1.9698692438885727e-05,
      "loss": 0.9119,
      "step": 5200
    },
    {
      "epoch": 7.243732590529248,
      "grad_norm": 2.515749931335449,
      "learning_rate": 1.9688743604320634e-05,
      "loss": 0.8842,
      "step": 5201
    },
    {
      "epoch": 7.245125348189415,
      "grad_norm": 2.6935107707977295,
      "learning_rate": 1.9678794769755542e-05,
      "loss": 0.7339,
      "step": 5202
    },
    {
      "epoch": 7.2465181058495824,
      "grad_norm": 2.112510919570923,
      "learning_rate": 1.9668845935190446e-05,
      "loss": 0.5913,
      "step": 5203
    },
    {
      "epoch": 7.247910863509749,
      "grad_norm": 2.314663887023926,
      "learning_rate": 1.9658897100625353e-05,
      "loss": 0.6436,
      "step": 5204
    },
    {
      "epoch": 7.249303621169917,
      "grad_norm": 2.8983373641967773,
      "learning_rate": 1.9648948266060257e-05,
      "loss": 0.7483,
      "step": 5205
    },
    {
      "epoch": 7.250696378830083,
      "grad_norm": 2.8968334197998047,
      "learning_rate": 1.9638999431495165e-05,
      "loss": 0.8495,
      "step": 5206
    },
    {
      "epoch": 7.252089136490251,
      "grad_norm": 2.4824650287628174,
      "learning_rate": 1.9629050596930072e-05,
      "loss": 0.684,
      "step": 5207
    },
    {
      "epoch": 7.2534818941504176,
      "grad_norm": 2.26605486869812,
      "learning_rate": 1.9619101762364976e-05,
      "loss": 0.7048,
      "step": 5208
    },
    {
      "epoch": 7.254874651810585,
      "grad_norm": 2.54241681098938,
      "learning_rate": 1.9609152927799884e-05,
      "loss": 0.8729,
      "step": 5209
    },
    {
      "epoch": 7.256267409470752,
      "grad_norm": 3.6656627655029297,
      "learning_rate": 1.9599204093234788e-05,
      "loss": 0.8071,
      "step": 5210
    },
    {
      "epoch": 7.257660167130919,
      "grad_norm": 3.096910238265991,
      "learning_rate": 1.9589255258669695e-05,
      "loss": 0.8771,
      "step": 5211
    },
    {
      "epoch": 7.259052924791086,
      "grad_norm": 2.730132579803467,
      "learning_rate": 1.9579306424104606e-05,
      "loss": 0.8259,
      "step": 5212
    },
    {
      "epoch": 7.2604456824512535,
      "grad_norm": 2.580861806869507,
      "learning_rate": 1.956935758953951e-05,
      "loss": 0.9026,
      "step": 5213
    },
    {
      "epoch": 7.26183844011142,
      "grad_norm": 2.4602959156036377,
      "learning_rate": 1.9559408754974418e-05,
      "loss": 0.7339,
      "step": 5214
    },
    {
      "epoch": 7.263231197771588,
      "grad_norm": 2.3018956184387207,
      "learning_rate": 1.9549459920409322e-05,
      "loss": 0.793,
      "step": 5215
    },
    {
      "epoch": 7.264623955431755,
      "grad_norm": 2.997117280960083,
      "learning_rate": 1.953951108584423e-05,
      "loss": 0.8779,
      "step": 5216
    },
    {
      "epoch": 7.266016713091922,
      "grad_norm": 2.6175875663757324,
      "learning_rate": 1.9529562251279137e-05,
      "loss": 0.7928,
      "step": 5217
    },
    {
      "epoch": 7.2674094707520895,
      "grad_norm": 3.443099021911621,
      "learning_rate": 1.951961341671404e-05,
      "loss": 0.7839,
      "step": 5218
    },
    {
      "epoch": 7.268802228412256,
      "grad_norm": 2.713618755340576,
      "learning_rate": 1.950966458214895e-05,
      "loss": 0.9359,
      "step": 5219
    },
    {
      "epoch": 7.270194986072424,
      "grad_norm": 2.643413543701172,
      "learning_rate": 1.9499715747583853e-05,
      "loss": 0.8584,
      "step": 5220
    },
    {
      "epoch": 7.27158774373259,
      "grad_norm": 2.323481559753418,
      "learning_rate": 1.948976691301876e-05,
      "loss": 0.7348,
      "step": 5221
    },
    {
      "epoch": 7.272980501392758,
      "grad_norm": 2.8908908367156982,
      "learning_rate": 1.9479818078453664e-05,
      "loss": 1.0852,
      "step": 5222
    },
    {
      "epoch": 7.274373259052925,
      "grad_norm": 2.4041051864624023,
      "learning_rate": 1.946986924388857e-05,
      "loss": 0.7248,
      "step": 5223
    },
    {
      "epoch": 7.275766016713092,
      "grad_norm": 2.178449869155884,
      "learning_rate": 1.945992040932348e-05,
      "loss": 0.7322,
      "step": 5224
    },
    {
      "epoch": 7.277158774373259,
      "grad_norm": 2.3139548301696777,
      "learning_rate": 1.9449971574758383e-05,
      "loss": 0.7602,
      "step": 5225
    },
    {
      "epoch": 7.278551532033426,
      "grad_norm": 2.421529531478882,
      "learning_rate": 1.944002274019329e-05,
      "loss": 0.6924,
      "step": 5226
    },
    {
      "epoch": 7.279944289693593,
      "grad_norm": 2.8418025970458984,
      "learning_rate": 1.9430073905628195e-05,
      "loss": 0.9891,
      "step": 5227
    },
    {
      "epoch": 7.281337047353761,
      "grad_norm": 2.5942490100860596,
      "learning_rate": 1.9420125071063102e-05,
      "loss": 0.8317,
      "step": 5228
    },
    {
      "epoch": 7.282729805013927,
      "grad_norm": 3.0093631744384766,
      "learning_rate": 1.941017623649801e-05,
      "loss": 0.9615,
      "step": 5229
    },
    {
      "epoch": 7.284122562674095,
      "grad_norm": 3.2489662170410156,
      "learning_rate": 1.9400227401932914e-05,
      "loss": 1.0558,
      "step": 5230
    },
    {
      "epoch": 7.2855153203342615,
      "grad_norm": 2.364912509918213,
      "learning_rate": 1.939027856736782e-05,
      "loss": 0.7061,
      "step": 5231
    },
    {
      "epoch": 7.286908077994429,
      "grad_norm": 3.0428390502929688,
      "learning_rate": 1.9380329732802725e-05,
      "loss": 0.9384,
      "step": 5232
    },
    {
      "epoch": 7.288300835654596,
      "grad_norm": 2.7181684970855713,
      "learning_rate": 1.9370380898237633e-05,
      "loss": 0.795,
      "step": 5233
    },
    {
      "epoch": 7.289693593314763,
      "grad_norm": 2.540064811706543,
      "learning_rate": 1.936043206367254e-05,
      "loss": 0.7905,
      "step": 5234
    },
    {
      "epoch": 7.29108635097493,
      "grad_norm": 2.6257853507995605,
      "learning_rate": 1.9350483229107444e-05,
      "loss": 1.0153,
      "step": 5235
    },
    {
      "epoch": 7.2924791086350975,
      "grad_norm": 2.388836145401001,
      "learning_rate": 1.934053439454235e-05,
      "loss": 0.6819,
      "step": 5236
    },
    {
      "epoch": 7.293871866295264,
      "grad_norm": 2.7063469886779785,
      "learning_rate": 1.9330585559977256e-05,
      "loss": 1.0169,
      "step": 5237
    },
    {
      "epoch": 7.295264623955432,
      "grad_norm": 2.6595163345336914,
      "learning_rate": 1.9320636725412163e-05,
      "loss": 0.7897,
      "step": 5238
    },
    {
      "epoch": 7.296657381615599,
      "grad_norm": 2.7297656536102295,
      "learning_rate": 1.9310687890847067e-05,
      "loss": 0.8888,
      "step": 5239
    },
    {
      "epoch": 7.298050139275766,
      "grad_norm": 2.5360867977142334,
      "learning_rate": 1.9300739056281978e-05,
      "loss": 0.7609,
      "step": 5240
    },
    {
      "epoch": 7.2994428969359335,
      "grad_norm": 2.6307790279388428,
      "learning_rate": 1.9290790221716886e-05,
      "loss": 0.6898,
      "step": 5241
    },
    {
      "epoch": 7.3008356545961,
      "grad_norm": 3.1441152095794678,
      "learning_rate": 1.928084138715179e-05,
      "loss": 1.0444,
      "step": 5242
    },
    {
      "epoch": 7.302228412256268,
      "grad_norm": 3.023423910140991,
      "learning_rate": 1.9270892552586697e-05,
      "loss": 0.9404,
      "step": 5243
    },
    {
      "epoch": 7.303621169916434,
      "grad_norm": 2.8225855827331543,
      "learning_rate": 1.92609437180216e-05,
      "loss": 0.8027,
      "step": 5244
    },
    {
      "epoch": 7.305013927576602,
      "grad_norm": 2.899827003479004,
      "learning_rate": 1.925099488345651e-05,
      "loss": 1.0238,
      "step": 5245
    },
    {
      "epoch": 7.306406685236769,
      "grad_norm": 3.0850961208343506,
      "learning_rate": 1.9241046048891416e-05,
      "loss": 0.9852,
      "step": 5246
    },
    {
      "epoch": 7.307799442896936,
      "grad_norm": 3.0236339569091797,
      "learning_rate": 1.923109721432632e-05,
      "loss": 0.9381,
      "step": 5247
    },
    {
      "epoch": 7.309192200557103,
      "grad_norm": 3.084923267364502,
      "learning_rate": 1.9221148379761228e-05,
      "loss": 0.8701,
      "step": 5248
    },
    {
      "epoch": 7.31058495821727,
      "grad_norm": 3.112299919128418,
      "learning_rate": 1.921119954519613e-05,
      "loss": 0.8863,
      "step": 5249
    },
    {
      "epoch": 7.311977715877437,
      "grad_norm": 2.5260541439056396,
      "learning_rate": 1.920125071063104e-05,
      "loss": 0.6948,
      "step": 5250
    },
    {
      "epoch": 7.313370473537605,
      "grad_norm": 2.752619981765747,
      "learning_rate": 1.9191301876065947e-05,
      "loss": 0.833,
      "step": 5251
    },
    {
      "epoch": 7.314763231197771,
      "grad_norm": 2.4809863567352295,
      "learning_rate": 1.918135304150085e-05,
      "loss": 0.6482,
      "step": 5252
    },
    {
      "epoch": 7.316155988857939,
      "grad_norm": 2.7999536991119385,
      "learning_rate": 1.9171404206935758e-05,
      "loss": 0.7211,
      "step": 5253
    },
    {
      "epoch": 7.3175487465181055,
      "grad_norm": 3.0541927814483643,
      "learning_rate": 1.9161455372370662e-05,
      "loss": 1.1089,
      "step": 5254
    },
    {
      "epoch": 7.318941504178273,
      "grad_norm": 2.6121320724487305,
      "learning_rate": 1.915150653780557e-05,
      "loss": 0.8468,
      "step": 5255
    },
    {
      "epoch": 7.32033426183844,
      "grad_norm": 2.517087936401367,
      "learning_rate": 1.9141557703240477e-05,
      "loss": 0.7409,
      "step": 5256
    },
    {
      "epoch": 7.321727019498607,
      "grad_norm": 2.7593190670013428,
      "learning_rate": 1.913160886867538e-05,
      "loss": 0.9637,
      "step": 5257
    },
    {
      "epoch": 7.323119777158775,
      "grad_norm": 2.8267886638641357,
      "learning_rate": 1.912166003411029e-05,
      "loss": 0.7191,
      "step": 5258
    },
    {
      "epoch": 7.3245125348189415,
      "grad_norm": 2.4519171714782715,
      "learning_rate": 1.9111711199545193e-05,
      "loss": 0.6571,
      "step": 5259
    },
    {
      "epoch": 7.325905292479109,
      "grad_norm": 2.661771774291992,
      "learning_rate": 1.91017623649801e-05,
      "loss": 0.774,
      "step": 5260
    },
    {
      "epoch": 7.327298050139276,
      "grad_norm": 3.1244726181030273,
      "learning_rate": 1.9091813530415004e-05,
      "loss": 1.0265,
      "step": 5261
    },
    {
      "epoch": 7.328690807799443,
      "grad_norm": 2.6667983531951904,
      "learning_rate": 1.9081864695849912e-05,
      "loss": 0.871,
      "step": 5262
    },
    {
      "epoch": 7.33008356545961,
      "grad_norm": 2.4615135192871094,
      "learning_rate": 1.907191586128482e-05,
      "loss": 0.703,
      "step": 5263
    },
    {
      "epoch": 7.3314763231197775,
      "grad_norm": 2.2512056827545166,
      "learning_rate": 1.9061967026719723e-05,
      "loss": 0.7559,
      "step": 5264
    },
    {
      "epoch": 7.332869080779944,
      "grad_norm": 2.763364553451538,
      "learning_rate": 1.905201819215463e-05,
      "loss": 0.8774,
      "step": 5265
    },
    {
      "epoch": 7.334261838440112,
      "grad_norm": 2.2631046772003174,
      "learning_rate": 1.9042069357589535e-05,
      "loss": 0.664,
      "step": 5266
    },
    {
      "epoch": 7.335654596100278,
      "grad_norm": 2.8672075271606445,
      "learning_rate": 1.9032120523024442e-05,
      "loss": 0.9558,
      "step": 5267
    },
    {
      "epoch": 7.337047353760446,
      "grad_norm": 2.6277668476104736,
      "learning_rate": 1.9022171688459353e-05,
      "loss": 0.7777,
      "step": 5268
    },
    {
      "epoch": 7.338440111420613,
      "grad_norm": 2.4149839878082275,
      "learning_rate": 1.9012222853894257e-05,
      "loss": 0.8391,
      "step": 5269
    },
    {
      "epoch": 7.33983286908078,
      "grad_norm": 2.5299253463745117,
      "learning_rate": 1.9002274019329165e-05,
      "loss": 0.8632,
      "step": 5270
    },
    {
      "epoch": 7.341225626740947,
      "grad_norm": 2.412696123123169,
      "learning_rate": 1.899232518476407e-05,
      "loss": 0.6032,
      "step": 5271
    },
    {
      "epoch": 7.342618384401114,
      "grad_norm": 2.2152018547058105,
      "learning_rate": 1.8982376350198976e-05,
      "loss": 0.5657,
      "step": 5272
    },
    {
      "epoch": 7.344011142061281,
      "grad_norm": 3.163825273513794,
      "learning_rate": 1.8972427515633884e-05,
      "loss": 0.8335,
      "step": 5273
    },
    {
      "epoch": 7.345403899721449,
      "grad_norm": 2.466665029525757,
      "learning_rate": 1.8962478681068788e-05,
      "loss": 0.776,
      "step": 5274
    },
    {
      "epoch": 7.346796657381615,
      "grad_norm": 2.686173439025879,
      "learning_rate": 1.8952529846503695e-05,
      "loss": 0.8854,
      "step": 5275
    },
    {
      "epoch": 7.348189415041783,
      "grad_norm": 2.6543776988983154,
      "learning_rate": 1.89425810119386e-05,
      "loss": 0.7509,
      "step": 5276
    },
    {
      "epoch": 7.3495821727019495,
      "grad_norm": 2.4722702503204346,
      "learning_rate": 1.8932632177373507e-05,
      "loss": 0.7713,
      "step": 5277
    },
    {
      "epoch": 7.350974930362117,
      "grad_norm": 2.601783037185669,
      "learning_rate": 1.8922683342808414e-05,
      "loss": 0.8249,
      "step": 5278
    },
    {
      "epoch": 7.352367688022284,
      "grad_norm": 2.874080181121826,
      "learning_rate": 1.8912734508243318e-05,
      "loss": 0.8515,
      "step": 5279
    },
    {
      "epoch": 7.353760445682451,
      "grad_norm": 2.882293462753296,
      "learning_rate": 1.8902785673678226e-05,
      "loss": 0.8188,
      "step": 5280
    },
    {
      "epoch": 7.355153203342619,
      "grad_norm": 2.140166997909546,
      "learning_rate": 1.889283683911313e-05,
      "loss": 0.5261,
      "step": 5281
    },
    {
      "epoch": 7.3565459610027855,
      "grad_norm": 2.9048073291778564,
      "learning_rate": 1.8882888004548037e-05,
      "loss": 0.8111,
      "step": 5282
    },
    {
      "epoch": 7.357938718662953,
      "grad_norm": 2.34552001953125,
      "learning_rate": 1.887293916998294e-05,
      "loss": 0.7812,
      "step": 5283
    },
    {
      "epoch": 7.35933147632312,
      "grad_norm": 2.349907159805298,
      "learning_rate": 1.886299033541785e-05,
      "loss": 0.6517,
      "step": 5284
    },
    {
      "epoch": 7.360724233983287,
      "grad_norm": 3.0905449390411377,
      "learning_rate": 1.8853041500852756e-05,
      "loss": 0.9429,
      "step": 5285
    },
    {
      "epoch": 7.362116991643454,
      "grad_norm": 2.25343918800354,
      "learning_rate": 1.884309266628766e-05,
      "loss": 0.6692,
      "step": 5286
    },
    {
      "epoch": 7.3635097493036215,
      "grad_norm": 2.5146708488464355,
      "learning_rate": 1.8833143831722568e-05,
      "loss": 0.9155,
      "step": 5287
    },
    {
      "epoch": 7.364902506963788,
      "grad_norm": 2.932051420211792,
      "learning_rate": 1.8823194997157472e-05,
      "loss": 0.7189,
      "step": 5288
    },
    {
      "epoch": 7.366295264623956,
      "grad_norm": 2.626209020614624,
      "learning_rate": 1.881324616259238e-05,
      "loss": 0.7896,
      "step": 5289
    },
    {
      "epoch": 7.367688022284122,
      "grad_norm": 2.668407440185547,
      "learning_rate": 1.8803297328027287e-05,
      "loss": 0.8976,
      "step": 5290
    },
    {
      "epoch": 7.36908077994429,
      "grad_norm": 2.6616017818450928,
      "learning_rate": 1.879334849346219e-05,
      "loss": 0.8108,
      "step": 5291
    },
    {
      "epoch": 7.370473537604457,
      "grad_norm": 2.352094888687134,
      "learning_rate": 1.87833996588971e-05,
      "loss": 0.7182,
      "step": 5292
    },
    {
      "epoch": 7.371866295264624,
      "grad_norm": 4.22088098526001,
      "learning_rate": 1.8773450824332002e-05,
      "loss": 1.0058,
      "step": 5293
    },
    {
      "epoch": 7.373259052924791,
      "grad_norm": 2.9428882598876953,
      "learning_rate": 1.876350198976691e-05,
      "loss": 0.8682,
      "step": 5294
    },
    {
      "epoch": 7.374651810584958,
      "grad_norm": 2.534942865371704,
      "learning_rate": 1.875355315520182e-05,
      "loss": 0.7611,
      "step": 5295
    },
    {
      "epoch": 7.376044568245125,
      "grad_norm": 2.342184543609619,
      "learning_rate": 1.874360432063672e-05,
      "loss": 0.695,
      "step": 5296
    },
    {
      "epoch": 7.3774373259052926,
      "grad_norm": 2.9026925563812256,
      "learning_rate": 1.8733655486071632e-05,
      "loss": 0.7989,
      "step": 5297
    },
    {
      "epoch": 7.378830083565459,
      "grad_norm": 2.5647268295288086,
      "learning_rate": 1.8723706651506536e-05,
      "loss": 0.7226,
      "step": 5298
    },
    {
      "epoch": 7.380222841225627,
      "grad_norm": 2.5180187225341797,
      "learning_rate": 1.8713757816941444e-05,
      "loss": 0.7938,
      "step": 5299
    },
    {
      "epoch": 7.381615598885794,
      "grad_norm": 2.7397608757019043,
      "learning_rate": 1.870380898237635e-05,
      "loss": 0.6367,
      "step": 5300
    },
    {
      "epoch": 7.383008356545961,
      "grad_norm": 2.5732569694519043,
      "learning_rate": 1.8693860147811255e-05,
      "loss": 0.7301,
      "step": 5301
    },
    {
      "epoch": 7.3844011142061285,
      "grad_norm": 2.7441160678863525,
      "learning_rate": 1.8683911313246163e-05,
      "loss": 0.8464,
      "step": 5302
    },
    {
      "epoch": 7.385793871866295,
      "grad_norm": 3.1986751556396484,
      "learning_rate": 1.8673962478681067e-05,
      "loss": 0.9428,
      "step": 5303
    },
    {
      "epoch": 7.387186629526463,
      "grad_norm": 2.3839962482452393,
      "learning_rate": 1.8664013644115974e-05,
      "loss": 0.6604,
      "step": 5304
    },
    {
      "epoch": 7.388579387186629,
      "grad_norm": 2.7603166103363037,
      "learning_rate": 1.865406480955088e-05,
      "loss": 0.9198,
      "step": 5305
    },
    {
      "epoch": 7.389972144846797,
      "grad_norm": 2.3966386318206787,
      "learning_rate": 1.8644115974985786e-05,
      "loss": 0.7653,
      "step": 5306
    },
    {
      "epoch": 7.391364902506964,
      "grad_norm": 2.5830929279327393,
      "learning_rate": 1.8634167140420693e-05,
      "loss": 0.8256,
      "step": 5307
    },
    {
      "epoch": 7.392757660167131,
      "grad_norm": 3.0383450984954834,
      "learning_rate": 1.8624218305855597e-05,
      "loss": 0.9586,
      "step": 5308
    },
    {
      "epoch": 7.394150417827298,
      "grad_norm": 2.53407883644104,
      "learning_rate": 1.8614269471290505e-05,
      "loss": 0.675,
      "step": 5309
    },
    {
      "epoch": 7.395543175487465,
      "grad_norm": 2.7543201446533203,
      "learning_rate": 1.860432063672541e-05,
      "loss": 0.9026,
      "step": 5310
    },
    {
      "epoch": 7.396935933147632,
      "grad_norm": 2.485504627227783,
      "learning_rate": 1.8594371802160316e-05,
      "loss": 0.8323,
      "step": 5311
    },
    {
      "epoch": 7.3983286908078,
      "grad_norm": 2.7948977947235107,
      "learning_rate": 1.8584422967595224e-05,
      "loss": 0.8652,
      "step": 5312
    },
    {
      "epoch": 7.399721448467966,
      "grad_norm": 2.5813651084899902,
      "learning_rate": 1.8574474133030128e-05,
      "loss": 0.8066,
      "step": 5313
    },
    {
      "epoch": 7.401114206128134,
      "grad_norm": 2.6509640216827393,
      "learning_rate": 1.8564525298465035e-05,
      "loss": 0.7535,
      "step": 5314
    },
    {
      "epoch": 7.4025069637883005,
      "grad_norm": 2.642306089401245,
      "learning_rate": 1.855457646389994e-05,
      "loss": 0.8696,
      "step": 5315
    },
    {
      "epoch": 7.403899721448468,
      "grad_norm": 2.6423439979553223,
      "learning_rate": 1.8544627629334847e-05,
      "loss": 0.6899,
      "step": 5316
    },
    {
      "epoch": 7.405292479108635,
      "grad_norm": 2.9575910568237305,
      "learning_rate": 1.8534678794769754e-05,
      "loss": 1.0177,
      "step": 5317
    },
    {
      "epoch": 7.406685236768802,
      "grad_norm": 2.517331123352051,
      "learning_rate": 1.852472996020466e-05,
      "loss": 0.7426,
      "step": 5318
    },
    {
      "epoch": 7.408077994428969,
      "grad_norm": 2.621168851852417,
      "learning_rate": 1.8514781125639566e-05,
      "loss": 0.7847,
      "step": 5319
    },
    {
      "epoch": 7.4094707520891365,
      "grad_norm": 2.5754759311676025,
      "learning_rate": 1.850483229107447e-05,
      "loss": 0.7888,
      "step": 5320
    },
    {
      "epoch": 7.410863509749303,
      "grad_norm": 3.159336805343628,
      "learning_rate": 1.8494883456509378e-05,
      "loss": 0.7203,
      "step": 5321
    },
    {
      "epoch": 7.412256267409471,
      "grad_norm": 2.5982353687286377,
      "learning_rate": 1.848493462194429e-05,
      "loss": 0.7717,
      "step": 5322
    },
    {
      "epoch": 7.413649025069638,
      "grad_norm": 2.3548851013183594,
      "learning_rate": 1.847498578737919e-05,
      "loss": 0.6459,
      "step": 5323
    },
    {
      "epoch": 7.415041782729805,
      "grad_norm": 2.8021745681762695,
      "learning_rate": 1.84650369528141e-05,
      "loss": 0.906,
      "step": 5324
    },
    {
      "epoch": 7.4164345403899725,
      "grad_norm": 3.013658285140991,
      "learning_rate": 1.8455088118249004e-05,
      "loss": 0.8091,
      "step": 5325
    },
    {
      "epoch": 7.417827298050139,
      "grad_norm": 2.7148406505584717,
      "learning_rate": 1.844513928368391e-05,
      "loss": 0.7998,
      "step": 5326
    },
    {
      "epoch": 7.419220055710307,
      "grad_norm": 2.475613594055176,
      "learning_rate": 1.8435190449118816e-05,
      "loss": 0.7506,
      "step": 5327
    },
    {
      "epoch": 7.420612813370473,
      "grad_norm": 2.788057804107666,
      "learning_rate": 1.8425241614553723e-05,
      "loss": 0.7888,
      "step": 5328
    },
    {
      "epoch": 7.422005571030641,
      "grad_norm": 2.3782296180725098,
      "learning_rate": 1.841529277998863e-05,
      "loss": 0.6336,
      "step": 5329
    },
    {
      "epoch": 7.423398328690808,
      "grad_norm": 2.264065980911255,
      "learning_rate": 1.8405343945423535e-05,
      "loss": 0.6319,
      "step": 5330
    },
    {
      "epoch": 7.424791086350975,
      "grad_norm": 2.715252637863159,
      "learning_rate": 1.8395395110858442e-05,
      "loss": 0.8039,
      "step": 5331
    },
    {
      "epoch": 7.426183844011142,
      "grad_norm": 2.309450626373291,
      "learning_rate": 1.8385446276293346e-05,
      "loss": 0.6296,
      "step": 5332
    },
    {
      "epoch": 7.427576601671309,
      "grad_norm": 2.5203335285186768,
      "learning_rate": 1.8375497441728254e-05,
      "loss": 0.8469,
      "step": 5333
    },
    {
      "epoch": 7.428969359331476,
      "grad_norm": 4.202627182006836,
      "learning_rate": 1.836554860716316e-05,
      "loss": 0.819,
      "step": 5334
    },
    {
      "epoch": 7.430362116991644,
      "grad_norm": 2.361088514328003,
      "learning_rate": 1.8355599772598065e-05,
      "loss": 0.7707,
      "step": 5335
    },
    {
      "epoch": 7.43175487465181,
      "grad_norm": 2.1766088008880615,
      "learning_rate": 1.8345650938032973e-05,
      "loss": 0.6533,
      "step": 5336
    },
    {
      "epoch": 7.433147632311978,
      "grad_norm": 3.1537468433380127,
      "learning_rate": 1.8335702103467877e-05,
      "loss": 0.9132,
      "step": 5337
    },
    {
      "epoch": 7.4345403899721445,
      "grad_norm": 2.5507802963256836,
      "learning_rate": 1.8325753268902784e-05,
      "loss": 0.8233,
      "step": 5338
    },
    {
      "epoch": 7.435933147632312,
      "grad_norm": 2.8491318225860596,
      "learning_rate": 1.831580443433769e-05,
      "loss": 0.7669,
      "step": 5339
    },
    {
      "epoch": 7.437325905292479,
      "grad_norm": 2.901625394821167,
      "learning_rate": 1.8305855599772596e-05,
      "loss": 0.8406,
      "step": 5340
    },
    {
      "epoch": 7.438718662952646,
      "grad_norm": 3.1791460514068604,
      "learning_rate": 1.8295906765207503e-05,
      "loss": 0.7758,
      "step": 5341
    },
    {
      "epoch": 7.440111420612814,
      "grad_norm": 3.0664150714874268,
      "learning_rate": 1.8285957930642407e-05,
      "loss": 0.842,
      "step": 5342
    },
    {
      "epoch": 7.4415041782729805,
      "grad_norm": 3.25173020362854,
      "learning_rate": 1.8276009096077315e-05,
      "loss": 0.9466,
      "step": 5343
    },
    {
      "epoch": 7.442896935933147,
      "grad_norm": 2.729762554168701,
      "learning_rate": 1.8266060261512222e-05,
      "loss": 0.846,
      "step": 5344
    },
    {
      "epoch": 7.444289693593315,
      "grad_norm": 2.436539888381958,
      "learning_rate": 1.8256111426947126e-05,
      "loss": 0.6487,
      "step": 5345
    },
    {
      "epoch": 7.445682451253482,
      "grad_norm": 2.9036285877227783,
      "learning_rate": 1.8246162592382034e-05,
      "loss": 0.8984,
      "step": 5346
    },
    {
      "epoch": 7.447075208913649,
      "grad_norm": 2.7688663005828857,
      "learning_rate": 1.8236213757816938e-05,
      "loss": 0.7661,
      "step": 5347
    },
    {
      "epoch": 7.4484679665738165,
      "grad_norm": 2.9770824909210205,
      "learning_rate": 1.8226264923251845e-05,
      "loss": 1.0991,
      "step": 5348
    },
    {
      "epoch": 7.449860724233983,
      "grad_norm": 2.889711856842041,
      "learning_rate": 1.821631608868675e-05,
      "loss": 1.0009,
      "step": 5349
    },
    {
      "epoch": 7.451253481894151,
      "grad_norm": 2.694047212600708,
      "learning_rate": 1.8206367254121657e-05,
      "loss": 0.6695,
      "step": 5350
    },
    {
      "epoch": 7.452646239554317,
      "grad_norm": 2.5336525440216064,
      "learning_rate": 1.8196418419556568e-05,
      "loss": 0.8706,
      "step": 5351
    },
    {
      "epoch": 7.454038997214485,
      "grad_norm": 3.0518062114715576,
      "learning_rate": 1.8186469584991468e-05,
      "loss": 0.9837,
      "step": 5352
    },
    {
      "epoch": 7.455431754874652,
      "grad_norm": 2.8290791511535645,
      "learning_rate": 1.817652075042638e-05,
      "loss": 0.7497,
      "step": 5353
    },
    {
      "epoch": 7.456824512534819,
      "grad_norm": 2.8035120964050293,
      "learning_rate": 1.8166571915861283e-05,
      "loss": 0.8461,
      "step": 5354
    },
    {
      "epoch": 7.458217270194986,
      "grad_norm": 2.4490153789520264,
      "learning_rate": 1.815662308129619e-05,
      "loss": 0.7488,
      "step": 5355
    },
    {
      "epoch": 7.459610027855153,
      "grad_norm": 2.6081271171569824,
      "learning_rate": 1.8146674246731098e-05,
      "loss": 0.7607,
      "step": 5356
    },
    {
      "epoch": 7.46100278551532,
      "grad_norm": 2.4156997203826904,
      "learning_rate": 1.8136725412166002e-05,
      "loss": 0.7671,
      "step": 5357
    },
    {
      "epoch": 7.462395543175488,
      "grad_norm": 2.8010993003845215,
      "learning_rate": 1.812677657760091e-05,
      "loss": 0.7032,
      "step": 5358
    },
    {
      "epoch": 7.463788300835654,
      "grad_norm": 2.6486268043518066,
      "learning_rate": 1.8116827743035814e-05,
      "loss": 0.714,
      "step": 5359
    },
    {
      "epoch": 7.465181058495822,
      "grad_norm": 2.284186840057373,
      "learning_rate": 1.810687890847072e-05,
      "loss": 0.6935,
      "step": 5360
    },
    {
      "epoch": 7.4665738161559885,
      "grad_norm": 7.3192853927612305,
      "learning_rate": 1.809693007390563e-05,
      "loss": 0.631,
      "step": 5361
    },
    {
      "epoch": 7.467966573816156,
      "grad_norm": 3.7031171321868896,
      "learning_rate": 1.8086981239340533e-05,
      "loss": 0.9192,
      "step": 5362
    },
    {
      "epoch": 7.469359331476323,
      "grad_norm": 2.848693370819092,
      "learning_rate": 1.807703240477544e-05,
      "loss": 0.9586,
      "step": 5363
    },
    {
      "epoch": 7.47075208913649,
      "grad_norm": 2.508500814437866,
      "learning_rate": 1.8067083570210344e-05,
      "loss": 0.7331,
      "step": 5364
    },
    {
      "epoch": 7.472144846796658,
      "grad_norm": 2.4824063777923584,
      "learning_rate": 1.805713473564525e-05,
      "loss": 0.7332,
      "step": 5365
    },
    {
      "epoch": 7.4735376044568245,
      "grad_norm": 2.697272300720215,
      "learning_rate": 1.804718590108016e-05,
      "loss": 1.0376,
      "step": 5366
    },
    {
      "epoch": 7.474930362116992,
      "grad_norm": 2.364535331726074,
      "learning_rate": 1.8037237066515063e-05,
      "loss": 0.6744,
      "step": 5367
    },
    {
      "epoch": 7.476323119777159,
      "grad_norm": 2.8499128818511963,
      "learning_rate": 1.802728823194997e-05,
      "loss": 0.9371,
      "step": 5368
    },
    {
      "epoch": 7.477715877437326,
      "grad_norm": 2.528367519378662,
      "learning_rate": 1.8017339397384875e-05,
      "loss": 0.8048,
      "step": 5369
    },
    {
      "epoch": 7.479108635097493,
      "grad_norm": 2.3179337978363037,
      "learning_rate": 1.8007390562819782e-05,
      "loss": 0.6478,
      "step": 5370
    },
    {
      "epoch": 7.4805013927576605,
      "grad_norm": 2.8684136867523193,
      "learning_rate": 1.7997441728254686e-05,
      "loss": 1.0433,
      "step": 5371
    },
    {
      "epoch": 7.481894150417827,
      "grad_norm": 2.4741365909576416,
      "learning_rate": 1.7987492893689594e-05,
      "loss": 0.7669,
      "step": 5372
    },
    {
      "epoch": 7.483286908077995,
      "grad_norm": 3.0352306365966797,
      "learning_rate": 1.79775440591245e-05,
      "loss": 1.1622,
      "step": 5373
    },
    {
      "epoch": 7.484679665738161,
      "grad_norm": 2.81583571434021,
      "learning_rate": 1.7967595224559405e-05,
      "loss": 0.9287,
      "step": 5374
    },
    {
      "epoch": 7.486072423398329,
      "grad_norm": 2.670285701751709,
      "learning_rate": 1.7957646389994313e-05,
      "loss": 0.8969,
      "step": 5375
    },
    {
      "epoch": 7.487465181058496,
      "grad_norm": 2.5652999877929688,
      "learning_rate": 1.7947697555429217e-05,
      "loss": 0.955,
      "step": 5376
    },
    {
      "epoch": 7.488857938718663,
      "grad_norm": 2.609499216079712,
      "learning_rate": 1.7937748720864124e-05,
      "loss": 0.9204,
      "step": 5377
    },
    {
      "epoch": 7.49025069637883,
      "grad_norm": 2.5414206981658936,
      "learning_rate": 1.7927799886299035e-05,
      "loss": 0.6906,
      "step": 5378
    },
    {
      "epoch": 7.491643454038997,
      "grad_norm": 2.760133743286133,
      "learning_rate": 1.7917851051733936e-05,
      "loss": 0.8683,
      "step": 5379
    },
    {
      "epoch": 7.493036211699164,
      "grad_norm": 2.682169198989868,
      "learning_rate": 1.7907902217168847e-05,
      "loss": 0.8578,
      "step": 5380
    },
    {
      "epoch": 7.494428969359332,
      "grad_norm": 2.098057270050049,
      "learning_rate": 1.789795338260375e-05,
      "loss": 0.5941,
      "step": 5381
    },
    {
      "epoch": 7.495821727019498,
      "grad_norm": 3.0723142623901367,
      "learning_rate": 1.7888004548038658e-05,
      "loss": 0.9611,
      "step": 5382
    },
    {
      "epoch": 7.497214484679666,
      "grad_norm": 2.3826351165771484,
      "learning_rate": 1.7878055713473566e-05,
      "loss": 0.6693,
      "step": 5383
    },
    {
      "epoch": 7.498607242339833,
      "grad_norm": 2.4851551055908203,
      "learning_rate": 1.786810687890847e-05,
      "loss": 0.8697,
      "step": 5384
    },
    {
      "epoch": 7.5,
      "grad_norm": 2.494851589202881,
      "learning_rate": 1.7858158044343377e-05,
      "loss": 0.7803,
      "step": 5385
    },
    {
      "epoch": 7.501392757660167,
      "grad_norm": 2.3112099170684814,
      "learning_rate": 1.784820920977828e-05,
      "loss": 0.8052,
      "step": 5386
    },
    {
      "epoch": 7.502785515320334,
      "grad_norm": 2.399679660797119,
      "learning_rate": 1.783826037521319e-05,
      "loss": 0.73,
      "step": 5387
    },
    {
      "epoch": 7.504178272980502,
      "grad_norm": 2.62286639213562,
      "learning_rate": 1.7828311540648096e-05,
      "loss": 0.87,
      "step": 5388
    },
    {
      "epoch": 7.505571030640668,
      "grad_norm": 3.6493852138519287,
      "learning_rate": 1.7818362706083e-05,
      "loss": 0.6422,
      "step": 5389
    },
    {
      "epoch": 7.506963788300836,
      "grad_norm": 2.447942018508911,
      "learning_rate": 1.7808413871517908e-05,
      "loss": 0.7403,
      "step": 5390
    },
    {
      "epoch": 7.508356545961003,
      "grad_norm": 2.873213768005371,
      "learning_rate": 1.7798465036952812e-05,
      "loss": 0.7787,
      "step": 5391
    },
    {
      "epoch": 7.50974930362117,
      "grad_norm": 2.470106840133667,
      "learning_rate": 1.778851620238772e-05,
      "loss": 0.8113,
      "step": 5392
    },
    {
      "epoch": 7.511142061281337,
      "grad_norm": 2.59989070892334,
      "learning_rate": 1.7778567367822623e-05,
      "loss": 0.7656,
      "step": 5393
    },
    {
      "epoch": 7.512534818941504,
      "grad_norm": 2.6054182052612305,
      "learning_rate": 1.776861853325753e-05,
      "loss": 0.6932,
      "step": 5394
    },
    {
      "epoch": 7.513927576601671,
      "grad_norm": 2.237776041030884,
      "learning_rate": 1.775866969869244e-05,
      "loss": 0.6018,
      "step": 5395
    },
    {
      "epoch": 7.515320334261839,
      "grad_norm": 2.5836374759674072,
      "learning_rate": 1.7748720864127342e-05,
      "loss": 0.7404,
      "step": 5396
    },
    {
      "epoch": 7.516713091922005,
      "grad_norm": 2.885637044906616,
      "learning_rate": 1.773877202956225e-05,
      "loss": 0.7564,
      "step": 5397
    },
    {
      "epoch": 7.518105849582173,
      "grad_norm": 2.3777401447296143,
      "learning_rate": 1.7728823194997154e-05,
      "loss": 0.8696,
      "step": 5398
    },
    {
      "epoch": 7.5194986072423395,
      "grad_norm": 2.781888484954834,
      "learning_rate": 1.771887436043206e-05,
      "loss": 0.9645,
      "step": 5399
    },
    {
      "epoch": 7.520891364902507,
      "grad_norm": 2.661517381668091,
      "learning_rate": 1.770892552586697e-05,
      "loss": 0.8297,
      "step": 5400
    },
    {
      "epoch": 7.522284122562674,
      "grad_norm": 2.6107850074768066,
      "learning_rate": 1.7698976691301873e-05,
      "loss": 0.8516,
      "step": 5401
    },
    {
      "epoch": 7.523676880222841,
      "grad_norm": 2.5141971111297607,
      "learning_rate": 1.768902785673678e-05,
      "loss": 0.8158,
      "step": 5402
    },
    {
      "epoch": 7.525069637883008,
      "grad_norm": 3.4384422302246094,
      "learning_rate": 1.7679079022171684e-05,
      "loss": 0.8487,
      "step": 5403
    },
    {
      "epoch": 7.5264623955431755,
      "grad_norm": 2.5683579444885254,
      "learning_rate": 1.7669130187606592e-05,
      "loss": 0.8595,
      "step": 5404
    },
    {
      "epoch": 7.527855153203342,
      "grad_norm": 2.845391035079956,
      "learning_rate": 1.76591813530415e-05,
      "loss": 0.8601,
      "step": 5405
    },
    {
      "epoch": 7.52924791086351,
      "grad_norm": 2.3391435146331787,
      "learning_rate": 1.7649232518476403e-05,
      "loss": 0.6663,
      "step": 5406
    },
    {
      "epoch": 7.530640668523677,
      "grad_norm": 2.5470023155212402,
      "learning_rate": 1.7639283683911314e-05,
      "loss": 0.824,
      "step": 5407
    },
    {
      "epoch": 7.532033426183844,
      "grad_norm": 2.9565205574035645,
      "learning_rate": 1.7629334849346215e-05,
      "loss": 0.9813,
      "step": 5408
    },
    {
      "epoch": 7.5334261838440115,
      "grad_norm": 2.5947036743164062,
      "learning_rate": 1.7619386014781126e-05,
      "loss": 0.8166,
      "step": 5409
    },
    {
      "epoch": 7.534818941504178,
      "grad_norm": 2.6413090229034424,
      "learning_rate": 1.7609437180216033e-05,
      "loss": 0.6797,
      "step": 5410
    },
    {
      "epoch": 7.536211699164346,
      "grad_norm": 2.1500332355499268,
      "learning_rate": 1.7599488345650937e-05,
      "loss": 0.555,
      "step": 5411
    },
    {
      "epoch": 7.537604456824512,
      "grad_norm": 2.5348427295684814,
      "learning_rate": 1.7589539511085845e-05,
      "loss": 0.7153,
      "step": 5412
    },
    {
      "epoch": 7.53899721448468,
      "grad_norm": 2.384018898010254,
      "learning_rate": 1.757959067652075e-05,
      "loss": 0.7399,
      "step": 5413
    },
    {
      "epoch": 7.540389972144847,
      "grad_norm": 2.929490566253662,
      "learning_rate": 1.7569641841955656e-05,
      "loss": 0.8884,
      "step": 5414
    },
    {
      "epoch": 7.541782729805014,
      "grad_norm": 2.9986398220062256,
      "learning_rate": 1.755969300739056e-05,
      "loss": 0.6664,
      "step": 5415
    },
    {
      "epoch": 7.543175487465181,
      "grad_norm": 2.725666046142578,
      "learning_rate": 1.7549744172825468e-05,
      "loss": 0.8639,
      "step": 5416
    },
    {
      "epoch": 7.544568245125348,
      "grad_norm": 2.7333314418792725,
      "learning_rate": 1.7539795338260375e-05,
      "loss": 0.8251,
      "step": 5417
    },
    {
      "epoch": 7.545961002785515,
      "grad_norm": 2.9461610317230225,
      "learning_rate": 1.752984650369528e-05,
      "loss": 0.8615,
      "step": 5418
    },
    {
      "epoch": 7.547353760445683,
      "grad_norm": 2.54337477684021,
      "learning_rate": 1.7519897669130187e-05,
      "loss": 0.8794,
      "step": 5419
    },
    {
      "epoch": 7.548746518105849,
      "grad_norm": 6.415521621704102,
      "learning_rate": 1.750994883456509e-05,
      "loss": 0.7321,
      "step": 5420
    },
    {
      "epoch": 7.550139275766017,
      "grad_norm": 1.941736102104187,
      "learning_rate": 1.75e-05,
      "loss": 0.5779,
      "step": 5421
    },
    {
      "epoch": 7.5515320334261835,
      "grad_norm": 2.8460235595703125,
      "learning_rate": 1.7490051165434906e-05,
      "loss": 0.8093,
      "step": 5422
    },
    {
      "epoch": 7.552924791086351,
      "grad_norm": 2.7201473712921143,
      "learning_rate": 1.748010233086981e-05,
      "loss": 0.8864,
      "step": 5423
    },
    {
      "epoch": 7.554317548746518,
      "grad_norm": 3.1468052864074707,
      "learning_rate": 1.7470153496304717e-05,
      "loss": 0.6753,
      "step": 5424
    },
    {
      "epoch": 7.555710306406685,
      "grad_norm": 2.8339996337890625,
      "learning_rate": 1.7460204661739625e-05,
      "loss": 0.8595,
      "step": 5425
    },
    {
      "epoch": 7.557103064066853,
      "grad_norm": 2.769394874572754,
      "learning_rate": 1.745025582717453e-05,
      "loss": 0.9072,
      "step": 5426
    },
    {
      "epoch": 7.5584958217270195,
      "grad_norm": 2.3809988498687744,
      "learning_rate": 1.7440306992609436e-05,
      "loss": 0.6453,
      "step": 5427
    },
    {
      "epoch": 7.559888579387186,
      "grad_norm": 2.7767210006713867,
      "learning_rate": 1.743035815804434e-05,
      "loss": 0.823,
      "step": 5428
    },
    {
      "epoch": 7.561281337047354,
      "grad_norm": 2.5831570625305176,
      "learning_rate": 1.7420409323479248e-05,
      "loss": 0.7418,
      "step": 5429
    },
    {
      "epoch": 7.562674094707521,
      "grad_norm": 3.282571315765381,
      "learning_rate": 1.7410460488914155e-05,
      "loss": 1.0822,
      "step": 5430
    },
    {
      "epoch": 7.564066852367688,
      "grad_norm": 2.7830049991607666,
      "learning_rate": 1.740051165434906e-05,
      "loss": 0.9058,
      "step": 5431
    },
    {
      "epoch": 7.5654596100278555,
      "grad_norm": 2.606900691986084,
      "learning_rate": 1.7390562819783967e-05,
      "loss": 0.7775,
      "step": 5432
    },
    {
      "epoch": 7.566852367688022,
      "grad_norm": 2.7287631034851074,
      "learning_rate": 1.738061398521887e-05,
      "loss": 0.9519,
      "step": 5433
    },
    {
      "epoch": 7.56824512534819,
      "grad_norm": 2.7785258293151855,
      "learning_rate": 1.737066515065378e-05,
      "loss": 0.8458,
      "step": 5434
    },
    {
      "epoch": 7.569637883008356,
      "grad_norm": 2.9038608074188232,
      "learning_rate": 1.7360716316088686e-05,
      "loss": 1.0261,
      "step": 5435
    },
    {
      "epoch": 7.571030640668524,
      "grad_norm": 2.3288161754608154,
      "learning_rate": 1.7350767481523593e-05,
      "loss": 0.643,
      "step": 5436
    },
    {
      "epoch": 7.572423398328691,
      "grad_norm": 2.881404161453247,
      "learning_rate": 1.7340818646958498e-05,
      "loss": 0.8509,
      "step": 5437
    },
    {
      "epoch": 7.573816155988858,
      "grad_norm": 2.6631224155426025,
      "learning_rate": 1.7330869812393405e-05,
      "loss": 0.5832,
      "step": 5438
    },
    {
      "epoch": 7.575208913649025,
      "grad_norm": 2.52154278755188,
      "learning_rate": 1.732092097782831e-05,
      "loss": 0.8464,
      "step": 5439
    },
    {
      "epoch": 7.576601671309192,
      "grad_norm": 2.851818561553955,
      "learning_rate": 1.7310972143263217e-05,
      "loss": 0.7793,
      "step": 5440
    },
    {
      "epoch": 7.577994428969359,
      "grad_norm": 2.672607421875,
      "learning_rate": 1.7301023308698124e-05,
      "loss": 0.7822,
      "step": 5441
    },
    {
      "epoch": 7.579387186629527,
      "grad_norm": 2.48591685295105,
      "learning_rate": 1.7291074474133028e-05,
      "loss": 0.7613,
      "step": 5442
    },
    {
      "epoch": 7.580779944289693,
      "grad_norm": 2.7468457221984863,
      "learning_rate": 1.7281125639567936e-05,
      "loss": 0.8192,
      "step": 5443
    },
    {
      "epoch": 7.582172701949861,
      "grad_norm": 2.4549994468688965,
      "learning_rate": 1.727117680500284e-05,
      "loss": 0.6692,
      "step": 5444
    },
    {
      "epoch": 7.5835654596100275,
      "grad_norm": 3.481426477432251,
      "learning_rate": 1.7261227970437747e-05,
      "loss": 0.8286,
      "step": 5445
    },
    {
      "epoch": 7.584958217270195,
      "grad_norm": 2.9519455432891846,
      "learning_rate": 1.7251279135872655e-05,
      "loss": 0.9889,
      "step": 5446
    },
    {
      "epoch": 7.586350974930362,
      "grad_norm": 4.644660949707031,
      "learning_rate": 1.724133030130756e-05,
      "loss": 0.961,
      "step": 5447
    },
    {
      "epoch": 7.587743732590529,
      "grad_norm": 2.657379388809204,
      "learning_rate": 1.7231381466742466e-05,
      "loss": 0.7251,
      "step": 5448
    },
    {
      "epoch": 7.589136490250697,
      "grad_norm": 2.662083148956299,
      "learning_rate": 1.722143263217737e-05,
      "loss": 0.7343,
      "step": 5449
    },
    {
      "epoch": 7.5905292479108635,
      "grad_norm": 2.9999523162841797,
      "learning_rate": 1.7211483797612278e-05,
      "loss": 0.9716,
      "step": 5450
    },
    {
      "epoch": 7.591922005571031,
      "grad_norm": 2.435251474380493,
      "learning_rate": 1.7201534963047185e-05,
      "loss": 0.6349,
      "step": 5451
    },
    {
      "epoch": 7.593314763231198,
      "grad_norm": 2.6481752395629883,
      "learning_rate": 1.7191586128482093e-05,
      "loss": 0.7385,
      "step": 5452
    },
    {
      "epoch": 7.594707520891365,
      "grad_norm": 2.7415566444396973,
      "learning_rate": 1.7181637293916997e-05,
      "loss": 1.0074,
      "step": 5453
    },
    {
      "epoch": 7.596100278551532,
      "grad_norm": 2.5755679607391357,
      "learning_rate": 1.7171688459351904e-05,
      "loss": 0.7337,
      "step": 5454
    },
    {
      "epoch": 7.5974930362116995,
      "grad_norm": 2.529637575149536,
      "learning_rate": 1.7161739624786808e-05,
      "loss": 0.7974,
      "step": 5455
    },
    {
      "epoch": 7.598885793871866,
      "grad_norm": 2.3905506134033203,
      "learning_rate": 1.7151790790221716e-05,
      "loss": 0.7598,
      "step": 5456
    },
    {
      "epoch": 7.600278551532034,
      "grad_norm": 2.273625612258911,
      "learning_rate": 1.7141841955656623e-05,
      "loss": 0.6601,
      "step": 5457
    },
    {
      "epoch": 7.6016713091922,
      "grad_norm": 2.8713133335113525,
      "learning_rate": 1.7131893121091527e-05,
      "loss": 1.0206,
      "step": 5458
    },
    {
      "epoch": 7.603064066852368,
      "grad_norm": 2.36533784866333,
      "learning_rate": 1.7121944286526435e-05,
      "loss": 0.6262,
      "step": 5459
    },
    {
      "epoch": 7.604456824512535,
      "grad_norm": 2.8157126903533936,
      "learning_rate": 1.711199545196134e-05,
      "loss": 0.7364,
      "step": 5460
    },
    {
      "epoch": 7.605849582172702,
      "grad_norm": 2.8216311931610107,
      "learning_rate": 1.7102046617396246e-05,
      "loss": 0.7561,
      "step": 5461
    },
    {
      "epoch": 7.607242339832869,
      "grad_norm": 2.4692447185516357,
      "learning_rate": 1.709209778283115e-05,
      "loss": 0.8601,
      "step": 5462
    },
    {
      "epoch": 7.608635097493036,
      "grad_norm": 2.7294795513153076,
      "learning_rate": 1.708214894826606e-05,
      "loss": 0.735,
      "step": 5463
    },
    {
      "epoch": 7.610027855153203,
      "grad_norm": 2.7090420722961426,
      "learning_rate": 1.7072200113700965e-05,
      "loss": 0.7506,
      "step": 5464
    },
    {
      "epoch": 7.611420612813371,
      "grad_norm": 2.555973768234253,
      "learning_rate": 1.7062251279135873e-05,
      "loss": 0.6659,
      "step": 5465
    },
    {
      "epoch": 7.612813370473537,
      "grad_norm": 2.3970787525177,
      "learning_rate": 1.7052302444570777e-05,
      "loss": 0.6477,
      "step": 5466
    },
    {
      "epoch": 7.614206128133705,
      "grad_norm": 2.8345634937286377,
      "learning_rate": 1.7042353610005684e-05,
      "loss": 0.9087,
      "step": 5467
    },
    {
      "epoch": 7.615598885793872,
      "grad_norm": 2.9403674602508545,
      "learning_rate": 1.703240477544059e-05,
      "loss": 0.8148,
      "step": 5468
    },
    {
      "epoch": 7.616991643454039,
      "grad_norm": 3.1836977005004883,
      "learning_rate": 1.7022455940875496e-05,
      "loss": 0.7056,
      "step": 5469
    },
    {
      "epoch": 7.618384401114206,
      "grad_norm": 2.4465065002441406,
      "learning_rate": 1.7012507106310403e-05,
      "loss": 0.728,
      "step": 5470
    },
    {
      "epoch": 7.619777158774373,
      "grad_norm": 2.708162307739258,
      "learning_rate": 1.7002558271745307e-05,
      "loss": 0.8091,
      "step": 5471
    },
    {
      "epoch": 7.621169916434541,
      "grad_norm": 2.887277603149414,
      "learning_rate": 1.6992609437180215e-05,
      "loss": 0.9227,
      "step": 5472
    },
    {
      "epoch": 7.6225626740947074,
      "grad_norm": 2.8316197395324707,
      "learning_rate": 1.698266060261512e-05,
      "loss": 0.9379,
      "step": 5473
    },
    {
      "epoch": 7.623955431754875,
      "grad_norm": 3.01243257522583,
      "learning_rate": 1.6972711768050026e-05,
      "loss": 1.0515,
      "step": 5474
    },
    {
      "epoch": 7.625348189415042,
      "grad_norm": 2.652139186859131,
      "learning_rate": 1.6962762933484934e-05,
      "loss": 0.7844,
      "step": 5475
    },
    {
      "epoch": 7.626740947075209,
      "grad_norm": 1.931072473526001,
      "learning_rate": 1.6952814098919838e-05,
      "loss": 0.5464,
      "step": 5476
    },
    {
      "epoch": 7.628133704735376,
      "grad_norm": 2.547229528427124,
      "learning_rate": 1.6942865264354745e-05,
      "loss": 0.5843,
      "step": 5477
    },
    {
      "epoch": 7.629526462395543,
      "grad_norm": 2.6638574600219727,
      "learning_rate": 1.6932916429789653e-05,
      "loss": 0.9491,
      "step": 5478
    },
    {
      "epoch": 7.63091922005571,
      "grad_norm": 2.8186087608337402,
      "learning_rate": 1.692296759522456e-05,
      "loss": 0.9055,
      "step": 5479
    },
    {
      "epoch": 7.632311977715878,
      "grad_norm": 2.598853349685669,
      "learning_rate": 1.6913018760659464e-05,
      "loss": 0.6766,
      "step": 5480
    },
    {
      "epoch": 7.633704735376044,
      "grad_norm": 2.5343191623687744,
      "learning_rate": 1.6903069926094372e-05,
      "loss": 0.6529,
      "step": 5481
    },
    {
      "epoch": 7.635097493036212,
      "grad_norm": 2.8590381145477295,
      "learning_rate": 1.6893121091529276e-05,
      "loss": 0.98,
      "step": 5482
    },
    {
      "epoch": 7.6364902506963785,
      "grad_norm": 2.413297653198242,
      "learning_rate": 1.6883172256964183e-05,
      "loss": 0.8172,
      "step": 5483
    },
    {
      "epoch": 7.637883008356546,
      "grad_norm": 2.6304757595062256,
      "learning_rate": 1.6873223422399087e-05,
      "loss": 0.8413,
      "step": 5484
    },
    {
      "epoch": 7.639275766016713,
      "grad_norm": 2.8911056518554688,
      "learning_rate": 1.6863274587833995e-05,
      "loss": 0.9724,
      "step": 5485
    },
    {
      "epoch": 7.64066852367688,
      "grad_norm": 3.1047544479370117,
      "learning_rate": 1.6853325753268902e-05,
      "loss": 0.8266,
      "step": 5486
    },
    {
      "epoch": 7.642061281337047,
      "grad_norm": 2.703200340270996,
      "learning_rate": 1.6843376918703806e-05,
      "loss": 0.9002,
      "step": 5487
    },
    {
      "epoch": 7.6434540389972145,
      "grad_norm": 2.464510440826416,
      "learning_rate": 1.6833428084138714e-05,
      "loss": 0.6793,
      "step": 5488
    },
    {
      "epoch": 7.644846796657381,
      "grad_norm": 3.0875465869903564,
      "learning_rate": 1.6823479249573618e-05,
      "loss": 0.8917,
      "step": 5489
    },
    {
      "epoch": 7.646239554317549,
      "grad_norm": 1.9251242876052856,
      "learning_rate": 1.681353041500853e-05,
      "loss": 0.4798,
      "step": 5490
    },
    {
      "epoch": 7.647632311977716,
      "grad_norm": 2.643867254257202,
      "learning_rate": 1.6803581580443433e-05,
      "loss": 0.8798,
      "step": 5491
    },
    {
      "epoch": 7.649025069637883,
      "grad_norm": 2.3868167400360107,
      "learning_rate": 1.679363274587834e-05,
      "loss": 0.8146,
      "step": 5492
    },
    {
      "epoch": 7.65041782729805,
      "grad_norm": 2.5355112552642822,
      "learning_rate": 1.6783683911313244e-05,
      "loss": 0.7204,
      "step": 5493
    },
    {
      "epoch": 7.651810584958217,
      "grad_norm": 3.13287091255188,
      "learning_rate": 1.6773735076748152e-05,
      "loss": 1.0552,
      "step": 5494
    },
    {
      "epoch": 7.653203342618385,
      "grad_norm": 2.4425160884857178,
      "learning_rate": 1.6763786242183056e-05,
      "loss": 0.8627,
      "step": 5495
    },
    {
      "epoch": 7.654596100278551,
      "grad_norm": 2.3554141521453857,
      "learning_rate": 1.6753837407617963e-05,
      "loss": 0.6306,
      "step": 5496
    },
    {
      "epoch": 7.655988857938719,
      "grad_norm": 2.2212038040161133,
      "learning_rate": 1.674388857305287e-05,
      "loss": 0.6204,
      "step": 5497
    },
    {
      "epoch": 7.657381615598886,
      "grad_norm": 2.7523720264434814,
      "learning_rate": 1.6733939738487775e-05,
      "loss": 0.796,
      "step": 5498
    },
    {
      "epoch": 7.658774373259053,
      "grad_norm": 2.4031729698181152,
      "learning_rate": 1.6723990903922682e-05,
      "loss": 0.8283,
      "step": 5499
    },
    {
      "epoch": 7.66016713091922,
      "grad_norm": 2.490022659301758,
      "learning_rate": 1.6714042069357586e-05,
      "loss": 0.6693,
      "step": 5500
    },
    {
      "epoch": 7.661559888579387,
      "grad_norm": 2.371617078781128,
      "learning_rate": 1.6704093234792494e-05,
      "loss": 0.6041,
      "step": 5501
    },
    {
      "epoch": 7.662952646239554,
      "grad_norm": 2.6778881549835205,
      "learning_rate": 1.66941444002274e-05,
      "loss": 0.6822,
      "step": 5502
    },
    {
      "epoch": 7.664345403899722,
      "grad_norm": 2.798565149307251,
      "learning_rate": 1.6684195565662305e-05,
      "loss": 0.8008,
      "step": 5503
    },
    {
      "epoch": 7.665738161559888,
      "grad_norm": 2.421142101287842,
      "learning_rate": 1.6674246731097213e-05,
      "loss": 0.7211,
      "step": 5504
    },
    {
      "epoch": 7.667130919220056,
      "grad_norm": 2.256908416748047,
      "learning_rate": 1.6664297896532117e-05,
      "loss": 0.5448,
      "step": 5505
    },
    {
      "epoch": 7.6685236768802225,
      "grad_norm": 2.430297374725342,
      "learning_rate": 1.6654349061967024e-05,
      "loss": 0.674,
      "step": 5506
    },
    {
      "epoch": 7.66991643454039,
      "grad_norm": 2.560138463973999,
      "learning_rate": 1.6644400227401932e-05,
      "loss": 0.7353,
      "step": 5507
    },
    {
      "epoch": 7.671309192200557,
      "grad_norm": 2.754990577697754,
      "learning_rate": 1.663445139283684e-05,
      "loss": 0.7594,
      "step": 5508
    },
    {
      "epoch": 7.672701949860724,
      "grad_norm": 2.3859455585479736,
      "learning_rate": 1.6624502558271743e-05,
      "loss": 0.6086,
      "step": 5509
    },
    {
      "epoch": 7.674094707520892,
      "grad_norm": 2.340557098388672,
      "learning_rate": 1.661455372370665e-05,
      "loss": 0.6464,
      "step": 5510
    },
    {
      "epoch": 7.6754874651810585,
      "grad_norm": 2.4551546573638916,
      "learning_rate": 1.6604604889141555e-05,
      "loss": 0.6357,
      "step": 5511
    },
    {
      "epoch": 7.676880222841225,
      "grad_norm": 2.8065202236175537,
      "learning_rate": 1.6594656054576462e-05,
      "loss": 0.8873,
      "step": 5512
    },
    {
      "epoch": 7.678272980501393,
      "grad_norm": 2.6035358905792236,
      "learning_rate": 1.658470722001137e-05,
      "loss": 0.7739,
      "step": 5513
    },
    {
      "epoch": 7.67966573816156,
      "grad_norm": 2.8600263595581055,
      "learning_rate": 1.6574758385446274e-05,
      "loss": 0.9284,
      "step": 5514
    },
    {
      "epoch": 7.681058495821727,
      "grad_norm": 2.302539587020874,
      "learning_rate": 1.656480955088118e-05,
      "loss": 0.7256,
      "step": 5515
    },
    {
      "epoch": 7.6824512534818945,
      "grad_norm": 3.3458428382873535,
      "learning_rate": 1.6554860716316085e-05,
      "loss": 0.8755,
      "step": 5516
    },
    {
      "epoch": 7.683844011142061,
      "grad_norm": 2.4608819484710693,
      "learning_rate": 1.6544911881750993e-05,
      "loss": 0.7326,
      "step": 5517
    },
    {
      "epoch": 7.685236768802229,
      "grad_norm": 2.4221181869506836,
      "learning_rate": 1.65349630471859e-05,
      "loss": 0.6242,
      "step": 5518
    },
    {
      "epoch": 7.686629526462395,
      "grad_norm": 2.9095842838287354,
      "learning_rate": 1.6525014212620808e-05,
      "loss": 0.9393,
      "step": 5519
    },
    {
      "epoch": 7.688022284122563,
      "grad_norm": 2.33331298828125,
      "learning_rate": 1.6515065378055712e-05,
      "loss": 0.6507,
      "step": 5520
    },
    {
      "epoch": 7.68941504178273,
      "grad_norm": 2.502535820007324,
      "learning_rate": 1.650511654349062e-05,
      "loss": 0.7312,
      "step": 5521
    },
    {
      "epoch": 7.690807799442897,
      "grad_norm": 2.9326870441436768,
      "learning_rate": 1.6495167708925523e-05,
      "loss": 1.0834,
      "step": 5522
    },
    {
      "epoch": 7.692200557103064,
      "grad_norm": 2.883514165878296,
      "learning_rate": 1.648521887436043e-05,
      "loss": 0.7977,
      "step": 5523
    },
    {
      "epoch": 7.693593314763231,
      "grad_norm": 2.428057909011841,
      "learning_rate": 1.647527003979534e-05,
      "loss": 0.8281,
      "step": 5524
    },
    {
      "epoch": 7.694986072423398,
      "grad_norm": 2.342104434967041,
      "learning_rate": 1.6465321205230242e-05,
      "loss": 0.6763,
      "step": 5525
    },
    {
      "epoch": 7.696378830083566,
      "grad_norm": 2.424226999282837,
      "learning_rate": 1.645537237066515e-05,
      "loss": 0.6285,
      "step": 5526
    },
    {
      "epoch": 7.697771587743732,
      "grad_norm": 2.470301389694214,
      "learning_rate": 1.6445423536100054e-05,
      "loss": 0.819,
      "step": 5527
    },
    {
      "epoch": 7.6991643454039,
      "grad_norm": 2.714398145675659,
      "learning_rate": 1.643547470153496e-05,
      "loss": 0.8148,
      "step": 5528
    },
    {
      "epoch": 7.7005571030640665,
      "grad_norm": 2.368927001953125,
      "learning_rate": 1.642552586696987e-05,
      "loss": 0.7107,
      "step": 5529
    },
    {
      "epoch": 7.701949860724234,
      "grad_norm": 2.540170669555664,
      "learning_rate": 1.6415577032404773e-05,
      "loss": 0.8445,
      "step": 5530
    },
    {
      "epoch": 7.703342618384401,
      "grad_norm": 2.7775514125823975,
      "learning_rate": 1.640562819783968e-05,
      "loss": 0.7228,
      "step": 5531
    },
    {
      "epoch": 7.704735376044568,
      "grad_norm": 2.2772164344787598,
      "learning_rate": 1.6395679363274585e-05,
      "loss": 0.806,
      "step": 5532
    },
    {
      "epoch": 7.706128133704736,
      "grad_norm": 2.8098080158233643,
      "learning_rate": 1.6385730528709492e-05,
      "loss": 0.8471,
      "step": 5533
    },
    {
      "epoch": 7.7075208913649025,
      "grad_norm": 2.630612850189209,
      "learning_rate": 1.63757816941444e-05,
      "loss": 0.7689,
      "step": 5534
    },
    {
      "epoch": 7.708913649025069,
      "grad_norm": 2.955442190170288,
      "learning_rate": 1.6365832859579307e-05,
      "loss": 0.7642,
      "step": 5535
    },
    {
      "epoch": 7.710306406685237,
      "grad_norm": 3.060309648513794,
      "learning_rate": 1.635588402501421e-05,
      "loss": 0.7033,
      "step": 5536
    },
    {
      "epoch": 7.711699164345404,
      "grad_norm": 2.650979518890381,
      "learning_rate": 1.634593519044912e-05,
      "loss": 0.6853,
      "step": 5537
    },
    {
      "epoch": 7.713091922005571,
      "grad_norm": 2.5225212574005127,
      "learning_rate": 1.6335986355884023e-05,
      "loss": 0.7362,
      "step": 5538
    },
    {
      "epoch": 7.7144846796657385,
      "grad_norm": 2.7491307258605957,
      "learning_rate": 1.632603752131893e-05,
      "loss": 0.8206,
      "step": 5539
    },
    {
      "epoch": 7.715877437325905,
      "grad_norm": 2.786600351333618,
      "learning_rate": 1.6316088686753837e-05,
      "loss": 0.8513,
      "step": 5540
    },
    {
      "epoch": 7.717270194986073,
      "grad_norm": 2.575843334197998,
      "learning_rate": 1.630613985218874e-05,
      "loss": 0.8263,
      "step": 5541
    },
    {
      "epoch": 7.718662952646239,
      "grad_norm": 2.4344546794891357,
      "learning_rate": 1.629619101762365e-05,
      "loss": 0.7748,
      "step": 5542
    },
    {
      "epoch": 7.720055710306407,
      "grad_norm": 2.481569290161133,
      "learning_rate": 1.6286242183058553e-05,
      "loss": 0.6716,
      "step": 5543
    },
    {
      "epoch": 7.721448467966574,
      "grad_norm": 2.69612979888916,
      "learning_rate": 1.627629334849346e-05,
      "loss": 0.9076,
      "step": 5544
    },
    {
      "epoch": 7.722841225626741,
      "grad_norm": 2.8241164684295654,
      "learning_rate": 1.6266344513928365e-05,
      "loss": 0.8286,
      "step": 5545
    },
    {
      "epoch": 7.724233983286908,
      "grad_norm": 2.8578298091888428,
      "learning_rate": 1.6256395679363275e-05,
      "loss": 0.7248,
      "step": 5546
    },
    {
      "epoch": 7.725626740947075,
      "grad_norm": 2.6153275966644287,
      "learning_rate": 1.624644684479818e-05,
      "loss": 0.9127,
      "step": 5547
    },
    {
      "epoch": 7.727019498607242,
      "grad_norm": 4.352839469909668,
      "learning_rate": 1.6236498010233087e-05,
      "loss": 0.7752,
      "step": 5548
    },
    {
      "epoch": 7.72841225626741,
      "grad_norm": 2.616942882537842,
      "learning_rate": 1.622654917566799e-05,
      "loss": 0.7326,
      "step": 5549
    },
    {
      "epoch": 7.729805013927576,
      "grad_norm": 2.4175174236297607,
      "learning_rate": 1.62166003411029e-05,
      "loss": 0.7237,
      "step": 5550
    },
    {
      "epoch": 7.731197771587744,
      "grad_norm": 3.226278781890869,
      "learning_rate": 1.6206651506537806e-05,
      "loss": 0.861,
      "step": 5551
    },
    {
      "epoch": 7.732590529247911,
      "grad_norm": 3.5635745525360107,
      "learning_rate": 1.619670267197271e-05,
      "loss": 0.7789,
      "step": 5552
    },
    {
      "epoch": 7.733983286908078,
      "grad_norm": 2.7078680992126465,
      "learning_rate": 1.6186753837407618e-05,
      "loss": 0.7734,
      "step": 5553
    },
    {
      "epoch": 7.735376044568245,
      "grad_norm": 2.684532403945923,
      "learning_rate": 1.617680500284252e-05,
      "loss": 0.7954,
      "step": 5554
    },
    {
      "epoch": 7.736768802228412,
      "grad_norm": 2.4593377113342285,
      "learning_rate": 1.616685616827743e-05,
      "loss": 0.7353,
      "step": 5555
    },
    {
      "epoch": 7.73816155988858,
      "grad_norm": 2.7976765632629395,
      "learning_rate": 1.6156907333712333e-05,
      "loss": 0.7252,
      "step": 5556
    },
    {
      "epoch": 7.7395543175487465,
      "grad_norm": 2.5154716968536377,
      "learning_rate": 1.614695849914724e-05,
      "loss": 0.7299,
      "step": 5557
    },
    {
      "epoch": 7.740947075208914,
      "grad_norm": 2.710759162902832,
      "learning_rate": 1.6137009664582148e-05,
      "loss": 0.884,
      "step": 5558
    },
    {
      "epoch": 7.742339832869081,
      "grad_norm": 2.633753538131714,
      "learning_rate": 1.6127060830017052e-05,
      "loss": 0.7493,
      "step": 5559
    },
    {
      "epoch": 7.743732590529248,
      "grad_norm": 2.786937713623047,
      "learning_rate": 1.611711199545196e-05,
      "loss": 0.8833,
      "step": 5560
    },
    {
      "epoch": 7.745125348189415,
      "grad_norm": 2.3996152877807617,
      "learning_rate": 1.6107163160886864e-05,
      "loss": 0.5208,
      "step": 5561
    },
    {
      "epoch": 7.7465181058495824,
      "grad_norm": 2.7960236072540283,
      "learning_rate": 1.6097214326321775e-05,
      "loss": 0.8819,
      "step": 5562
    },
    {
      "epoch": 7.747910863509749,
      "grad_norm": 2.499798536300659,
      "learning_rate": 1.608726549175668e-05,
      "loss": 0.6926,
      "step": 5563
    },
    {
      "epoch": 7.749303621169917,
      "grad_norm": 2.8297054767608643,
      "learning_rate": 1.6077316657191586e-05,
      "loss": 0.8726,
      "step": 5564
    },
    {
      "epoch": 7.750696378830083,
      "grad_norm": 2.852834701538086,
      "learning_rate": 1.606736782262649e-05,
      "loss": 0.9254,
      "step": 5565
    },
    {
      "epoch": 7.752089136490251,
      "grad_norm": 2.609112501144409,
      "learning_rate": 1.6057418988061398e-05,
      "loss": 0.6887,
      "step": 5566
    },
    {
      "epoch": 7.7534818941504176,
      "grad_norm": 2.426393985748291,
      "learning_rate": 1.6047470153496302e-05,
      "loss": 0.7099,
      "step": 5567
    },
    {
      "epoch": 7.754874651810585,
      "grad_norm": 3.0347394943237305,
      "learning_rate": 1.603752131893121e-05,
      "loss": 0.834,
      "step": 5568
    },
    {
      "epoch": 7.756267409470752,
      "grad_norm": 2.8130438327789307,
      "learning_rate": 1.6027572484366117e-05,
      "loss": 0.7637,
      "step": 5569
    },
    {
      "epoch": 7.757660167130919,
      "grad_norm": 2.989290714263916,
      "learning_rate": 1.601762364980102e-05,
      "loss": 1.0875,
      "step": 5570
    },
    {
      "epoch": 7.759052924791086,
      "grad_norm": 2.9280595779418945,
      "learning_rate": 1.6007674815235928e-05,
      "loss": 0.7164,
      "step": 5571
    },
    {
      "epoch": 7.7604456824512535,
      "grad_norm": 2.8934755325317383,
      "learning_rate": 1.5997725980670832e-05,
      "loss": 0.8064,
      "step": 5572
    },
    {
      "epoch": 7.76183844011142,
      "grad_norm": 2.6276254653930664,
      "learning_rate": 1.598777714610574e-05,
      "loss": 0.6988,
      "step": 5573
    },
    {
      "epoch": 7.763231197771588,
      "grad_norm": 3.078321933746338,
      "learning_rate": 1.5977828311540647e-05,
      "loss": 0.8105,
      "step": 5574
    },
    {
      "epoch": 7.764623955431755,
      "grad_norm": 2.2949001789093018,
      "learning_rate": 1.5967879476975555e-05,
      "loss": 0.5527,
      "step": 5575
    },
    {
      "epoch": 7.766016713091922,
      "grad_norm": 2.639017105102539,
      "learning_rate": 1.595793064241046e-05,
      "loss": 0.737,
      "step": 5576
    },
    {
      "epoch": 7.767409470752089,
      "grad_norm": 3.0754311084747314,
      "learning_rate": 1.5947981807845366e-05,
      "loss": 0.8296,
      "step": 5577
    },
    {
      "epoch": 7.768802228412256,
      "grad_norm": 2.674020767211914,
      "learning_rate": 1.593803297328027e-05,
      "loss": 0.8421,
      "step": 5578
    },
    {
      "epoch": 7.770194986072424,
      "grad_norm": 2.3705127239227295,
      "learning_rate": 1.5928084138715178e-05,
      "loss": 0.6161,
      "step": 5579
    },
    {
      "epoch": 7.77158774373259,
      "grad_norm": 2.5708835124969482,
      "learning_rate": 1.5918135304150085e-05,
      "loss": 0.7274,
      "step": 5580
    },
    {
      "epoch": 7.772980501392758,
      "grad_norm": 2.8865108489990234,
      "learning_rate": 1.590818646958499e-05,
      "loss": 0.7965,
      "step": 5581
    },
    {
      "epoch": 7.774373259052925,
      "grad_norm": 2.9721262454986572,
      "learning_rate": 1.5898237635019897e-05,
      "loss": 0.784,
      "step": 5582
    },
    {
      "epoch": 7.775766016713092,
      "grad_norm": 2.682227849960327,
      "learning_rate": 1.58882888004548e-05,
      "loss": 0.6563,
      "step": 5583
    },
    {
      "epoch": 7.777158774373259,
      "grad_norm": 2.9040980339050293,
      "learning_rate": 1.5878339965889708e-05,
      "loss": 0.7591,
      "step": 5584
    },
    {
      "epoch": 7.778551532033426,
      "grad_norm": 2.747191905975342,
      "learning_rate": 1.5868391131324616e-05,
      "loss": 0.7726,
      "step": 5585
    },
    {
      "epoch": 7.779944289693593,
      "grad_norm": 2.908496856689453,
      "learning_rate": 1.585844229675952e-05,
      "loss": 0.7806,
      "step": 5586
    },
    {
      "epoch": 7.781337047353761,
      "grad_norm": 2.671966552734375,
      "learning_rate": 1.5848493462194427e-05,
      "loss": 0.6197,
      "step": 5587
    },
    {
      "epoch": 7.782729805013927,
      "grad_norm": 2.7716825008392334,
      "learning_rate": 1.583854462762933e-05,
      "loss": 0.9544,
      "step": 5588
    },
    {
      "epoch": 7.784122562674095,
      "grad_norm": 3.0952935218811035,
      "learning_rate": 1.582859579306424e-05,
      "loss": 0.8225,
      "step": 5589
    },
    {
      "epoch": 7.7855153203342615,
      "grad_norm": 2.783839464187622,
      "learning_rate": 1.5818646958499146e-05,
      "loss": 0.6937,
      "step": 5590
    },
    {
      "epoch": 7.786908077994429,
      "grad_norm": 2.6238903999328613,
      "learning_rate": 1.5808698123934054e-05,
      "loss": 0.7029,
      "step": 5591
    },
    {
      "epoch": 7.788300835654596,
      "grad_norm": 2.4679155349731445,
      "learning_rate": 1.5798749289368958e-05,
      "loss": 0.7577,
      "step": 5592
    },
    {
      "epoch": 7.789693593314763,
      "grad_norm": 2.842315912246704,
      "learning_rate": 1.5788800454803865e-05,
      "loss": 0.7747,
      "step": 5593
    },
    {
      "epoch": 7.791086350974931,
      "grad_norm": 2.250731945037842,
      "learning_rate": 1.577885162023877e-05,
      "loss": 0.586,
      "step": 5594
    },
    {
      "epoch": 7.7924791086350975,
      "grad_norm": 2.5972139835357666,
      "learning_rate": 1.5768902785673677e-05,
      "loss": 0.7453,
      "step": 5595
    },
    {
      "epoch": 7.793871866295264,
      "grad_norm": 3.002807140350342,
      "learning_rate": 1.5758953951108584e-05,
      "loss": 0.8393,
      "step": 5596
    },
    {
      "epoch": 7.795264623955432,
      "grad_norm": 2.551990032196045,
      "learning_rate": 1.574900511654349e-05,
      "loss": 0.7232,
      "step": 5597
    },
    {
      "epoch": 7.796657381615599,
      "grad_norm": 2.2206497192382812,
      "learning_rate": 1.5739056281978396e-05,
      "loss": 0.6792,
      "step": 5598
    },
    {
      "epoch": 7.798050139275766,
      "grad_norm": 2.9718410968780518,
      "learning_rate": 1.57291074474133e-05,
      "loss": 1.0455,
      "step": 5599
    },
    {
      "epoch": 7.7994428969359335,
      "grad_norm": 3.1451773643493652,
      "learning_rate": 1.5719158612848207e-05,
      "loss": 0.7256,
      "step": 5600
    },
    {
      "epoch": 7.8008356545961,
      "grad_norm": 2.6781413555145264,
      "learning_rate": 1.5709209778283115e-05,
      "loss": 0.869,
      "step": 5601
    },
    {
      "epoch": 7.802228412256268,
      "grad_norm": 2.681276798248291,
      "learning_rate": 1.569926094371802e-05,
      "loss": 0.8421,
      "step": 5602
    },
    {
      "epoch": 7.803621169916434,
      "grad_norm": 2.325745105743408,
      "learning_rate": 1.5689312109152926e-05,
      "loss": 0.6422,
      "step": 5603
    },
    {
      "epoch": 7.805013927576602,
      "grad_norm": 3.1258108615875244,
      "learning_rate": 1.5679363274587834e-05,
      "loss": 0.8142,
      "step": 5604
    },
    {
      "epoch": 7.806406685236769,
      "grad_norm": 2.8076539039611816,
      "learning_rate": 1.5669414440022738e-05,
      "loss": 0.9406,
      "step": 5605
    },
    {
      "epoch": 7.807799442896936,
      "grad_norm": 2.98014497756958,
      "learning_rate": 1.5659465605457645e-05,
      "loss": 0.82,
      "step": 5606
    },
    {
      "epoch": 7.809192200557103,
      "grad_norm": 2.5118582248687744,
      "learning_rate": 1.5649516770892553e-05,
      "loss": 0.7875,
      "step": 5607
    },
    {
      "epoch": 7.81058495821727,
      "grad_norm": 2.427443265914917,
      "learning_rate": 1.5639567936327457e-05,
      "loss": 0.7352,
      "step": 5608
    },
    {
      "epoch": 7.811977715877437,
      "grad_norm": 2.578648567199707,
      "learning_rate": 1.5629619101762364e-05,
      "loss": 0.6233,
      "step": 5609
    },
    {
      "epoch": 7.813370473537605,
      "grad_norm": 2.2706801891326904,
      "learning_rate": 1.561967026719727e-05,
      "loss": 0.6984,
      "step": 5610
    },
    {
      "epoch": 7.814763231197771,
      "grad_norm": 2.4212236404418945,
      "learning_rate": 1.5609721432632176e-05,
      "loss": 0.6463,
      "step": 5611
    },
    {
      "epoch": 7.816155988857939,
      "grad_norm": 2.941241502761841,
      "learning_rate": 1.5599772598067083e-05,
      "loss": 0.8984,
      "step": 5612
    },
    {
      "epoch": 7.8175487465181055,
      "grad_norm": 2.364943265914917,
      "learning_rate": 1.5589823763501987e-05,
      "loss": 0.7095,
      "step": 5613
    },
    {
      "epoch": 7.818941504178273,
      "grad_norm": 2.514819383621216,
      "learning_rate": 1.5579874928936895e-05,
      "loss": 0.8956,
      "step": 5614
    },
    {
      "epoch": 7.82033426183844,
      "grad_norm": 2.2634267807006836,
      "learning_rate": 1.55699260943718e-05,
      "loss": 0.7432,
      "step": 5615
    },
    {
      "epoch": 7.821727019498607,
      "grad_norm": 2.539466619491577,
      "learning_rate": 1.5559977259806706e-05,
      "loss": 0.745,
      "step": 5616
    },
    {
      "epoch": 7.823119777158775,
      "grad_norm": 3.175222158432007,
      "learning_rate": 1.5550028425241614e-05,
      "loss": 0.7331,
      "step": 5617
    },
    {
      "epoch": 7.8245125348189415,
      "grad_norm": 2.66514253616333,
      "learning_rate": 1.554007959067652e-05,
      "loss": 0.646,
      "step": 5618
    },
    {
      "epoch": 7.825905292479108,
      "grad_norm": 4.708824157714844,
      "learning_rate": 1.5530130756111425e-05,
      "loss": 0.6814,
      "step": 5619
    },
    {
      "epoch": 7.827298050139276,
      "grad_norm": 2.8916537761688232,
      "learning_rate": 1.5520181921546333e-05,
      "loss": 0.9828,
      "step": 5620
    },
    {
      "epoch": 7.828690807799443,
      "grad_norm": 2.7039146423339844,
      "learning_rate": 1.5510233086981237e-05,
      "loss": 1.0111,
      "step": 5621
    },
    {
      "epoch": 7.83008356545961,
      "grad_norm": 2.8030519485473633,
      "learning_rate": 1.5500284252416144e-05,
      "loss": 0.8601,
      "step": 5622
    },
    {
      "epoch": 7.8314763231197775,
      "grad_norm": 2.383009672164917,
      "learning_rate": 1.5490335417851052e-05,
      "loss": 0.8488,
      "step": 5623
    },
    {
      "epoch": 7.832869080779944,
      "grad_norm": 2.9474103450775146,
      "learning_rate": 1.5480386583285956e-05,
      "loss": 0.88,
      "step": 5624
    },
    {
      "epoch": 7.834261838440112,
      "grad_norm": 2.38859224319458,
      "learning_rate": 1.5470437748720863e-05,
      "loss": 0.7978,
      "step": 5625
    },
    {
      "epoch": 7.835654596100278,
      "grad_norm": 2.316699743270874,
      "learning_rate": 1.5460488914155768e-05,
      "loss": 0.6718,
      "step": 5626
    },
    {
      "epoch": 7.837047353760446,
      "grad_norm": 2.29079008102417,
      "learning_rate": 1.5450540079590675e-05,
      "loss": 0.661,
      "step": 5627
    },
    {
      "epoch": 7.838440111420613,
      "grad_norm": 2.701103925704956,
      "learning_rate": 1.5440591245025582e-05,
      "loss": 0.8497,
      "step": 5628
    },
    {
      "epoch": 7.83983286908078,
      "grad_norm": 3.1637492179870605,
      "learning_rate": 1.5430642410460487e-05,
      "loss": 1.0143,
      "step": 5629
    },
    {
      "epoch": 7.841225626740947,
      "grad_norm": 2.374556064605713,
      "learning_rate": 1.5420693575895394e-05,
      "loss": 0.6001,
      "step": 5630
    },
    {
      "epoch": 7.842618384401114,
      "grad_norm": 2.5755841732025146,
      "learning_rate": 1.54107447413303e-05,
      "loss": 0.7803,
      "step": 5631
    },
    {
      "epoch": 7.844011142061281,
      "grad_norm": 2.7818307876586914,
      "learning_rate": 1.5400795906765206e-05,
      "loss": 0.9121,
      "step": 5632
    },
    {
      "epoch": 7.845403899721449,
      "grad_norm": 2.5337882041931152,
      "learning_rate": 1.5390847072200113e-05,
      "loss": 0.7806,
      "step": 5633
    },
    {
      "epoch": 7.846796657381615,
      "grad_norm": 3.340794086456299,
      "learning_rate": 1.538089823763502e-05,
      "loss": 0.9883,
      "step": 5634
    },
    {
      "epoch": 7.848189415041783,
      "grad_norm": 4.154270172119141,
      "learning_rate": 1.5370949403069925e-05,
      "loss": 0.8184,
      "step": 5635
    },
    {
      "epoch": 7.84958217270195,
      "grad_norm": 2.594151496887207,
      "learning_rate": 1.5361000568504832e-05,
      "loss": 0.6705,
      "step": 5636
    },
    {
      "epoch": 7.850974930362117,
      "grad_norm": 2.5860893726348877,
      "learning_rate": 1.5351051733939736e-05,
      "loss": 0.8204,
      "step": 5637
    },
    {
      "epoch": 7.852367688022284,
      "grad_norm": 2.2906734943389893,
      "learning_rate": 1.5341102899374644e-05,
      "loss": 0.7963,
      "step": 5638
    },
    {
      "epoch": 7.853760445682451,
      "grad_norm": 2.1849262714385986,
      "learning_rate": 1.533115406480955e-05,
      "loss": 0.7335,
      "step": 5639
    },
    {
      "epoch": 7.855153203342619,
      "grad_norm": 2.508540153503418,
      "learning_rate": 1.5321205230244455e-05,
      "loss": 0.7183,
      "step": 5640
    },
    {
      "epoch": 7.8565459610027855,
      "grad_norm": 2.3108091354370117,
      "learning_rate": 1.5311256395679363e-05,
      "loss": 0.6723,
      "step": 5641
    },
    {
      "epoch": 7.857938718662953,
      "grad_norm": 2.8738508224487305,
      "learning_rate": 1.5301307561114267e-05,
      "loss": 0.8377,
      "step": 5642
    },
    {
      "epoch": 7.85933147632312,
      "grad_norm": 2.4582111835479736,
      "learning_rate": 1.5291358726549174e-05,
      "loss": 0.7304,
      "step": 5643
    },
    {
      "epoch": 7.860724233983287,
      "grad_norm": 2.4296915531158447,
      "learning_rate": 1.5281409891984078e-05,
      "loss": 0.7116,
      "step": 5644
    },
    {
      "epoch": 7.862116991643454,
      "grad_norm": 2.100569248199463,
      "learning_rate": 1.527146105741899e-05,
      "loss": 0.5974,
      "step": 5645
    },
    {
      "epoch": 7.8635097493036215,
      "grad_norm": 2.504051923751831,
      "learning_rate": 1.5261512222853893e-05,
      "loss": 0.7527,
      "step": 5646
    },
    {
      "epoch": 7.864902506963788,
      "grad_norm": 2.7811291217803955,
      "learning_rate": 1.5251563388288799e-05,
      "loss": 0.8601,
      "step": 5647
    },
    {
      "epoch": 7.866295264623956,
      "grad_norm": 2.9225103855133057,
      "learning_rate": 1.5241614553723705e-05,
      "loss": 0.8525,
      "step": 5648
    },
    {
      "epoch": 7.867688022284122,
      "grad_norm": 2.3884072303771973,
      "learning_rate": 1.523166571915861e-05,
      "loss": 0.6826,
      "step": 5649
    },
    {
      "epoch": 7.86908077994429,
      "grad_norm": 2.5493416786193848,
      "learning_rate": 1.5221716884593518e-05,
      "loss": 0.5761,
      "step": 5650
    },
    {
      "epoch": 7.870473537604457,
      "grad_norm": 2.43230938911438,
      "learning_rate": 1.5211768050028425e-05,
      "loss": 0.7446,
      "step": 5651
    },
    {
      "epoch": 7.871866295264624,
      "grad_norm": 4.126525402069092,
      "learning_rate": 1.5201819215463331e-05,
      "loss": 0.7879,
      "step": 5652
    },
    {
      "epoch": 7.873259052924791,
      "grad_norm": 2.607642889022827,
      "learning_rate": 1.5191870380898237e-05,
      "loss": 0.8421,
      "step": 5653
    },
    {
      "epoch": 7.874651810584958,
      "grad_norm": 2.85125732421875,
      "learning_rate": 1.5181921546333143e-05,
      "loss": 0.7928,
      "step": 5654
    },
    {
      "epoch": 7.876044568245125,
      "grad_norm": 2.496875762939453,
      "learning_rate": 1.5171972711768048e-05,
      "loss": 0.785,
      "step": 5655
    },
    {
      "epoch": 7.8774373259052926,
      "grad_norm": 2.5948171615600586,
      "learning_rate": 1.5162023877202956e-05,
      "loss": 0.7341,
      "step": 5656
    },
    {
      "epoch": 7.878830083565459,
      "grad_norm": 3.0469987392425537,
      "learning_rate": 1.5152075042637862e-05,
      "loss": 0.8603,
      "step": 5657
    },
    {
      "epoch": 7.880222841225627,
      "grad_norm": 2.4342963695526123,
      "learning_rate": 1.5142126208072767e-05,
      "loss": 0.6999,
      "step": 5658
    },
    {
      "epoch": 7.881615598885794,
      "grad_norm": 2.8552112579345703,
      "learning_rate": 1.5132177373507673e-05,
      "loss": 0.8601,
      "step": 5659
    },
    {
      "epoch": 7.883008356545961,
      "grad_norm": 2.7369091510772705,
      "learning_rate": 1.5122228538942579e-05,
      "loss": 0.8949,
      "step": 5660
    },
    {
      "epoch": 7.884401114206128,
      "grad_norm": 2.396972179412842,
      "learning_rate": 1.5112279704377486e-05,
      "loss": 0.7546,
      "step": 5661
    },
    {
      "epoch": 7.885793871866295,
      "grad_norm": 2.2595653533935547,
      "learning_rate": 1.5102330869812392e-05,
      "loss": 0.6355,
      "step": 5662
    },
    {
      "epoch": 7.887186629526463,
      "grad_norm": 2.827040195465088,
      "learning_rate": 1.5092382035247298e-05,
      "loss": 0.8552,
      "step": 5663
    },
    {
      "epoch": 7.888579387186629,
      "grad_norm": 2.8968186378479004,
      "learning_rate": 1.5082433200682204e-05,
      "loss": 0.8445,
      "step": 5664
    },
    {
      "epoch": 7.889972144846797,
      "grad_norm": 2.5456926822662354,
      "learning_rate": 1.507248436611711e-05,
      "loss": 0.7291,
      "step": 5665
    },
    {
      "epoch": 7.891364902506964,
      "grad_norm": 2.5830819606781006,
      "learning_rate": 1.5062535531552015e-05,
      "loss": 0.7048,
      "step": 5666
    },
    {
      "epoch": 7.892757660167131,
      "grad_norm": 2.4604547023773193,
      "learning_rate": 1.5052586696986924e-05,
      "loss": 0.7161,
      "step": 5667
    },
    {
      "epoch": 7.894150417827298,
      "grad_norm": 2.7419490814208984,
      "learning_rate": 1.504263786242183e-05,
      "loss": 1.0447,
      "step": 5668
    },
    {
      "epoch": 7.895543175487465,
      "grad_norm": 2.6495437622070312,
      "learning_rate": 1.5032689027856736e-05,
      "loss": 0.7987,
      "step": 5669
    },
    {
      "epoch": 7.896935933147632,
      "grad_norm": 2.556122303009033,
      "learning_rate": 1.5022740193291642e-05,
      "loss": 0.8719,
      "step": 5670
    },
    {
      "epoch": 7.8983286908078,
      "grad_norm": 2.394429922103882,
      "learning_rate": 1.5012791358726547e-05,
      "loss": 0.7168,
      "step": 5671
    },
    {
      "epoch": 7.899721448467966,
      "grad_norm": 2.569481134414673,
      "learning_rate": 1.5002842524161455e-05,
      "loss": 0.7791,
      "step": 5672
    },
    {
      "epoch": 7.901114206128134,
      "grad_norm": 2.329503059387207,
      "learning_rate": 1.499289368959636e-05,
      "loss": 0.609,
      "step": 5673
    },
    {
      "epoch": 7.9025069637883005,
      "grad_norm": 2.459470272064209,
      "learning_rate": 1.4982944855031266e-05,
      "loss": 0.7467,
      "step": 5674
    },
    {
      "epoch": 7.903899721448468,
      "grad_norm": 2.417170524597168,
      "learning_rate": 1.4972996020466172e-05,
      "loss": 0.7591,
      "step": 5675
    },
    {
      "epoch": 7.905292479108635,
      "grad_norm": 2.2791519165039062,
      "learning_rate": 1.4963047185901078e-05,
      "loss": 0.6549,
      "step": 5676
    },
    {
      "epoch": 7.906685236768802,
      "grad_norm": 2.9026219844818115,
      "learning_rate": 1.4953098351335984e-05,
      "loss": 0.7479,
      "step": 5677
    },
    {
      "epoch": 7.908077994428969,
      "grad_norm": 2.889512300491333,
      "learning_rate": 1.4943149516770891e-05,
      "loss": 0.8028,
      "step": 5678
    },
    {
      "epoch": 7.9094707520891365,
      "grad_norm": 2.6889922618865967,
      "learning_rate": 1.4933200682205799e-05,
      "loss": 0.7674,
      "step": 5679
    },
    {
      "epoch": 7.910863509749303,
      "grad_norm": 2.6476426124572754,
      "learning_rate": 1.4923251847640704e-05,
      "loss": 0.6657,
      "step": 5680
    },
    {
      "epoch": 7.912256267409471,
      "grad_norm": 2.9579553604125977,
      "learning_rate": 1.491330301307561e-05,
      "loss": 0.8853,
      "step": 5681
    },
    {
      "epoch": 7.913649025069638,
      "grad_norm": 1.990238904953003,
      "learning_rate": 1.4903354178510516e-05,
      "loss": 0.4864,
      "step": 5682
    },
    {
      "epoch": 7.915041782729805,
      "grad_norm": 2.6816718578338623,
      "learning_rate": 1.4893405343945423e-05,
      "loss": 0.8325,
      "step": 5683
    },
    {
      "epoch": 7.9164345403899725,
      "grad_norm": 2.970919132232666,
      "learning_rate": 1.488345650938033e-05,
      "loss": 0.8256,
      "step": 5684
    },
    {
      "epoch": 7.917827298050139,
      "grad_norm": 2.6956913471221924,
      "learning_rate": 1.4873507674815235e-05,
      "loss": 0.7907,
      "step": 5685
    },
    {
      "epoch": 7.919220055710307,
      "grad_norm": 2.150650978088379,
      "learning_rate": 1.486355884025014e-05,
      "loss": 0.7931,
      "step": 5686
    },
    {
      "epoch": 7.920612813370473,
      "grad_norm": 2.735020399093628,
      "learning_rate": 1.4853610005685047e-05,
      "loss": 0.7559,
      "step": 5687
    },
    {
      "epoch": 7.922005571030641,
      "grad_norm": 2.7567038536071777,
      "learning_rate": 1.4843661171119952e-05,
      "loss": 0.9638,
      "step": 5688
    },
    {
      "epoch": 7.923398328690808,
      "grad_norm": 2.8808114528656006,
      "learning_rate": 1.483371233655486e-05,
      "loss": 0.7945,
      "step": 5689
    },
    {
      "epoch": 7.924791086350975,
      "grad_norm": 2.803769111633301,
      "learning_rate": 1.4823763501989766e-05,
      "loss": 1.0416,
      "step": 5690
    },
    {
      "epoch": 7.926183844011142,
      "grad_norm": 2.9110865592956543,
      "learning_rate": 1.4813814667424671e-05,
      "loss": 0.7346,
      "step": 5691
    },
    {
      "epoch": 7.927576601671309,
      "grad_norm": 3.0096850395202637,
      "learning_rate": 1.4803865832859577e-05,
      "loss": 0.7868,
      "step": 5692
    },
    {
      "epoch": 7.928969359331476,
      "grad_norm": 2.6202094554901123,
      "learning_rate": 1.4793916998294483e-05,
      "loss": 0.8138,
      "step": 5693
    },
    {
      "epoch": 7.930362116991644,
      "grad_norm": 2.9771244525909424,
      "learning_rate": 1.4783968163729392e-05,
      "loss": 0.864,
      "step": 5694
    },
    {
      "epoch": 7.93175487465181,
      "grad_norm": 2.3767271041870117,
      "learning_rate": 1.4774019329164298e-05,
      "loss": 0.6459,
      "step": 5695
    },
    {
      "epoch": 7.933147632311978,
      "grad_norm": 2.8269755840301514,
      "learning_rate": 1.4764070494599204e-05,
      "loss": 0.8445,
      "step": 5696
    },
    {
      "epoch": 7.9345403899721445,
      "grad_norm": 2.45312237739563,
      "learning_rate": 1.475412166003411e-05,
      "loss": 0.7114,
      "step": 5697
    },
    {
      "epoch": 7.935933147632312,
      "grad_norm": 2.465182065963745,
      "learning_rate": 1.4744172825469015e-05,
      "loss": 0.6793,
      "step": 5698
    },
    {
      "epoch": 7.937325905292479,
      "grad_norm": 2.35618257522583,
      "learning_rate": 1.473422399090392e-05,
      "loss": 0.7439,
      "step": 5699
    },
    {
      "epoch": 7.938718662952646,
      "grad_norm": 2.613288402557373,
      "learning_rate": 1.4724275156338828e-05,
      "loss": 0.8685,
      "step": 5700
    },
    {
      "epoch": 7.940111420612814,
      "grad_norm": 2.3746984004974365,
      "learning_rate": 1.4714326321773734e-05,
      "loss": 0.778,
      "step": 5701
    },
    {
      "epoch": 7.9415041782729805,
      "grad_norm": 2.941166400909424,
      "learning_rate": 1.470437748720864e-05,
      "loss": 0.6549,
      "step": 5702
    },
    {
      "epoch": 7.942896935933147,
      "grad_norm": 2.9136996269226074,
      "learning_rate": 1.4694428652643546e-05,
      "loss": 0.9501,
      "step": 5703
    },
    {
      "epoch": 7.944289693593315,
      "grad_norm": 3.1199958324432373,
      "learning_rate": 1.4684479818078451e-05,
      "loss": 0.931,
      "step": 5704
    },
    {
      "epoch": 7.945682451253482,
      "grad_norm": 2.5881006717681885,
      "learning_rate": 1.4674530983513359e-05,
      "loss": 0.7387,
      "step": 5705
    },
    {
      "epoch": 7.947075208913649,
      "grad_norm": 2.805751085281372,
      "learning_rate": 1.4664582148948265e-05,
      "loss": 0.8509,
      "step": 5706
    },
    {
      "epoch": 7.9484679665738165,
      "grad_norm": 2.4799911975860596,
      "learning_rate": 1.4654633314383172e-05,
      "loss": 0.7261,
      "step": 5707
    },
    {
      "epoch": 7.949860724233983,
      "grad_norm": 2.4309656620025635,
      "learning_rate": 1.4644684479818078e-05,
      "loss": 0.7461,
      "step": 5708
    },
    {
      "epoch": 7.951253481894151,
      "grad_norm": 2.6114726066589355,
      "learning_rate": 1.4634735645252984e-05,
      "loss": 0.7763,
      "step": 5709
    },
    {
      "epoch": 7.952646239554317,
      "grad_norm": 2.3870186805725098,
      "learning_rate": 1.462478681068789e-05,
      "loss": 0.6467,
      "step": 5710
    },
    {
      "epoch": 7.954038997214485,
      "grad_norm": 2.8836123943328857,
      "learning_rate": 1.4614837976122797e-05,
      "loss": 0.9972,
      "step": 5711
    },
    {
      "epoch": 7.955431754874652,
      "grad_norm": 3.082615375518799,
      "learning_rate": 1.4604889141557703e-05,
      "loss": 0.9587,
      "step": 5712
    },
    {
      "epoch": 7.956824512534819,
      "grad_norm": 2.9020183086395264,
      "learning_rate": 1.4594940306992608e-05,
      "loss": 0.9053,
      "step": 5713
    },
    {
      "epoch": 7.958217270194986,
      "grad_norm": 2.556014060974121,
      "learning_rate": 1.4584991472427514e-05,
      "loss": 0.7685,
      "step": 5714
    },
    {
      "epoch": 7.959610027855153,
      "grad_norm": 2.619906425476074,
      "learning_rate": 1.457504263786242e-05,
      "loss": 0.7352,
      "step": 5715
    },
    {
      "epoch": 7.96100278551532,
      "grad_norm": 2.594573497772217,
      "learning_rate": 1.4565093803297326e-05,
      "loss": 0.743,
      "step": 5716
    },
    {
      "epoch": 7.962395543175488,
      "grad_norm": 2.6292269229888916,
      "learning_rate": 1.4555144968732233e-05,
      "loss": 0.6509,
      "step": 5717
    },
    {
      "epoch": 7.963788300835654,
      "grad_norm": 2.802565336227417,
      "learning_rate": 1.4545196134167139e-05,
      "loss": 0.9216,
      "step": 5718
    },
    {
      "epoch": 7.965181058495822,
      "grad_norm": 2.444504737854004,
      "learning_rate": 1.4535247299602045e-05,
      "loss": 0.7161,
      "step": 5719
    },
    {
      "epoch": 7.9665738161559885,
      "grad_norm": 2.265277862548828,
      "learning_rate": 1.452529846503695e-05,
      "loss": 0.5789,
      "step": 5720
    },
    {
      "epoch": 7.967966573816156,
      "grad_norm": 2.3342485427856445,
      "learning_rate": 1.4515349630471856e-05,
      "loss": 0.6536,
      "step": 5721
    },
    {
      "epoch": 7.969359331476323,
      "grad_norm": 2.153320550918579,
      "learning_rate": 1.4505400795906765e-05,
      "loss": 0.6483,
      "step": 5722
    },
    {
      "epoch": 7.97075208913649,
      "grad_norm": 2.940307140350342,
      "learning_rate": 1.4495451961341671e-05,
      "loss": 0.9323,
      "step": 5723
    },
    {
      "epoch": 7.972144846796658,
      "grad_norm": 2.91743803024292,
      "learning_rate": 1.4485503126776577e-05,
      "loss": 0.8105,
      "step": 5724
    },
    {
      "epoch": 7.9735376044568245,
      "grad_norm": 2.5661327838897705,
      "learning_rate": 1.4475554292211483e-05,
      "loss": 0.8149,
      "step": 5725
    },
    {
      "epoch": 7.974930362116992,
      "grad_norm": 2.6780052185058594,
      "learning_rate": 1.4465605457646388e-05,
      "loss": 0.9606,
      "step": 5726
    },
    {
      "epoch": 7.976323119777159,
      "grad_norm": 2.502511501312256,
      "learning_rate": 1.4455656623081294e-05,
      "loss": 0.6927,
      "step": 5727
    },
    {
      "epoch": 7.977715877437326,
      "grad_norm": 2.9799442291259766,
      "learning_rate": 1.4445707788516202e-05,
      "loss": 0.8756,
      "step": 5728
    },
    {
      "epoch": 7.979108635097493,
      "grad_norm": 2.6761505603790283,
      "learning_rate": 1.4435758953951107e-05,
      "loss": 0.8301,
      "step": 5729
    },
    {
      "epoch": 7.9805013927576605,
      "grad_norm": 3.159679889678955,
      "learning_rate": 1.4425810119386013e-05,
      "loss": 0.7773,
      "step": 5730
    },
    {
      "epoch": 7.981894150417827,
      "grad_norm": 2.164862632751465,
      "learning_rate": 1.4415861284820919e-05,
      "loss": 0.7532,
      "step": 5731
    },
    {
      "epoch": 7.983286908077995,
      "grad_norm": 3.3656914234161377,
      "learning_rate": 1.4405912450255825e-05,
      "loss": 0.9429,
      "step": 5732
    },
    {
      "epoch": 7.984679665738161,
      "grad_norm": 2.541602373123169,
      "learning_rate": 1.4395963615690732e-05,
      "loss": 0.7399,
      "step": 5733
    },
    {
      "epoch": 7.986072423398329,
      "grad_norm": 2.8899080753326416,
      "learning_rate": 1.4386014781125638e-05,
      "loss": 0.8363,
      "step": 5734
    },
    {
      "epoch": 7.987465181058496,
      "grad_norm": 2.7722439765930176,
      "learning_rate": 1.4376065946560545e-05,
      "loss": 0.8408,
      "step": 5735
    },
    {
      "epoch": 7.988857938718663,
      "grad_norm": 2.579982280731201,
      "learning_rate": 1.4366117111995451e-05,
      "loss": 0.7021,
      "step": 5736
    },
    {
      "epoch": 7.99025069637883,
      "grad_norm": 2.7417590618133545,
      "learning_rate": 1.4356168277430357e-05,
      "loss": 0.8916,
      "step": 5737
    },
    {
      "epoch": 7.991643454038997,
      "grad_norm": 2.7365336418151855,
      "learning_rate": 1.4346219442865263e-05,
      "loss": 0.9358,
      "step": 5738
    },
    {
      "epoch": 7.993036211699164,
      "grad_norm": 2.475696325302124,
      "learning_rate": 1.433627060830017e-05,
      "loss": 0.7145,
      "step": 5739
    },
    {
      "epoch": 7.994428969359332,
      "grad_norm": 2.2998290061950684,
      "learning_rate": 1.4326321773735076e-05,
      "loss": 0.575,
      "step": 5740
    },
    {
      "epoch": 7.995821727019498,
      "grad_norm": 2.5487418174743652,
      "learning_rate": 1.4316372939169982e-05,
      "loss": 0.785,
      "step": 5741
    },
    {
      "epoch": 7.997214484679666,
      "grad_norm": 2.654986619949341,
      "learning_rate": 1.4306424104604888e-05,
      "loss": 0.6867,
      "step": 5742
    },
    {
      "epoch": 7.998607242339833,
      "grad_norm": 2.1407663822174072,
      "learning_rate": 1.4296475270039793e-05,
      "loss": 0.6288,
      "step": 5743
    },
    {
      "epoch": 8.0,
      "grad_norm": 2.55751633644104,
      "learning_rate": 1.42865264354747e-05,
      "loss": 0.7525,
      "step": 5744
    },
    {
      "epoch": 8.001392757660167,
      "grad_norm": 2.391916275024414,
      "learning_rate": 1.4276577600909607e-05,
      "loss": 0.7063,
      "step": 5745
    },
    {
      "epoch": 8.002785515320335,
      "grad_norm": 2.0450918674468994,
      "learning_rate": 1.4266628766344512e-05,
      "loss": 0.4889,
      "step": 5746
    },
    {
      "epoch": 8.004178272980502,
      "grad_norm": 2.7029154300689697,
      "learning_rate": 1.4256679931779418e-05,
      "loss": 0.7572,
      "step": 5747
    },
    {
      "epoch": 8.005571030640668,
      "grad_norm": 2.054926872253418,
      "learning_rate": 1.4246731097214324e-05,
      "loss": 0.5982,
      "step": 5748
    },
    {
      "epoch": 8.006963788300835,
      "grad_norm": 2.607769727706909,
      "learning_rate": 1.423678226264923e-05,
      "loss": 0.7556,
      "step": 5749
    },
    {
      "epoch": 8.008356545961004,
      "grad_norm": 2.3893003463745117,
      "learning_rate": 1.4226833428084139e-05,
      "loss": 0.6554,
      "step": 5750
    },
    {
      "epoch": 8.00974930362117,
      "grad_norm": 2.1952672004699707,
      "learning_rate": 1.4216884593519045e-05,
      "loss": 0.5623,
      "step": 5751
    },
    {
      "epoch": 8.011142061281337,
      "grad_norm": 2.2442495822906494,
      "learning_rate": 1.420693575895395e-05,
      "loss": 0.5921,
      "step": 5752
    },
    {
      "epoch": 8.012534818941504,
      "grad_norm": 2.343412160873413,
      "learning_rate": 1.4196986924388856e-05,
      "loss": 0.6303,
      "step": 5753
    },
    {
      "epoch": 8.013927576601672,
      "grad_norm": 2.431013822555542,
      "learning_rate": 1.4187038089823762e-05,
      "loss": 0.5914,
      "step": 5754
    },
    {
      "epoch": 8.015320334261839,
      "grad_norm": 2.8709309101104736,
      "learning_rate": 1.417708925525867e-05,
      "loss": 0.677,
      "step": 5755
    },
    {
      "epoch": 8.016713091922005,
      "grad_norm": 2.6854491233825684,
      "learning_rate": 1.4167140420693575e-05,
      "loss": 0.9044,
      "step": 5756
    },
    {
      "epoch": 8.018105849582172,
      "grad_norm": 2.2322912216186523,
      "learning_rate": 1.4157191586128481e-05,
      "loss": 0.5336,
      "step": 5757
    },
    {
      "epoch": 8.01949860724234,
      "grad_norm": 2.5030367374420166,
      "learning_rate": 1.4147242751563387e-05,
      "loss": 0.6665,
      "step": 5758
    },
    {
      "epoch": 8.020891364902507,
      "grad_norm": 2.7106118202209473,
      "learning_rate": 1.4137293916998292e-05,
      "loss": 0.6792,
      "step": 5759
    },
    {
      "epoch": 8.022284122562674,
      "grad_norm": 3.1940295696258545,
      "learning_rate": 1.4127345082433198e-05,
      "loss": 0.6484,
      "step": 5760
    },
    {
      "epoch": 8.02367688022284,
      "grad_norm": 2.362699508666992,
      "learning_rate": 1.4117396247868106e-05,
      "loss": 0.729,
      "step": 5761
    },
    {
      "epoch": 8.025069637883009,
      "grad_norm": 2.236466407775879,
      "learning_rate": 1.4107447413303011e-05,
      "loss": 0.5975,
      "step": 5762
    },
    {
      "epoch": 8.026462395543176,
      "grad_norm": 2.5531554222106934,
      "learning_rate": 1.4097498578737917e-05,
      "loss": 0.9051,
      "step": 5763
    },
    {
      "epoch": 8.027855153203342,
      "grad_norm": 2.501363754272461,
      "learning_rate": 1.4087549744172825e-05,
      "loss": 0.7443,
      "step": 5764
    },
    {
      "epoch": 8.029247910863509,
      "grad_norm": 2.6491715908050537,
      "learning_rate": 1.407760090960773e-05,
      "loss": 0.7844,
      "step": 5765
    },
    {
      "epoch": 8.030640668523677,
      "grad_norm": 2.756535291671753,
      "learning_rate": 1.4067652075042638e-05,
      "loss": 0.7348,
      "step": 5766
    },
    {
      "epoch": 8.032033426183844,
      "grad_norm": 2.7806143760681152,
      "learning_rate": 1.4057703240477544e-05,
      "loss": 0.6787,
      "step": 5767
    },
    {
      "epoch": 8.03342618384401,
      "grad_norm": 2.6109886169433594,
      "learning_rate": 1.404775440591245e-05,
      "loss": 0.8259,
      "step": 5768
    },
    {
      "epoch": 8.034818941504179,
      "grad_norm": 2.5723721981048584,
      "learning_rate": 1.4037805571347355e-05,
      "loss": 0.911,
      "step": 5769
    },
    {
      "epoch": 8.036211699164346,
      "grad_norm": 2.319932460784912,
      "learning_rate": 1.4027856736782261e-05,
      "loss": 0.552,
      "step": 5770
    },
    {
      "epoch": 8.037604456824512,
      "grad_norm": 2.5703868865966797,
      "learning_rate": 1.4017907902217167e-05,
      "loss": 0.6775,
      "step": 5771
    },
    {
      "epoch": 8.038997214484679,
      "grad_norm": 2.100165605545044,
      "learning_rate": 1.4007959067652074e-05,
      "loss": 0.5726,
      "step": 5772
    },
    {
      "epoch": 8.040389972144848,
      "grad_norm": 2.1826250553131104,
      "learning_rate": 1.399801023308698e-05,
      "loss": 0.7185,
      "step": 5773
    },
    {
      "epoch": 8.041782729805014,
      "grad_norm": 2.448641061782837,
      "learning_rate": 1.3988061398521886e-05,
      "loss": 0.7163,
      "step": 5774
    },
    {
      "epoch": 8.04317548746518,
      "grad_norm": 2.128293037414551,
      "learning_rate": 1.3978112563956791e-05,
      "loss": 0.6839,
      "step": 5775
    },
    {
      "epoch": 8.044568245125348,
      "grad_norm": 2.3341121673583984,
      "learning_rate": 1.3968163729391697e-05,
      "loss": 0.6366,
      "step": 5776
    },
    {
      "epoch": 8.045961002785516,
      "grad_norm": 2.6132967472076416,
      "learning_rate": 1.3958214894826606e-05,
      "loss": 0.6322,
      "step": 5777
    },
    {
      "epoch": 8.047353760445683,
      "grad_norm": 2.2238409519195557,
      "learning_rate": 1.3948266060261512e-05,
      "loss": 0.5845,
      "step": 5778
    },
    {
      "epoch": 8.04874651810585,
      "grad_norm": 2.2329790592193604,
      "learning_rate": 1.3938317225696418e-05,
      "loss": 0.6621,
      "step": 5779
    },
    {
      "epoch": 8.050139275766016,
      "grad_norm": 2.3248584270477295,
      "learning_rate": 1.3928368391131324e-05,
      "loss": 0.7721,
      "step": 5780
    },
    {
      "epoch": 8.051532033426184,
      "grad_norm": 2.487091302871704,
      "learning_rate": 1.391841955656623e-05,
      "loss": 0.7894,
      "step": 5781
    },
    {
      "epoch": 8.052924791086351,
      "grad_norm": 2.1566736698150635,
      "learning_rate": 1.3908470722001135e-05,
      "loss": 0.621,
      "step": 5782
    },
    {
      "epoch": 8.054317548746518,
      "grad_norm": 2.67290997505188,
      "learning_rate": 1.3898521887436043e-05,
      "loss": 0.8812,
      "step": 5783
    },
    {
      "epoch": 8.055710306406684,
      "grad_norm": 2.1292243003845215,
      "learning_rate": 1.3888573052870948e-05,
      "loss": 0.5414,
      "step": 5784
    },
    {
      "epoch": 8.057103064066853,
      "grad_norm": 4.09104061126709,
      "learning_rate": 1.3878624218305854e-05,
      "loss": 0.7092,
      "step": 5785
    },
    {
      "epoch": 8.05849582172702,
      "grad_norm": 2.084872245788574,
      "learning_rate": 1.386867538374076e-05,
      "loss": 0.5472,
      "step": 5786
    },
    {
      "epoch": 8.059888579387186,
      "grad_norm": 1.9917781352996826,
      "learning_rate": 1.3858726549175666e-05,
      "loss": 0.5196,
      "step": 5787
    },
    {
      "epoch": 8.061281337047355,
      "grad_norm": 2.74969482421875,
      "learning_rate": 1.3848777714610573e-05,
      "loss": 0.7509,
      "step": 5788
    },
    {
      "epoch": 8.062674094707521,
      "grad_norm": 2.2629988193511963,
      "learning_rate": 1.3838828880045479e-05,
      "loss": 0.5508,
      "step": 5789
    },
    {
      "epoch": 8.064066852367688,
      "grad_norm": 3.313831090927124,
      "learning_rate": 1.3828880045480385e-05,
      "loss": 1.0226,
      "step": 5790
    },
    {
      "epoch": 8.065459610027855,
      "grad_norm": 2.245662212371826,
      "learning_rate": 1.381893121091529e-05,
      "loss": 0.6747,
      "step": 5791
    },
    {
      "epoch": 8.066852367688023,
      "grad_norm": 2.039762496948242,
      "learning_rate": 1.3808982376350198e-05,
      "loss": 0.6079,
      "step": 5792
    },
    {
      "epoch": 8.06824512534819,
      "grad_norm": 2.0505447387695312,
      "learning_rate": 1.3799033541785104e-05,
      "loss": 0.5687,
      "step": 5793
    },
    {
      "epoch": 8.069637883008356,
      "grad_norm": 2.716017484664917,
      "learning_rate": 1.3789084707220011e-05,
      "loss": 0.7565,
      "step": 5794
    },
    {
      "epoch": 8.071030640668523,
      "grad_norm": 2.498021125793457,
      "learning_rate": 1.3779135872654917e-05,
      "loss": 0.9454,
      "step": 5795
    },
    {
      "epoch": 8.072423398328691,
      "grad_norm": 2.0985682010650635,
      "learning_rate": 1.3769187038089823e-05,
      "loss": 0.6228,
      "step": 5796
    },
    {
      "epoch": 8.073816155988858,
      "grad_norm": 2.0225183963775635,
      "learning_rate": 1.3759238203524729e-05,
      "loss": 0.5725,
      "step": 5797
    },
    {
      "epoch": 8.075208913649025,
      "grad_norm": 2.3394439220428467,
      "learning_rate": 1.3749289368959634e-05,
      "loss": 0.6084,
      "step": 5798
    },
    {
      "epoch": 8.076601671309191,
      "grad_norm": 2.3332998752593994,
      "learning_rate": 1.3739340534394542e-05,
      "loss": 0.6034,
      "step": 5799
    },
    {
      "epoch": 8.07799442896936,
      "grad_norm": 2.2177438735961914,
      "learning_rate": 1.3729391699829448e-05,
      "loss": 0.5445,
      "step": 5800
    },
    {
      "epoch": 8.079387186629527,
      "grad_norm": 2.416872978210449,
      "learning_rate": 1.3719442865264353e-05,
      "loss": 0.6073,
      "step": 5801
    },
    {
      "epoch": 8.080779944289693,
      "grad_norm": 2.3437678813934326,
      "learning_rate": 1.3709494030699259e-05,
      "loss": 0.6007,
      "step": 5802
    },
    {
      "epoch": 8.08217270194986,
      "grad_norm": 2.0099823474884033,
      "learning_rate": 1.3699545196134165e-05,
      "loss": 0.5035,
      "step": 5803
    },
    {
      "epoch": 8.083565459610028,
      "grad_norm": 2.5457677841186523,
      "learning_rate": 1.368959636156907e-05,
      "loss": 0.7358,
      "step": 5804
    },
    {
      "epoch": 8.084958217270195,
      "grad_norm": 2.1337838172912598,
      "learning_rate": 1.367964752700398e-05,
      "loss": 0.6909,
      "step": 5805
    },
    {
      "epoch": 8.086350974930362,
      "grad_norm": 2.7705256938934326,
      "learning_rate": 1.3669698692438886e-05,
      "loss": 0.75,
      "step": 5806
    },
    {
      "epoch": 8.087743732590528,
      "grad_norm": 2.366692066192627,
      "learning_rate": 1.3659749857873791e-05,
      "loss": 0.6185,
      "step": 5807
    },
    {
      "epoch": 8.089136490250697,
      "grad_norm": 2.7063069343566895,
      "learning_rate": 1.3649801023308697e-05,
      "loss": 0.7608,
      "step": 5808
    },
    {
      "epoch": 8.090529247910863,
      "grad_norm": 2.7802040576934814,
      "learning_rate": 1.3639852188743603e-05,
      "loss": 0.7523,
      "step": 5809
    },
    {
      "epoch": 8.09192200557103,
      "grad_norm": 2.513087034225464,
      "learning_rate": 1.362990335417851e-05,
      "loss": 0.6017,
      "step": 5810
    },
    {
      "epoch": 8.093314763231199,
      "grad_norm": 2.6714537143707275,
      "learning_rate": 1.3619954519613416e-05,
      "loss": 0.863,
      "step": 5811
    },
    {
      "epoch": 8.094707520891365,
      "grad_norm": 2.186694860458374,
      "learning_rate": 1.3610005685048322e-05,
      "loss": 0.5695,
      "step": 5812
    },
    {
      "epoch": 8.096100278551532,
      "grad_norm": 2.306868314743042,
      "learning_rate": 1.3600056850483228e-05,
      "loss": 0.6054,
      "step": 5813
    },
    {
      "epoch": 8.097493036211699,
      "grad_norm": 2.0273630619049072,
      "learning_rate": 1.3590108015918133e-05,
      "loss": 0.5264,
      "step": 5814
    },
    {
      "epoch": 8.098885793871867,
      "grad_norm": 2.6661019325256348,
      "learning_rate": 1.358015918135304e-05,
      "loss": 0.7713,
      "step": 5815
    },
    {
      "epoch": 8.100278551532034,
      "grad_norm": 2.276488780975342,
      "learning_rate": 1.3570210346787947e-05,
      "loss": 0.5874,
      "step": 5816
    },
    {
      "epoch": 8.1016713091922,
      "grad_norm": 2.4340615272521973,
      "learning_rate": 1.3560261512222852e-05,
      "loss": 0.7338,
      "step": 5817
    },
    {
      "epoch": 8.103064066852367,
      "grad_norm": 2.042086362838745,
      "learning_rate": 1.3550312677657758e-05,
      "loss": 0.5316,
      "step": 5818
    },
    {
      "epoch": 8.104456824512535,
      "grad_norm": 2.4385125637054443,
      "learning_rate": 1.3540363843092664e-05,
      "loss": 0.7456,
      "step": 5819
    },
    {
      "epoch": 8.105849582172702,
      "grad_norm": 2.796144723892212,
      "learning_rate": 1.3530415008527571e-05,
      "loss": 0.7485,
      "step": 5820
    },
    {
      "epoch": 8.107242339832869,
      "grad_norm": 2.8352746963500977,
      "learning_rate": 1.3520466173962479e-05,
      "loss": 0.6964,
      "step": 5821
    },
    {
      "epoch": 8.108635097493035,
      "grad_norm": 2.162991762161255,
      "learning_rate": 1.3510517339397385e-05,
      "loss": 0.6187,
      "step": 5822
    },
    {
      "epoch": 8.110027855153204,
      "grad_norm": 2.3581225872039795,
      "learning_rate": 1.350056850483229e-05,
      "loss": 0.7305,
      "step": 5823
    },
    {
      "epoch": 8.11142061281337,
      "grad_norm": 2.6001851558685303,
      "learning_rate": 1.3490619670267196e-05,
      "loss": 0.6513,
      "step": 5824
    },
    {
      "epoch": 8.112813370473537,
      "grad_norm": 2.750192165374756,
      "learning_rate": 1.3480670835702102e-05,
      "loss": 0.7932,
      "step": 5825
    },
    {
      "epoch": 8.114206128133704,
      "grad_norm": 2.251194953918457,
      "learning_rate": 1.3470722001137008e-05,
      "loss": 0.6281,
      "step": 5826
    },
    {
      "epoch": 8.115598885793872,
      "grad_norm": 2.7577006816864014,
      "learning_rate": 1.3460773166571915e-05,
      "loss": 0.712,
      "step": 5827
    },
    {
      "epoch": 8.116991643454039,
      "grad_norm": 2.456524133682251,
      "learning_rate": 1.3450824332006821e-05,
      "loss": 0.6428,
      "step": 5828
    },
    {
      "epoch": 8.118384401114206,
      "grad_norm": 2.1486616134643555,
      "learning_rate": 1.3440875497441727e-05,
      "loss": 0.6689,
      "step": 5829
    },
    {
      "epoch": 8.119777158774374,
      "grad_norm": 2.1123476028442383,
      "learning_rate": 1.3430926662876632e-05,
      "loss": 0.5007,
      "step": 5830
    },
    {
      "epoch": 8.12116991643454,
      "grad_norm": 2.353919506072998,
      "learning_rate": 1.3420977828311538e-05,
      "loss": 0.6885,
      "step": 5831
    },
    {
      "epoch": 8.122562674094707,
      "grad_norm": 2.1658499240875244,
      "learning_rate": 1.3411028993746447e-05,
      "loss": 0.5995,
      "step": 5832
    },
    {
      "epoch": 8.123955431754874,
      "grad_norm": 2.5123870372772217,
      "learning_rate": 1.3401080159181353e-05,
      "loss": 0.6298,
      "step": 5833
    },
    {
      "epoch": 8.125348189415043,
      "grad_norm": 2.5471677780151367,
      "learning_rate": 1.3391131324616259e-05,
      "loss": 0.6832,
      "step": 5834
    },
    {
      "epoch": 8.12674094707521,
      "grad_norm": 2.5444443225860596,
      "learning_rate": 1.3381182490051165e-05,
      "loss": 0.648,
      "step": 5835
    },
    {
      "epoch": 8.128133704735376,
      "grad_norm": 2.4492671489715576,
      "learning_rate": 1.337123365548607e-05,
      "loss": 0.6774,
      "step": 5836
    },
    {
      "epoch": 8.129526462395543,
      "grad_norm": 2.365010976791382,
      "learning_rate": 1.3361284820920976e-05,
      "loss": 0.7052,
      "step": 5837
    },
    {
      "epoch": 8.130919220055711,
      "grad_norm": 2.1995530128479004,
      "learning_rate": 1.3351335986355884e-05,
      "loss": 0.6682,
      "step": 5838
    },
    {
      "epoch": 8.132311977715878,
      "grad_norm": 2.2907214164733887,
      "learning_rate": 1.334138715179079e-05,
      "loss": 0.5913,
      "step": 5839
    },
    {
      "epoch": 8.133704735376044,
      "grad_norm": 2.1191372871398926,
      "learning_rate": 1.3331438317225695e-05,
      "loss": 0.5251,
      "step": 5840
    },
    {
      "epoch": 8.135097493036211,
      "grad_norm": 2.6210310459136963,
      "learning_rate": 1.3321489482660601e-05,
      "loss": 0.5285,
      "step": 5841
    },
    {
      "epoch": 8.13649025069638,
      "grad_norm": 2.2320189476013184,
      "learning_rate": 1.3311540648095507e-05,
      "loss": 0.6324,
      "step": 5842
    },
    {
      "epoch": 8.137883008356546,
      "grad_norm": 2.838926315307617,
      "learning_rate": 1.3301591813530414e-05,
      "loss": 0.6681,
      "step": 5843
    },
    {
      "epoch": 8.139275766016713,
      "grad_norm": 2.5873560905456543,
      "learning_rate": 1.329164297896532e-05,
      "loss": 0.7269,
      "step": 5844
    },
    {
      "epoch": 8.14066852367688,
      "grad_norm": 2.117415428161621,
      "learning_rate": 1.3281694144400226e-05,
      "loss": 0.6531,
      "step": 5845
    },
    {
      "epoch": 8.142061281337048,
      "grad_norm": 2.1007559299468994,
      "learning_rate": 1.3271745309835132e-05,
      "loss": 0.5534,
      "step": 5846
    },
    {
      "epoch": 8.143454038997215,
      "grad_norm": 2.65936279296875,
      "learning_rate": 1.3261796475270037e-05,
      "loss": 0.6624,
      "step": 5847
    },
    {
      "epoch": 8.144846796657381,
      "grad_norm": 2.128936290740967,
      "learning_rate": 1.3251847640704945e-05,
      "loss": 0.6995,
      "step": 5848
    },
    {
      "epoch": 8.14623955431755,
      "grad_norm": 2.5572454929351807,
      "learning_rate": 1.3241898806139852e-05,
      "loss": 0.7573,
      "step": 5849
    },
    {
      "epoch": 8.147632311977716,
      "grad_norm": 2.2091126441955566,
      "learning_rate": 1.3231949971574758e-05,
      "loss": 0.6345,
      "step": 5850
    },
    {
      "epoch": 8.149025069637883,
      "grad_norm": 2.3995816707611084,
      "learning_rate": 1.3222001137009664e-05,
      "loss": 0.7709,
      "step": 5851
    },
    {
      "epoch": 8.15041782729805,
      "grad_norm": 2.8272969722747803,
      "learning_rate": 1.321205230244457e-05,
      "loss": 0.6315,
      "step": 5852
    },
    {
      "epoch": 8.151810584958218,
      "grad_norm": 2.152749538421631,
      "learning_rate": 1.3202103467879475e-05,
      "loss": 0.645,
      "step": 5853
    },
    {
      "epoch": 8.153203342618385,
      "grad_norm": 2.56510853767395,
      "learning_rate": 1.3192154633314383e-05,
      "loss": 0.7223,
      "step": 5854
    },
    {
      "epoch": 8.154596100278551,
      "grad_norm": 2.3824665546417236,
      "learning_rate": 1.3182205798749289e-05,
      "loss": 0.85,
      "step": 5855
    },
    {
      "epoch": 8.155988857938718,
      "grad_norm": 2.052708864212036,
      "learning_rate": 1.3172256964184194e-05,
      "loss": 0.6054,
      "step": 5856
    },
    {
      "epoch": 8.157381615598887,
      "grad_norm": 2.3831207752227783,
      "learning_rate": 1.31623081296191e-05,
      "loss": 0.6628,
      "step": 5857
    },
    {
      "epoch": 8.158774373259053,
      "grad_norm": 2.5285627841949463,
      "learning_rate": 1.3152359295054006e-05,
      "loss": 0.91,
      "step": 5858
    },
    {
      "epoch": 8.16016713091922,
      "grad_norm": 2.5442399978637695,
      "learning_rate": 1.3142410460488912e-05,
      "loss": 0.5158,
      "step": 5859
    },
    {
      "epoch": 8.161559888579387,
      "grad_norm": 2.2371113300323486,
      "learning_rate": 1.313246162592382e-05,
      "loss": 0.6232,
      "step": 5860
    },
    {
      "epoch": 8.162952646239555,
      "grad_norm": 2.5069162845611572,
      "learning_rate": 1.3122512791358727e-05,
      "loss": 0.6472,
      "step": 5861
    },
    {
      "epoch": 8.164345403899722,
      "grad_norm": 1.9748362302780151,
      "learning_rate": 1.3112563956793632e-05,
      "loss": 0.4711,
      "step": 5862
    },
    {
      "epoch": 8.165738161559888,
      "grad_norm": 2.6397035121917725,
      "learning_rate": 1.3102615122228538e-05,
      "loss": 0.6641,
      "step": 5863
    },
    {
      "epoch": 8.167130919220055,
      "grad_norm": 2.3952383995056152,
      "learning_rate": 1.3092666287663444e-05,
      "loss": 0.6823,
      "step": 5864
    },
    {
      "epoch": 8.168523676880223,
      "grad_norm": 2.539874315261841,
      "learning_rate": 1.3082717453098351e-05,
      "loss": 0.7934,
      "step": 5865
    },
    {
      "epoch": 8.16991643454039,
      "grad_norm": 2.3489582538604736,
      "learning_rate": 1.3072768618533257e-05,
      "loss": 0.7011,
      "step": 5866
    },
    {
      "epoch": 8.171309192200557,
      "grad_norm": 2.5650250911712646,
      "learning_rate": 1.3062819783968163e-05,
      "loss": 0.6256,
      "step": 5867
    },
    {
      "epoch": 8.172701949860723,
      "grad_norm": 2.726438045501709,
      "learning_rate": 1.3052870949403069e-05,
      "loss": 0.7297,
      "step": 5868
    },
    {
      "epoch": 8.174094707520892,
      "grad_norm": 2.390582799911499,
      "learning_rate": 1.3042922114837974e-05,
      "loss": 0.6669,
      "step": 5869
    },
    {
      "epoch": 8.175487465181059,
      "grad_norm": 2.175386667251587,
      "learning_rate": 1.303297328027288e-05,
      "loss": 0.5627,
      "step": 5870
    },
    {
      "epoch": 8.176880222841225,
      "grad_norm": 2.3071563243865967,
      "learning_rate": 1.3023024445707788e-05,
      "loss": 0.6578,
      "step": 5871
    },
    {
      "epoch": 8.178272980501394,
      "grad_norm": 2.320700168609619,
      "learning_rate": 1.3013075611142693e-05,
      "loss": 0.6997,
      "step": 5872
    },
    {
      "epoch": 8.17966573816156,
      "grad_norm": 2.4623591899871826,
      "learning_rate": 1.30031267765776e-05,
      "loss": 0.707,
      "step": 5873
    },
    {
      "epoch": 8.181058495821727,
      "grad_norm": 1.9218674898147583,
      "learning_rate": 1.2993177942012505e-05,
      "loss": 0.4918,
      "step": 5874
    },
    {
      "epoch": 8.182451253481894,
      "grad_norm": 2.1007614135742188,
      "learning_rate": 1.298322910744741e-05,
      "loss": 0.5732,
      "step": 5875
    },
    {
      "epoch": 8.183844011142062,
      "grad_norm": 2.257274627685547,
      "learning_rate": 1.297328027288232e-05,
      "loss": 0.6475,
      "step": 5876
    },
    {
      "epoch": 8.185236768802229,
      "grad_norm": 2.255138635635376,
      "learning_rate": 1.2963331438317226e-05,
      "loss": 0.6226,
      "step": 5877
    },
    {
      "epoch": 8.186629526462395,
      "grad_norm": 2.243072271347046,
      "learning_rate": 1.2953382603752131e-05,
      "loss": 0.6013,
      "step": 5878
    },
    {
      "epoch": 8.188022284122562,
      "grad_norm": 2.5485637187957764,
      "learning_rate": 1.2943433769187037e-05,
      "loss": 0.6773,
      "step": 5879
    },
    {
      "epoch": 8.18941504178273,
      "grad_norm": 2.3386006355285645,
      "learning_rate": 1.2933484934621943e-05,
      "loss": 0.5755,
      "step": 5880
    },
    {
      "epoch": 8.190807799442897,
      "grad_norm": 2.494722843170166,
      "learning_rate": 1.2923536100056849e-05,
      "loss": 0.6814,
      "step": 5881
    },
    {
      "epoch": 8.192200557103064,
      "grad_norm": 2.4812235832214355,
      "learning_rate": 1.2913587265491756e-05,
      "loss": 0.7207,
      "step": 5882
    },
    {
      "epoch": 8.19359331476323,
      "grad_norm": 2.194302797317505,
      "learning_rate": 1.2903638430926662e-05,
      "loss": 0.6737,
      "step": 5883
    },
    {
      "epoch": 8.194986072423399,
      "grad_norm": 2.4026405811309814,
      "learning_rate": 1.2893689596361568e-05,
      "loss": 0.7586,
      "step": 5884
    },
    {
      "epoch": 8.196378830083566,
      "grad_norm": 2.3500783443450928,
      "learning_rate": 1.2883740761796473e-05,
      "loss": 0.6373,
      "step": 5885
    },
    {
      "epoch": 8.197771587743732,
      "grad_norm": 2.294128894805908,
      "learning_rate": 1.287379192723138e-05,
      "loss": 0.5818,
      "step": 5886
    },
    {
      "epoch": 8.199164345403899,
      "grad_norm": 3.788611888885498,
      "learning_rate": 1.2863843092666287e-05,
      "loss": 0.7514,
      "step": 5887
    },
    {
      "epoch": 8.200557103064067,
      "grad_norm": 2.2748560905456543,
      "learning_rate": 1.2853894258101192e-05,
      "loss": 0.6306,
      "step": 5888
    },
    {
      "epoch": 8.201949860724234,
      "grad_norm": 2.349860191345215,
      "learning_rate": 1.28439454235361e-05,
      "loss": 0.6139,
      "step": 5889
    },
    {
      "epoch": 8.2033426183844,
      "grad_norm": 2.2774569988250732,
      "learning_rate": 1.2833996588971006e-05,
      "loss": 0.649,
      "step": 5890
    },
    {
      "epoch": 8.204735376044567,
      "grad_norm": 2.288731813430786,
      "learning_rate": 1.2824047754405911e-05,
      "loss": 0.6084,
      "step": 5891
    },
    {
      "epoch": 8.206128133704736,
      "grad_norm": 3.4352667331695557,
      "learning_rate": 1.2814098919840817e-05,
      "loss": 0.4728,
      "step": 5892
    },
    {
      "epoch": 8.207520891364902,
      "grad_norm": 2.0849528312683105,
      "learning_rate": 1.2804150085275725e-05,
      "loss": 0.5846,
      "step": 5893
    },
    {
      "epoch": 8.20891364902507,
      "grad_norm": 3.286938428878784,
      "learning_rate": 1.279420125071063e-05,
      "loss": 0.5766,
      "step": 5894
    },
    {
      "epoch": 8.210306406685238,
      "grad_norm": 2.490229845046997,
      "learning_rate": 1.2784252416145536e-05,
      "loss": 0.7511,
      "step": 5895
    },
    {
      "epoch": 8.211699164345404,
      "grad_norm": 2.1200239658355713,
      "learning_rate": 1.2774303581580442e-05,
      "loss": 0.5537,
      "step": 5896
    },
    {
      "epoch": 8.213091922005571,
      "grad_norm": 2.2673428058624268,
      "learning_rate": 1.2764354747015348e-05,
      "loss": 0.6893,
      "step": 5897
    },
    {
      "epoch": 8.214484679665738,
      "grad_norm": 2.736982822418213,
      "learning_rate": 1.2754405912450255e-05,
      "loss": 0.7344,
      "step": 5898
    },
    {
      "epoch": 8.215877437325906,
      "grad_norm": 2.560347318649292,
      "learning_rate": 1.2744457077885161e-05,
      "loss": 0.7171,
      "step": 5899
    },
    {
      "epoch": 8.217270194986073,
      "grad_norm": 2.484302520751953,
      "learning_rate": 1.2734508243320067e-05,
      "loss": 0.6429,
      "step": 5900
    },
    {
      "epoch": 8.21866295264624,
      "grad_norm": 2.39678692817688,
      "learning_rate": 1.2724559408754973e-05,
      "loss": 0.6213,
      "step": 5901
    },
    {
      "epoch": 8.220055710306406,
      "grad_norm": 2.588151693344116,
      "learning_rate": 1.2714610574189878e-05,
      "loss": 0.738,
      "step": 5902
    },
    {
      "epoch": 8.221448467966574,
      "grad_norm": 2.280254364013672,
      "learning_rate": 1.2704661739624784e-05,
      "loss": 0.6531,
      "step": 5903
    },
    {
      "epoch": 8.222841225626741,
      "grad_norm": 2.2270069122314453,
      "learning_rate": 1.2694712905059693e-05,
      "loss": 0.6994,
      "step": 5904
    },
    {
      "epoch": 8.224233983286908,
      "grad_norm": 2.891584873199463,
      "learning_rate": 1.2684764070494599e-05,
      "loss": 0.6786,
      "step": 5905
    },
    {
      "epoch": 8.225626740947074,
      "grad_norm": 2.5043554306030273,
      "learning_rate": 1.2674815235929505e-05,
      "loss": 0.6914,
      "step": 5906
    },
    {
      "epoch": 8.227019498607243,
      "grad_norm": 2.258073329925537,
      "learning_rate": 1.266486640136441e-05,
      "loss": 0.6952,
      "step": 5907
    },
    {
      "epoch": 8.22841225626741,
      "grad_norm": 2.1068105697631836,
      "learning_rate": 1.2654917566799316e-05,
      "loss": 0.499,
      "step": 5908
    },
    {
      "epoch": 8.229805013927576,
      "grad_norm": 2.5303189754486084,
      "learning_rate": 1.2644968732234224e-05,
      "loss": 0.7266,
      "step": 5909
    },
    {
      "epoch": 8.231197771587743,
      "grad_norm": 2.7753899097442627,
      "learning_rate": 1.263501989766913e-05,
      "loss": 0.8295,
      "step": 5910
    },
    {
      "epoch": 8.232590529247911,
      "grad_norm": 2.471743106842041,
      "learning_rate": 1.2625071063104035e-05,
      "loss": 0.5109,
      "step": 5911
    },
    {
      "epoch": 8.233983286908078,
      "grad_norm": 2.314821481704712,
      "learning_rate": 1.2615122228538941e-05,
      "loss": 0.7335,
      "step": 5912
    },
    {
      "epoch": 8.235376044568245,
      "grad_norm": 2.624682903289795,
      "learning_rate": 1.2605173393973847e-05,
      "loss": 0.7699,
      "step": 5913
    },
    {
      "epoch": 8.236768802228413,
      "grad_norm": 2.130018711090088,
      "learning_rate": 1.2595224559408753e-05,
      "loss": 0.5058,
      "step": 5914
    },
    {
      "epoch": 8.23816155988858,
      "grad_norm": 2.4204928874969482,
      "learning_rate": 1.258527572484366e-05,
      "loss": 0.8315,
      "step": 5915
    },
    {
      "epoch": 8.239554317548746,
      "grad_norm": 2.3368215560913086,
      "learning_rate": 1.2575326890278566e-05,
      "loss": 0.5695,
      "step": 5916
    },
    {
      "epoch": 8.240947075208913,
      "grad_norm": 2.070472478866577,
      "learning_rate": 1.2565378055713473e-05,
      "loss": 0.5549,
      "step": 5917
    },
    {
      "epoch": 8.242339832869082,
      "grad_norm": 2.43178653717041,
      "learning_rate": 1.2555429221148379e-05,
      "loss": 0.7577,
      "step": 5918
    },
    {
      "epoch": 8.243732590529248,
      "grad_norm": 2.3918399810791016,
      "learning_rate": 1.2545480386583285e-05,
      "loss": 0.6415,
      "step": 5919
    },
    {
      "epoch": 8.245125348189415,
      "grad_norm": 2.3452882766723633,
      "learning_rate": 1.253553155201819e-05,
      "loss": 0.7346,
      "step": 5920
    },
    {
      "epoch": 8.246518105849582,
      "grad_norm": 2.5305464267730713,
      "learning_rate": 1.2525582717453098e-05,
      "loss": 0.7734,
      "step": 5921
    },
    {
      "epoch": 8.24791086350975,
      "grad_norm": 2.26212477684021,
      "learning_rate": 1.2515633882888004e-05,
      "loss": 0.478,
      "step": 5922
    },
    {
      "epoch": 8.249303621169917,
      "grad_norm": 2.480069637298584,
      "learning_rate": 1.250568504832291e-05,
      "loss": 0.6553,
      "step": 5923
    },
    {
      "epoch": 8.250696378830083,
      "grad_norm": 2.397341012954712,
      "learning_rate": 1.2495736213757815e-05,
      "loss": 0.6801,
      "step": 5924
    },
    {
      "epoch": 8.25208913649025,
      "grad_norm": 2.522189140319824,
      "learning_rate": 1.2485787379192721e-05,
      "loss": 0.6564,
      "step": 5925
    },
    {
      "epoch": 8.253481894150418,
      "grad_norm": 2.4263949394226074,
      "learning_rate": 1.2475838544627629e-05,
      "loss": 0.6019,
      "step": 5926
    },
    {
      "epoch": 8.254874651810585,
      "grad_norm": 2.0276315212249756,
      "learning_rate": 1.2465889710062534e-05,
      "loss": 0.6907,
      "step": 5927
    },
    {
      "epoch": 8.256267409470752,
      "grad_norm": 2.6812098026275635,
      "learning_rate": 1.245594087549744e-05,
      "loss": 0.6119,
      "step": 5928
    },
    {
      "epoch": 8.257660167130918,
      "grad_norm": 1.9010913372039795,
      "learning_rate": 1.2445992040932346e-05,
      "loss": 0.4513,
      "step": 5929
    },
    {
      "epoch": 8.259052924791087,
      "grad_norm": 2.417888879776001,
      "learning_rate": 1.2436043206367252e-05,
      "loss": 0.7542,
      "step": 5930
    },
    {
      "epoch": 8.260445682451254,
      "grad_norm": 2.392244577407837,
      "learning_rate": 1.2426094371802157e-05,
      "loss": 0.7648,
      "step": 5931
    },
    {
      "epoch": 8.26183844011142,
      "grad_norm": 2.4280102252960205,
      "learning_rate": 1.2416145537237067e-05,
      "loss": 0.6441,
      "step": 5932
    },
    {
      "epoch": 8.263231197771589,
      "grad_norm": 2.130587100982666,
      "learning_rate": 1.2406196702671972e-05,
      "loss": 0.6038,
      "step": 5933
    },
    {
      "epoch": 8.264623955431755,
      "grad_norm": 2.398505687713623,
      "learning_rate": 1.2396247868106878e-05,
      "loss": 0.7898,
      "step": 5934
    },
    {
      "epoch": 8.266016713091922,
      "grad_norm": 2.954123020172119,
      "learning_rate": 1.2386299033541784e-05,
      "loss": 0.7322,
      "step": 5935
    },
    {
      "epoch": 8.267409470752089,
      "grad_norm": 2.5999250411987305,
      "learning_rate": 1.237635019897669e-05,
      "loss": 0.7961,
      "step": 5936
    },
    {
      "epoch": 8.268802228412257,
      "grad_norm": 2.0471620559692383,
      "learning_rate": 1.2366401364411597e-05,
      "loss": 0.5758,
      "step": 5937
    },
    {
      "epoch": 8.270194986072424,
      "grad_norm": 2.710904598236084,
      "learning_rate": 1.2356452529846503e-05,
      "loss": 0.6447,
      "step": 5938
    },
    {
      "epoch": 8.27158774373259,
      "grad_norm": 2.6580443382263184,
      "learning_rate": 1.2346503695281409e-05,
      "loss": 0.5315,
      "step": 5939
    },
    {
      "epoch": 8.272980501392757,
      "grad_norm": 2.3212594985961914,
      "learning_rate": 1.2336554860716315e-05,
      "loss": 0.5587,
      "step": 5940
    },
    {
      "epoch": 8.274373259052926,
      "grad_norm": 2.2812185287475586,
      "learning_rate": 1.232660602615122e-05,
      "loss": 0.7328,
      "step": 5941
    },
    {
      "epoch": 8.275766016713092,
      "grad_norm": 2.3596303462982178,
      "learning_rate": 1.2316657191586126e-05,
      "loss": 0.7359,
      "step": 5942
    },
    {
      "epoch": 8.277158774373259,
      "grad_norm": 2.8402531147003174,
      "learning_rate": 1.2306708357021034e-05,
      "loss": 0.7279,
      "step": 5943
    },
    {
      "epoch": 8.278551532033426,
      "grad_norm": 2.4832041263580322,
      "learning_rate": 1.229675952245594e-05,
      "loss": 0.7613,
      "step": 5944
    },
    {
      "epoch": 8.279944289693594,
      "grad_norm": 2.0726089477539062,
      "learning_rate": 1.2286810687890847e-05,
      "loss": 0.6109,
      "step": 5945
    },
    {
      "epoch": 8.28133704735376,
      "grad_norm": 2.011716604232788,
      "learning_rate": 1.2276861853325753e-05,
      "loss": 0.5422,
      "step": 5946
    },
    {
      "epoch": 8.282729805013927,
      "grad_norm": 2.4055304527282715,
      "learning_rate": 1.2266913018760658e-05,
      "loss": 0.6343,
      "step": 5947
    },
    {
      "epoch": 8.284122562674094,
      "grad_norm": 1.9171804189682007,
      "learning_rate": 1.2256964184195566e-05,
      "loss": 0.4508,
      "step": 5948
    },
    {
      "epoch": 8.285515320334262,
      "grad_norm": 2.1899428367614746,
      "learning_rate": 1.2247015349630472e-05,
      "loss": 0.6534,
      "step": 5949
    },
    {
      "epoch": 8.286908077994429,
      "grad_norm": 2.282148838043213,
      "learning_rate": 1.2237066515065377e-05,
      "loss": 0.5948,
      "step": 5950
    },
    {
      "epoch": 8.288300835654596,
      "grad_norm": 2.289850950241089,
      "learning_rate": 1.2227117680500283e-05,
      "loss": 0.5804,
      "step": 5951
    },
    {
      "epoch": 8.289693593314762,
      "grad_norm": 2.1792290210723877,
      "learning_rate": 1.2217168845935189e-05,
      "loss": 0.5808,
      "step": 5952
    },
    {
      "epoch": 8.29108635097493,
      "grad_norm": 2.6128134727478027,
      "learning_rate": 1.2207220011370095e-05,
      "loss": 0.7482,
      "step": 5953
    },
    {
      "epoch": 8.292479108635098,
      "grad_norm": 2.1930418014526367,
      "learning_rate": 1.2197271176805002e-05,
      "loss": 0.6192,
      "step": 5954
    },
    {
      "epoch": 8.293871866295264,
      "grad_norm": 2.435877561569214,
      "learning_rate": 1.2187322342239908e-05,
      "loss": 0.7605,
      "step": 5955
    },
    {
      "epoch": 8.295264623955433,
      "grad_norm": 2.295370101928711,
      "learning_rate": 1.2177373507674814e-05,
      "loss": 0.6078,
      "step": 5956
    },
    {
      "epoch": 8.2966573816156,
      "grad_norm": 2.4461095333099365,
      "learning_rate": 1.216742467310972e-05,
      "loss": 0.621,
      "step": 5957
    },
    {
      "epoch": 8.298050139275766,
      "grad_norm": 2.2723653316497803,
      "learning_rate": 1.2157475838544625e-05,
      "loss": 0.5849,
      "step": 5958
    },
    {
      "epoch": 8.299442896935933,
      "grad_norm": 1.9850846529006958,
      "learning_rate": 1.2147527003979534e-05,
      "loss": 0.5557,
      "step": 5959
    },
    {
      "epoch": 8.300835654596101,
      "grad_norm": 2.227482557296753,
      "learning_rate": 1.213757816941444e-05,
      "loss": 0.5069,
      "step": 5960
    },
    {
      "epoch": 8.302228412256268,
      "grad_norm": 2.16172456741333,
      "learning_rate": 1.2127629334849346e-05,
      "loss": 0.5123,
      "step": 5961
    },
    {
      "epoch": 8.303621169916434,
      "grad_norm": 2.513965368270874,
      "learning_rate": 1.2117680500284252e-05,
      "loss": 0.6231,
      "step": 5962
    },
    {
      "epoch": 8.305013927576601,
      "grad_norm": 2.4063384532928467,
      "learning_rate": 1.2107731665719157e-05,
      "loss": 0.7008,
      "step": 5963
    },
    {
      "epoch": 8.30640668523677,
      "grad_norm": 2.155329465866089,
      "learning_rate": 1.2097782831154063e-05,
      "loss": 0.6408,
      "step": 5964
    },
    {
      "epoch": 8.307799442896936,
      "grad_norm": 2.879249095916748,
      "learning_rate": 1.208783399658897e-05,
      "loss": 0.6533,
      "step": 5965
    },
    {
      "epoch": 8.309192200557103,
      "grad_norm": 3.0566322803497314,
      "learning_rate": 1.2077885162023876e-05,
      "loss": 0.8159,
      "step": 5966
    },
    {
      "epoch": 8.31058495821727,
      "grad_norm": 2.2032434940338135,
      "learning_rate": 1.2067936327458782e-05,
      "loss": 0.6026,
      "step": 5967
    },
    {
      "epoch": 8.311977715877438,
      "grad_norm": 2.7518277168273926,
      "learning_rate": 1.2057987492893688e-05,
      "loss": 0.8638,
      "step": 5968
    },
    {
      "epoch": 8.313370473537605,
      "grad_norm": 2.5776665210723877,
      "learning_rate": 1.2048038658328594e-05,
      "loss": 0.7632,
      "step": 5969
    },
    {
      "epoch": 8.314763231197771,
      "grad_norm": 2.0614118576049805,
      "learning_rate": 1.2038089823763501e-05,
      "loss": 0.4663,
      "step": 5970
    },
    {
      "epoch": 8.316155988857938,
      "grad_norm": 2.5430283546447754,
      "learning_rate": 1.2028140989198407e-05,
      "loss": 0.6766,
      "step": 5971
    },
    {
      "epoch": 8.317548746518106,
      "grad_norm": 2.0144097805023193,
      "learning_rate": 1.2018192154633313e-05,
      "loss": 0.573,
      "step": 5972
    },
    {
      "epoch": 8.318941504178273,
      "grad_norm": 2.3663172721862793,
      "learning_rate": 1.200824332006822e-05,
      "loss": 0.7452,
      "step": 5973
    },
    {
      "epoch": 8.32033426183844,
      "grad_norm": 2.3292689323425293,
      "learning_rate": 1.1998294485503126e-05,
      "loss": 0.4725,
      "step": 5974
    },
    {
      "epoch": 8.321727019498606,
      "grad_norm": 2.3262789249420166,
      "learning_rate": 1.1988345650938032e-05,
      "loss": 0.6988,
      "step": 5975
    },
    {
      "epoch": 8.323119777158775,
      "grad_norm": 2.0353403091430664,
      "learning_rate": 1.1978396816372939e-05,
      "loss": 0.5519,
      "step": 5976
    },
    {
      "epoch": 8.324512534818941,
      "grad_norm": 2.522467851638794,
      "learning_rate": 1.1968447981807845e-05,
      "loss": 0.6967,
      "step": 5977
    },
    {
      "epoch": 8.325905292479108,
      "grad_norm": 2.53556489944458,
      "learning_rate": 1.195849914724275e-05,
      "loss": 0.7762,
      "step": 5978
    },
    {
      "epoch": 8.327298050139277,
      "grad_norm": 2.6604411602020264,
      "learning_rate": 1.1948550312677656e-05,
      "loss": 0.838,
      "step": 5979
    },
    {
      "epoch": 8.328690807799443,
      "grad_norm": 2.690307378768921,
      "learning_rate": 1.1938601478112562e-05,
      "loss": 0.7921,
      "step": 5980
    },
    {
      "epoch": 8.33008356545961,
      "grad_norm": 2.9018197059631348,
      "learning_rate": 1.192865264354747e-05,
      "loss": 0.8936,
      "step": 5981
    },
    {
      "epoch": 8.331476323119777,
      "grad_norm": 2.4915647506713867,
      "learning_rate": 1.1918703808982375e-05,
      "loss": 0.6134,
      "step": 5982
    },
    {
      "epoch": 8.332869080779945,
      "grad_norm": 2.345963716506958,
      "learning_rate": 1.1908754974417281e-05,
      "loss": 0.7294,
      "step": 5983
    },
    {
      "epoch": 8.334261838440112,
      "grad_norm": 2.099630832672119,
      "learning_rate": 1.1898806139852187e-05,
      "loss": 0.6348,
      "step": 5984
    },
    {
      "epoch": 8.335654596100278,
      "grad_norm": 2.1514601707458496,
      "learning_rate": 1.1888857305287093e-05,
      "loss": 0.6229,
      "step": 5985
    },
    {
      "epoch": 8.337047353760445,
      "grad_norm": 2.3876328468322754,
      "learning_rate": 1.1878908470721999e-05,
      "loss": 0.6337,
      "step": 5986
    },
    {
      "epoch": 8.338440111420613,
      "grad_norm": 2.335765838623047,
      "learning_rate": 1.1868959636156908e-05,
      "loss": 0.632,
      "step": 5987
    },
    {
      "epoch": 8.33983286908078,
      "grad_norm": 2.3674325942993164,
      "learning_rate": 1.1859010801591813e-05,
      "loss": 0.6469,
      "step": 5988
    },
    {
      "epoch": 8.341225626740947,
      "grad_norm": 2.791717529296875,
      "learning_rate": 1.184906196702672e-05,
      "loss": 0.7054,
      "step": 5989
    },
    {
      "epoch": 8.342618384401113,
      "grad_norm": 2.028801679611206,
      "learning_rate": 1.1839113132461625e-05,
      "loss": 0.5468,
      "step": 5990
    },
    {
      "epoch": 8.344011142061282,
      "grad_norm": 2.0309956073760986,
      "learning_rate": 1.182916429789653e-05,
      "loss": 0.5663,
      "step": 5991
    },
    {
      "epoch": 8.345403899721449,
      "grad_norm": 2.292309522628784,
      "learning_rate": 1.1819215463331438e-05,
      "loss": 0.5529,
      "step": 5992
    },
    {
      "epoch": 8.346796657381615,
      "grad_norm": 2.561776876449585,
      "learning_rate": 1.1809266628766344e-05,
      "loss": 0.6528,
      "step": 5993
    },
    {
      "epoch": 8.348189415041782,
      "grad_norm": 2.0604283809661865,
      "learning_rate": 1.179931779420125e-05,
      "loss": 0.514,
      "step": 5994
    },
    {
      "epoch": 8.34958217270195,
      "grad_norm": 2.6872153282165527,
      "learning_rate": 1.1789368959636156e-05,
      "loss": 0.7647,
      "step": 5995
    },
    {
      "epoch": 8.350974930362117,
      "grad_norm": 2.186255693435669,
      "learning_rate": 1.1779420125071061e-05,
      "loss": 0.5133,
      "step": 5996
    },
    {
      "epoch": 8.352367688022284,
      "grad_norm": 2.1541671752929688,
      "learning_rate": 1.1769471290505967e-05,
      "loss": 0.5094,
      "step": 5997
    },
    {
      "epoch": 8.35376044568245,
      "grad_norm": 2.272148847579956,
      "learning_rate": 1.1759522455940875e-05,
      "loss": 0.5915,
      "step": 5998
    },
    {
      "epoch": 8.355153203342619,
      "grad_norm": 2.8752074241638184,
      "learning_rate": 1.174957362137578e-05,
      "loss": 0.6677,
      "step": 5999
    },
    {
      "epoch": 8.356545961002785,
      "grad_norm": 2.014286756515503,
      "learning_rate": 1.1739624786810686e-05,
      "loss": 0.4691,
      "step": 6000
    },
    {
      "epoch": 8.357938718662952,
      "grad_norm": 2.2888171672821045,
      "learning_rate": 1.1729675952245594e-05,
      "loss": 0.5322,
      "step": 6001
    },
    {
      "epoch": 8.35933147632312,
      "grad_norm": 2.31115460395813,
      "learning_rate": 1.17197271176805e-05,
      "loss": 0.7171,
      "step": 6002
    },
    {
      "epoch": 8.360724233983287,
      "grad_norm": 4.947230815887451,
      "learning_rate": 1.1709778283115407e-05,
      "loss": 0.4356,
      "step": 6003
    },
    {
      "epoch": 8.362116991643454,
      "grad_norm": 2.268806219100952,
      "learning_rate": 1.1699829448550313e-05,
      "loss": 0.5276,
      "step": 6004
    },
    {
      "epoch": 8.36350974930362,
      "grad_norm": 2.08506178855896,
      "learning_rate": 1.1689880613985218e-05,
      "loss": 0.5546,
      "step": 6005
    },
    {
      "epoch": 8.364902506963789,
      "grad_norm": 2.5294394493103027,
      "learning_rate": 1.1679931779420124e-05,
      "loss": 0.7976,
      "step": 6006
    },
    {
      "epoch": 8.366295264623956,
      "grad_norm": 2.3474745750427246,
      "learning_rate": 1.166998294485503e-05,
      "loss": 0.7576,
      "step": 6007
    },
    {
      "epoch": 8.367688022284122,
      "grad_norm": 2.291719913482666,
      "learning_rate": 1.1660034110289936e-05,
      "loss": 0.6335,
      "step": 6008
    },
    {
      "epoch": 8.369080779944289,
      "grad_norm": 2.443223237991333,
      "learning_rate": 1.1650085275724843e-05,
      "loss": 0.6618,
      "step": 6009
    },
    {
      "epoch": 8.370473537604457,
      "grad_norm": 2.556466817855835,
      "learning_rate": 1.1640136441159749e-05,
      "loss": 0.5996,
      "step": 6010
    },
    {
      "epoch": 8.371866295264624,
      "grad_norm": 2.3878207206726074,
      "learning_rate": 1.1630187606594655e-05,
      "loss": 0.6798,
      "step": 6011
    },
    {
      "epoch": 8.37325905292479,
      "grad_norm": 2.4887008666992188,
      "learning_rate": 1.162023877202956e-05,
      "loss": 0.6105,
      "step": 6012
    },
    {
      "epoch": 8.374651810584957,
      "grad_norm": 2.777446985244751,
      "learning_rate": 1.1610289937464466e-05,
      "loss": 0.6818,
      "step": 6013
    },
    {
      "epoch": 8.376044568245126,
      "grad_norm": 2.743042469024658,
      "learning_rate": 1.1600341102899375e-05,
      "loss": 0.8038,
      "step": 6014
    },
    {
      "epoch": 8.377437325905293,
      "grad_norm": 1.8929640054702759,
      "learning_rate": 1.1590392268334281e-05,
      "loss": 0.518,
      "step": 6015
    },
    {
      "epoch": 8.37883008356546,
      "grad_norm": 2.203227996826172,
      "learning_rate": 1.1580443433769187e-05,
      "loss": 0.5197,
      "step": 6016
    },
    {
      "epoch": 8.380222841225626,
      "grad_norm": 2.3382623195648193,
      "learning_rate": 1.1570494599204093e-05,
      "loss": 0.5744,
      "step": 6017
    },
    {
      "epoch": 8.381615598885794,
      "grad_norm": 2.7288174629211426,
      "learning_rate": 1.1560545764638998e-05,
      "loss": 0.8261,
      "step": 6018
    },
    {
      "epoch": 8.383008356545961,
      "grad_norm": 2.1930718421936035,
      "learning_rate": 1.1550596930073904e-05,
      "loss": 0.5449,
      "step": 6019
    },
    {
      "epoch": 8.384401114206128,
      "grad_norm": 2.059731960296631,
      "learning_rate": 1.1540648095508812e-05,
      "loss": 0.5807,
      "step": 6020
    },
    {
      "epoch": 8.385793871866296,
      "grad_norm": 2.0126492977142334,
      "learning_rate": 1.1530699260943717e-05,
      "loss": 0.479,
      "step": 6021
    },
    {
      "epoch": 8.387186629526463,
      "grad_norm": 3.5227551460266113,
      "learning_rate": 1.1520750426378623e-05,
      "loss": 0.6927,
      "step": 6022
    },
    {
      "epoch": 8.38857938718663,
      "grad_norm": 2.895655393600464,
      "learning_rate": 1.1510801591813529e-05,
      "loss": 0.8958,
      "step": 6023
    },
    {
      "epoch": 8.389972144846796,
      "grad_norm": 2.062605381011963,
      "learning_rate": 1.1500852757248435e-05,
      "loss": 0.5053,
      "step": 6024
    },
    {
      "epoch": 8.391364902506965,
      "grad_norm": 2.0187153816223145,
      "learning_rate": 1.1490903922683342e-05,
      "loss": 0.5552,
      "step": 6025
    },
    {
      "epoch": 8.392757660167131,
      "grad_norm": 2.3866779804229736,
      "learning_rate": 1.1480955088118248e-05,
      "loss": 0.6162,
      "step": 6026
    },
    {
      "epoch": 8.394150417827298,
      "grad_norm": 2.3278214931488037,
      "learning_rate": 1.1471006253553154e-05,
      "loss": 0.6794,
      "step": 6027
    },
    {
      "epoch": 8.395543175487465,
      "grad_norm": 2.476162910461426,
      "learning_rate": 1.146105741898806e-05,
      "loss": 0.682,
      "step": 6028
    },
    {
      "epoch": 8.396935933147633,
      "grad_norm": 2.229062080383301,
      "learning_rate": 1.1451108584422967e-05,
      "loss": 0.6901,
      "step": 6029
    },
    {
      "epoch": 8.3983286908078,
      "grad_norm": 2.519990921020508,
      "learning_rate": 1.1441159749857873e-05,
      "loss": 0.6159,
      "step": 6030
    },
    {
      "epoch": 8.399721448467966,
      "grad_norm": 1.9492111206054688,
      "learning_rate": 1.143121091529278e-05,
      "loss": 0.5927,
      "step": 6031
    },
    {
      "epoch": 8.401114206128133,
      "grad_norm": 2.318389415740967,
      "learning_rate": 1.1421262080727686e-05,
      "loss": 0.6085,
      "step": 6032
    },
    {
      "epoch": 8.402506963788301,
      "grad_norm": 2.561988115310669,
      "learning_rate": 1.1411313246162592e-05,
      "loss": 0.7195,
      "step": 6033
    },
    {
      "epoch": 8.403899721448468,
      "grad_norm": 2.4790420532226562,
      "learning_rate": 1.1401364411597497e-05,
      "loss": 0.8115,
      "step": 6034
    },
    {
      "epoch": 8.405292479108635,
      "grad_norm": 2.3170247077941895,
      "learning_rate": 1.1391415577032403e-05,
      "loss": 0.711,
      "step": 6035
    },
    {
      "epoch": 8.406685236768801,
      "grad_norm": 2.3760390281677246,
      "learning_rate": 1.138146674246731e-05,
      "loss": 0.6746,
      "step": 6036
    },
    {
      "epoch": 8.40807799442897,
      "grad_norm": 2.3536174297332764,
      "learning_rate": 1.1371517907902216e-05,
      "loss": 0.7556,
      "step": 6037
    },
    {
      "epoch": 8.409470752089137,
      "grad_norm": 2.398081064224243,
      "learning_rate": 1.1361569073337122e-05,
      "loss": 0.577,
      "step": 6038
    },
    {
      "epoch": 8.410863509749303,
      "grad_norm": 2.490631341934204,
      "learning_rate": 1.1351620238772028e-05,
      "loss": 0.776,
      "step": 6039
    },
    {
      "epoch": 8.412256267409472,
      "grad_norm": 2.167231798171997,
      "learning_rate": 1.1341671404206934e-05,
      "loss": 0.6009,
      "step": 6040
    },
    {
      "epoch": 8.413649025069638,
      "grad_norm": 3.3253467082977295,
      "learning_rate": 1.133172256964184e-05,
      "loss": 0.7999,
      "step": 6041
    },
    {
      "epoch": 8.415041782729805,
      "grad_norm": 2.2623608112335205,
      "learning_rate": 1.1321773735076749e-05,
      "loss": 0.5736,
      "step": 6042
    },
    {
      "epoch": 8.416434540389972,
      "grad_norm": 2.054021120071411,
      "learning_rate": 1.1311824900511654e-05,
      "loss": 0.5795,
      "step": 6043
    },
    {
      "epoch": 8.41782729805014,
      "grad_norm": 2.448329448699951,
      "learning_rate": 1.130187606594656e-05,
      "loss": 0.8009,
      "step": 6044
    },
    {
      "epoch": 8.419220055710307,
      "grad_norm": 2.1401994228363037,
      "learning_rate": 1.1291927231381466e-05,
      "loss": 0.597,
      "step": 6045
    },
    {
      "epoch": 8.420612813370473,
      "grad_norm": 2.19303035736084,
      "learning_rate": 1.1281978396816372e-05,
      "loss": 0.6198,
      "step": 6046
    },
    {
      "epoch": 8.42200557103064,
      "grad_norm": 2.4928131103515625,
      "learning_rate": 1.127202956225128e-05,
      "loss": 0.5572,
      "step": 6047
    },
    {
      "epoch": 8.423398328690809,
      "grad_norm": 3.331573009490967,
      "learning_rate": 1.1262080727686185e-05,
      "loss": 0.9076,
      "step": 6048
    },
    {
      "epoch": 8.424791086350975,
      "grad_norm": 2.0832619667053223,
      "learning_rate": 1.125213189312109e-05,
      "loss": 0.5759,
      "step": 6049
    },
    {
      "epoch": 8.426183844011142,
      "grad_norm": 2.446834087371826,
      "learning_rate": 1.1242183058555997e-05,
      "loss": 0.7688,
      "step": 6050
    },
    {
      "epoch": 8.427576601671309,
      "grad_norm": 2.683854818344116,
      "learning_rate": 1.1232234223990902e-05,
      "loss": 0.8399,
      "step": 6051
    },
    {
      "epoch": 8.428969359331477,
      "grad_norm": 2.356480121612549,
      "learning_rate": 1.1222285389425808e-05,
      "loss": 0.6126,
      "step": 6052
    },
    {
      "epoch": 8.430362116991644,
      "grad_norm": 2.6480844020843506,
      "learning_rate": 1.1212336554860716e-05,
      "loss": 0.6399,
      "step": 6053
    },
    {
      "epoch": 8.43175487465181,
      "grad_norm": 2.059575319290161,
      "learning_rate": 1.1202387720295621e-05,
      "loss": 0.5944,
      "step": 6054
    },
    {
      "epoch": 8.433147632311977,
      "grad_norm": 2.2821314334869385,
      "learning_rate": 1.1192438885730527e-05,
      "loss": 0.5851,
      "step": 6055
    },
    {
      "epoch": 8.434540389972145,
      "grad_norm": 2.48276948928833,
      "learning_rate": 1.1182490051165433e-05,
      "loss": 0.6728,
      "step": 6056
    },
    {
      "epoch": 8.435933147632312,
      "grad_norm": 2.7306902408599854,
      "learning_rate": 1.117254121660034e-05,
      "loss": 0.7469,
      "step": 6057
    },
    {
      "epoch": 8.437325905292479,
      "grad_norm": 2.428895950317383,
      "learning_rate": 1.1162592382035248e-05,
      "loss": 0.6445,
      "step": 6058
    },
    {
      "epoch": 8.438718662952645,
      "grad_norm": 2.553884744644165,
      "learning_rate": 1.1152643547470154e-05,
      "loss": 0.7236,
      "step": 6059
    },
    {
      "epoch": 8.440111420612814,
      "grad_norm": 2.110875129699707,
      "learning_rate": 1.114269471290506e-05,
      "loss": 0.5479,
      "step": 6060
    },
    {
      "epoch": 8.44150417827298,
      "grad_norm": 2.2131245136260986,
      "learning_rate": 1.1132745878339965e-05,
      "loss": 0.6372,
      "step": 6061
    },
    {
      "epoch": 8.442896935933147,
      "grad_norm": 2.3978965282440186,
      "learning_rate": 1.112279704377487e-05,
      "loss": 0.6745,
      "step": 6062
    },
    {
      "epoch": 8.444289693593316,
      "grad_norm": 2.78007435798645,
      "learning_rate": 1.1112848209209777e-05,
      "loss": 0.7903,
      "step": 6063
    },
    {
      "epoch": 8.445682451253482,
      "grad_norm": 2.0242226123809814,
      "learning_rate": 1.1102899374644684e-05,
      "loss": 0.5913,
      "step": 6064
    },
    {
      "epoch": 8.447075208913649,
      "grad_norm": 2.1386795043945312,
      "learning_rate": 1.109295054007959e-05,
      "loss": 0.5568,
      "step": 6065
    },
    {
      "epoch": 8.448467966573816,
      "grad_norm": 2.5141165256500244,
      "learning_rate": 1.1083001705514496e-05,
      "loss": 0.7348,
      "step": 6066
    },
    {
      "epoch": 8.449860724233984,
      "grad_norm": 2.2028284072875977,
      "learning_rate": 1.1073052870949401e-05,
      "loss": 0.5872,
      "step": 6067
    },
    {
      "epoch": 8.45125348189415,
      "grad_norm": 2.050466299057007,
      "learning_rate": 1.1063104036384307e-05,
      "loss": 0.5879,
      "step": 6068
    },
    {
      "epoch": 8.452646239554317,
      "grad_norm": 2.748208999633789,
      "learning_rate": 1.1053155201819215e-05,
      "loss": 0.9008,
      "step": 6069
    },
    {
      "epoch": 8.454038997214484,
      "grad_norm": 2.5057647228240967,
      "learning_rate": 1.1043206367254122e-05,
      "loss": 0.5594,
      "step": 6070
    },
    {
      "epoch": 8.455431754874652,
      "grad_norm": 2.995257616043091,
      "learning_rate": 1.1033257532689028e-05,
      "loss": 0.6864,
      "step": 6071
    },
    {
      "epoch": 8.45682451253482,
      "grad_norm": 2.420017957687378,
      "learning_rate": 1.1023308698123934e-05,
      "loss": 0.5537,
      "step": 6072
    },
    {
      "epoch": 8.458217270194986,
      "grad_norm": 2.578761100769043,
      "learning_rate": 1.101335986355884e-05,
      "loss": 0.7239,
      "step": 6073
    },
    {
      "epoch": 8.459610027855152,
      "grad_norm": 2.409921884536743,
      "learning_rate": 1.1003411028993745e-05,
      "loss": 0.6184,
      "step": 6074
    },
    {
      "epoch": 8.461002785515321,
      "grad_norm": 2.3940603733062744,
      "learning_rate": 1.0993462194428653e-05,
      "loss": 0.7035,
      "step": 6075
    },
    {
      "epoch": 8.462395543175488,
      "grad_norm": 2.53452730178833,
      "learning_rate": 1.0983513359863558e-05,
      "loss": 0.8525,
      "step": 6076
    },
    {
      "epoch": 8.463788300835654,
      "grad_norm": 2.4821276664733887,
      "learning_rate": 1.0973564525298464e-05,
      "loss": 0.6698,
      "step": 6077
    },
    {
      "epoch": 8.465181058495821,
      "grad_norm": 2.6680209636688232,
      "learning_rate": 1.096361569073337e-05,
      "loss": 0.8518,
      "step": 6078
    },
    {
      "epoch": 8.46657381615599,
      "grad_norm": 2.228168487548828,
      "learning_rate": 1.0953666856168276e-05,
      "loss": 0.6164,
      "step": 6079
    },
    {
      "epoch": 8.467966573816156,
      "grad_norm": 2.6836867332458496,
      "learning_rate": 1.0943718021603183e-05,
      "loss": 0.8314,
      "step": 6080
    },
    {
      "epoch": 8.469359331476323,
      "grad_norm": 2.4073612689971924,
      "learning_rate": 1.0933769187038089e-05,
      "loss": 0.6607,
      "step": 6081
    },
    {
      "epoch": 8.47075208913649,
      "grad_norm": 2.2259182929992676,
      "learning_rate": 1.0923820352472995e-05,
      "loss": 0.6018,
      "step": 6082
    },
    {
      "epoch": 8.472144846796658,
      "grad_norm": 2.0883285999298096,
      "learning_rate": 1.09138715179079e-05,
      "loss": 0.5973,
      "step": 6083
    },
    {
      "epoch": 8.473537604456824,
      "grad_norm": 2.151155948638916,
      "learning_rate": 1.0903922683342806e-05,
      "loss": 0.6066,
      "step": 6084
    },
    {
      "epoch": 8.474930362116991,
      "grad_norm": 2.8162882328033447,
      "learning_rate": 1.0893973848777714e-05,
      "loss": 0.6641,
      "step": 6085
    },
    {
      "epoch": 8.47632311977716,
      "grad_norm": 2.4459331035614014,
      "learning_rate": 1.0884025014212621e-05,
      "loss": 0.6417,
      "step": 6086
    },
    {
      "epoch": 8.477715877437326,
      "grad_norm": 2.411317825317383,
      "learning_rate": 1.0874076179647527e-05,
      "loss": 0.6312,
      "step": 6087
    },
    {
      "epoch": 8.479108635097493,
      "grad_norm": 2.0946240425109863,
      "learning_rate": 1.0864127345082433e-05,
      "loss": 0.487,
      "step": 6088
    },
    {
      "epoch": 8.48050139275766,
      "grad_norm": 2.4434168338775635,
      "learning_rate": 1.0854178510517338e-05,
      "loss": 0.7161,
      "step": 6089
    },
    {
      "epoch": 8.481894150417828,
      "grad_norm": 2.4298110008239746,
      "learning_rate": 1.0844229675952244e-05,
      "loss": 0.7625,
      "step": 6090
    },
    {
      "epoch": 8.483286908077995,
      "grad_norm": 1.8792526721954346,
      "learning_rate": 1.0834280841387152e-05,
      "loss": 0.484,
      "step": 6091
    },
    {
      "epoch": 8.484679665738161,
      "grad_norm": 2.4808316230773926,
      "learning_rate": 1.0824332006822057e-05,
      "loss": 0.6855,
      "step": 6092
    },
    {
      "epoch": 8.486072423398328,
      "grad_norm": 2.2407937049865723,
      "learning_rate": 1.0814383172256963e-05,
      "loss": 0.6131,
      "step": 6093
    },
    {
      "epoch": 8.487465181058496,
      "grad_norm": 2.2950949668884277,
      "learning_rate": 1.0804434337691869e-05,
      "loss": 0.6918,
      "step": 6094
    },
    {
      "epoch": 8.488857938718663,
      "grad_norm": 2.2543063163757324,
      "learning_rate": 1.0794485503126775e-05,
      "loss": 0.6633,
      "step": 6095
    },
    {
      "epoch": 8.49025069637883,
      "grad_norm": 2.020587205886841,
      "learning_rate": 1.078453666856168e-05,
      "loss": 0.517,
      "step": 6096
    },
    {
      "epoch": 8.491643454038996,
      "grad_norm": 2.5152981281280518,
      "learning_rate": 1.0774587833996588e-05,
      "loss": 0.7744,
      "step": 6097
    },
    {
      "epoch": 8.493036211699165,
      "grad_norm": 2.577939987182617,
      "learning_rate": 1.0764638999431495e-05,
      "loss": 0.7159,
      "step": 6098
    },
    {
      "epoch": 8.494428969359332,
      "grad_norm": 2.560106039047241,
      "learning_rate": 1.0754690164866401e-05,
      "loss": 0.724,
      "step": 6099
    },
    {
      "epoch": 8.495821727019498,
      "grad_norm": 2.466357469558716,
      "learning_rate": 1.0744741330301307e-05,
      "loss": 0.7571,
      "step": 6100
    },
    {
      "epoch": 8.497214484679665,
      "grad_norm": 2.45121169090271,
      "learning_rate": 1.0734792495736213e-05,
      "loss": 0.7114,
      "step": 6101
    },
    {
      "epoch": 8.498607242339833,
      "grad_norm": 2.4164838790893555,
      "learning_rate": 1.072484366117112e-05,
      "loss": 0.687,
      "step": 6102
    },
    {
      "epoch": 8.5,
      "grad_norm": 2.3164844512939453,
      "learning_rate": 1.0714894826606026e-05,
      "loss": 0.6274,
      "step": 6103
    },
    {
      "epoch": 8.501392757660167,
      "grad_norm": 2.7290728092193604,
      "learning_rate": 1.0704945992040932e-05,
      "loss": 0.6255,
      "step": 6104
    },
    {
      "epoch": 8.502785515320333,
      "grad_norm": 2.267289400100708,
      "learning_rate": 1.0694997157475838e-05,
      "loss": 0.578,
      "step": 6105
    },
    {
      "epoch": 8.504178272980502,
      "grad_norm": 2.3413429260253906,
      "learning_rate": 1.0685048322910743e-05,
      "loss": 0.5592,
      "step": 6106
    },
    {
      "epoch": 8.505571030640668,
      "grad_norm": 1.9599863290786743,
      "learning_rate": 1.0675099488345649e-05,
      "loss": 0.4539,
      "step": 6107
    },
    {
      "epoch": 8.506963788300835,
      "grad_norm": 2.0619986057281494,
      "learning_rate": 1.0665150653780557e-05,
      "loss": 0.5147,
      "step": 6108
    },
    {
      "epoch": 8.508356545961004,
      "grad_norm": 2.3537232875823975,
      "learning_rate": 1.0655201819215462e-05,
      "loss": 0.638,
      "step": 6109
    },
    {
      "epoch": 8.50974930362117,
      "grad_norm": 2.3712711334228516,
      "learning_rate": 1.0645252984650368e-05,
      "loss": 0.7834,
      "step": 6110
    },
    {
      "epoch": 8.511142061281337,
      "grad_norm": 1.855421543121338,
      "learning_rate": 1.0635304150085274e-05,
      "loss": 0.5217,
      "step": 6111
    },
    {
      "epoch": 8.512534818941504,
      "grad_norm": 2.5245094299316406,
      "learning_rate": 1.062535531552018e-05,
      "loss": 0.6808,
      "step": 6112
    },
    {
      "epoch": 8.513927576601672,
      "grad_norm": 2.324967384338379,
      "learning_rate": 1.0615406480955085e-05,
      "loss": 0.751,
      "step": 6113
    },
    {
      "epoch": 8.515320334261839,
      "grad_norm": 2.984036922454834,
      "learning_rate": 1.0605457646389995e-05,
      "loss": 0.8782,
      "step": 6114
    },
    {
      "epoch": 8.516713091922005,
      "grad_norm": 2.8980345726013184,
      "learning_rate": 1.05955088118249e-05,
      "loss": 0.7493,
      "step": 6115
    },
    {
      "epoch": 8.518105849582172,
      "grad_norm": 2.2524449825286865,
      "learning_rate": 1.0585559977259806e-05,
      "loss": 0.6959,
      "step": 6116
    },
    {
      "epoch": 8.51949860724234,
      "grad_norm": 2.000723361968994,
      "learning_rate": 1.0575611142694712e-05,
      "loss": 0.514,
      "step": 6117
    },
    {
      "epoch": 8.520891364902507,
      "grad_norm": 2.113044500350952,
      "learning_rate": 1.0565662308129618e-05,
      "loss": 0.6553,
      "step": 6118
    },
    {
      "epoch": 8.522284122562674,
      "grad_norm": 1.9838714599609375,
      "learning_rate": 1.0555713473564525e-05,
      "loss": 0.533,
      "step": 6119
    },
    {
      "epoch": 8.52367688022284,
      "grad_norm": 2.3189139366149902,
      "learning_rate": 1.0545764638999431e-05,
      "loss": 0.5607,
      "step": 6120
    },
    {
      "epoch": 8.525069637883009,
      "grad_norm": 2.4928135871887207,
      "learning_rate": 1.0535815804434337e-05,
      "loss": 0.7224,
      "step": 6121
    },
    {
      "epoch": 8.526462395543176,
      "grad_norm": 1.9980363845825195,
      "learning_rate": 1.0525866969869242e-05,
      "loss": 0.4732,
      "step": 6122
    },
    {
      "epoch": 8.527855153203342,
      "grad_norm": 2.1374313831329346,
      "learning_rate": 1.0515918135304148e-05,
      "loss": 0.6443,
      "step": 6123
    },
    {
      "epoch": 8.52924791086351,
      "grad_norm": 2.423882246017456,
      "learning_rate": 1.0505969300739054e-05,
      "loss": 0.7051,
      "step": 6124
    },
    {
      "epoch": 8.530640668523677,
      "grad_norm": 2.315556526184082,
      "learning_rate": 1.0496020466173961e-05,
      "loss": 0.5865,
      "step": 6125
    },
    {
      "epoch": 8.532033426183844,
      "grad_norm": 2.257117509841919,
      "learning_rate": 1.0486071631608869e-05,
      "loss": 0.6034,
      "step": 6126
    },
    {
      "epoch": 8.53342618384401,
      "grad_norm": 2.1576497554779053,
      "learning_rate": 1.0476122797043775e-05,
      "loss": 0.5834,
      "step": 6127
    },
    {
      "epoch": 8.534818941504179,
      "grad_norm": 2.4187119007110596,
      "learning_rate": 1.046617396247868e-05,
      "loss": 0.7332,
      "step": 6128
    },
    {
      "epoch": 8.536211699164346,
      "grad_norm": 2.3867664337158203,
      "learning_rate": 1.0456225127913586e-05,
      "loss": 0.6255,
      "step": 6129
    },
    {
      "epoch": 8.537604456824512,
      "grad_norm": 2.352771520614624,
      "learning_rate": 1.0446276293348494e-05,
      "loss": 0.6072,
      "step": 6130
    },
    {
      "epoch": 8.538997214484679,
      "grad_norm": 2.334275722503662,
      "learning_rate": 1.04363274587834e-05,
      "loss": 0.7366,
      "step": 6131
    },
    {
      "epoch": 8.540389972144848,
      "grad_norm": 2.602456569671631,
      "learning_rate": 1.0426378624218305e-05,
      "loss": 0.7915,
      "step": 6132
    },
    {
      "epoch": 8.541782729805014,
      "grad_norm": 2.212661027908325,
      "learning_rate": 1.0416429789653211e-05,
      "loss": 0.6547,
      "step": 6133
    },
    {
      "epoch": 8.54317548746518,
      "grad_norm": 2.5381555557250977,
      "learning_rate": 1.0406480955088117e-05,
      "loss": 0.7952,
      "step": 6134
    },
    {
      "epoch": 8.544568245125348,
      "grad_norm": 2.6080639362335205,
      "learning_rate": 1.0396532120523022e-05,
      "loss": 0.8796,
      "step": 6135
    },
    {
      "epoch": 8.545961002785516,
      "grad_norm": 2.1430845260620117,
      "learning_rate": 1.038658328595793e-05,
      "loss": 0.702,
      "step": 6136
    },
    {
      "epoch": 8.547353760445683,
      "grad_norm": 2.3446338176727295,
      "learning_rate": 1.0376634451392836e-05,
      "loss": 0.5499,
      "step": 6137
    },
    {
      "epoch": 8.54874651810585,
      "grad_norm": 2.168344736099243,
      "learning_rate": 1.0366685616827741e-05,
      "loss": 0.536,
      "step": 6138
    },
    {
      "epoch": 8.550139275766016,
      "grad_norm": 2.585874319076538,
      "learning_rate": 1.0356736782262647e-05,
      "loss": 0.6948,
      "step": 6139
    },
    {
      "epoch": 8.551532033426184,
      "grad_norm": 2.2352216243743896,
      "learning_rate": 1.0346787947697553e-05,
      "loss": 0.609,
      "step": 6140
    },
    {
      "epoch": 8.552924791086351,
      "grad_norm": 3.1932296752929688,
      "learning_rate": 1.0336839113132462e-05,
      "loss": 0.6917,
      "step": 6141
    },
    {
      "epoch": 8.554317548746518,
      "grad_norm": 2.3370463848114014,
      "learning_rate": 1.0326890278567368e-05,
      "loss": 0.5339,
      "step": 6142
    },
    {
      "epoch": 8.555710306406684,
      "grad_norm": 2.7332851886749268,
      "learning_rate": 1.0316941444002274e-05,
      "loss": 0.8413,
      "step": 6143
    },
    {
      "epoch": 8.557103064066853,
      "grad_norm": 2.489694595336914,
      "learning_rate": 1.030699260943718e-05,
      "loss": 0.7553,
      "step": 6144
    },
    {
      "epoch": 8.55849582172702,
      "grad_norm": 2.6534366607666016,
      "learning_rate": 1.0297043774872085e-05,
      "loss": 0.8061,
      "step": 6145
    },
    {
      "epoch": 8.559888579387186,
      "grad_norm": 2.3838272094726562,
      "learning_rate": 1.0287094940306991e-05,
      "loss": 0.6884,
      "step": 6146
    },
    {
      "epoch": 8.561281337047355,
      "grad_norm": 2.1057419776916504,
      "learning_rate": 1.0277146105741898e-05,
      "loss": 0.5985,
      "step": 6147
    },
    {
      "epoch": 8.562674094707521,
      "grad_norm": 2.3954782485961914,
      "learning_rate": 1.0267197271176804e-05,
      "loss": 0.5937,
      "step": 6148
    },
    {
      "epoch": 8.564066852367688,
      "grad_norm": 2.503591299057007,
      "learning_rate": 1.025724843661171e-05,
      "loss": 0.774,
      "step": 6149
    },
    {
      "epoch": 8.565459610027855,
      "grad_norm": 2.894399404525757,
      "learning_rate": 1.0247299602046616e-05,
      "loss": 0.5723,
      "step": 6150
    },
    {
      "epoch": 8.566852367688023,
      "grad_norm": 2.4418928623199463,
      "learning_rate": 1.0237350767481522e-05,
      "loss": 0.7451,
      "step": 6151
    },
    {
      "epoch": 8.56824512534819,
      "grad_norm": 2.3711578845977783,
      "learning_rate": 1.0227401932916429e-05,
      "loss": 0.7051,
      "step": 6152
    },
    {
      "epoch": 8.569637883008356,
      "grad_norm": 2.750633716583252,
      "learning_rate": 1.0217453098351335e-05,
      "loss": 0.7105,
      "step": 6153
    },
    {
      "epoch": 8.571030640668523,
      "grad_norm": 2.5556020736694336,
      "learning_rate": 1.0207504263786242e-05,
      "loss": 0.7731,
      "step": 6154
    },
    {
      "epoch": 8.572423398328691,
      "grad_norm": 2.362330913543701,
      "learning_rate": 1.0197555429221148e-05,
      "loss": 0.7308,
      "step": 6155
    },
    {
      "epoch": 8.573816155988858,
      "grad_norm": 2.430133581161499,
      "learning_rate": 1.0187606594656054e-05,
      "loss": 0.6906,
      "step": 6156
    },
    {
      "epoch": 8.575208913649025,
      "grad_norm": 2.301302194595337,
      "learning_rate": 1.017765776009096e-05,
      "loss": 0.5844,
      "step": 6157
    },
    {
      "epoch": 8.576601671309191,
      "grad_norm": 2.2957372665405273,
      "learning_rate": 1.0167708925525867e-05,
      "loss": 0.6841,
      "step": 6158
    },
    {
      "epoch": 8.57799442896936,
      "grad_norm": 2.3437983989715576,
      "learning_rate": 1.0157760090960773e-05,
      "loss": 0.6135,
      "step": 6159
    },
    {
      "epoch": 8.579387186629527,
      "grad_norm": 2.098742961883545,
      "learning_rate": 1.0147811256395679e-05,
      "loss": 0.4693,
      "step": 6160
    },
    {
      "epoch": 8.580779944289693,
      "grad_norm": 2.3117828369140625,
      "learning_rate": 1.0137862421830584e-05,
      "loss": 0.5804,
      "step": 6161
    },
    {
      "epoch": 8.58217270194986,
      "grad_norm": 2.6008477210998535,
      "learning_rate": 1.012791358726549e-05,
      "loss": 0.672,
      "step": 6162
    },
    {
      "epoch": 8.583565459610028,
      "grad_norm": 3.014817476272583,
      "learning_rate": 1.0117964752700398e-05,
      "loss": 0.5709,
      "step": 6163
    },
    {
      "epoch": 8.584958217270195,
      "grad_norm": 2.3694300651550293,
      "learning_rate": 1.0108015918135303e-05,
      "loss": 0.6873,
      "step": 6164
    },
    {
      "epoch": 8.586350974930362,
      "grad_norm": 2.311758279800415,
      "learning_rate": 1.0098067083570209e-05,
      "loss": 0.5831,
      "step": 6165
    },
    {
      "epoch": 8.587743732590528,
      "grad_norm": 2.4471564292907715,
      "learning_rate": 1.0088118249005115e-05,
      "loss": 0.6819,
      "step": 6166
    },
    {
      "epoch": 8.589136490250697,
      "grad_norm": 2.028956413269043,
      "learning_rate": 1.007816941444002e-05,
      "loss": 0.5429,
      "step": 6167
    },
    {
      "epoch": 8.590529247910863,
      "grad_norm": 2.548349380493164,
      "learning_rate": 1.0068220579874926e-05,
      "loss": 0.8417,
      "step": 6168
    },
    {
      "epoch": 8.59192200557103,
      "grad_norm": 1.9181674718856812,
      "learning_rate": 1.0058271745309836e-05,
      "loss": 0.5545,
      "step": 6169
    },
    {
      "epoch": 8.593314763231199,
      "grad_norm": 1.998604416847229,
      "learning_rate": 1.0048322910744741e-05,
      "loss": 0.4737,
      "step": 6170
    },
    {
      "epoch": 8.594707520891365,
      "grad_norm": 1.970977783203125,
      "learning_rate": 1.0038374076179647e-05,
      "loss": 0.5835,
      "step": 6171
    },
    {
      "epoch": 8.596100278551532,
      "grad_norm": 2.541653871536255,
      "learning_rate": 1.0028425241614553e-05,
      "loss": 0.7278,
      "step": 6172
    },
    {
      "epoch": 8.597493036211699,
      "grad_norm": 2.5394837856292725,
      "learning_rate": 1.0018476407049459e-05,
      "loss": 0.7412,
      "step": 6173
    },
    {
      "epoch": 8.598885793871867,
      "grad_norm": 2.08638596534729,
      "learning_rate": 1.0008527572484366e-05,
      "loss": 0.5682,
      "step": 6174
    },
    {
      "epoch": 8.600278551532034,
      "grad_norm": 2.239345073699951,
      "learning_rate": 9.998578737919272e-06,
      "loss": 0.6036,
      "step": 6175
    },
    {
      "epoch": 8.6016713091922,
      "grad_norm": 2.568204641342163,
      "learning_rate": 9.988629903354178e-06,
      "loss": 0.8099,
      "step": 6176
    },
    {
      "epoch": 8.603064066852367,
      "grad_norm": 2.748413324356079,
      "learning_rate": 9.978681068789083e-06,
      "loss": 0.9198,
      "step": 6177
    },
    {
      "epoch": 8.604456824512535,
      "grad_norm": 2.369399309158325,
      "learning_rate": 9.96873223422399e-06,
      "loss": 0.6536,
      "step": 6178
    },
    {
      "epoch": 8.605849582172702,
      "grad_norm": 1.9755353927612305,
      "learning_rate": 9.958783399658895e-06,
      "loss": 0.6128,
      "step": 6179
    },
    {
      "epoch": 8.607242339832869,
      "grad_norm": 2.522010087966919,
      "learning_rate": 9.948834565093802e-06,
      "loss": 0.501,
      "step": 6180
    },
    {
      "epoch": 8.608635097493035,
      "grad_norm": 2.2480692863464355,
      "learning_rate": 9.938885730528708e-06,
      "loss": 0.4739,
      "step": 6181
    },
    {
      "epoch": 8.610027855153204,
      "grad_norm": 2.5440099239349365,
      "learning_rate": 9.928936895963616e-06,
      "loss": 0.6548,
      "step": 6182
    },
    {
      "epoch": 8.61142061281337,
      "grad_norm": 2.194700002670288,
      "learning_rate": 9.918988061398521e-06,
      "loss": 0.7249,
      "step": 6183
    },
    {
      "epoch": 8.612813370473537,
      "grad_norm": 2.014458179473877,
      "learning_rate": 9.909039226833427e-06,
      "loss": 0.5552,
      "step": 6184
    },
    {
      "epoch": 8.614206128133706,
      "grad_norm": 2.3420283794403076,
      "learning_rate": 9.899090392268335e-06,
      "loss": 0.7262,
      "step": 6185
    },
    {
      "epoch": 8.615598885793872,
      "grad_norm": 2.8859808444976807,
      "learning_rate": 9.88914155770324e-06,
      "loss": 0.998,
      "step": 6186
    },
    {
      "epoch": 8.616991643454039,
      "grad_norm": 2.879088878631592,
      "learning_rate": 9.879192723138146e-06,
      "loss": 0.8435,
      "step": 6187
    },
    {
      "epoch": 8.618384401114206,
      "grad_norm": 2.1142868995666504,
      "learning_rate": 9.869243888573052e-06,
      "loss": 0.5091,
      "step": 6188
    },
    {
      "epoch": 8.619777158774372,
      "grad_norm": 2.2945990562438965,
      "learning_rate": 9.859295054007958e-06,
      "loss": 0.682,
      "step": 6189
    },
    {
      "epoch": 8.62116991643454,
      "grad_norm": 2.226578712463379,
      "learning_rate": 9.849346219442863e-06,
      "loss": 0.6142,
      "step": 6190
    },
    {
      "epoch": 8.622562674094707,
      "grad_norm": 2.5178959369659424,
      "learning_rate": 9.839397384877771e-06,
      "loss": 0.6461,
      "step": 6191
    },
    {
      "epoch": 8.623955431754874,
      "grad_norm": 2.4162328243255615,
      "learning_rate": 9.829448550312677e-06,
      "loss": 0.6262,
      "step": 6192
    },
    {
      "epoch": 8.625348189415043,
      "grad_norm": 2.248405933380127,
      "learning_rate": 9.819499715747582e-06,
      "loss": 0.639,
      "step": 6193
    },
    {
      "epoch": 8.62674094707521,
      "grad_norm": 2.142740249633789,
      "learning_rate": 9.809550881182488e-06,
      "loss": 0.5656,
      "step": 6194
    },
    {
      "epoch": 8.628133704735376,
      "grad_norm": 11.2222318649292,
      "learning_rate": 9.799602046617394e-06,
      "loss": 0.6877,
      "step": 6195
    },
    {
      "epoch": 8.629526462395543,
      "grad_norm": 2.5588018894195557,
      "learning_rate": 9.789653212052303e-06,
      "loss": 0.7678,
      "step": 6196
    },
    {
      "epoch": 8.630919220055711,
      "grad_norm": 2.0188772678375244,
      "learning_rate": 9.779704377487209e-06,
      "loss": 0.5143,
      "step": 6197
    },
    {
      "epoch": 8.632311977715878,
      "grad_norm": 2.4895517826080322,
      "learning_rate": 9.769755542922115e-06,
      "loss": 0.741,
      "step": 6198
    },
    {
      "epoch": 8.633704735376044,
      "grad_norm": 2.028364658355713,
      "learning_rate": 9.75980670835702e-06,
      "loss": 0.5493,
      "step": 6199
    },
    {
      "epoch": 8.635097493036211,
      "grad_norm": 2.5568647384643555,
      "learning_rate": 9.749857873791926e-06,
      "loss": 0.6039,
      "step": 6200
    },
    {
      "epoch": 8.63649025069638,
      "grad_norm": 2.497607707977295,
      "learning_rate": 9.739909039226832e-06,
      "loss": 0.6868,
      "step": 6201
    },
    {
      "epoch": 8.637883008356546,
      "grad_norm": 2.6691572666168213,
      "learning_rate": 9.72996020466174e-06,
      "loss": 0.8506,
      "step": 6202
    },
    {
      "epoch": 8.639275766016713,
      "grad_norm": 2.588259220123291,
      "learning_rate": 9.720011370096645e-06,
      "loss": 0.6523,
      "step": 6203
    },
    {
      "epoch": 8.64066852367688,
      "grad_norm": 2.6926379203796387,
      "learning_rate": 9.710062535531551e-06,
      "loss": 0.7702,
      "step": 6204
    },
    {
      "epoch": 8.642061281337048,
      "grad_norm": 2.3415701389312744,
      "learning_rate": 9.700113700966457e-06,
      "loss": 0.7117,
      "step": 6205
    },
    {
      "epoch": 8.643454038997215,
      "grad_norm": 2.263221263885498,
      "learning_rate": 9.690164866401363e-06,
      "loss": 0.6067,
      "step": 6206
    },
    {
      "epoch": 8.644846796657381,
      "grad_norm": 3.0019490718841553,
      "learning_rate": 9.68021603183627e-06,
      "loss": 0.6614,
      "step": 6207
    },
    {
      "epoch": 8.64623955431755,
      "grad_norm": 2.1951980590820312,
      "learning_rate": 9.670267197271176e-06,
      "loss": 0.5938,
      "step": 6208
    },
    {
      "epoch": 8.647632311977716,
      "grad_norm": 2.2621564865112305,
      "learning_rate": 9.660318362706082e-06,
      "loss": 0.6477,
      "step": 6209
    },
    {
      "epoch": 8.649025069637883,
      "grad_norm": 2.6822450160980225,
      "learning_rate": 9.650369528140989e-06,
      "loss": 0.7119,
      "step": 6210
    },
    {
      "epoch": 8.65041782729805,
      "grad_norm": 2.1923351287841797,
      "learning_rate": 9.640420693575895e-06,
      "loss": 0.6461,
      "step": 6211
    },
    {
      "epoch": 8.651810584958218,
      "grad_norm": 2.5055742263793945,
      "learning_rate": 9.6304718590108e-06,
      "loss": 0.7208,
      "step": 6212
    },
    {
      "epoch": 8.653203342618385,
      "grad_norm": 2.129638433456421,
      "learning_rate": 9.620523024445708e-06,
      "loss": 0.6233,
      "step": 6213
    },
    {
      "epoch": 8.654596100278551,
      "grad_norm": 2.059680461883545,
      "learning_rate": 9.610574189880614e-06,
      "loss": 0.5662,
      "step": 6214
    },
    {
      "epoch": 8.655988857938718,
      "grad_norm": 2.1819725036621094,
      "learning_rate": 9.60062535531552e-06,
      "loss": 0.5014,
      "step": 6215
    },
    {
      "epoch": 8.657381615598887,
      "grad_norm": 2.555784225463867,
      "learning_rate": 9.590676520750425e-06,
      "loss": 0.6371,
      "step": 6216
    },
    {
      "epoch": 8.658774373259053,
      "grad_norm": 2.4387550354003906,
      "learning_rate": 9.580727686185331e-06,
      "loss": 0.6615,
      "step": 6217
    },
    {
      "epoch": 8.66016713091922,
      "grad_norm": 1.9884040355682373,
      "learning_rate": 9.570778851620239e-06,
      "loss": 0.5488,
      "step": 6218
    },
    {
      "epoch": 8.661559888579387,
      "grad_norm": 2.188859224319458,
      "learning_rate": 9.560830017055144e-06,
      "loss": 0.5665,
      "step": 6219
    },
    {
      "epoch": 8.662952646239555,
      "grad_norm": 2.0982136726379395,
      "learning_rate": 9.55088118249005e-06,
      "loss": 0.6464,
      "step": 6220
    },
    {
      "epoch": 8.664345403899722,
      "grad_norm": 1.977082371711731,
      "learning_rate": 9.540932347924956e-06,
      "loss": 0.4598,
      "step": 6221
    },
    {
      "epoch": 8.665738161559888,
      "grad_norm": 2.4717345237731934,
      "learning_rate": 9.530983513359862e-06,
      "loss": 0.6632,
      "step": 6222
    },
    {
      "epoch": 8.667130919220055,
      "grad_norm": 2.65958571434021,
      "learning_rate": 9.521034678794767e-06,
      "loss": 0.7021,
      "step": 6223
    },
    {
      "epoch": 8.668523676880223,
      "grad_norm": 1.9559311866760254,
      "learning_rate": 9.511085844229677e-06,
      "loss": 0.4133,
      "step": 6224
    },
    {
      "epoch": 8.66991643454039,
      "grad_norm": 2.413815975189209,
      "learning_rate": 9.501137009664582e-06,
      "loss": 0.6967,
      "step": 6225
    },
    {
      "epoch": 8.671309192200557,
      "grad_norm": 2.0799384117126465,
      "learning_rate": 9.491188175099488e-06,
      "loss": 0.6857,
      "step": 6226
    },
    {
      "epoch": 8.672701949860723,
      "grad_norm": 2.4632363319396973,
      "learning_rate": 9.481239340534394e-06,
      "loss": 0.69,
      "step": 6227
    },
    {
      "epoch": 8.674094707520892,
      "grad_norm": 2.4808237552642822,
      "learning_rate": 9.4712905059693e-06,
      "loss": 0.6757,
      "step": 6228
    },
    {
      "epoch": 8.675487465181059,
      "grad_norm": 2.4378230571746826,
      "learning_rate": 9.461341671404207e-06,
      "loss": 0.6528,
      "step": 6229
    },
    {
      "epoch": 8.676880222841225,
      "grad_norm": 3.099447250366211,
      "learning_rate": 9.451392836839113e-06,
      "loss": 0.6603,
      "step": 6230
    },
    {
      "epoch": 8.678272980501394,
      "grad_norm": 2.001962900161743,
      "learning_rate": 9.441444002274019e-06,
      "loss": 0.4916,
      "step": 6231
    },
    {
      "epoch": 8.67966573816156,
      "grad_norm": 2.306098222732544,
      "learning_rate": 9.431495167708924e-06,
      "loss": 0.6936,
      "step": 6232
    },
    {
      "epoch": 8.681058495821727,
      "grad_norm": 2.3302414417266846,
      "learning_rate": 9.42154633314383e-06,
      "loss": 0.6671,
      "step": 6233
    },
    {
      "epoch": 8.682451253481894,
      "grad_norm": 2.6579461097717285,
      "learning_rate": 9.411597498578736e-06,
      "loss": 0.5523,
      "step": 6234
    },
    {
      "epoch": 8.683844011142062,
      "grad_norm": 2.4258522987365723,
      "learning_rate": 9.401648664013643e-06,
      "loss": 0.6678,
      "step": 6235
    },
    {
      "epoch": 8.685236768802229,
      "grad_norm": 2.312063694000244,
      "learning_rate": 9.39169982944855e-06,
      "loss": 0.6703,
      "step": 6236
    },
    {
      "epoch": 8.686629526462395,
      "grad_norm": 2.2572250366210938,
      "learning_rate": 9.381750994883455e-06,
      "loss": 0.6261,
      "step": 6237
    },
    {
      "epoch": 8.688022284122562,
      "grad_norm": 2.1370370388031006,
      "learning_rate": 9.37180216031836e-06,
      "loss": 0.6779,
      "step": 6238
    },
    {
      "epoch": 8.68941504178273,
      "grad_norm": 2.361268997192383,
      "learning_rate": 9.361853325753268e-06,
      "loss": 0.6468,
      "step": 6239
    },
    {
      "epoch": 8.690807799442897,
      "grad_norm": 2.149677038192749,
      "learning_rate": 9.351904491188176e-06,
      "loss": 0.601,
      "step": 6240
    },
    {
      "epoch": 8.692200557103064,
      "grad_norm": 1.8978216648101807,
      "learning_rate": 9.341955656623081e-06,
      "loss": 0.4258,
      "step": 6241
    },
    {
      "epoch": 8.69359331476323,
      "grad_norm": 2.113858461380005,
      "learning_rate": 9.332006822057987e-06,
      "loss": 0.5342,
      "step": 6242
    },
    {
      "epoch": 8.694986072423399,
      "grad_norm": 2.1423451900482178,
      "learning_rate": 9.322057987492893e-06,
      "loss": 0.5411,
      "step": 6243
    },
    {
      "epoch": 8.696378830083566,
      "grad_norm": 2.2512307167053223,
      "learning_rate": 9.312109152927799e-06,
      "loss": 0.6263,
      "step": 6244
    },
    {
      "epoch": 8.697771587743732,
      "grad_norm": 2.957404375076294,
      "learning_rate": 9.302160318362704e-06,
      "loss": 0.7694,
      "step": 6245
    },
    {
      "epoch": 8.699164345403899,
      "grad_norm": 2.7469513416290283,
      "learning_rate": 9.292211483797612e-06,
      "loss": 0.6972,
      "step": 6246
    },
    {
      "epoch": 8.700557103064067,
      "grad_norm": 2.0505714416503906,
      "learning_rate": 9.282262649232518e-06,
      "loss": 0.5273,
      "step": 6247
    },
    {
      "epoch": 8.701949860724234,
      "grad_norm": 2.007658004760742,
      "learning_rate": 9.272313814667423e-06,
      "loss": 0.5611,
      "step": 6248
    },
    {
      "epoch": 8.7033426183844,
      "grad_norm": 1.9458181858062744,
      "learning_rate": 9.26236498010233e-06,
      "loss": 0.4343,
      "step": 6249
    },
    {
      "epoch": 8.704735376044567,
      "grad_norm": 2.614433765411377,
      "learning_rate": 9.252416145537235e-06,
      "loss": 0.879,
      "step": 6250
    },
    {
      "epoch": 8.706128133704736,
      "grad_norm": 2.5961923599243164,
      "learning_rate": 9.242467310972144e-06,
      "loss": 0.7532,
      "step": 6251
    },
    {
      "epoch": 8.707520891364902,
      "grad_norm": 1.9644964933395386,
      "learning_rate": 9.23251847640705e-06,
      "loss": 0.4588,
      "step": 6252
    },
    {
      "epoch": 8.70891364902507,
      "grad_norm": 2.656649351119995,
      "learning_rate": 9.222569641841956e-06,
      "loss": 0.9547,
      "step": 6253
    },
    {
      "epoch": 8.710306406685238,
      "grad_norm": 2.868112802505493,
      "learning_rate": 9.212620807276861e-06,
      "loss": 0.7435,
      "step": 6254
    },
    {
      "epoch": 8.711699164345404,
      "grad_norm": 2.573540687561035,
      "learning_rate": 9.202671972711767e-06,
      "loss": 0.7418,
      "step": 6255
    },
    {
      "epoch": 8.713091922005571,
      "grad_norm": 2.7416908740997314,
      "learning_rate": 9.192723138146673e-06,
      "loss": 0.6664,
      "step": 6256
    },
    {
      "epoch": 8.714484679665738,
      "grad_norm": 2.4586341381073,
      "learning_rate": 9.18277430358158e-06,
      "loss": 0.701,
      "step": 6257
    },
    {
      "epoch": 8.715877437325906,
      "grad_norm": 2.4834976196289062,
      "learning_rate": 9.172825469016486e-06,
      "loss": 0.7483,
      "step": 6258
    },
    {
      "epoch": 8.717270194986073,
      "grad_norm": 2.4113335609436035,
      "learning_rate": 9.162876634451392e-06,
      "loss": 0.5699,
      "step": 6259
    },
    {
      "epoch": 8.71866295264624,
      "grad_norm": 2.408886194229126,
      "learning_rate": 9.152927799886298e-06,
      "loss": 0.5494,
      "step": 6260
    },
    {
      "epoch": 8.720055710306406,
      "grad_norm": 2.3952715396881104,
      "learning_rate": 9.142978965321204e-06,
      "loss": 0.7711,
      "step": 6261
    },
    {
      "epoch": 8.721448467966574,
      "grad_norm": 2.3807334899902344,
      "learning_rate": 9.133030130756111e-06,
      "loss": 0.4661,
      "step": 6262
    },
    {
      "epoch": 8.722841225626741,
      "grad_norm": 3.1220288276672363,
      "learning_rate": 9.123081296191017e-06,
      "loss": 0.7197,
      "step": 6263
    },
    {
      "epoch": 8.724233983286908,
      "grad_norm": 2.0899627208709717,
      "learning_rate": 9.113132461625923e-06,
      "loss": 0.545,
      "step": 6264
    },
    {
      "epoch": 8.725626740947074,
      "grad_norm": 2.251028060913086,
      "learning_rate": 9.103183627060828e-06,
      "loss": 0.6706,
      "step": 6265
    },
    {
      "epoch": 8.727019498607243,
      "grad_norm": 2.0511791706085205,
      "learning_rate": 9.093234792495734e-06,
      "loss": 0.6125,
      "step": 6266
    },
    {
      "epoch": 8.72841225626741,
      "grad_norm": 1.962691307067871,
      "learning_rate": 9.083285957930642e-06,
      "loss": 0.455,
      "step": 6267
    },
    {
      "epoch": 8.729805013927576,
      "grad_norm": 2.496546983718872,
      "learning_rate": 9.073337123365549e-06,
      "loss": 0.8317,
      "step": 6268
    },
    {
      "epoch": 8.731197771587743,
      "grad_norm": 2.121131181716919,
      "learning_rate": 9.063388288800455e-06,
      "loss": 0.5414,
      "step": 6269
    },
    {
      "epoch": 8.732590529247911,
      "grad_norm": 2.086209774017334,
      "learning_rate": 9.05343945423536e-06,
      "loss": 0.532,
      "step": 6270
    },
    {
      "epoch": 8.733983286908078,
      "grad_norm": 1.8627957105636597,
      "learning_rate": 9.043490619670266e-06,
      "loss": 0.4795,
      "step": 6271
    },
    {
      "epoch": 8.735376044568245,
      "grad_norm": 3.083702564239502,
      "learning_rate": 9.033541785105172e-06,
      "loss": 0.6021,
      "step": 6272
    },
    {
      "epoch": 8.736768802228411,
      "grad_norm": 2.2250123023986816,
      "learning_rate": 9.02359295054008e-06,
      "loss": 0.6004,
      "step": 6273
    },
    {
      "epoch": 8.73816155988858,
      "grad_norm": 2.6575896739959717,
      "learning_rate": 9.013644115974985e-06,
      "loss": 0.706,
      "step": 6274
    },
    {
      "epoch": 8.739554317548746,
      "grad_norm": 2.206486463546753,
      "learning_rate": 9.003695281409891e-06,
      "loss": 0.5474,
      "step": 6275
    },
    {
      "epoch": 8.740947075208913,
      "grad_norm": 2.0687928199768066,
      "learning_rate": 8.993746446844797e-06,
      "loss": 0.4435,
      "step": 6276
    },
    {
      "epoch": 8.742339832869082,
      "grad_norm": 2.3520524501800537,
      "learning_rate": 8.983797612279703e-06,
      "loss": 0.5865,
      "step": 6277
    },
    {
      "epoch": 8.743732590529248,
      "grad_norm": 2.687025547027588,
      "learning_rate": 8.973848777714608e-06,
      "loss": 0.8384,
      "step": 6278
    },
    {
      "epoch": 8.745125348189415,
      "grad_norm": 2.443650484085083,
      "learning_rate": 8.963899943149518e-06,
      "loss": 0.7113,
      "step": 6279
    },
    {
      "epoch": 8.746518105849582,
      "grad_norm": 4.449234485626221,
      "learning_rate": 8.953951108584423e-06,
      "loss": 0.7642,
      "step": 6280
    },
    {
      "epoch": 8.74791086350975,
      "grad_norm": 2.805539131164551,
      "learning_rate": 8.944002274019329e-06,
      "loss": 0.5477,
      "step": 6281
    },
    {
      "epoch": 8.749303621169917,
      "grad_norm": 2.4428164958953857,
      "learning_rate": 8.934053439454235e-06,
      "loss": 0.5837,
      "step": 6282
    },
    {
      "epoch": 8.750696378830083,
      "grad_norm": 2.272738456726074,
      "learning_rate": 8.92410460488914e-06,
      "loss": 0.5737,
      "step": 6283
    },
    {
      "epoch": 8.75208913649025,
      "grad_norm": 2.465785503387451,
      "learning_rate": 8.914155770324048e-06,
      "loss": 0.6352,
      "step": 6284
    },
    {
      "epoch": 8.753481894150418,
      "grad_norm": 2.5750913619995117,
      "learning_rate": 8.904206935758954e-06,
      "loss": 0.7051,
      "step": 6285
    },
    {
      "epoch": 8.754874651810585,
      "grad_norm": 2.47235107421875,
      "learning_rate": 8.89425810119386e-06,
      "loss": 0.6843,
      "step": 6286
    },
    {
      "epoch": 8.756267409470752,
      "grad_norm": 2.425645112991333,
      "learning_rate": 8.884309266628765e-06,
      "loss": 0.563,
      "step": 6287
    },
    {
      "epoch": 8.757660167130918,
      "grad_norm": 2.2004497051239014,
      "learning_rate": 8.874360432063671e-06,
      "loss": 0.6022,
      "step": 6288
    },
    {
      "epoch": 8.759052924791087,
      "grad_norm": 2.0673093795776367,
      "learning_rate": 8.864411597498577e-06,
      "loss": 0.5229,
      "step": 6289
    },
    {
      "epoch": 8.760445682451254,
      "grad_norm": 2.6072375774383545,
      "learning_rate": 8.854462762933484e-06,
      "loss": 0.8253,
      "step": 6290
    },
    {
      "epoch": 8.76183844011142,
      "grad_norm": 2.562784433364868,
      "learning_rate": 8.84451392836839e-06,
      "loss": 0.6664,
      "step": 6291
    },
    {
      "epoch": 8.763231197771589,
      "grad_norm": 2.4072632789611816,
      "learning_rate": 8.834565093803296e-06,
      "loss": 0.6328,
      "step": 6292
    },
    {
      "epoch": 8.764623955431755,
      "grad_norm": 2.034414529800415,
      "learning_rate": 8.824616259238202e-06,
      "loss": 0.5251,
      "step": 6293
    },
    {
      "epoch": 8.766016713091922,
      "grad_norm": 2.5559253692626953,
      "learning_rate": 8.814667424673107e-06,
      "loss": 0.7352,
      "step": 6294
    },
    {
      "epoch": 8.767409470752089,
      "grad_norm": 9.379827499389648,
      "learning_rate": 8.804718590108017e-06,
      "loss": 0.6291,
      "step": 6295
    },
    {
      "epoch": 8.768802228412257,
      "grad_norm": 2.123366117477417,
      "learning_rate": 8.794769755542922e-06,
      "loss": 0.5206,
      "step": 6296
    },
    {
      "epoch": 8.770194986072424,
      "grad_norm": 2.862316608428955,
      "learning_rate": 8.784820920977828e-06,
      "loss": 0.6797,
      "step": 6297
    },
    {
      "epoch": 8.77158774373259,
      "grad_norm": 2.3365907669067383,
      "learning_rate": 8.774872086412734e-06,
      "loss": 0.5556,
      "step": 6298
    },
    {
      "epoch": 8.772980501392757,
      "grad_norm": 2.0204622745513916,
      "learning_rate": 8.76492325184764e-06,
      "loss": 0.5291,
      "step": 6299
    },
    {
      "epoch": 8.774373259052926,
      "grad_norm": 2.288923978805542,
      "learning_rate": 8.754974417282546e-06,
      "loss": 0.6139,
      "step": 6300
    },
    {
      "epoch": 8.775766016713092,
      "grad_norm": 2.1300530433654785,
      "learning_rate": 8.745025582717453e-06,
      "loss": 0.6121,
      "step": 6301
    },
    {
      "epoch": 8.777158774373259,
      "grad_norm": 2.426448106765747,
      "learning_rate": 8.735076748152359e-06,
      "loss": 0.8317,
      "step": 6302
    },
    {
      "epoch": 8.778551532033426,
      "grad_norm": 1.941066026687622,
      "learning_rate": 8.725127913587265e-06,
      "loss": 0.5266,
      "step": 6303
    },
    {
      "epoch": 8.779944289693594,
      "grad_norm": 2.1124699115753174,
      "learning_rate": 8.71517907902217e-06,
      "loss": 0.5945,
      "step": 6304
    },
    {
      "epoch": 8.78133704735376,
      "grad_norm": 2.294548749923706,
      "learning_rate": 8.705230244457078e-06,
      "loss": 0.5947,
      "step": 6305
    },
    {
      "epoch": 8.782729805013927,
      "grad_norm": 2.6320042610168457,
      "learning_rate": 8.695281409891984e-06,
      "loss": 0.6269,
      "step": 6306
    },
    {
      "epoch": 8.784122562674094,
      "grad_norm": 2.093122720718384,
      "learning_rate": 8.68533257532689e-06,
      "loss": 0.3532,
      "step": 6307
    },
    {
      "epoch": 8.785515320334262,
      "grad_norm": 2.3614604473114014,
      "learning_rate": 8.675383740761797e-06,
      "loss": 0.6072,
      "step": 6308
    },
    {
      "epoch": 8.786908077994429,
      "grad_norm": 2.1654114723205566,
      "learning_rate": 8.665434906196703e-06,
      "loss": 0.4391,
      "step": 6309
    },
    {
      "epoch": 8.788300835654596,
      "grad_norm": 2.2236990928649902,
      "learning_rate": 8.655486071631608e-06,
      "loss": 0.5794,
      "step": 6310
    },
    {
      "epoch": 8.789693593314762,
      "grad_norm": 2.972788095474243,
      "learning_rate": 8.645537237066514e-06,
      "loss": 0.6736,
      "step": 6311
    },
    {
      "epoch": 8.79108635097493,
      "grad_norm": 1.9443541765213013,
      "learning_rate": 8.63558840250142e-06,
      "loss": 0.5474,
      "step": 6312
    },
    {
      "epoch": 8.792479108635098,
      "grad_norm": 3.0000555515289307,
      "learning_rate": 8.625639567936327e-06,
      "loss": 0.8171,
      "step": 6313
    },
    {
      "epoch": 8.793871866295264,
      "grad_norm": 2.4785919189453125,
      "learning_rate": 8.615690733371233e-06,
      "loss": 0.6326,
      "step": 6314
    },
    {
      "epoch": 8.795264623955433,
      "grad_norm": 2.1436679363250732,
      "learning_rate": 8.605741898806139e-06,
      "loss": 0.6217,
      "step": 6315
    },
    {
      "epoch": 8.7966573816156,
      "grad_norm": 2.8069963455200195,
      "learning_rate": 8.595793064241046e-06,
      "loss": 0.7795,
      "step": 6316
    },
    {
      "epoch": 8.798050139275766,
      "grad_norm": 2.0932371616363525,
      "learning_rate": 8.585844229675952e-06,
      "loss": 0.6109,
      "step": 6317
    },
    {
      "epoch": 8.799442896935933,
      "grad_norm": 1.9473953247070312,
      "learning_rate": 8.575895395110858e-06,
      "loss": 0.5751,
      "step": 6318
    },
    {
      "epoch": 8.800835654596101,
      "grad_norm": 2.075078248977661,
      "learning_rate": 8.565946560545764e-06,
      "loss": 0.6307,
      "step": 6319
    },
    {
      "epoch": 8.802228412256268,
      "grad_norm": 2.0488667488098145,
      "learning_rate": 8.55599772598067e-06,
      "loss": 0.4987,
      "step": 6320
    },
    {
      "epoch": 8.803621169916434,
      "grad_norm": 2.2753357887268066,
      "learning_rate": 8.546048891415575e-06,
      "loss": 0.6262,
      "step": 6321
    },
    {
      "epoch": 8.805013927576601,
      "grad_norm": 5.119298934936523,
      "learning_rate": 8.536100056850483e-06,
      "loss": 0.8127,
      "step": 6322
    },
    {
      "epoch": 8.80640668523677,
      "grad_norm": 2.296943187713623,
      "learning_rate": 8.526151222285388e-06,
      "loss": 0.6588,
      "step": 6323
    },
    {
      "epoch": 8.807799442896936,
      "grad_norm": 2.4178757667541504,
      "learning_rate": 8.516202387720296e-06,
      "loss": 0.4879,
      "step": 6324
    },
    {
      "epoch": 8.809192200557103,
      "grad_norm": 2.0936036109924316,
      "learning_rate": 8.506253553155202e-06,
      "loss": 0.55,
      "step": 6325
    },
    {
      "epoch": 8.81058495821727,
      "grad_norm": 1.953722596168518,
      "learning_rate": 8.496304718590107e-06,
      "loss": 0.4707,
      "step": 6326
    },
    {
      "epoch": 8.811977715877438,
      "grad_norm": 2.4794445037841797,
      "learning_rate": 8.486355884025013e-06,
      "loss": 0.7046,
      "step": 6327
    },
    {
      "epoch": 8.813370473537605,
      "grad_norm": 2.4205312728881836,
      "learning_rate": 8.476407049459919e-06,
      "loss": 0.7026,
      "step": 6328
    },
    {
      "epoch": 8.814763231197771,
      "grad_norm": 2.315272569656372,
      "learning_rate": 8.466458214894826e-06,
      "loss": 0.6319,
      "step": 6329
    },
    {
      "epoch": 8.816155988857938,
      "grad_norm": 2.3596596717834473,
      "learning_rate": 8.456509380329732e-06,
      "loss": 0.6448,
      "step": 6330
    },
    {
      "epoch": 8.817548746518106,
      "grad_norm": 2.025387763977051,
      "learning_rate": 8.446560545764638e-06,
      "loss": 0.5503,
      "step": 6331
    },
    {
      "epoch": 8.818941504178273,
      "grad_norm": 1.8706268072128296,
      "learning_rate": 8.436611711199544e-06,
      "loss": 0.5552,
      "step": 6332
    },
    {
      "epoch": 8.82033426183844,
      "grad_norm": 2.4644529819488525,
      "learning_rate": 8.426662876634451e-06,
      "loss": 0.6394,
      "step": 6333
    },
    {
      "epoch": 8.821727019498606,
      "grad_norm": 2.3819756507873535,
      "learning_rate": 8.416714042069357e-06,
      "loss": 0.7558,
      "step": 6334
    },
    {
      "epoch": 8.823119777158775,
      "grad_norm": 2.2065489292144775,
      "learning_rate": 8.406765207504264e-06,
      "loss": 0.4688,
      "step": 6335
    },
    {
      "epoch": 8.824512534818941,
      "grad_norm": 2.437697172164917,
      "learning_rate": 8.39681637293917e-06,
      "loss": 0.6055,
      "step": 6336
    },
    {
      "epoch": 8.825905292479108,
      "grad_norm": 1.6681405305862427,
      "learning_rate": 8.386867538374076e-06,
      "loss": 0.4151,
      "step": 6337
    },
    {
      "epoch": 8.827298050139277,
      "grad_norm": 2.163010597229004,
      "learning_rate": 8.376918703808982e-06,
      "loss": 0.5549,
      "step": 6338
    },
    {
      "epoch": 8.828690807799443,
      "grad_norm": 3.0924649238586426,
      "learning_rate": 8.366969869243887e-06,
      "loss": 0.5241,
      "step": 6339
    },
    {
      "epoch": 8.83008356545961,
      "grad_norm": 2.328601598739624,
      "learning_rate": 8.357021034678793e-06,
      "loss": 0.5889,
      "step": 6340
    },
    {
      "epoch": 8.831476323119777,
      "grad_norm": 2.064568281173706,
      "learning_rate": 8.3470722001137e-06,
      "loss": 0.5624,
      "step": 6341
    },
    {
      "epoch": 8.832869080779945,
      "grad_norm": 2.0899322032928467,
      "learning_rate": 8.337123365548606e-06,
      "loss": 0.5557,
      "step": 6342
    },
    {
      "epoch": 8.834261838440112,
      "grad_norm": 2.094360589981079,
      "learning_rate": 8.327174530983512e-06,
      "loss": 0.5485,
      "step": 6343
    },
    {
      "epoch": 8.835654596100278,
      "grad_norm": 2.1241538524627686,
      "learning_rate": 8.31722569641842e-06,
      "loss": 0.6834,
      "step": 6344
    },
    {
      "epoch": 8.837047353760445,
      "grad_norm": 2.5731396675109863,
      "learning_rate": 8.307276861853325e-06,
      "loss": 0.7289,
      "step": 6345
    },
    {
      "epoch": 8.838440111420613,
      "grad_norm": 2.2893941402435303,
      "learning_rate": 8.297328027288231e-06,
      "loss": 0.6117,
      "step": 6346
    },
    {
      "epoch": 8.83983286908078,
      "grad_norm": 2.296159505844116,
      "learning_rate": 8.287379192723137e-06,
      "loss": 0.5925,
      "step": 6347
    },
    {
      "epoch": 8.841225626740947,
      "grad_norm": 2.214094877243042,
      "learning_rate": 8.277430358158043e-06,
      "loss": 0.6989,
      "step": 6348
    },
    {
      "epoch": 8.842618384401113,
      "grad_norm": 2.2275171279907227,
      "learning_rate": 8.26748152359295e-06,
      "loss": 0.5738,
      "step": 6349
    },
    {
      "epoch": 8.844011142061282,
      "grad_norm": 2.639855146408081,
      "learning_rate": 8.257532689027856e-06,
      "loss": 0.6824,
      "step": 6350
    },
    {
      "epoch": 8.845403899721449,
      "grad_norm": 2.3448219299316406,
      "learning_rate": 8.247583854462762e-06,
      "loss": 0.6128,
      "step": 6351
    },
    {
      "epoch": 8.846796657381615,
      "grad_norm": 2.3506157398223877,
      "learning_rate": 8.23763501989767e-06,
      "loss": 0.7248,
      "step": 6352
    },
    {
      "epoch": 8.848189415041782,
      "grad_norm": 2.2729196548461914,
      "learning_rate": 8.227686185332575e-06,
      "loss": 0.6088,
      "step": 6353
    },
    {
      "epoch": 8.84958217270195,
      "grad_norm": 2.17976975440979,
      "learning_rate": 8.21773735076748e-06,
      "loss": 0.5869,
      "step": 6354
    },
    {
      "epoch": 8.850974930362117,
      "grad_norm": 2.1733644008636475,
      "learning_rate": 8.207788516202387e-06,
      "loss": 0.6588,
      "step": 6355
    },
    {
      "epoch": 8.852367688022284,
      "grad_norm": 2.554656982421875,
      "learning_rate": 8.197839681637292e-06,
      "loss": 0.6929,
      "step": 6356
    },
    {
      "epoch": 8.85376044568245,
      "grad_norm": 1.9149433374404907,
      "learning_rate": 8.1878908470722e-06,
      "loss": 0.5418,
      "step": 6357
    },
    {
      "epoch": 8.855153203342619,
      "grad_norm": 2.480768918991089,
      "learning_rate": 8.177942012507106e-06,
      "loss": 0.6574,
      "step": 6358
    },
    {
      "epoch": 8.856545961002785,
      "grad_norm": 2.1273303031921387,
      "learning_rate": 8.167993177942011e-06,
      "loss": 0.5279,
      "step": 6359
    },
    {
      "epoch": 8.857938718662952,
      "grad_norm": 2.612752914428711,
      "learning_rate": 8.158044343376919e-06,
      "loss": 0.8142,
      "step": 6360
    },
    {
      "epoch": 8.85933147632312,
      "grad_norm": 2.2889060974121094,
      "learning_rate": 8.148095508811825e-06,
      "loss": 0.6762,
      "step": 6361
    },
    {
      "epoch": 8.860724233983287,
      "grad_norm": 2.2247323989868164,
      "learning_rate": 8.13814667424673e-06,
      "loss": 0.5483,
      "step": 6362
    },
    {
      "epoch": 8.862116991643454,
      "grad_norm": 1.9490143060684204,
      "learning_rate": 8.128197839681638e-06,
      "loss": 0.5258,
      "step": 6363
    },
    {
      "epoch": 8.86350974930362,
      "grad_norm": 1.9298173189163208,
      "learning_rate": 8.118249005116544e-06,
      "loss": 0.4203,
      "step": 6364
    },
    {
      "epoch": 8.864902506963789,
      "grad_norm": 2.6802420616149902,
      "learning_rate": 8.10830017055145e-06,
      "loss": 0.7453,
      "step": 6365
    },
    {
      "epoch": 8.866295264623956,
      "grad_norm": 2.622260093688965,
      "learning_rate": 8.098351335986355e-06,
      "loss": 0.7084,
      "step": 6366
    },
    {
      "epoch": 8.867688022284122,
      "grad_norm": 2.229451894760132,
      "learning_rate": 8.08840250142126e-06,
      "loss": 0.6352,
      "step": 6367
    },
    {
      "epoch": 8.869080779944289,
      "grad_norm": 2.427901268005371,
      "learning_rate": 8.078453666856167e-06,
      "loss": 0.609,
      "step": 6368
    },
    {
      "epoch": 8.870473537604457,
      "grad_norm": 2.5861806869506836,
      "learning_rate": 8.068504832291074e-06,
      "loss": 0.8659,
      "step": 6369
    },
    {
      "epoch": 8.871866295264624,
      "grad_norm": 2.042874336242676,
      "learning_rate": 8.05855599772598e-06,
      "loss": 0.5616,
      "step": 6370
    },
    {
      "epoch": 8.87325905292479,
      "grad_norm": 2.3502142429351807,
      "learning_rate": 8.048607163160887e-06,
      "loss": 0.6585,
      "step": 6371
    },
    {
      "epoch": 8.874651810584957,
      "grad_norm": 2.520488977432251,
      "learning_rate": 8.038658328595793e-06,
      "loss": 0.7801,
      "step": 6372
    },
    {
      "epoch": 8.876044568245126,
      "grad_norm": 2.4678702354431152,
      "learning_rate": 8.028709494030699e-06,
      "loss": 0.6454,
      "step": 6373
    },
    {
      "epoch": 8.877437325905293,
      "grad_norm": 2.1264488697052,
      "learning_rate": 8.018760659465605e-06,
      "loss": 0.6294,
      "step": 6374
    },
    {
      "epoch": 8.87883008356546,
      "grad_norm": 2.327361583709717,
      "learning_rate": 8.00881182490051e-06,
      "loss": 0.7171,
      "step": 6375
    },
    {
      "epoch": 8.880222841225628,
      "grad_norm": 2.119594097137451,
      "learning_rate": 7.998862990335416e-06,
      "loss": 0.5455,
      "step": 6376
    },
    {
      "epoch": 8.881615598885794,
      "grad_norm": 2.0090112686157227,
      "learning_rate": 7.988914155770324e-06,
      "loss": 0.5778,
      "step": 6377
    },
    {
      "epoch": 8.883008356545961,
      "grad_norm": 2.290604591369629,
      "learning_rate": 7.97896532120523e-06,
      "loss": 0.7082,
      "step": 6378
    },
    {
      "epoch": 8.884401114206128,
      "grad_norm": 2.384650468826294,
      "learning_rate": 7.969016486640135e-06,
      "loss": 0.6976,
      "step": 6379
    },
    {
      "epoch": 8.885793871866294,
      "grad_norm": 2.0418272018432617,
      "learning_rate": 7.959067652075043e-06,
      "loss": 0.4512,
      "step": 6380
    },
    {
      "epoch": 8.887186629526463,
      "grad_norm": 4.094249725341797,
      "learning_rate": 7.949118817509948e-06,
      "loss": 0.519,
      "step": 6381
    },
    {
      "epoch": 8.88857938718663,
      "grad_norm": 2.6078884601593018,
      "learning_rate": 7.939169982944854e-06,
      "loss": 0.8992,
      "step": 6382
    },
    {
      "epoch": 8.889972144846796,
      "grad_norm": 2.191352367401123,
      "learning_rate": 7.92922114837976e-06,
      "loss": 0.6294,
      "step": 6383
    },
    {
      "epoch": 8.891364902506965,
      "grad_norm": 2.573664903640747,
      "learning_rate": 7.919272313814666e-06,
      "loss": 0.6933,
      "step": 6384
    },
    {
      "epoch": 8.892757660167131,
      "grad_norm": 1.7536120414733887,
      "learning_rate": 7.909323479249573e-06,
      "loss": 0.4474,
      "step": 6385
    },
    {
      "epoch": 8.894150417827298,
      "grad_norm": 1.9037634134292603,
      "learning_rate": 7.899374644684479e-06,
      "loss": 0.427,
      "step": 6386
    },
    {
      "epoch": 8.895543175487465,
      "grad_norm": 2.3941051959991455,
      "learning_rate": 7.889425810119385e-06,
      "loss": 0.6907,
      "step": 6387
    },
    {
      "epoch": 8.896935933147633,
      "grad_norm": 2.87742018699646,
      "learning_rate": 7.879476975554292e-06,
      "loss": 0.7522,
      "step": 6388
    },
    {
      "epoch": 8.8983286908078,
      "grad_norm": 2.0602097511291504,
      "learning_rate": 7.869528140989198e-06,
      "loss": 0.5668,
      "step": 6389
    },
    {
      "epoch": 8.899721448467966,
      "grad_norm": 2.61000657081604,
      "learning_rate": 7.859579306424104e-06,
      "loss": 0.6381,
      "step": 6390
    },
    {
      "epoch": 8.901114206128133,
      "grad_norm": 2.4334514141082764,
      "learning_rate": 7.84963047185901e-06,
      "loss": 0.5955,
      "step": 6391
    },
    {
      "epoch": 8.902506963788301,
      "grad_norm": 2.498460292816162,
      "learning_rate": 7.839681637293917e-06,
      "loss": 0.7454,
      "step": 6392
    },
    {
      "epoch": 8.903899721448468,
      "grad_norm": 2.1475067138671875,
      "learning_rate": 7.829732802728823e-06,
      "loss": 0.5761,
      "step": 6393
    },
    {
      "epoch": 8.905292479108635,
      "grad_norm": 2.2519261837005615,
      "learning_rate": 7.819783968163728e-06,
      "loss": 0.6272,
      "step": 6394
    },
    {
      "epoch": 8.906685236768801,
      "grad_norm": 2.2598562240600586,
      "learning_rate": 7.809835133598634e-06,
      "loss": 0.7187,
      "step": 6395
    },
    {
      "epoch": 8.90807799442897,
      "grad_norm": 2.552800416946411,
      "learning_rate": 7.799886299033542e-06,
      "loss": 0.5676,
      "step": 6396
    },
    {
      "epoch": 8.909470752089137,
      "grad_norm": 2.866748809814453,
      "learning_rate": 7.789937464468447e-06,
      "loss": 0.9511,
      "step": 6397
    },
    {
      "epoch": 8.910863509749303,
      "grad_norm": 2.2814176082611084,
      "learning_rate": 7.779988629903353e-06,
      "loss": 0.688,
      "step": 6398
    },
    {
      "epoch": 8.912256267409472,
      "grad_norm": 2.148735523223877,
      "learning_rate": 7.77003979533826e-06,
      "loss": 0.4889,
      "step": 6399
    },
    {
      "epoch": 8.913649025069638,
      "grad_norm": 2.1620900630950928,
      "learning_rate": 7.760090960773166e-06,
      "loss": 0.5491,
      "step": 6400
    },
    {
      "epoch": 8.915041782729805,
      "grad_norm": 2.202448844909668,
      "learning_rate": 7.750142126208072e-06,
      "loss": 0.6079,
      "step": 6401
    },
    {
      "epoch": 8.916434540389972,
      "grad_norm": 2.5117387771606445,
      "learning_rate": 7.740193291642978e-06,
      "loss": 0.7127,
      "step": 6402
    },
    {
      "epoch": 8.91782729805014,
      "grad_norm": 2.3837361335754395,
      "learning_rate": 7.730244457077884e-06,
      "loss": 0.6081,
      "step": 6403
    },
    {
      "epoch": 8.919220055710307,
      "grad_norm": 2.114670991897583,
      "learning_rate": 7.720295622512791e-06,
      "loss": 0.4769,
      "step": 6404
    },
    {
      "epoch": 8.920612813370473,
      "grad_norm": 2.271439552307129,
      "learning_rate": 7.710346787947697e-06,
      "loss": 0.637,
      "step": 6405
    },
    {
      "epoch": 8.92200557103064,
      "grad_norm": 1.810683012008667,
      "learning_rate": 7.700397953382603e-06,
      "loss": 0.5446,
      "step": 6406
    },
    {
      "epoch": 8.923398328690809,
      "grad_norm": 2.686549425125122,
      "learning_rate": 7.69044911881751e-06,
      "loss": 0.7877,
      "step": 6407
    },
    {
      "epoch": 8.924791086350975,
      "grad_norm": 2.7595272064208984,
      "learning_rate": 7.680500284252416e-06,
      "loss": 0.7785,
      "step": 6408
    },
    {
      "epoch": 8.926183844011142,
      "grad_norm": 2.541989803314209,
      "learning_rate": 7.670551449687322e-06,
      "loss": 0.6519,
      "step": 6409
    },
    {
      "epoch": 8.927576601671309,
      "grad_norm": 2.749061346054077,
      "learning_rate": 7.660602615122228e-06,
      "loss": 0.8212,
      "step": 6410
    },
    {
      "epoch": 8.928969359331477,
      "grad_norm": 2.9554824829101562,
      "learning_rate": 7.650653780557133e-06,
      "loss": 0.645,
      "step": 6411
    },
    {
      "epoch": 8.930362116991644,
      "grad_norm": 2.175075054168701,
      "learning_rate": 7.640704945992039e-06,
      "loss": 0.5324,
      "step": 6412
    },
    {
      "epoch": 8.93175487465181,
      "grad_norm": 1.6839343309402466,
      "learning_rate": 7.630756111426947e-06,
      "loss": 0.4137,
      "step": 6413
    },
    {
      "epoch": 8.933147632311977,
      "grad_norm": 2.240232467651367,
      "learning_rate": 7.620807276861852e-06,
      "loss": 0.5484,
      "step": 6414
    },
    {
      "epoch": 8.934540389972145,
      "grad_norm": 2.220383882522583,
      "learning_rate": 7.610858442296759e-06,
      "loss": 0.5923,
      "step": 6415
    },
    {
      "epoch": 8.935933147632312,
      "grad_norm": 2.379972457885742,
      "learning_rate": 7.6009096077316655e-06,
      "loss": 0.7216,
      "step": 6416
    },
    {
      "epoch": 8.937325905292479,
      "grad_norm": 2.165274143218994,
      "learning_rate": 7.590960773166571e-06,
      "loss": 0.497,
      "step": 6417
    },
    {
      "epoch": 8.938718662952645,
      "grad_norm": 2.3317317962646484,
      "learning_rate": 7.581011938601478e-06,
      "loss": 0.6365,
      "step": 6418
    },
    {
      "epoch": 8.940111420612814,
      "grad_norm": 2.669895887374878,
      "learning_rate": 7.571063104036384e-06,
      "loss": 0.7133,
      "step": 6419
    },
    {
      "epoch": 8.94150417827298,
      "grad_norm": 2.572749137878418,
      "learning_rate": 7.5611142694712895e-06,
      "loss": 0.7673,
      "step": 6420
    },
    {
      "epoch": 8.942896935933147,
      "grad_norm": 2.1923329830169678,
      "learning_rate": 7.551165434906196e-06,
      "loss": 0.5866,
      "step": 6421
    },
    {
      "epoch": 8.944289693593316,
      "grad_norm": 2.577394962310791,
      "learning_rate": 7.541216600341102e-06,
      "loss": 0.6512,
      "step": 6422
    },
    {
      "epoch": 8.945682451253482,
      "grad_norm": 2.8903701305389404,
      "learning_rate": 7.531267765776008e-06,
      "loss": 0.9225,
      "step": 6423
    },
    {
      "epoch": 8.947075208913649,
      "grad_norm": 2.316950559616089,
      "learning_rate": 7.521318931210915e-06,
      "loss": 0.5731,
      "step": 6424
    },
    {
      "epoch": 8.948467966573816,
      "grad_norm": 2.4975435733795166,
      "learning_rate": 7.511370096645821e-06,
      "loss": 0.5957,
      "step": 6425
    },
    {
      "epoch": 8.949860724233984,
      "grad_norm": 2.2214558124542236,
      "learning_rate": 7.5014212620807275e-06,
      "loss": 0.5169,
      "step": 6426
    },
    {
      "epoch": 8.95125348189415,
      "grad_norm": 2.169856309890747,
      "learning_rate": 7.491472427515633e-06,
      "loss": 0.4889,
      "step": 6427
    },
    {
      "epoch": 8.952646239554317,
      "grad_norm": 1.9657492637634277,
      "learning_rate": 7.481523592950539e-06,
      "loss": 0.4941,
      "step": 6428
    },
    {
      "epoch": 8.954038997214484,
      "grad_norm": 2.2034287452697754,
      "learning_rate": 7.471574758385446e-06,
      "loss": 0.59,
      "step": 6429
    },
    {
      "epoch": 8.955431754874652,
      "grad_norm": 2.387701988220215,
      "learning_rate": 7.461625923820352e-06,
      "loss": 0.7226,
      "step": 6430
    },
    {
      "epoch": 8.95682451253482,
      "grad_norm": 2.715440273284912,
      "learning_rate": 7.451677089255258e-06,
      "loss": 0.7354,
      "step": 6431
    },
    {
      "epoch": 8.958217270194986,
      "grad_norm": 2.4074158668518066,
      "learning_rate": 7.441728254690165e-06,
      "loss": 0.8206,
      "step": 6432
    },
    {
      "epoch": 8.959610027855152,
      "grad_norm": 2.1775918006896973,
      "learning_rate": 7.43177942012507e-06,
      "loss": 0.6853,
      "step": 6433
    },
    {
      "epoch": 8.961002785515321,
      "grad_norm": 2.364708185195923,
      "learning_rate": 7.421830585559976e-06,
      "loss": 0.6133,
      "step": 6434
    },
    {
      "epoch": 8.962395543175488,
      "grad_norm": 2.233593225479126,
      "learning_rate": 7.411881750994883e-06,
      "loss": 0.5081,
      "step": 6435
    },
    {
      "epoch": 8.963788300835654,
      "grad_norm": 2.5921921730041504,
      "learning_rate": 7.4019329164297885e-06,
      "loss": 0.7114,
      "step": 6436
    },
    {
      "epoch": 8.965181058495821,
      "grad_norm": 2.1150667667388916,
      "learning_rate": 7.391984081864696e-06,
      "loss": 0.5276,
      "step": 6437
    },
    {
      "epoch": 8.96657381615599,
      "grad_norm": 2.1465353965759277,
      "learning_rate": 7.382035247299602e-06,
      "loss": 0.5026,
      "step": 6438
    },
    {
      "epoch": 8.967966573816156,
      "grad_norm": 2.005256414413452,
      "learning_rate": 7.3720864127345075e-06,
      "loss": 0.4283,
      "step": 6439
    },
    {
      "epoch": 8.969359331476323,
      "grad_norm": 2.601466655731201,
      "learning_rate": 7.362137578169414e-06,
      "loss": 0.8654,
      "step": 6440
    },
    {
      "epoch": 8.97075208913649,
      "grad_norm": 2.86332631111145,
      "learning_rate": 7.35218874360432e-06,
      "loss": 0.7487,
      "step": 6441
    },
    {
      "epoch": 8.972144846796658,
      "grad_norm": 2.120083808898926,
      "learning_rate": 7.342239909039226e-06,
      "loss": 0.46,
      "step": 6442
    },
    {
      "epoch": 8.973537604456824,
      "grad_norm": 2.6671149730682373,
      "learning_rate": 7.332291074474132e-06,
      "loss": 0.6688,
      "step": 6443
    },
    {
      "epoch": 8.974930362116991,
      "grad_norm": 2.4337751865386963,
      "learning_rate": 7.322342239909039e-06,
      "loss": 0.6878,
      "step": 6444
    },
    {
      "epoch": 8.97632311977716,
      "grad_norm": 2.029585123062134,
      "learning_rate": 7.312393405343945e-06,
      "loss": 0.5332,
      "step": 6445
    },
    {
      "epoch": 8.977715877437326,
      "grad_norm": 1.9723401069641113,
      "learning_rate": 7.302444570778851e-06,
      "loss": 0.5103,
      "step": 6446
    },
    {
      "epoch": 8.979108635097493,
      "grad_norm": 2.330951452255249,
      "learning_rate": 7.292495736213757e-06,
      "loss": 0.6988,
      "step": 6447
    },
    {
      "epoch": 8.98050139275766,
      "grad_norm": 2.130052328109741,
      "learning_rate": 7.282546901648663e-06,
      "loss": 0.5799,
      "step": 6448
    },
    {
      "epoch": 8.981894150417828,
      "grad_norm": 2.8521018028259277,
      "learning_rate": 7.2725980670835695e-06,
      "loss": 0.8852,
      "step": 6449
    },
    {
      "epoch": 8.983286908077995,
      "grad_norm": 2.0426878929138184,
      "learning_rate": 7.262649232518475e-06,
      "loss": 0.5008,
      "step": 6450
    },
    {
      "epoch": 8.984679665738161,
      "grad_norm": 2.6988277435302734,
      "learning_rate": 7.252700397953383e-06,
      "loss": 0.6231,
      "step": 6451
    },
    {
      "epoch": 8.986072423398328,
      "grad_norm": 2.195624828338623,
      "learning_rate": 7.2427515633882885e-06,
      "loss": 0.7232,
      "step": 6452
    },
    {
      "epoch": 8.987465181058496,
      "grad_norm": 2.369209051132202,
      "learning_rate": 7.232802728823194e-06,
      "loss": 0.5215,
      "step": 6453
    },
    {
      "epoch": 8.988857938718663,
      "grad_norm": 2.2820773124694824,
      "learning_rate": 7.222853894258101e-06,
      "loss": 0.5163,
      "step": 6454
    },
    {
      "epoch": 8.99025069637883,
      "grad_norm": 2.16398549079895,
      "learning_rate": 7.212905059693007e-06,
      "loss": 0.5011,
      "step": 6455
    },
    {
      "epoch": 8.991643454038996,
      "grad_norm": 2.217278003692627,
      "learning_rate": 7.202956225127912e-06,
      "loss": 0.6777,
      "step": 6456
    },
    {
      "epoch": 8.993036211699165,
      "grad_norm": 2.1077592372894287,
      "learning_rate": 7.193007390562819e-06,
      "loss": 0.4697,
      "step": 6457
    },
    {
      "epoch": 8.994428969359332,
      "grad_norm": 2.4509129524230957,
      "learning_rate": 7.183058555997726e-06,
      "loss": 0.7212,
      "step": 6458
    },
    {
      "epoch": 8.995821727019498,
      "grad_norm": 2.3753774166107178,
      "learning_rate": 7.173109721432631e-06,
      "loss": 0.5633,
      "step": 6459
    },
    {
      "epoch": 8.997214484679667,
      "grad_norm": 2.4649858474731445,
      "learning_rate": 7.163160886867538e-06,
      "loss": 0.6322,
      "step": 6460
    },
    {
      "epoch": 8.998607242339833,
      "grad_norm": 1.9081060886383057,
      "learning_rate": 7.153212052302444e-06,
      "loss": 0.4937,
      "step": 6461
    },
    {
      "epoch": 9.0,
      "grad_norm": 2.2499141693115234,
      "learning_rate": 7.14326321773735e-06,
      "loss": 0.5931,
      "step": 6462
    },
    {
      "epoch": 9.001392757660167,
      "grad_norm": 2.136467456817627,
      "learning_rate": 7.133314383172256e-06,
      "loss": 0.4405,
      "step": 6463
    },
    {
      "epoch": 9.002785515320335,
      "grad_norm": 2.313202142715454,
      "learning_rate": 7.123365548607162e-06,
      "loss": 0.6688,
      "step": 6464
    },
    {
      "epoch": 9.004178272980502,
      "grad_norm": 2.090893030166626,
      "learning_rate": 7.113416714042069e-06,
      "loss": 0.4964,
      "step": 6465
    },
    {
      "epoch": 9.005571030640668,
      "grad_norm": 2.3237464427948,
      "learning_rate": 7.103467879476975e-06,
      "loss": 0.4757,
      "step": 6466
    },
    {
      "epoch": 9.006963788300835,
      "grad_norm": 1.8766562938690186,
      "learning_rate": 7.093519044911881e-06,
      "loss": 0.5428,
      "step": 6467
    },
    {
      "epoch": 9.008356545961004,
      "grad_norm": 2.2717623710632324,
      "learning_rate": 7.0835702103467875e-06,
      "loss": 0.6654,
      "step": 6468
    },
    {
      "epoch": 9.00974930362117,
      "grad_norm": 2.110006332397461,
      "learning_rate": 7.073621375781693e-06,
      "loss": 0.5555,
      "step": 6469
    },
    {
      "epoch": 9.011142061281337,
      "grad_norm": 2.2815473079681396,
      "learning_rate": 7.063672541216599e-06,
      "loss": 0.595,
      "step": 6470
    },
    {
      "epoch": 9.012534818941504,
      "grad_norm": 2.1525909900665283,
      "learning_rate": 7.053723706651506e-06,
      "loss": 0.5173,
      "step": 6471
    },
    {
      "epoch": 9.013927576601672,
      "grad_norm": 2.469860792160034,
      "learning_rate": 7.043774872086412e-06,
      "loss": 0.6452,
      "step": 6472
    },
    {
      "epoch": 9.015320334261839,
      "grad_norm": 1.8618592023849487,
      "learning_rate": 7.033826037521319e-06,
      "loss": 0.5143,
      "step": 6473
    },
    {
      "epoch": 9.016713091922005,
      "grad_norm": 2.2681984901428223,
      "learning_rate": 7.023877202956225e-06,
      "loss": 0.6183,
      "step": 6474
    },
    {
      "epoch": 9.018105849582172,
      "grad_norm": 2.1257636547088623,
      "learning_rate": 7.0139283683911305e-06,
      "loss": 0.6137,
      "step": 6475
    },
    {
      "epoch": 9.01949860724234,
      "grad_norm": 2.035341739654541,
      "learning_rate": 7.003979533826037e-06,
      "loss": 0.5467,
      "step": 6476
    },
    {
      "epoch": 9.020891364902507,
      "grad_norm": 2.1111531257629395,
      "learning_rate": 6.994030699260943e-06,
      "loss": 0.5556,
      "step": 6477
    },
    {
      "epoch": 9.022284122562674,
      "grad_norm": 2.399853229522705,
      "learning_rate": 6.984081864695849e-06,
      "loss": 0.7357,
      "step": 6478
    },
    {
      "epoch": 9.02367688022284,
      "grad_norm": 1.8878366947174072,
      "learning_rate": 6.974133030130756e-06,
      "loss": 0.3905,
      "step": 6479
    },
    {
      "epoch": 9.025069637883009,
      "grad_norm": 1.9461958408355713,
      "learning_rate": 6.964184195565662e-06,
      "loss": 0.5955,
      "step": 6480
    },
    {
      "epoch": 9.026462395543176,
      "grad_norm": 2.0812342166900635,
      "learning_rate": 6.954235361000568e-06,
      "loss": 0.599,
      "step": 6481
    },
    {
      "epoch": 9.027855153203342,
      "grad_norm": 2.3478505611419678,
      "learning_rate": 6.944286526435474e-06,
      "loss": 0.6056,
      "step": 6482
    },
    {
      "epoch": 9.029247910863509,
      "grad_norm": 2.2185094356536865,
      "learning_rate": 6.93433769187038e-06,
      "loss": 0.656,
      "step": 6483
    },
    {
      "epoch": 9.030640668523677,
      "grad_norm": 2.0745649337768555,
      "learning_rate": 6.924388857305287e-06,
      "loss": 0.5492,
      "step": 6484
    },
    {
      "epoch": 9.032033426183844,
      "grad_norm": 2.6149706840515137,
      "learning_rate": 6.914440022740192e-06,
      "loss": 0.7426,
      "step": 6485
    },
    {
      "epoch": 9.03342618384401,
      "grad_norm": 2.519866704940796,
      "learning_rate": 6.904491188175099e-06,
      "loss": 0.6276,
      "step": 6486
    },
    {
      "epoch": 9.034818941504179,
      "grad_norm": 1.6276168823242188,
      "learning_rate": 6.894542353610006e-06,
      "loss": 0.4102,
      "step": 6487
    },
    {
      "epoch": 9.036211699164346,
      "grad_norm": 1.8854120969772339,
      "learning_rate": 6.884593519044911e-06,
      "loss": 0.4847,
      "step": 6488
    },
    {
      "epoch": 9.037604456824512,
      "grad_norm": 2.044466733932495,
      "learning_rate": 6.874644684479817e-06,
      "loss": 0.6135,
      "step": 6489
    },
    {
      "epoch": 9.038997214484679,
      "grad_norm": 2.0263922214508057,
      "learning_rate": 6.864695849914724e-06,
      "loss": 0.5746,
      "step": 6490
    },
    {
      "epoch": 9.040389972144848,
      "grad_norm": 2.099761486053467,
      "learning_rate": 6.8547470153496295e-06,
      "loss": 0.5026,
      "step": 6491
    },
    {
      "epoch": 9.041782729805014,
      "grad_norm": 1.8517907857894897,
      "learning_rate": 6.844798180784535e-06,
      "loss": 0.4566,
      "step": 6492
    },
    {
      "epoch": 9.04317548746518,
      "grad_norm": 2.01879620552063,
      "learning_rate": 6.834849346219443e-06,
      "loss": 0.6084,
      "step": 6493
    },
    {
      "epoch": 9.044568245125348,
      "grad_norm": 1.533672571182251,
      "learning_rate": 6.8249005116543485e-06,
      "loss": 0.3441,
      "step": 6494
    },
    {
      "epoch": 9.045961002785516,
      "grad_norm": 1.9208685159683228,
      "learning_rate": 6.814951677089255e-06,
      "loss": 0.43,
      "step": 6495
    },
    {
      "epoch": 9.047353760445683,
      "grad_norm": 2.1694424152374268,
      "learning_rate": 6.805002842524161e-06,
      "loss": 0.6528,
      "step": 6496
    },
    {
      "epoch": 9.04874651810585,
      "grad_norm": 1.8260539770126343,
      "learning_rate": 6.795054007959067e-06,
      "loss": 0.4115,
      "step": 6497
    },
    {
      "epoch": 9.050139275766016,
      "grad_norm": 2.137845754623413,
      "learning_rate": 6.785105173393973e-06,
      "loss": 0.6022,
      "step": 6498
    },
    {
      "epoch": 9.051532033426184,
      "grad_norm": 2.4480721950531006,
      "learning_rate": 6.775156338828879e-06,
      "loss": 0.8338,
      "step": 6499
    },
    {
      "epoch": 9.052924791086351,
      "grad_norm": 2.1186275482177734,
      "learning_rate": 6.765207504263786e-06,
      "loss": 0.5907,
      "step": 6500
    },
    {
      "epoch": 9.054317548746518,
      "grad_norm": 2.262209415435791,
      "learning_rate": 6.755258669698692e-06,
      "loss": 0.5248,
      "step": 6501
    },
    {
      "epoch": 9.055710306406684,
      "grad_norm": 2.148144483566284,
      "learning_rate": 6.745309835133598e-06,
      "loss": 0.5823,
      "step": 6502
    },
    {
      "epoch": 9.057103064066853,
      "grad_norm": 2.2517294883728027,
      "learning_rate": 6.735361000568504e-06,
      "loss": 0.455,
      "step": 6503
    },
    {
      "epoch": 9.05849582172702,
      "grad_norm": 1.8921222686767578,
      "learning_rate": 6.7254121660034105e-06,
      "loss": 0.4669,
      "step": 6504
    },
    {
      "epoch": 9.059888579387186,
      "grad_norm": 2.537829637527466,
      "learning_rate": 6.715463331438316e-06,
      "loss": 0.6485,
      "step": 6505
    },
    {
      "epoch": 9.061281337047355,
      "grad_norm": 2.7535204887390137,
      "learning_rate": 6.705514496873224e-06,
      "loss": 0.6671,
      "step": 6506
    },
    {
      "epoch": 9.062674094707521,
      "grad_norm": 2.0473270416259766,
      "learning_rate": 6.6955656623081295e-06,
      "loss": 0.5458,
      "step": 6507
    },
    {
      "epoch": 9.064066852367688,
      "grad_norm": 1.9972840547561646,
      "learning_rate": 6.685616827743035e-06,
      "loss": 0.5426,
      "step": 6508
    },
    {
      "epoch": 9.065459610027855,
      "grad_norm": 2.451608657836914,
      "learning_rate": 6.675667993177942e-06,
      "loss": 0.5775,
      "step": 6509
    },
    {
      "epoch": 9.066852367688023,
      "grad_norm": 2.119060754776001,
      "learning_rate": 6.665719158612848e-06,
      "loss": 0.5437,
      "step": 6510
    },
    {
      "epoch": 9.06824512534819,
      "grad_norm": 2.03897762298584,
      "learning_rate": 6.655770324047753e-06,
      "loss": 0.6292,
      "step": 6511
    },
    {
      "epoch": 9.069637883008356,
      "grad_norm": 1.972775936126709,
      "learning_rate": 6.64582148948266e-06,
      "loss": 0.5501,
      "step": 6512
    },
    {
      "epoch": 9.071030640668523,
      "grad_norm": 1.6815879344940186,
      "learning_rate": 6.635872654917566e-06,
      "loss": 0.4543,
      "step": 6513
    },
    {
      "epoch": 9.072423398328691,
      "grad_norm": 2.2197539806365967,
      "learning_rate": 6.625923820352472e-06,
      "loss": 0.6066,
      "step": 6514
    },
    {
      "epoch": 9.073816155988858,
      "grad_norm": 2.0606441497802734,
      "learning_rate": 6.615974985787379e-06,
      "loss": 0.4188,
      "step": 6515
    },
    {
      "epoch": 9.075208913649025,
      "grad_norm": 2.4126522541046143,
      "learning_rate": 6.606026151222285e-06,
      "loss": 0.679,
      "step": 6516
    },
    {
      "epoch": 9.076601671309191,
      "grad_norm": 2.097259283065796,
      "learning_rate": 6.596077316657191e-06,
      "loss": 0.4659,
      "step": 6517
    },
    {
      "epoch": 9.07799442896936,
      "grad_norm": 1.8794084787368774,
      "learning_rate": 6.586128482092097e-06,
      "loss": 0.4761,
      "step": 6518
    },
    {
      "epoch": 9.079387186629527,
      "grad_norm": 2.0499138832092285,
      "learning_rate": 6.576179647527003e-06,
      "loss": 0.5466,
      "step": 6519
    },
    {
      "epoch": 9.080779944289693,
      "grad_norm": 2.3997960090637207,
      "learning_rate": 6.56623081296191e-06,
      "loss": 0.7621,
      "step": 6520
    },
    {
      "epoch": 9.08217270194986,
      "grad_norm": 2.227309465408325,
      "learning_rate": 6.556281978396816e-06,
      "loss": 0.5626,
      "step": 6521
    },
    {
      "epoch": 9.083565459610028,
      "grad_norm": 2.1702988147735596,
      "learning_rate": 6.546333143831722e-06,
      "loss": 0.571,
      "step": 6522
    },
    {
      "epoch": 9.084958217270195,
      "grad_norm": 1.8819431066513062,
      "learning_rate": 6.5363843092666286e-06,
      "loss": 0.4433,
      "step": 6523
    },
    {
      "epoch": 9.086350974930362,
      "grad_norm": 2.238947629928589,
      "learning_rate": 6.526435474701534e-06,
      "loss": 0.662,
      "step": 6524
    },
    {
      "epoch": 9.087743732590528,
      "grad_norm": 2.6882331371307373,
      "learning_rate": 6.51648664013644e-06,
      "loss": 0.6829,
      "step": 6525
    },
    {
      "epoch": 9.089136490250697,
      "grad_norm": 2.0446624755859375,
      "learning_rate": 6.506537805571347e-06,
      "loss": 0.5858,
      "step": 6526
    },
    {
      "epoch": 9.090529247910863,
      "grad_norm": 1.9051544666290283,
      "learning_rate": 6.4965889710062525e-06,
      "loss": 0.4415,
      "step": 6527
    },
    {
      "epoch": 9.09192200557103,
      "grad_norm": 2.035181999206543,
      "learning_rate": 6.48664013644116e-06,
      "loss": 0.497,
      "step": 6528
    },
    {
      "epoch": 9.093314763231199,
      "grad_norm": 2.2840728759765625,
      "learning_rate": 6.476691301876066e-06,
      "loss": 0.729,
      "step": 6529
    },
    {
      "epoch": 9.094707520891365,
      "grad_norm": 2.1286890506744385,
      "learning_rate": 6.4667424673109715e-06,
      "loss": 0.5436,
      "step": 6530
    },
    {
      "epoch": 9.096100278551532,
      "grad_norm": 2.1142804622650146,
      "learning_rate": 6.456793632745878e-06,
      "loss": 0.5353,
      "step": 6531
    },
    {
      "epoch": 9.097493036211699,
      "grad_norm": 1.9269057512283325,
      "learning_rate": 6.446844798180784e-06,
      "loss": 0.4997,
      "step": 6532
    },
    {
      "epoch": 9.098885793871867,
      "grad_norm": 2.1926798820495605,
      "learning_rate": 6.43689596361569e-06,
      "loss": 0.5438,
      "step": 6533
    },
    {
      "epoch": 9.100278551532034,
      "grad_norm": 2.042038917541504,
      "learning_rate": 6.426947129050596e-06,
      "loss": 0.508,
      "step": 6534
    },
    {
      "epoch": 9.1016713091922,
      "grad_norm": 1.7788965702056885,
      "learning_rate": 6.416998294485503e-06,
      "loss": 0.4407,
      "step": 6535
    },
    {
      "epoch": 9.103064066852367,
      "grad_norm": 2.239969253540039,
      "learning_rate": 6.407049459920409e-06,
      "loss": 0.6697,
      "step": 6536
    },
    {
      "epoch": 9.104456824512535,
      "grad_norm": 2.0588786602020264,
      "learning_rate": 6.397100625355315e-06,
      "loss": 0.582,
      "step": 6537
    },
    {
      "epoch": 9.105849582172702,
      "grad_norm": 1.8924956321716309,
      "learning_rate": 6.387151790790221e-06,
      "loss": 0.4345,
      "step": 6538
    },
    {
      "epoch": 9.107242339832869,
      "grad_norm": 2.6685168743133545,
      "learning_rate": 6.377202956225128e-06,
      "loss": 0.5076,
      "step": 6539
    },
    {
      "epoch": 9.108635097493035,
      "grad_norm": 1.8214253187179565,
      "learning_rate": 6.367254121660033e-06,
      "loss": 0.5259,
      "step": 6540
    },
    {
      "epoch": 9.110027855153204,
      "grad_norm": 1.6199214458465576,
      "learning_rate": 6.357305287094939e-06,
      "loss": 0.4196,
      "step": 6541
    },
    {
      "epoch": 9.11142061281337,
      "grad_norm": 1.9098056554794312,
      "learning_rate": 6.347356452529847e-06,
      "loss": 0.5348,
      "step": 6542
    },
    {
      "epoch": 9.112813370473537,
      "grad_norm": 2.2226037979125977,
      "learning_rate": 6.337407617964752e-06,
      "loss": 0.5473,
      "step": 6543
    },
    {
      "epoch": 9.114206128133704,
      "grad_norm": 2.042248487472534,
      "learning_rate": 6.327458783399658e-06,
      "loss": 0.4278,
      "step": 6544
    },
    {
      "epoch": 9.115598885793872,
      "grad_norm": 2.474745988845825,
      "learning_rate": 6.317509948834565e-06,
      "loss": 0.6941,
      "step": 6545
    },
    {
      "epoch": 9.116991643454039,
      "grad_norm": 2.1232476234436035,
      "learning_rate": 6.3075611142694706e-06,
      "loss": 0.5255,
      "step": 6546
    },
    {
      "epoch": 9.118384401114206,
      "grad_norm": 2.09792423248291,
      "learning_rate": 6.297612279704376e-06,
      "loss": 0.509,
      "step": 6547
    },
    {
      "epoch": 9.119777158774374,
      "grad_norm": 2.5328168869018555,
      "learning_rate": 6.287663445139283e-06,
      "loss": 0.6558,
      "step": 6548
    },
    {
      "epoch": 9.12116991643454,
      "grad_norm": 2.478328227996826,
      "learning_rate": 6.2777146105741896e-06,
      "loss": 0.8837,
      "step": 6549
    },
    {
      "epoch": 9.122562674094707,
      "grad_norm": 2.2600510120391846,
      "learning_rate": 6.267765776009095e-06,
      "loss": 0.6868,
      "step": 6550
    },
    {
      "epoch": 9.123955431754874,
      "grad_norm": 2.1773951053619385,
      "learning_rate": 6.257816941444002e-06,
      "loss": 0.6119,
      "step": 6551
    },
    {
      "epoch": 9.125348189415043,
      "grad_norm": 2.062988519668579,
      "learning_rate": 6.247868106878908e-06,
      "loss": 0.5197,
      "step": 6552
    },
    {
      "epoch": 9.12674094707521,
      "grad_norm": 2.266681432723999,
      "learning_rate": 6.237919272313814e-06,
      "loss": 0.4914,
      "step": 6553
    },
    {
      "epoch": 9.128133704735376,
      "grad_norm": 2.474975824356079,
      "learning_rate": 6.22797043774872e-06,
      "loss": 0.4814,
      "step": 6554
    },
    {
      "epoch": 9.129526462395543,
      "grad_norm": 1.9066749811172485,
      "learning_rate": 6.218021603183626e-06,
      "loss": 0.5745,
      "step": 6555
    },
    {
      "epoch": 9.130919220055711,
      "grad_norm": 2.4156877994537354,
      "learning_rate": 6.208072768618533e-06,
      "loss": 0.6683,
      "step": 6556
    },
    {
      "epoch": 9.132311977715878,
      "grad_norm": 1.9416658878326416,
      "learning_rate": 6.198123934053439e-06,
      "loss": 0.4561,
      "step": 6557
    },
    {
      "epoch": 9.133704735376044,
      "grad_norm": 1.8224204778671265,
      "learning_rate": 6.188175099488345e-06,
      "loss": 0.4899,
      "step": 6558
    },
    {
      "epoch": 9.135097493036211,
      "grad_norm": 2.1181678771972656,
      "learning_rate": 6.1782262649232515e-06,
      "loss": 0.6304,
      "step": 6559
    },
    {
      "epoch": 9.13649025069638,
      "grad_norm": 2.207368850708008,
      "learning_rate": 6.168277430358157e-06,
      "loss": 0.5457,
      "step": 6560
    },
    {
      "epoch": 9.137883008356546,
      "grad_norm": 2.40140438079834,
      "learning_rate": 6.158328595793063e-06,
      "loss": 0.5149,
      "step": 6561
    },
    {
      "epoch": 9.139275766016713,
      "grad_norm": 2.3443496227264404,
      "learning_rate": 6.14837976122797e-06,
      "loss": 0.5534,
      "step": 6562
    },
    {
      "epoch": 9.14066852367688,
      "grad_norm": 2.1881232261657715,
      "learning_rate": 6.138430926662876e-06,
      "loss": 0.5862,
      "step": 6563
    },
    {
      "epoch": 9.142061281337048,
      "grad_norm": 2.1861469745635986,
      "learning_rate": 6.128482092097783e-06,
      "loss": 0.7328,
      "step": 6564
    },
    {
      "epoch": 9.143454038997215,
      "grad_norm": 2.315837860107422,
      "learning_rate": 6.118533257532689e-06,
      "loss": 0.6537,
      "step": 6565
    },
    {
      "epoch": 9.144846796657381,
      "grad_norm": 2.104191303253174,
      "learning_rate": 6.108584422967594e-06,
      "loss": 0.581,
      "step": 6566
    },
    {
      "epoch": 9.14623955431755,
      "grad_norm": 2.272362470626831,
      "learning_rate": 6.098635588402501e-06,
      "loss": 0.6664,
      "step": 6567
    },
    {
      "epoch": 9.147632311977716,
      "grad_norm": 2.103689193725586,
      "learning_rate": 6.088686753837407e-06,
      "loss": 0.5321,
      "step": 6568
    },
    {
      "epoch": 9.149025069637883,
      "grad_norm": 2.0636544227600098,
      "learning_rate": 6.0787379192723126e-06,
      "loss": 0.5396,
      "step": 6569
    },
    {
      "epoch": 9.15041782729805,
      "grad_norm": 2.2445733547210693,
      "learning_rate": 6.06878908470722e-06,
      "loss": 0.529,
      "step": 6570
    },
    {
      "epoch": 9.151810584958218,
      "grad_norm": 1.8259060382843018,
      "learning_rate": 6.058840250142126e-06,
      "loss": 0.4758,
      "step": 6571
    },
    {
      "epoch": 9.153203342618385,
      "grad_norm": 2.264030694961548,
      "learning_rate": 6.0488914155770316e-06,
      "loss": 0.603,
      "step": 6572
    },
    {
      "epoch": 9.154596100278551,
      "grad_norm": 2.1737589836120605,
      "learning_rate": 6.038942581011938e-06,
      "loss": 0.5718,
      "step": 6573
    },
    {
      "epoch": 9.155988857938718,
      "grad_norm": 2.5338075160980225,
      "learning_rate": 6.028993746446844e-06,
      "loss": 0.8101,
      "step": 6574
    },
    {
      "epoch": 9.157381615598887,
      "grad_norm": 2.3226256370544434,
      "learning_rate": 6.0190449118817506e-06,
      "loss": 0.6567,
      "step": 6575
    },
    {
      "epoch": 9.158774373259053,
      "grad_norm": 1.9931647777557373,
      "learning_rate": 6.009096077316656e-06,
      "loss": 0.5562,
      "step": 6576
    },
    {
      "epoch": 9.16016713091922,
      "grad_norm": 2.3122940063476562,
      "learning_rate": 5.999147242751563e-06,
      "loss": 0.5887,
      "step": 6577
    },
    {
      "epoch": 9.161559888579387,
      "grad_norm": 1.7064476013183594,
      "learning_rate": 5.9891984081864696e-06,
      "loss": 0.3999,
      "step": 6578
    },
    {
      "epoch": 9.162952646239555,
      "grad_norm": 1.9466005563735962,
      "learning_rate": 5.979249573621375e-06,
      "loss": 0.4791,
      "step": 6579
    },
    {
      "epoch": 9.164345403899722,
      "grad_norm": 2.7882297039031982,
      "learning_rate": 5.969300739056281e-06,
      "loss": 0.6314,
      "step": 6580
    },
    {
      "epoch": 9.165738161559888,
      "grad_norm": 2.1359176635742188,
      "learning_rate": 5.959351904491188e-06,
      "loss": 0.6299,
      "step": 6581
    },
    {
      "epoch": 9.167130919220055,
      "grad_norm": 2.3436279296875,
      "learning_rate": 5.9494030699260935e-06,
      "loss": 0.5784,
      "step": 6582
    },
    {
      "epoch": 9.168523676880223,
      "grad_norm": 2.3204479217529297,
      "learning_rate": 5.939454235360999e-06,
      "loss": 0.7025,
      "step": 6583
    },
    {
      "epoch": 9.16991643454039,
      "grad_norm": 1.9044424295425415,
      "learning_rate": 5.929505400795907e-06,
      "loss": 0.5345,
      "step": 6584
    },
    {
      "epoch": 9.171309192200557,
      "grad_norm": 2.173640012741089,
      "learning_rate": 5.9195565662308125e-06,
      "loss": 0.5031,
      "step": 6585
    },
    {
      "epoch": 9.172701949860723,
      "grad_norm": 2.1724770069122314,
      "learning_rate": 5.909607731665719e-06,
      "loss": 0.5905,
      "step": 6586
    },
    {
      "epoch": 9.174094707520892,
      "grad_norm": 5.491093158721924,
      "learning_rate": 5.899658897100625e-06,
      "loss": 0.5465,
      "step": 6587
    },
    {
      "epoch": 9.175487465181059,
      "grad_norm": 2.1875393390655518,
      "learning_rate": 5.889710062535531e-06,
      "loss": 0.5646,
      "step": 6588
    },
    {
      "epoch": 9.176880222841225,
      "grad_norm": 2.554903507232666,
      "learning_rate": 5.879761227970437e-06,
      "loss": 0.6221,
      "step": 6589
    },
    {
      "epoch": 9.178272980501394,
      "grad_norm": 2.604617118835449,
      "learning_rate": 5.869812393405343e-06,
      "loss": 0.8008,
      "step": 6590
    },
    {
      "epoch": 9.17966573816156,
      "grad_norm": 2.177938222885132,
      "learning_rate": 5.85986355884025e-06,
      "loss": 0.5306,
      "step": 6591
    },
    {
      "epoch": 9.181058495821727,
      "grad_norm": 1.9429033994674683,
      "learning_rate": 5.849914724275156e-06,
      "loss": 0.4525,
      "step": 6592
    },
    {
      "epoch": 9.182451253481894,
      "grad_norm": 2.2409863471984863,
      "learning_rate": 5.839965889710062e-06,
      "loss": 0.6216,
      "step": 6593
    },
    {
      "epoch": 9.183844011142062,
      "grad_norm": 2.0866949558258057,
      "learning_rate": 5.830017055144968e-06,
      "loss": 0.4791,
      "step": 6594
    },
    {
      "epoch": 9.185236768802229,
      "grad_norm": 1.870818018913269,
      "learning_rate": 5.820068220579874e-06,
      "loss": 0.4292,
      "step": 6595
    },
    {
      "epoch": 9.186629526462395,
      "grad_norm": 2.2606542110443115,
      "learning_rate": 5.81011938601478e-06,
      "loss": 0.599,
      "step": 6596
    },
    {
      "epoch": 9.188022284122562,
      "grad_norm": 2.318547010421753,
      "learning_rate": 5.800170551449688e-06,
      "loss": 0.4828,
      "step": 6597
    },
    {
      "epoch": 9.18941504178273,
      "grad_norm": 2.084038019180298,
      "learning_rate": 5.790221716884593e-06,
      "loss": 0.5901,
      "step": 6598
    },
    {
      "epoch": 9.190807799442897,
      "grad_norm": 2.06911563873291,
      "learning_rate": 5.780272882319499e-06,
      "loss": 0.5821,
      "step": 6599
    },
    {
      "epoch": 9.192200557103064,
      "grad_norm": 3.8607749938964844,
      "learning_rate": 5.770324047754406e-06,
      "loss": 0.612,
      "step": 6600
    },
    {
      "epoch": 9.19359331476323,
      "grad_norm": 2.122738838195801,
      "learning_rate": 5.7603752131893116e-06,
      "loss": 0.58,
      "step": 6601
    },
    {
      "epoch": 9.194986072423399,
      "grad_norm": 2.52789568901062,
      "learning_rate": 5.750426378624217e-06,
      "loss": 0.7067,
      "step": 6602
    },
    {
      "epoch": 9.196378830083566,
      "grad_norm": 2.0914368629455566,
      "learning_rate": 5.740477544059124e-06,
      "loss": 0.543,
      "step": 6603
    },
    {
      "epoch": 9.197771587743732,
      "grad_norm": 2.113074779510498,
      "learning_rate": 5.73052870949403e-06,
      "loss": 0.5929,
      "step": 6604
    },
    {
      "epoch": 9.199164345403899,
      "grad_norm": 2.141188621520996,
      "learning_rate": 5.720579874928936e-06,
      "loss": 0.5811,
      "step": 6605
    },
    {
      "epoch": 9.200557103064067,
      "grad_norm": 2.0961978435516357,
      "learning_rate": 5.710631040363843e-06,
      "loss": 0.6667,
      "step": 6606
    },
    {
      "epoch": 9.201949860724234,
      "grad_norm": 2.0160956382751465,
      "learning_rate": 5.700682205798749e-06,
      "loss": 0.5359,
      "step": 6607
    },
    {
      "epoch": 9.2033426183844,
      "grad_norm": 2.2770156860351562,
      "learning_rate": 5.690733371233655e-06,
      "loss": 0.6647,
      "step": 6608
    },
    {
      "epoch": 9.204735376044567,
      "grad_norm": 2.2007791996002197,
      "learning_rate": 5.680784536668561e-06,
      "loss": 0.6502,
      "step": 6609
    },
    {
      "epoch": 9.206128133704736,
      "grad_norm": 2.4591970443725586,
      "learning_rate": 5.670835702103467e-06,
      "loss": 0.5854,
      "step": 6610
    },
    {
      "epoch": 9.207520891364902,
      "grad_norm": 2.0855774879455566,
      "learning_rate": 5.660886867538374e-06,
      "loss": 0.5477,
      "step": 6611
    },
    {
      "epoch": 9.20891364902507,
      "grad_norm": 1.9735785722732544,
      "learning_rate": 5.65093803297328e-06,
      "loss": 0.5616,
      "step": 6612
    },
    {
      "epoch": 9.210306406685238,
      "grad_norm": 1.9095104932785034,
      "learning_rate": 5.640989198408186e-06,
      "loss": 0.4562,
      "step": 6613
    },
    {
      "epoch": 9.211699164345404,
      "grad_norm": 2.0682218074798584,
      "learning_rate": 5.6310403638430925e-06,
      "loss": 0.5367,
      "step": 6614
    },
    {
      "epoch": 9.213091922005571,
      "grad_norm": 2.425248861312866,
      "learning_rate": 5.621091529277998e-06,
      "loss": 0.6349,
      "step": 6615
    },
    {
      "epoch": 9.214484679665738,
      "grad_norm": 1.8182013034820557,
      "learning_rate": 5.611142694712904e-06,
      "loss": 0.4559,
      "step": 6616
    },
    {
      "epoch": 9.215877437325906,
      "grad_norm": 2.4161479473114014,
      "learning_rate": 5.601193860147811e-06,
      "loss": 0.6224,
      "step": 6617
    },
    {
      "epoch": 9.217270194986073,
      "grad_norm": 2.2463033199310303,
      "learning_rate": 5.591245025582716e-06,
      "loss": 0.6994,
      "step": 6618
    },
    {
      "epoch": 9.21866295264624,
      "grad_norm": 2.5108425617218018,
      "learning_rate": 5.581296191017624e-06,
      "loss": 0.5931,
      "step": 6619
    },
    {
      "epoch": 9.220055710306406,
      "grad_norm": 1.9878058433532715,
      "learning_rate": 5.57134735645253e-06,
      "loss": 0.5535,
      "step": 6620
    },
    {
      "epoch": 9.221448467966574,
      "grad_norm": 2.05832839012146,
      "learning_rate": 5.561398521887435e-06,
      "loss": 0.4756,
      "step": 6621
    },
    {
      "epoch": 9.222841225626741,
      "grad_norm": 1.9694043397903442,
      "learning_rate": 5.551449687322342e-06,
      "loss": 0.4976,
      "step": 6622
    },
    {
      "epoch": 9.224233983286908,
      "grad_norm": 2.2591166496276855,
      "learning_rate": 5.541500852757248e-06,
      "loss": 0.5794,
      "step": 6623
    },
    {
      "epoch": 9.225626740947074,
      "grad_norm": 1.9439704418182373,
      "learning_rate": 5.5315520181921536e-06,
      "loss": 0.4655,
      "step": 6624
    },
    {
      "epoch": 9.227019498607243,
      "grad_norm": 2.149198532104492,
      "learning_rate": 5.521603183627061e-06,
      "loss": 0.6658,
      "step": 6625
    },
    {
      "epoch": 9.22841225626741,
      "grad_norm": 1.7805689573287964,
      "learning_rate": 5.511654349061967e-06,
      "loss": 0.4119,
      "step": 6626
    },
    {
      "epoch": 9.229805013927576,
      "grad_norm": 1.9849166870117188,
      "learning_rate": 5.5017055144968726e-06,
      "loss": 0.5005,
      "step": 6627
    },
    {
      "epoch": 9.231197771587743,
      "grad_norm": 2.055591344833374,
      "learning_rate": 5.491756679931779e-06,
      "loss": 0.5138,
      "step": 6628
    },
    {
      "epoch": 9.232590529247911,
      "grad_norm": 1.8095927238464355,
      "learning_rate": 5.481807845366685e-06,
      "loss": 0.4968,
      "step": 6629
    },
    {
      "epoch": 9.233983286908078,
      "grad_norm": 1.74369478225708,
      "learning_rate": 5.4718590108015916e-06,
      "loss": 0.471,
      "step": 6630
    },
    {
      "epoch": 9.235376044568245,
      "grad_norm": 2.1644845008850098,
      "learning_rate": 5.461910176236497e-06,
      "loss": 0.4978,
      "step": 6631
    },
    {
      "epoch": 9.236768802228413,
      "grad_norm": 1.9678279161453247,
      "learning_rate": 5.451961341671403e-06,
      "loss": 0.4752,
      "step": 6632
    },
    {
      "epoch": 9.23816155988858,
      "grad_norm": 2.2192115783691406,
      "learning_rate": 5.4420125071063106e-06,
      "loss": 0.4683,
      "step": 6633
    },
    {
      "epoch": 9.239554317548746,
      "grad_norm": 2.171332836151123,
      "learning_rate": 5.432063672541216e-06,
      "loss": 0.6633,
      "step": 6634
    },
    {
      "epoch": 9.240947075208913,
      "grad_norm": 2.0996429920196533,
      "learning_rate": 5.422114837976122e-06,
      "loss": 0.5084,
      "step": 6635
    },
    {
      "epoch": 9.242339832869082,
      "grad_norm": 7.0336594581604,
      "learning_rate": 5.412166003411029e-06,
      "loss": 0.5198,
      "step": 6636
    },
    {
      "epoch": 9.243732590529248,
      "grad_norm": 2.24898099899292,
      "learning_rate": 5.4022171688459345e-06,
      "loss": 0.5915,
      "step": 6637
    },
    {
      "epoch": 9.245125348189415,
      "grad_norm": 1.914724588394165,
      "learning_rate": 5.39226833428084e-06,
      "loss": 0.5765,
      "step": 6638
    },
    {
      "epoch": 9.246518105849582,
      "grad_norm": 2.361132860183716,
      "learning_rate": 5.382319499715748e-06,
      "loss": 0.6781,
      "step": 6639
    },
    {
      "epoch": 9.24791086350975,
      "grad_norm": 2.349442720413208,
      "learning_rate": 5.3723706651506535e-06,
      "loss": 0.7101,
      "step": 6640
    },
    {
      "epoch": 9.249303621169917,
      "grad_norm": 2.002004384994507,
      "learning_rate": 5.36242183058556e-06,
      "loss": 0.5796,
      "step": 6641
    },
    {
      "epoch": 9.250696378830083,
      "grad_norm": 2.1309149265289307,
      "learning_rate": 5.352472996020466e-06,
      "loss": 0.5567,
      "step": 6642
    },
    {
      "epoch": 9.25208913649025,
      "grad_norm": 2.001441478729248,
      "learning_rate": 5.342524161455372e-06,
      "loss": 0.4673,
      "step": 6643
    },
    {
      "epoch": 9.253481894150418,
      "grad_norm": 2.335261821746826,
      "learning_rate": 5.332575326890278e-06,
      "loss": 0.596,
      "step": 6644
    },
    {
      "epoch": 9.254874651810585,
      "grad_norm": 2.1615543365478516,
      "learning_rate": 5.322626492325184e-06,
      "loss": 0.5647,
      "step": 6645
    },
    {
      "epoch": 9.256267409470752,
      "grad_norm": 2.391345977783203,
      "learning_rate": 5.31267765776009e-06,
      "loss": 0.7754,
      "step": 6646
    },
    {
      "epoch": 9.257660167130918,
      "grad_norm": 2.632402181625366,
      "learning_rate": 5.302728823194997e-06,
      "loss": 0.6171,
      "step": 6647
    },
    {
      "epoch": 9.259052924791087,
      "grad_norm": 2.426246404647827,
      "learning_rate": 5.292779988629903e-06,
      "loss": 0.7527,
      "step": 6648
    },
    {
      "epoch": 9.260445682451254,
      "grad_norm": 1.877597451210022,
      "learning_rate": 5.282831154064809e-06,
      "loss": 0.4197,
      "step": 6649
    },
    {
      "epoch": 9.26183844011142,
      "grad_norm": 2.0923333168029785,
      "learning_rate": 5.2728823194997154e-06,
      "loss": 0.5272,
      "step": 6650
    },
    {
      "epoch": 9.263231197771589,
      "grad_norm": 2.482633352279663,
      "learning_rate": 5.262933484934621e-06,
      "loss": 0.7377,
      "step": 6651
    },
    {
      "epoch": 9.264623955431755,
      "grad_norm": 2.743605136871338,
      "learning_rate": 5.252984650369527e-06,
      "loss": 0.6502,
      "step": 6652
    },
    {
      "epoch": 9.266016713091922,
      "grad_norm": 1.9745889902114868,
      "learning_rate": 5.2430358158044344e-06,
      "loss": 0.6234,
      "step": 6653
    },
    {
      "epoch": 9.267409470752089,
      "grad_norm": 1.8544808626174927,
      "learning_rate": 5.23308698123934e-06,
      "loss": 0.4923,
      "step": 6654
    },
    {
      "epoch": 9.268802228412257,
      "grad_norm": 2.3628005981445312,
      "learning_rate": 5.223138146674247e-06,
      "loss": 0.7013,
      "step": 6655
    },
    {
      "epoch": 9.270194986072424,
      "grad_norm": 2.123011827468872,
      "learning_rate": 5.213189312109153e-06,
      "loss": 0.6657,
      "step": 6656
    },
    {
      "epoch": 9.27158774373259,
      "grad_norm": 2.0028347969055176,
      "learning_rate": 5.203240477544058e-06,
      "loss": 0.4596,
      "step": 6657
    },
    {
      "epoch": 9.272980501392757,
      "grad_norm": 2.028984546661377,
      "learning_rate": 5.193291642978965e-06,
      "loss": 0.5634,
      "step": 6658
    },
    {
      "epoch": 9.274373259052926,
      "grad_norm": 2.5099196434020996,
      "learning_rate": 5.183342808413871e-06,
      "loss": 0.8089,
      "step": 6659
    },
    {
      "epoch": 9.275766016713092,
      "grad_norm": 2.106228828430176,
      "learning_rate": 5.1733939738487765e-06,
      "loss": 0.5925,
      "step": 6660
    },
    {
      "epoch": 9.277158774373259,
      "grad_norm": 1.9736827611923218,
      "learning_rate": 5.163445139283684e-06,
      "loss": 0.5531,
      "step": 6661
    },
    {
      "epoch": 9.278551532033426,
      "grad_norm": 2.2319600582122803,
      "learning_rate": 5.15349630471859e-06,
      "loss": 0.613,
      "step": 6662
    },
    {
      "epoch": 9.279944289693594,
      "grad_norm": 2.1375272274017334,
      "learning_rate": 5.1435474701534955e-06,
      "loss": 0.594,
      "step": 6663
    },
    {
      "epoch": 9.28133704735376,
      "grad_norm": 5.751987934112549,
      "learning_rate": 5.133598635588402e-06,
      "loss": 0.4896,
      "step": 6664
    },
    {
      "epoch": 9.282729805013927,
      "grad_norm": 2.2547049522399902,
      "learning_rate": 5.123649801023308e-06,
      "loss": 0.6529,
      "step": 6665
    },
    {
      "epoch": 9.284122562674094,
      "grad_norm": 1.929680585861206,
      "learning_rate": 5.1137009664582145e-06,
      "loss": 0.4584,
      "step": 6666
    },
    {
      "epoch": 9.285515320334262,
      "grad_norm": 2.387620210647583,
      "learning_rate": 5.103752131893121e-06,
      "loss": 0.545,
      "step": 6667
    },
    {
      "epoch": 9.286908077994429,
      "grad_norm": 1.8755158185958862,
      "learning_rate": 5.093803297328027e-06,
      "loss": 0.4759,
      "step": 6668
    },
    {
      "epoch": 9.288300835654596,
      "grad_norm": 2.2386081218719482,
      "learning_rate": 5.0838544627629335e-06,
      "loss": 0.5903,
      "step": 6669
    },
    {
      "epoch": 9.289693593314762,
      "grad_norm": 2.1858484745025635,
      "learning_rate": 5.073905628197839e-06,
      "loss": 0.4471,
      "step": 6670
    },
    {
      "epoch": 9.29108635097493,
      "grad_norm": 2.3394856452941895,
      "learning_rate": 5.063956793632745e-06,
      "loss": 0.5806,
      "step": 6671
    },
    {
      "epoch": 9.292479108635098,
      "grad_norm": 2.412019729614258,
      "learning_rate": 5.054007959067652e-06,
      "loss": 0.6955,
      "step": 6672
    },
    {
      "epoch": 9.293871866295264,
      "grad_norm": 1.8427023887634277,
      "learning_rate": 5.0440591245025574e-06,
      "loss": 0.5396,
      "step": 6673
    },
    {
      "epoch": 9.295264623955433,
      "grad_norm": 2.1877055168151855,
      "learning_rate": 5.034110289937463e-06,
      "loss": 0.5349,
      "step": 6674
    },
    {
      "epoch": 9.2966573816156,
      "grad_norm": 2.2356770038604736,
      "learning_rate": 5.024161455372371e-06,
      "loss": 0.6146,
      "step": 6675
    },
    {
      "epoch": 9.298050139275766,
      "grad_norm": 2.124812126159668,
      "learning_rate": 5.0142126208072764e-06,
      "loss": 0.5767,
      "step": 6676
    },
    {
      "epoch": 9.299442896935933,
      "grad_norm": 2.352794647216797,
      "learning_rate": 5.004263786242183e-06,
      "loss": 0.66,
      "step": 6677
    },
    {
      "epoch": 9.300835654596101,
      "grad_norm": 2.1764092445373535,
      "learning_rate": 4.994314951677089e-06,
      "loss": 0.7739,
      "step": 6678
    },
    {
      "epoch": 9.302228412256268,
      "grad_norm": 2.2855031490325928,
      "learning_rate": 4.984366117111995e-06,
      "loss": 0.6507,
      "step": 6679
    },
    {
      "epoch": 9.303621169916434,
      "grad_norm": 1.9761897325515747,
      "learning_rate": 4.974417282546901e-06,
      "loss": 0.6233,
      "step": 6680
    },
    {
      "epoch": 9.305013927576601,
      "grad_norm": 2.1757235527038574,
      "learning_rate": 4.964468447981808e-06,
      "loss": 0.5873,
      "step": 6681
    },
    {
      "epoch": 9.30640668523677,
      "grad_norm": 1.8421281576156616,
      "learning_rate": 4.954519613416714e-06,
      "loss": 0.54,
      "step": 6682
    },
    {
      "epoch": 9.307799442896936,
      "grad_norm": 1.8912826776504517,
      "learning_rate": 4.94457077885162e-06,
      "loss": 0.4241,
      "step": 6683
    },
    {
      "epoch": 9.309192200557103,
      "grad_norm": 2.3895020484924316,
      "learning_rate": 4.934621944286526e-06,
      "loss": 0.6684,
      "step": 6684
    },
    {
      "epoch": 9.31058495821727,
      "grad_norm": 2.4576313495635986,
      "learning_rate": 4.924673109721432e-06,
      "loss": 0.6911,
      "step": 6685
    },
    {
      "epoch": 9.311977715877438,
      "grad_norm": 2.165168523788452,
      "learning_rate": 4.914724275156338e-06,
      "loss": 0.6065,
      "step": 6686
    },
    {
      "epoch": 9.313370473537605,
      "grad_norm": 1.9459168910980225,
      "learning_rate": 4.904775440591244e-06,
      "loss": 0.443,
      "step": 6687
    },
    {
      "epoch": 9.314763231197771,
      "grad_norm": 2.1418256759643555,
      "learning_rate": 4.894826606026152e-06,
      "loss": 0.4688,
      "step": 6688
    },
    {
      "epoch": 9.316155988857938,
      "grad_norm": 2.76389741897583,
      "learning_rate": 4.884877771461057e-06,
      "loss": 0.9413,
      "step": 6689
    },
    {
      "epoch": 9.317548746518106,
      "grad_norm": 2.0555858612060547,
      "learning_rate": 4.874928936895963e-06,
      "loss": 0.5609,
      "step": 6690
    },
    {
      "epoch": 9.318941504178273,
      "grad_norm": 2.319624185562134,
      "learning_rate": 4.86498010233087e-06,
      "loss": 0.5781,
      "step": 6691
    },
    {
      "epoch": 9.32033426183844,
      "grad_norm": 1.983708381652832,
      "learning_rate": 4.8550312677657755e-06,
      "loss": 0.5228,
      "step": 6692
    },
    {
      "epoch": 9.321727019498606,
      "grad_norm": 2.338756799697876,
      "learning_rate": 4.845082433200681e-06,
      "loss": 0.6649,
      "step": 6693
    },
    {
      "epoch": 9.323119777158775,
      "grad_norm": 2.617812156677246,
      "learning_rate": 4.835133598635588e-06,
      "loss": 0.6386,
      "step": 6694
    },
    {
      "epoch": 9.324512534818941,
      "grad_norm": 1.9342000484466553,
      "learning_rate": 4.8251847640704945e-06,
      "loss": 0.5451,
      "step": 6695
    },
    {
      "epoch": 9.325905292479108,
      "grad_norm": 2.455490827560425,
      "learning_rate": 4.8152359295054e-06,
      "loss": 0.6154,
      "step": 6696
    },
    {
      "epoch": 9.327298050139277,
      "grad_norm": 2.2828080654144287,
      "learning_rate": 4.805287094940307e-06,
      "loss": 0.6426,
      "step": 6697
    },
    {
      "epoch": 9.328690807799443,
      "grad_norm": 2.1832237243652344,
      "learning_rate": 4.795338260375213e-06,
      "loss": 0.5574,
      "step": 6698
    },
    {
      "epoch": 9.33008356545961,
      "grad_norm": 2.169520378112793,
      "learning_rate": 4.785389425810119e-06,
      "loss": 0.666,
      "step": 6699
    },
    {
      "epoch": 9.331476323119777,
      "grad_norm": 2.042065143585205,
      "learning_rate": 4.775440591245025e-06,
      "loss": 0.4648,
      "step": 6700
    },
    {
      "epoch": 9.332869080779945,
      "grad_norm": 2.5215961933135986,
      "learning_rate": 4.765491756679931e-06,
      "loss": 0.6618,
      "step": 6701
    },
    {
      "epoch": 9.334261838440112,
      "grad_norm": 2.3360421657562256,
      "learning_rate": 4.755542922114838e-06,
      "loss": 0.628,
      "step": 6702
    },
    {
      "epoch": 9.335654596100278,
      "grad_norm": 3.0754506587982178,
      "learning_rate": 4.745594087549744e-06,
      "loss": 0.5901,
      "step": 6703
    },
    {
      "epoch": 9.337047353760445,
      "grad_norm": 2.697869062423706,
      "learning_rate": 4.73564525298465e-06,
      "loss": 0.545,
      "step": 6704
    },
    {
      "epoch": 9.338440111420613,
      "grad_norm": 1.9758132696151733,
      "learning_rate": 4.7256964184195564e-06,
      "loss": 0.5344,
      "step": 6705
    },
    {
      "epoch": 9.33983286908078,
      "grad_norm": 2.1251981258392334,
      "learning_rate": 4.715747583854462e-06,
      "loss": 0.3737,
      "step": 6706
    },
    {
      "epoch": 9.341225626740947,
      "grad_norm": 1.9647719860076904,
      "learning_rate": 4.705798749289368e-06,
      "loss": 0.4168,
      "step": 6707
    },
    {
      "epoch": 9.342618384401113,
      "grad_norm": 1.8448026180267334,
      "learning_rate": 4.695849914724275e-06,
      "loss": 0.4246,
      "step": 6708
    },
    {
      "epoch": 9.344011142061282,
      "grad_norm": 2.073429584503174,
      "learning_rate": 4.68590108015918e-06,
      "loss": 0.4977,
      "step": 6709
    },
    {
      "epoch": 9.345403899721449,
      "grad_norm": 2.2559731006622314,
      "learning_rate": 4.675952245594088e-06,
      "loss": 0.5836,
      "step": 6710
    },
    {
      "epoch": 9.346796657381615,
      "grad_norm": 1.941716194152832,
      "learning_rate": 4.666003411028994e-06,
      "loss": 0.5033,
      "step": 6711
    },
    {
      "epoch": 9.348189415041782,
      "grad_norm": 1.8942596912384033,
      "learning_rate": 4.656054576463899e-06,
      "loss": 0.4844,
      "step": 6712
    },
    {
      "epoch": 9.34958217270195,
      "grad_norm": 2.08406662940979,
      "learning_rate": 4.646105741898806e-06,
      "loss": 0.5139,
      "step": 6713
    },
    {
      "epoch": 9.350974930362117,
      "grad_norm": 2.2924787998199463,
      "learning_rate": 4.636156907333712e-06,
      "loss": 0.5734,
      "step": 6714
    },
    {
      "epoch": 9.352367688022284,
      "grad_norm": 2.0736114978790283,
      "learning_rate": 4.6262080727686175e-06,
      "loss": 0.5575,
      "step": 6715
    },
    {
      "epoch": 9.35376044568245,
      "grad_norm": 1.6771340370178223,
      "learning_rate": 4.616259238203525e-06,
      "loss": 0.3436,
      "step": 6716
    },
    {
      "epoch": 9.355153203342619,
      "grad_norm": 2.115764856338501,
      "learning_rate": 4.606310403638431e-06,
      "loss": 0.4658,
      "step": 6717
    },
    {
      "epoch": 9.356545961002785,
      "grad_norm": 2.183945894241333,
      "learning_rate": 4.5963615690733365e-06,
      "loss": 0.6343,
      "step": 6718
    },
    {
      "epoch": 9.357938718662952,
      "grad_norm": 2.2778444290161133,
      "learning_rate": 4.586412734508243e-06,
      "loss": 0.7001,
      "step": 6719
    },
    {
      "epoch": 9.35933147632312,
      "grad_norm": 2.308894395828247,
      "learning_rate": 4.576463899943149e-06,
      "loss": 0.6485,
      "step": 6720
    },
    {
      "epoch": 9.360724233983287,
      "grad_norm": 2.110804796218872,
      "learning_rate": 4.5665150653780555e-06,
      "loss": 0.5745,
      "step": 6721
    },
    {
      "epoch": 9.362116991643454,
      "grad_norm": 1.934041976928711,
      "learning_rate": 4.556566230812961e-06,
      "loss": 0.5032,
      "step": 6722
    },
    {
      "epoch": 9.36350974930362,
      "grad_norm": 1.87200129032135,
      "learning_rate": 4.546617396247867e-06,
      "loss": 0.4351,
      "step": 6723
    },
    {
      "epoch": 9.364902506963789,
      "grad_norm": 1.953367829322815,
      "learning_rate": 4.5366685616827745e-06,
      "loss": 0.5016,
      "step": 6724
    },
    {
      "epoch": 9.366295264623956,
      "grad_norm": 2.0533089637756348,
      "learning_rate": 4.52671972711768e-06,
      "loss": 0.6008,
      "step": 6725
    },
    {
      "epoch": 9.367688022284122,
      "grad_norm": 2.586660385131836,
      "learning_rate": 4.516770892552586e-06,
      "loss": 0.6865,
      "step": 6726
    },
    {
      "epoch": 9.369080779944289,
      "grad_norm": 2.1469037532806396,
      "learning_rate": 4.506822057987493e-06,
      "loss": 0.5693,
      "step": 6727
    },
    {
      "epoch": 9.370473537604457,
      "grad_norm": 2.256702423095703,
      "learning_rate": 4.4968732234223984e-06,
      "loss": 0.5153,
      "step": 6728
    },
    {
      "epoch": 9.371866295264624,
      "grad_norm": 1.728001356124878,
      "learning_rate": 4.486924388857304e-06,
      "loss": 0.4704,
      "step": 6729
    },
    {
      "epoch": 9.37325905292479,
      "grad_norm": 2.113049268722534,
      "learning_rate": 4.476975554292212e-06,
      "loss": 0.5075,
      "step": 6730
    },
    {
      "epoch": 9.374651810584957,
      "grad_norm": 2.0204129219055176,
      "learning_rate": 4.4670267197271174e-06,
      "loss": 0.4786,
      "step": 6731
    },
    {
      "epoch": 9.376044568245126,
      "grad_norm": 1.8315682411193848,
      "learning_rate": 4.457077885162024e-06,
      "loss": 0.4078,
      "step": 6732
    },
    {
      "epoch": 9.377437325905293,
      "grad_norm": 2.0842456817626953,
      "learning_rate": 4.44712905059693e-06,
      "loss": 0.6208,
      "step": 6733
    },
    {
      "epoch": 9.37883008356546,
      "grad_norm": 2.3982343673706055,
      "learning_rate": 4.437180216031836e-06,
      "loss": 0.6758,
      "step": 6734
    },
    {
      "epoch": 9.380222841225626,
      "grad_norm": 1.9927856922149658,
      "learning_rate": 4.427231381466742e-06,
      "loss": 0.4766,
      "step": 6735
    },
    {
      "epoch": 9.381615598885794,
      "grad_norm": 2.199819803237915,
      "learning_rate": 4.417282546901648e-06,
      "loss": 0.6303,
      "step": 6736
    },
    {
      "epoch": 9.383008356545961,
      "grad_norm": 1.983742594718933,
      "learning_rate": 4.407333712336554e-06,
      "loss": 0.5537,
      "step": 6737
    },
    {
      "epoch": 9.384401114206128,
      "grad_norm": 2.1473169326782227,
      "learning_rate": 4.397384877771461e-06,
      "loss": 0.6254,
      "step": 6738
    },
    {
      "epoch": 9.385793871866296,
      "grad_norm": 2.208550453186035,
      "learning_rate": 4.387436043206367e-06,
      "loss": 0.5505,
      "step": 6739
    },
    {
      "epoch": 9.387186629526463,
      "grad_norm": 2.168860673904419,
      "learning_rate": 4.377487208641273e-06,
      "loss": 0.4925,
      "step": 6740
    },
    {
      "epoch": 9.38857938718663,
      "grad_norm": 2.385719060897827,
      "learning_rate": 4.367538374076179e-06,
      "loss": 0.6254,
      "step": 6741
    },
    {
      "epoch": 9.389972144846796,
      "grad_norm": 1.8590692281723022,
      "learning_rate": 4.357589539511085e-06,
      "loss": 0.5158,
      "step": 6742
    },
    {
      "epoch": 9.391364902506965,
      "grad_norm": 2.19563364982605,
      "learning_rate": 4.347640704945992e-06,
      "loss": 0.5806,
      "step": 6743
    },
    {
      "epoch": 9.392757660167131,
      "grad_norm": 1.9516217708587646,
      "learning_rate": 4.337691870380898e-06,
      "loss": 0.5567,
      "step": 6744
    },
    {
      "epoch": 9.394150417827298,
      "grad_norm": 2.4490139484405518,
      "learning_rate": 4.327743035815804e-06,
      "loss": 0.617,
      "step": 6745
    },
    {
      "epoch": 9.395543175487465,
      "grad_norm": 1.9458948373794556,
      "learning_rate": 4.31779420125071e-06,
      "loss": 0.5444,
      "step": 6746
    },
    {
      "epoch": 9.396935933147633,
      "grad_norm": 2.157991647720337,
      "learning_rate": 4.3078453666856165e-06,
      "loss": 0.5913,
      "step": 6747
    },
    {
      "epoch": 9.3983286908078,
      "grad_norm": 2.0901591777801514,
      "learning_rate": 4.297896532120523e-06,
      "loss": 0.4922,
      "step": 6748
    },
    {
      "epoch": 9.399721448467966,
      "grad_norm": 2.1302778720855713,
      "learning_rate": 4.287947697555429e-06,
      "loss": 0.5691,
      "step": 6749
    },
    {
      "epoch": 9.401114206128133,
      "grad_norm": 2.671572685241699,
      "learning_rate": 4.277998862990335e-06,
      "loss": 0.7691,
      "step": 6750
    },
    {
      "epoch": 9.402506963788301,
      "grad_norm": 1.9120711088180542,
      "learning_rate": 4.268050028425241e-06,
      "loss": 0.5165,
      "step": 6751
    },
    {
      "epoch": 9.403899721448468,
      "grad_norm": 2.2509145736694336,
      "learning_rate": 4.258101193860148e-06,
      "loss": 0.6263,
      "step": 6752
    },
    {
      "epoch": 9.405292479108635,
      "grad_norm": 2.2784059047698975,
      "learning_rate": 4.248152359295054e-06,
      "loss": 0.5518,
      "step": 6753
    },
    {
      "epoch": 9.406685236768801,
      "grad_norm": 1.7827047109603882,
      "learning_rate": 4.2382035247299594e-06,
      "loss": 0.426,
      "step": 6754
    },
    {
      "epoch": 9.40807799442897,
      "grad_norm": 2.2618050575256348,
      "learning_rate": 4.228254690164866e-06,
      "loss": 0.6541,
      "step": 6755
    },
    {
      "epoch": 9.409470752089137,
      "grad_norm": 2.3271522521972656,
      "learning_rate": 4.218305855599772e-06,
      "loss": 0.5396,
      "step": 6756
    },
    {
      "epoch": 9.410863509749303,
      "grad_norm": 2.2148826122283936,
      "learning_rate": 4.2083570210346784e-06,
      "loss": 0.6406,
      "step": 6757
    },
    {
      "epoch": 9.412256267409472,
      "grad_norm": 2.005791425704956,
      "learning_rate": 4.198408186469585e-06,
      "loss": 0.4294,
      "step": 6758
    },
    {
      "epoch": 9.413649025069638,
      "grad_norm": 1.8397232294082642,
      "learning_rate": 4.188459351904491e-06,
      "loss": 0.4662,
      "step": 6759
    },
    {
      "epoch": 9.415041782729805,
      "grad_norm": 1.7420717477798462,
      "learning_rate": 4.178510517339397e-06,
      "loss": 0.3784,
      "step": 6760
    },
    {
      "epoch": 9.416434540389972,
      "grad_norm": 2.056633234024048,
      "learning_rate": 4.168561682774303e-06,
      "loss": 0.5443,
      "step": 6761
    },
    {
      "epoch": 9.41782729805014,
      "grad_norm": 1.816161036491394,
      "learning_rate": 4.15861284820921e-06,
      "loss": 0.4245,
      "step": 6762
    },
    {
      "epoch": 9.419220055710307,
      "grad_norm": 2.3265507221221924,
      "learning_rate": 4.148664013644116e-06,
      "loss": 0.6085,
      "step": 6763
    },
    {
      "epoch": 9.420612813370473,
      "grad_norm": 1.9604753255844116,
      "learning_rate": 4.138715179079021e-06,
      "loss": 0.5458,
      "step": 6764
    },
    {
      "epoch": 9.42200557103064,
      "grad_norm": 1.7473949193954468,
      "learning_rate": 4.128766344513928e-06,
      "loss": 0.4521,
      "step": 6765
    },
    {
      "epoch": 9.423398328690809,
      "grad_norm": 1.8409470319747925,
      "learning_rate": 4.118817509948835e-06,
      "loss": 0.4751,
      "step": 6766
    },
    {
      "epoch": 9.424791086350975,
      "grad_norm": 2.09124755859375,
      "learning_rate": 4.10886867538374e-06,
      "loss": 0.5173,
      "step": 6767
    },
    {
      "epoch": 9.426183844011142,
      "grad_norm": 1.998836636543274,
      "learning_rate": 4.098919840818646e-06,
      "loss": 0.6055,
      "step": 6768
    },
    {
      "epoch": 9.427576601671309,
      "grad_norm": 2.2272489070892334,
      "learning_rate": 4.088971006253553e-06,
      "loss": 0.5947,
      "step": 6769
    },
    {
      "epoch": 9.428969359331477,
      "grad_norm": 2.0613739490509033,
      "learning_rate": 4.079022171688459e-06,
      "loss": 0.513,
      "step": 6770
    },
    {
      "epoch": 9.430362116991644,
      "grad_norm": 2.181584119796753,
      "learning_rate": 4.069073337123365e-06,
      "loss": 0.5332,
      "step": 6771
    },
    {
      "epoch": 9.43175487465181,
      "grad_norm": 2.1950395107269287,
      "learning_rate": 4.059124502558272e-06,
      "loss": 0.6239,
      "step": 6772
    },
    {
      "epoch": 9.433147632311977,
      "grad_norm": 2.2424991130828857,
      "learning_rate": 4.0491756679931775e-06,
      "loss": 0.5076,
      "step": 6773
    },
    {
      "epoch": 9.434540389972145,
      "grad_norm": 2.6309731006622314,
      "learning_rate": 4.039226833428083e-06,
      "loss": 0.7419,
      "step": 6774
    },
    {
      "epoch": 9.435933147632312,
      "grad_norm": 2.420696258544922,
      "learning_rate": 4.02927799886299e-06,
      "loss": 0.651,
      "step": 6775
    },
    {
      "epoch": 9.437325905292479,
      "grad_norm": 2.2619361877441406,
      "learning_rate": 4.0193291642978965e-06,
      "loss": 0.6268,
      "step": 6776
    },
    {
      "epoch": 9.438718662952645,
      "grad_norm": 2.0040783882141113,
      "learning_rate": 4.009380329732802e-06,
      "loss": 0.4365,
      "step": 6777
    },
    {
      "epoch": 9.440111420612814,
      "grad_norm": 2.1652371883392334,
      "learning_rate": 3.999431495167708e-06,
      "loss": 0.6763,
      "step": 6778
    },
    {
      "epoch": 9.44150417827298,
      "grad_norm": 1.946468472480774,
      "learning_rate": 3.989482660602615e-06,
      "loss": 0.531,
      "step": 6779
    },
    {
      "epoch": 9.442896935933147,
      "grad_norm": 2.0985682010650635,
      "learning_rate": 3.979533826037521e-06,
      "loss": 0.5662,
      "step": 6780
    },
    {
      "epoch": 9.444289693593316,
      "grad_norm": 2.9044721126556396,
      "learning_rate": 3.969584991472427e-06,
      "loss": 0.7111,
      "step": 6781
    },
    {
      "epoch": 9.445682451253482,
      "grad_norm": 1.901016354560852,
      "learning_rate": 3.959636156907333e-06,
      "loss": 0.424,
      "step": 6782
    },
    {
      "epoch": 9.447075208913649,
      "grad_norm": 2.362870693206787,
      "learning_rate": 3.9496873223422395e-06,
      "loss": 0.4705,
      "step": 6783
    },
    {
      "epoch": 9.448467966573816,
      "grad_norm": 1.503623366355896,
      "learning_rate": 3.939738487777146e-06,
      "loss": 0.3077,
      "step": 6784
    },
    {
      "epoch": 9.449860724233984,
      "grad_norm": 1.8384147882461548,
      "learning_rate": 3.929789653212052e-06,
      "loss": 0.3329,
      "step": 6785
    },
    {
      "epoch": 9.45125348189415,
      "grad_norm": 1.9747810363769531,
      "learning_rate": 3.9198408186469585e-06,
      "loss": 0.5828,
      "step": 6786
    },
    {
      "epoch": 9.452646239554317,
      "grad_norm": 1.9929118156433105,
      "learning_rate": 3.909891984081864e-06,
      "loss": 0.4236,
      "step": 6787
    },
    {
      "epoch": 9.454038997214484,
      "grad_norm": 2.373485803604126,
      "learning_rate": 3.899943149516771e-06,
      "loss": 0.5345,
      "step": 6788
    },
    {
      "epoch": 9.455431754874652,
      "grad_norm": 1.934966802597046,
      "learning_rate": 3.889994314951677e-06,
      "loss": 0.5323,
      "step": 6789
    },
    {
      "epoch": 9.45682451253482,
      "grad_norm": 2.5214738845825195,
      "learning_rate": 3.880045480386583e-06,
      "loss": 0.7866,
      "step": 6790
    },
    {
      "epoch": 9.458217270194986,
      "grad_norm": 1.8218555450439453,
      "learning_rate": 3.870096645821489e-06,
      "loss": 0.4144,
      "step": 6791
    },
    {
      "epoch": 9.459610027855152,
      "grad_norm": 1.9606658220291138,
      "learning_rate": 3.860147811256396e-06,
      "loss": 0.5411,
      "step": 6792
    },
    {
      "epoch": 9.461002785515321,
      "grad_norm": 1.7190555334091187,
      "learning_rate": 3.850198976691301e-06,
      "loss": 0.4121,
      "step": 6793
    },
    {
      "epoch": 9.462395543175488,
      "grad_norm": 2.5075480937957764,
      "learning_rate": 3.840250142126208e-06,
      "loss": 0.5883,
      "step": 6794
    },
    {
      "epoch": 9.463788300835654,
      "grad_norm": 2.277167320251465,
      "learning_rate": 3.830301307561114e-06,
      "loss": 0.5809,
      "step": 6795
    },
    {
      "epoch": 9.465181058495821,
      "grad_norm": 1.7093855142593384,
      "learning_rate": 3.8203524729960195e-06,
      "loss": 0.4828,
      "step": 6796
    },
    {
      "epoch": 9.46657381615599,
      "grad_norm": 1.923637866973877,
      "learning_rate": 3.810403638430926e-06,
      "loss": 0.53,
      "step": 6797
    },
    {
      "epoch": 9.467966573816156,
      "grad_norm": 2.2378780841827393,
      "learning_rate": 3.8004548038658328e-06,
      "loss": 0.5775,
      "step": 6798
    },
    {
      "epoch": 9.469359331476323,
      "grad_norm": 2.268008232116699,
      "learning_rate": 3.790505969300739e-06,
      "loss": 0.533,
      "step": 6799
    },
    {
      "epoch": 9.47075208913649,
      "grad_norm": 2.2728209495544434,
      "learning_rate": 3.7805571347356447e-06,
      "loss": 0.6456,
      "step": 6800
    },
    {
      "epoch": 9.472144846796658,
      "grad_norm": 2.1043710708618164,
      "learning_rate": 3.770608300170551e-06,
      "loss": 0.5893,
      "step": 6801
    },
    {
      "epoch": 9.473537604456824,
      "grad_norm": 1.8332468271255493,
      "learning_rate": 3.7606594656054575e-06,
      "loss": 0.4217,
      "step": 6802
    },
    {
      "epoch": 9.474930362116991,
      "grad_norm": 1.8673499822616577,
      "learning_rate": 3.7507106310403637e-06,
      "loss": 0.5118,
      "step": 6803
    },
    {
      "epoch": 9.47632311977716,
      "grad_norm": 2.1703574657440186,
      "learning_rate": 3.7407617964752695e-06,
      "loss": 0.7188,
      "step": 6804
    },
    {
      "epoch": 9.477715877437326,
      "grad_norm": 2.5039427280426025,
      "learning_rate": 3.730812961910176e-06,
      "loss": 0.5681,
      "step": 6805
    },
    {
      "epoch": 9.479108635097493,
      "grad_norm": 2.1298906803131104,
      "learning_rate": 3.7208641273450823e-06,
      "loss": 0.4972,
      "step": 6806
    },
    {
      "epoch": 9.48050139275766,
      "grad_norm": 1.9694808721542358,
      "learning_rate": 3.710915292779988e-06,
      "loss": 0.4645,
      "step": 6807
    },
    {
      "epoch": 9.481894150417828,
      "grad_norm": 2.109868049621582,
      "learning_rate": 3.7009664582148943e-06,
      "loss": 0.5743,
      "step": 6808
    },
    {
      "epoch": 9.483286908077995,
      "grad_norm": 2.0913796424865723,
      "learning_rate": 3.691017623649801e-06,
      "loss": 0.5902,
      "step": 6809
    },
    {
      "epoch": 9.484679665738161,
      "grad_norm": 2.116060972213745,
      "learning_rate": 3.681068789084707e-06,
      "loss": 0.6449,
      "step": 6810
    },
    {
      "epoch": 9.486072423398328,
      "grad_norm": 2.1506569385528564,
      "learning_rate": 3.671119954519613e-06,
      "loss": 0.6338,
      "step": 6811
    },
    {
      "epoch": 9.487465181058496,
      "grad_norm": 2.254997730255127,
      "learning_rate": 3.6611711199545195e-06,
      "loss": 0.513,
      "step": 6812
    },
    {
      "epoch": 9.488857938718663,
      "grad_norm": 2.3468072414398193,
      "learning_rate": 3.6512222853894257e-06,
      "loss": 0.6028,
      "step": 6813
    },
    {
      "epoch": 9.49025069637883,
      "grad_norm": 1.951341152191162,
      "learning_rate": 3.6412734508243314e-06,
      "loss": 0.5671,
      "step": 6814
    },
    {
      "epoch": 9.491643454038996,
      "grad_norm": 2.2778260707855225,
      "learning_rate": 3.6313246162592376e-06,
      "loss": 0.4638,
      "step": 6815
    },
    {
      "epoch": 9.493036211699165,
      "grad_norm": 2.0231311321258545,
      "learning_rate": 3.6213757816941442e-06,
      "loss": 0.5473,
      "step": 6816
    },
    {
      "epoch": 9.494428969359332,
      "grad_norm": 2.116697072982788,
      "learning_rate": 3.6114269471290504e-06,
      "loss": 0.4566,
      "step": 6817
    },
    {
      "epoch": 9.495821727019498,
      "grad_norm": 1.9370697736740112,
      "learning_rate": 3.601478112563956e-06,
      "loss": 0.4776,
      "step": 6818
    },
    {
      "epoch": 9.497214484679665,
      "grad_norm": 1.903659462928772,
      "learning_rate": 3.591529277998863e-06,
      "loss": 0.507,
      "step": 6819
    },
    {
      "epoch": 9.498607242339833,
      "grad_norm": 2.2818992137908936,
      "learning_rate": 3.581580443433769e-06,
      "loss": 0.7046,
      "step": 6820
    },
    {
      "epoch": 9.5,
      "grad_norm": 2.027432918548584,
      "learning_rate": 3.571631608868675e-06,
      "loss": 0.4643,
      "step": 6821
    },
    {
      "epoch": 9.501392757660167,
      "grad_norm": 2.352668046951294,
      "learning_rate": 3.561682774303581e-06,
      "loss": 0.7195,
      "step": 6822
    },
    {
      "epoch": 9.502785515320333,
      "grad_norm": 2.2707459926605225,
      "learning_rate": 3.5517339397384876e-06,
      "loss": 0.6042,
      "step": 6823
    },
    {
      "epoch": 9.504178272980502,
      "grad_norm": 2.524899959564209,
      "learning_rate": 3.5417851051733938e-06,
      "loss": 0.6264,
      "step": 6824
    },
    {
      "epoch": 9.505571030640668,
      "grad_norm": 1.987490177154541,
      "learning_rate": 3.5318362706082995e-06,
      "loss": 0.5492,
      "step": 6825
    },
    {
      "epoch": 9.506963788300835,
      "grad_norm": 1.8484636545181274,
      "learning_rate": 3.521887436043206e-06,
      "loss": 0.472,
      "step": 6826
    },
    {
      "epoch": 9.508356545961004,
      "grad_norm": 2.245314836502075,
      "learning_rate": 3.5119386014781123e-06,
      "loss": 0.6465,
      "step": 6827
    },
    {
      "epoch": 9.50974930362117,
      "grad_norm": 2.340188503265381,
      "learning_rate": 3.5019897669130185e-06,
      "loss": 0.6109,
      "step": 6828
    },
    {
      "epoch": 9.511142061281337,
      "grad_norm": 2.700007200241089,
      "learning_rate": 3.4920409323479243e-06,
      "loss": 0.6385,
      "step": 6829
    },
    {
      "epoch": 9.512534818941504,
      "grad_norm": 1.9856232404708862,
      "learning_rate": 3.482092097782831e-06,
      "loss": 0.4791,
      "step": 6830
    },
    {
      "epoch": 9.513927576601672,
      "grad_norm": 1.9098279476165771,
      "learning_rate": 3.472143263217737e-06,
      "loss": 0.4022,
      "step": 6831
    },
    {
      "epoch": 9.515320334261839,
      "grad_norm": 2.438234567642212,
      "learning_rate": 3.4621944286526433e-06,
      "loss": 0.7368,
      "step": 6832
    },
    {
      "epoch": 9.516713091922005,
      "grad_norm": 1.7444934844970703,
      "learning_rate": 3.4522455940875495e-06,
      "loss": 0.5008,
      "step": 6833
    },
    {
      "epoch": 9.518105849582172,
      "grad_norm": 2.0556321144104004,
      "learning_rate": 3.4422967595224557e-06,
      "loss": 0.4713,
      "step": 6834
    },
    {
      "epoch": 9.51949860724234,
      "grad_norm": 2.0487747192382812,
      "learning_rate": 3.432347924957362e-06,
      "loss": 0.5607,
      "step": 6835
    },
    {
      "epoch": 9.520891364902507,
      "grad_norm": 2.1954009532928467,
      "learning_rate": 3.4223990903922677e-06,
      "loss": 0.5943,
      "step": 6836
    },
    {
      "epoch": 9.522284122562674,
      "grad_norm": 2.11663556098938,
      "learning_rate": 3.4124502558271743e-06,
      "loss": 0.5555,
      "step": 6837
    },
    {
      "epoch": 9.52367688022284,
      "grad_norm": 2.2933733463287354,
      "learning_rate": 3.4025014212620805e-06,
      "loss": 0.689,
      "step": 6838
    },
    {
      "epoch": 9.525069637883009,
      "grad_norm": 2.073500156402588,
      "learning_rate": 3.3925525866969867e-06,
      "loss": 0.4693,
      "step": 6839
    },
    {
      "epoch": 9.526462395543176,
      "grad_norm": 1.6908445358276367,
      "learning_rate": 3.382603752131893e-06,
      "loss": 0.4207,
      "step": 6840
    },
    {
      "epoch": 9.527855153203342,
      "grad_norm": 2.208913803100586,
      "learning_rate": 3.372654917566799e-06,
      "loss": 0.6036,
      "step": 6841
    },
    {
      "epoch": 9.52924791086351,
      "grad_norm": 2.4213709831237793,
      "learning_rate": 3.3627060830017052e-06,
      "loss": 0.6669,
      "step": 6842
    },
    {
      "epoch": 9.530640668523677,
      "grad_norm": 2.1149485111236572,
      "learning_rate": 3.352757248436612e-06,
      "loss": 0.5837,
      "step": 6843
    },
    {
      "epoch": 9.532033426183844,
      "grad_norm": 1.683316707611084,
      "learning_rate": 3.3428084138715176e-06,
      "loss": 0.3735,
      "step": 6844
    },
    {
      "epoch": 9.53342618384401,
      "grad_norm": 2.1535401344299316,
      "learning_rate": 3.332859579306424e-06,
      "loss": 0.5639,
      "step": 6845
    },
    {
      "epoch": 9.534818941504179,
      "grad_norm": 2.3267405033111572,
      "learning_rate": 3.32291074474133e-06,
      "loss": 0.6312,
      "step": 6846
    },
    {
      "epoch": 9.536211699164346,
      "grad_norm": 2.0664336681365967,
      "learning_rate": 3.312961910176236e-06,
      "loss": 0.6014,
      "step": 6847
    },
    {
      "epoch": 9.537604456824512,
      "grad_norm": 2.2134811878204346,
      "learning_rate": 3.3030130756111424e-06,
      "loss": 0.548,
      "step": 6848
    },
    {
      "epoch": 9.538997214484679,
      "grad_norm": 2.397930145263672,
      "learning_rate": 3.2930642410460486e-06,
      "loss": 0.7653,
      "step": 6849
    },
    {
      "epoch": 9.540389972144848,
      "grad_norm": 1.931451439857483,
      "learning_rate": 3.283115406480955e-06,
      "loss": 0.5529,
      "step": 6850
    },
    {
      "epoch": 9.541782729805014,
      "grad_norm": 2.0318799018859863,
      "learning_rate": 3.273166571915861e-06,
      "loss": 0.3954,
      "step": 6851
    },
    {
      "epoch": 9.54317548746518,
      "grad_norm": 2.2449874877929688,
      "learning_rate": 3.263217737350767e-06,
      "loss": 0.5723,
      "step": 6852
    },
    {
      "epoch": 9.544568245125348,
      "grad_norm": 1.9648016691207886,
      "learning_rate": 3.2532689027856734e-06,
      "loss": 0.6213,
      "step": 6853
    },
    {
      "epoch": 9.545961002785516,
      "grad_norm": 1.7888097763061523,
      "learning_rate": 3.24332006822058e-06,
      "loss": 0.4491,
      "step": 6854
    },
    {
      "epoch": 9.547353760445683,
      "grad_norm": 2.4542691707611084,
      "learning_rate": 3.2333712336554857e-06,
      "loss": 0.6643,
      "step": 6855
    },
    {
      "epoch": 9.54874651810585,
      "grad_norm": 2.0693256855010986,
      "learning_rate": 3.223422399090392e-06,
      "loss": 0.4155,
      "step": 6856
    },
    {
      "epoch": 9.550139275766016,
      "grad_norm": 1.9995914697647095,
      "learning_rate": 3.213473564525298e-06,
      "loss": 0.492,
      "step": 6857
    },
    {
      "epoch": 9.551532033426184,
      "grad_norm": 2.548415184020996,
      "learning_rate": 3.2035247299602043e-06,
      "loss": 0.6386,
      "step": 6858
    },
    {
      "epoch": 9.552924791086351,
      "grad_norm": 2.3077518939971924,
      "learning_rate": 3.1935758953951105e-06,
      "loss": 0.7407,
      "step": 6859
    },
    {
      "epoch": 9.554317548746518,
      "grad_norm": 2.6892733573913574,
      "learning_rate": 3.1836270608300167e-06,
      "loss": 0.5772,
      "step": 6860
    },
    {
      "epoch": 9.555710306406684,
      "grad_norm": 2.148426055908203,
      "learning_rate": 3.1736782262649233e-06,
      "loss": 0.5233,
      "step": 6861
    },
    {
      "epoch": 9.557103064066853,
      "grad_norm": 1.9864565134048462,
      "learning_rate": 3.163729391699829e-06,
      "loss": 0.5015,
      "step": 6862
    },
    {
      "epoch": 9.55849582172702,
      "grad_norm": 2.461149215698242,
      "learning_rate": 3.1537805571347353e-06,
      "loss": 0.8165,
      "step": 6863
    },
    {
      "epoch": 9.559888579387186,
      "grad_norm": 2.3243892192840576,
      "learning_rate": 3.1438317225696415e-06,
      "loss": 0.6147,
      "step": 6864
    },
    {
      "epoch": 9.561281337047355,
      "grad_norm": 2.165677785873413,
      "learning_rate": 3.1338828880045477e-06,
      "loss": 0.4625,
      "step": 6865
    },
    {
      "epoch": 9.562674094707521,
      "grad_norm": 1.836151123046875,
      "learning_rate": 3.123934053439454e-06,
      "loss": 0.3987,
      "step": 6866
    },
    {
      "epoch": 9.564066852367688,
      "grad_norm": 2.061715602874756,
      "learning_rate": 3.11398521887436e-06,
      "loss": 0.609,
      "step": 6867
    },
    {
      "epoch": 9.565459610027855,
      "grad_norm": 1.9791380167007446,
      "learning_rate": 3.1040363843092667e-06,
      "loss": 0.5075,
      "step": 6868
    },
    {
      "epoch": 9.566852367688023,
      "grad_norm": 2.4446194171905518,
      "learning_rate": 3.0940875497441724e-06,
      "loss": 0.7393,
      "step": 6869
    },
    {
      "epoch": 9.56824512534819,
      "grad_norm": 2.0252926349639893,
      "learning_rate": 3.0841387151790786e-06,
      "loss": 0.5545,
      "step": 6870
    },
    {
      "epoch": 9.569637883008356,
      "grad_norm": 2.1388938426971436,
      "learning_rate": 3.074189880613985e-06,
      "loss": 0.6642,
      "step": 6871
    },
    {
      "epoch": 9.571030640668523,
      "grad_norm": 1.940800428390503,
      "learning_rate": 3.0642410460488914e-06,
      "loss": 0.4897,
      "step": 6872
    },
    {
      "epoch": 9.572423398328691,
      "grad_norm": 1.8035876750946045,
      "learning_rate": 3.054292211483797e-06,
      "loss": 0.4695,
      "step": 6873
    },
    {
      "epoch": 9.573816155988858,
      "grad_norm": 2.1073710918426514,
      "learning_rate": 3.0443433769187034e-06,
      "loss": 0.5661,
      "step": 6874
    },
    {
      "epoch": 9.575208913649025,
      "grad_norm": 2.249248504638672,
      "learning_rate": 3.03439454235361e-06,
      "loss": 0.7143,
      "step": 6875
    },
    {
      "epoch": 9.576601671309191,
      "grad_norm": 2.2620809078216553,
      "learning_rate": 3.0244457077885158e-06,
      "loss": 0.4422,
      "step": 6876
    },
    {
      "epoch": 9.57799442896936,
      "grad_norm": 2.0016865730285645,
      "learning_rate": 3.014496873223422e-06,
      "loss": 0.522,
      "step": 6877
    },
    {
      "epoch": 9.579387186629527,
      "grad_norm": 2.14836049079895,
      "learning_rate": 3.004548038658328e-06,
      "loss": 0.6262,
      "step": 6878
    },
    {
      "epoch": 9.580779944289693,
      "grad_norm": 2.43053936958313,
      "learning_rate": 2.9945992040932348e-06,
      "loss": 0.7094,
      "step": 6879
    },
    {
      "epoch": 9.58217270194986,
      "grad_norm": 2.0721077919006348,
      "learning_rate": 2.9846503695281406e-06,
      "loss": 0.4594,
      "step": 6880
    },
    {
      "epoch": 9.583565459610028,
      "grad_norm": 2.3775992393493652,
      "learning_rate": 2.9747015349630467e-06,
      "loss": 0.6097,
      "step": 6881
    },
    {
      "epoch": 9.584958217270195,
      "grad_norm": 1.7133889198303223,
      "learning_rate": 2.9647527003979534e-06,
      "loss": 0.4903,
      "step": 6882
    },
    {
      "epoch": 9.586350974930362,
      "grad_norm": 2.000835418701172,
      "learning_rate": 2.9548038658328596e-06,
      "loss": 0.5703,
      "step": 6883
    },
    {
      "epoch": 9.587743732590528,
      "grad_norm": 2.188347816467285,
      "learning_rate": 2.9448550312677653e-06,
      "loss": 0.5702,
      "step": 6884
    },
    {
      "epoch": 9.589136490250697,
      "grad_norm": 2.036536693572998,
      "learning_rate": 2.9349061967026715e-06,
      "loss": 0.5254,
      "step": 6885
    },
    {
      "epoch": 9.590529247910863,
      "grad_norm": 2.263777256011963,
      "learning_rate": 2.924957362137578e-06,
      "loss": 0.5976,
      "step": 6886
    },
    {
      "epoch": 9.59192200557103,
      "grad_norm": 1.9548321962356567,
      "learning_rate": 2.915008527572484e-06,
      "loss": 0.4607,
      "step": 6887
    },
    {
      "epoch": 9.593314763231199,
      "grad_norm": 2.230546236038208,
      "learning_rate": 2.90505969300739e-06,
      "loss": 0.7865,
      "step": 6888
    },
    {
      "epoch": 9.594707520891365,
      "grad_norm": 2.3613121509552,
      "learning_rate": 2.8951108584422967e-06,
      "loss": 0.7575,
      "step": 6889
    },
    {
      "epoch": 9.596100278551532,
      "grad_norm": 1.9772577285766602,
      "learning_rate": 2.885162023877203e-06,
      "loss": 0.5071,
      "step": 6890
    },
    {
      "epoch": 9.597493036211699,
      "grad_norm": 2.2921950817108154,
      "learning_rate": 2.8752131893121087e-06,
      "loss": 0.6333,
      "step": 6891
    },
    {
      "epoch": 9.598885793871867,
      "grad_norm": 2.2228214740753174,
      "learning_rate": 2.865264354747015e-06,
      "loss": 0.6499,
      "step": 6892
    },
    {
      "epoch": 9.600278551532034,
      "grad_norm": 1.8338245153427124,
      "learning_rate": 2.8553155201819215e-06,
      "loss": 0.5282,
      "step": 6893
    },
    {
      "epoch": 9.6016713091922,
      "grad_norm": 1.925538420677185,
      "learning_rate": 2.8453666856168277e-06,
      "loss": 0.4687,
      "step": 6894
    },
    {
      "epoch": 9.603064066852367,
      "grad_norm": 1.5537601709365845,
      "learning_rate": 2.8354178510517334e-06,
      "loss": 0.3136,
      "step": 6895
    },
    {
      "epoch": 9.604456824512535,
      "grad_norm": 1.8593311309814453,
      "learning_rate": 2.82546901648664e-06,
      "loss": 0.4402,
      "step": 6896
    },
    {
      "epoch": 9.605849582172702,
      "grad_norm": 2.3682188987731934,
      "learning_rate": 2.8155201819215462e-06,
      "loss": 0.6091,
      "step": 6897
    },
    {
      "epoch": 9.607242339832869,
      "grad_norm": 2.1733884811401367,
      "learning_rate": 2.805571347356452e-06,
      "loss": 0.71,
      "step": 6898
    },
    {
      "epoch": 9.608635097493035,
      "grad_norm": 2.267303466796875,
      "learning_rate": 2.795622512791358e-06,
      "loss": 0.5613,
      "step": 6899
    },
    {
      "epoch": 9.610027855153204,
      "grad_norm": 2.11245059967041,
      "learning_rate": 2.785673678226265e-06,
      "loss": 0.4584,
      "step": 6900
    },
    {
      "epoch": 9.61142061281337,
      "grad_norm": 1.9563874006271362,
      "learning_rate": 2.775724843661171e-06,
      "loss": 0.6084,
      "step": 6901
    },
    {
      "epoch": 9.612813370473537,
      "grad_norm": 2.390535831451416,
      "learning_rate": 2.7657760090960768e-06,
      "loss": 0.734,
      "step": 6902
    },
    {
      "epoch": 9.614206128133706,
      "grad_norm": 2.5485520362854004,
      "learning_rate": 2.7558271745309834e-06,
      "loss": 0.5524,
      "step": 6903
    },
    {
      "epoch": 9.615598885793872,
      "grad_norm": 2.0168912410736084,
      "learning_rate": 2.7458783399658896e-06,
      "loss": 0.5521,
      "step": 6904
    },
    {
      "epoch": 9.616991643454039,
      "grad_norm": 1.8310315608978271,
      "learning_rate": 2.7359295054007958e-06,
      "loss": 0.5109,
      "step": 6905
    },
    {
      "epoch": 9.618384401114206,
      "grad_norm": 2.0850415229797363,
      "learning_rate": 2.7259806708357016e-06,
      "loss": 0.516,
      "step": 6906
    },
    {
      "epoch": 9.619777158774372,
      "grad_norm": 2.121399164199829,
      "learning_rate": 2.716031836270608e-06,
      "loss": 0.4571,
      "step": 6907
    },
    {
      "epoch": 9.62116991643454,
      "grad_norm": 2.0826025009155273,
      "learning_rate": 2.7060830017055144e-06,
      "loss": 0.6349,
      "step": 6908
    },
    {
      "epoch": 9.622562674094707,
      "grad_norm": 2.023716688156128,
      "learning_rate": 2.69613416714042e-06,
      "loss": 0.5386,
      "step": 6909
    },
    {
      "epoch": 9.623955431754874,
      "grad_norm": 2.1747753620147705,
      "learning_rate": 2.6861853325753267e-06,
      "loss": 0.5967,
      "step": 6910
    },
    {
      "epoch": 9.625348189415043,
      "grad_norm": 1.7980483770370483,
      "learning_rate": 2.676236498010233e-06,
      "loss": 0.4663,
      "step": 6911
    },
    {
      "epoch": 9.62674094707521,
      "grad_norm": 2.0955042839050293,
      "learning_rate": 2.666287663445139e-06,
      "loss": 0.5562,
      "step": 6912
    },
    {
      "epoch": 9.628133704735376,
      "grad_norm": 2.251206636428833,
      "learning_rate": 2.656338828880045e-06,
      "loss": 0.5868,
      "step": 6913
    },
    {
      "epoch": 9.629526462395543,
      "grad_norm": 1.8673650026321411,
      "learning_rate": 2.6463899943149515e-06,
      "loss": 0.5288,
      "step": 6914
    },
    {
      "epoch": 9.630919220055711,
      "grad_norm": 2.5020244121551514,
      "learning_rate": 2.6364411597498577e-06,
      "loss": 0.6089,
      "step": 6915
    },
    {
      "epoch": 9.632311977715878,
      "grad_norm": 2.392106533050537,
      "learning_rate": 2.6264923251847635e-06,
      "loss": 0.6196,
      "step": 6916
    },
    {
      "epoch": 9.633704735376044,
      "grad_norm": 2.0079996585845947,
      "learning_rate": 2.61654349061967e-06,
      "loss": 0.5551,
      "step": 6917
    },
    {
      "epoch": 9.635097493036211,
      "grad_norm": 2.3050472736358643,
      "learning_rate": 2.6065946560545763e-06,
      "loss": 0.5626,
      "step": 6918
    },
    {
      "epoch": 9.63649025069638,
      "grad_norm": 2.0892646312713623,
      "learning_rate": 2.5966458214894825e-06,
      "loss": 0.5317,
      "step": 6919
    },
    {
      "epoch": 9.637883008356546,
      "grad_norm": 2.22564959526062,
      "learning_rate": 2.5866969869243883e-06,
      "loss": 0.5402,
      "step": 6920
    },
    {
      "epoch": 9.639275766016713,
      "grad_norm": 1.9598835706710815,
      "learning_rate": 2.576748152359295e-06,
      "loss": 0.4181,
      "step": 6921
    },
    {
      "epoch": 9.64066852367688,
      "grad_norm": 1.9309732913970947,
      "learning_rate": 2.566799317794201e-06,
      "loss": 0.5266,
      "step": 6922
    },
    {
      "epoch": 9.642061281337048,
      "grad_norm": 2.229978561401367,
      "learning_rate": 2.5568504832291073e-06,
      "loss": 0.524,
      "step": 6923
    },
    {
      "epoch": 9.643454038997215,
      "grad_norm": 2.2351856231689453,
      "learning_rate": 2.5469016486640134e-06,
      "loss": 0.6238,
      "step": 6924
    },
    {
      "epoch": 9.644846796657381,
      "grad_norm": 2.101581573486328,
      "learning_rate": 2.5369528140989196e-06,
      "loss": 0.5739,
      "step": 6925
    },
    {
      "epoch": 9.64623955431755,
      "grad_norm": 1.8099535703659058,
      "learning_rate": 2.527003979533826e-06,
      "loss": 0.4697,
      "step": 6926
    },
    {
      "epoch": 9.647632311977716,
      "grad_norm": 2.1131985187530518,
      "learning_rate": 2.5170551449687316e-06,
      "loss": 0.5731,
      "step": 6927
    },
    {
      "epoch": 9.649025069637883,
      "grad_norm": 2.122297763824463,
      "learning_rate": 2.5071063104036382e-06,
      "loss": 0.7187,
      "step": 6928
    },
    {
      "epoch": 9.65041782729805,
      "grad_norm": 2.3489110469818115,
      "learning_rate": 2.4971574758385444e-06,
      "loss": 0.8092,
      "step": 6929
    },
    {
      "epoch": 9.651810584958218,
      "grad_norm": 1.7810783386230469,
      "learning_rate": 2.4872086412734506e-06,
      "loss": 0.4758,
      "step": 6930
    },
    {
      "epoch": 9.653203342618385,
      "grad_norm": 2.2817227840423584,
      "learning_rate": 2.477259806708357e-06,
      "loss": 0.64,
      "step": 6931
    },
    {
      "epoch": 9.654596100278551,
      "grad_norm": 2.217214584350586,
      "learning_rate": 2.467310972143263e-06,
      "loss": 0.5357,
      "step": 6932
    },
    {
      "epoch": 9.655988857938718,
      "grad_norm": 2.1526618003845215,
      "learning_rate": 2.457362137578169e-06,
      "loss": 0.6023,
      "step": 6933
    },
    {
      "epoch": 9.657381615598887,
      "grad_norm": 2.233933925628662,
      "learning_rate": 2.447413303013076e-06,
      "loss": 0.5486,
      "step": 6934
    },
    {
      "epoch": 9.658774373259053,
      "grad_norm": 2.1165871620178223,
      "learning_rate": 2.4374644684479816e-06,
      "loss": 0.6014,
      "step": 6935
    },
    {
      "epoch": 9.66016713091922,
      "grad_norm": 1.994081735610962,
      "learning_rate": 2.4275156338828878e-06,
      "loss": 0.5341,
      "step": 6936
    },
    {
      "epoch": 9.661559888579387,
      "grad_norm": 2.0053889751434326,
      "learning_rate": 2.417566799317794e-06,
      "loss": 0.5132,
      "step": 6937
    },
    {
      "epoch": 9.662952646239555,
      "grad_norm": 2.163989305496216,
      "learning_rate": 2.4076179647527e-06,
      "loss": 0.6876,
      "step": 6938
    },
    {
      "epoch": 9.664345403899722,
      "grad_norm": 2.0896172523498535,
      "learning_rate": 2.3976691301876063e-06,
      "loss": 0.579,
      "step": 6939
    },
    {
      "epoch": 9.665738161559888,
      "grad_norm": 2.0439982414245605,
      "learning_rate": 2.3877202956225125e-06,
      "loss": 0.4529,
      "step": 6940
    },
    {
      "epoch": 9.667130919220055,
      "grad_norm": 1.9303287267684937,
      "learning_rate": 2.377771461057419e-06,
      "loss": 0.5237,
      "step": 6941
    },
    {
      "epoch": 9.668523676880223,
      "grad_norm": 2.6126015186309814,
      "learning_rate": 2.367822626492325e-06,
      "loss": 0.6362,
      "step": 6942
    },
    {
      "epoch": 9.66991643454039,
      "grad_norm": 2.1996190547943115,
      "learning_rate": 2.357873791927231e-06,
      "loss": 0.5479,
      "step": 6943
    },
    {
      "epoch": 9.671309192200557,
      "grad_norm": 1.8324912786483765,
      "learning_rate": 2.3479249573621373e-06,
      "loss": 0.4954,
      "step": 6944
    },
    {
      "epoch": 9.672701949860723,
      "grad_norm": 1.8486496210098267,
      "learning_rate": 2.337976122797044e-06,
      "loss": 0.4296,
      "step": 6945
    },
    {
      "epoch": 9.674094707520892,
      "grad_norm": 1.6869251728057861,
      "learning_rate": 2.3280272882319497e-06,
      "loss": 0.3979,
      "step": 6946
    },
    {
      "epoch": 9.675487465181059,
      "grad_norm": 2.0417392253875732,
      "learning_rate": 2.318078453666856e-06,
      "loss": 0.5187,
      "step": 6947
    },
    {
      "epoch": 9.676880222841225,
      "grad_norm": 2.3201656341552734,
      "learning_rate": 2.3081296191017625e-06,
      "loss": 0.5462,
      "step": 6948
    },
    {
      "epoch": 9.678272980501394,
      "grad_norm": 1.9886524677276611,
      "learning_rate": 2.2981807845366683e-06,
      "loss": 0.5588,
      "step": 6949
    },
    {
      "epoch": 9.67966573816156,
      "grad_norm": 2.0460445880889893,
      "learning_rate": 2.2882319499715745e-06,
      "loss": 0.4677,
      "step": 6950
    },
    {
      "epoch": 9.681058495821727,
      "grad_norm": 2.216416835784912,
      "learning_rate": 2.2782831154064806e-06,
      "loss": 0.615,
      "step": 6951
    },
    {
      "epoch": 9.682451253481894,
      "grad_norm": 2.2661871910095215,
      "learning_rate": 2.2683342808413873e-06,
      "loss": 0.6978,
      "step": 6952
    },
    {
      "epoch": 9.683844011142062,
      "grad_norm": 2.5274295806884766,
      "learning_rate": 2.258385446276293e-06,
      "loss": 0.6989,
      "step": 6953
    },
    {
      "epoch": 9.685236768802229,
      "grad_norm": 1.9905458688735962,
      "learning_rate": 2.2484366117111992e-06,
      "loss": 0.5145,
      "step": 6954
    },
    {
      "epoch": 9.686629526462395,
      "grad_norm": 2.3244149684906006,
      "learning_rate": 2.238487777146106e-06,
      "loss": 0.5652,
      "step": 6955
    },
    {
      "epoch": 9.688022284122562,
      "grad_norm": 2.5218443870544434,
      "learning_rate": 2.228538942581012e-06,
      "loss": 0.5619,
      "step": 6956
    },
    {
      "epoch": 9.68941504178273,
      "grad_norm": 2.1873762607574463,
      "learning_rate": 2.218590108015918e-06,
      "loss": 0.6727,
      "step": 6957
    },
    {
      "epoch": 9.690807799442897,
      "grad_norm": 2.5588035583496094,
      "learning_rate": 2.208641273450824e-06,
      "loss": 0.6758,
      "step": 6958
    },
    {
      "epoch": 9.692200557103064,
      "grad_norm": 2.6251144409179688,
      "learning_rate": 2.1986924388857306e-06,
      "loss": 0.5718,
      "step": 6959
    },
    {
      "epoch": 9.69359331476323,
      "grad_norm": 2.7621402740478516,
      "learning_rate": 2.1887436043206364e-06,
      "loss": 0.5315,
      "step": 6960
    },
    {
      "epoch": 9.694986072423399,
      "grad_norm": 2.1926655769348145,
      "learning_rate": 2.1787947697555426e-06,
      "loss": 0.6315,
      "step": 6961
    },
    {
      "epoch": 9.696378830083566,
      "grad_norm": 2.6438217163085938,
      "learning_rate": 2.168845935190449e-06,
      "loss": 0.6022,
      "step": 6962
    },
    {
      "epoch": 9.697771587743732,
      "grad_norm": 1.984354853630066,
      "learning_rate": 2.158897100625355e-06,
      "loss": 0.4491,
      "step": 6963
    },
    {
      "epoch": 9.699164345403899,
      "grad_norm": 2.20831298828125,
      "learning_rate": 2.1489482660602616e-06,
      "loss": 0.5286,
      "step": 6964
    },
    {
      "epoch": 9.700557103064067,
      "grad_norm": 2.2550482749938965,
      "learning_rate": 2.1389994314951673e-06,
      "loss": 0.7065,
      "step": 6965
    },
    {
      "epoch": 9.701949860724234,
      "grad_norm": 1.811528205871582,
      "learning_rate": 2.129050596930074e-06,
      "loss": 0.3716,
      "step": 6966
    },
    {
      "epoch": 9.7033426183844,
      "grad_norm": 2.439908504486084,
      "learning_rate": 2.1191017623649797e-06,
      "loss": 0.6874,
      "step": 6967
    },
    {
      "epoch": 9.704735376044567,
      "grad_norm": 2.070103168487549,
      "learning_rate": 2.109152927799886e-06,
      "loss": 0.6209,
      "step": 6968
    },
    {
      "epoch": 9.706128133704736,
      "grad_norm": 2.3902666568756104,
      "learning_rate": 2.0992040932347925e-06,
      "loss": 0.6534,
      "step": 6969
    },
    {
      "epoch": 9.707520891364902,
      "grad_norm": 2.2055246829986572,
      "learning_rate": 2.0892552586696983e-06,
      "loss": 0.5116,
      "step": 6970
    },
    {
      "epoch": 9.70891364902507,
      "grad_norm": 1.8873040676116943,
      "learning_rate": 2.079306424104605e-06,
      "loss": 0.496,
      "step": 6971
    },
    {
      "epoch": 9.710306406685238,
      "grad_norm": 2.046614646911621,
      "learning_rate": 2.0693575895395107e-06,
      "loss": 0.4669,
      "step": 6972
    },
    {
      "epoch": 9.711699164345404,
      "grad_norm": 1.6519601345062256,
      "learning_rate": 2.0594087549744173e-06,
      "loss": 0.4092,
      "step": 6973
    },
    {
      "epoch": 9.713091922005571,
      "grad_norm": 2.278075933456421,
      "learning_rate": 2.049459920409323e-06,
      "loss": 0.7151,
      "step": 6974
    },
    {
      "epoch": 9.714484679665738,
      "grad_norm": 2.32436203956604,
      "learning_rate": 2.0395110858442297e-06,
      "loss": 0.5953,
      "step": 6975
    },
    {
      "epoch": 9.715877437325906,
      "grad_norm": 1.7162848711013794,
      "learning_rate": 2.029562251279136e-06,
      "loss": 0.4261,
      "step": 6976
    },
    {
      "epoch": 9.717270194986073,
      "grad_norm": 1.9412891864776611,
      "learning_rate": 2.0196134167140416e-06,
      "loss": 0.4663,
      "step": 6977
    },
    {
      "epoch": 9.71866295264624,
      "grad_norm": 2.11315655708313,
      "learning_rate": 2.0096645821489483e-06,
      "loss": 0.6713,
      "step": 6978
    },
    {
      "epoch": 9.720055710306406,
      "grad_norm": 2.1294631958007812,
      "learning_rate": 1.999715747583854e-06,
      "loss": 0.5247,
      "step": 6979
    },
    {
      "epoch": 9.721448467966574,
      "grad_norm": 2.0747945308685303,
      "learning_rate": 1.9897669130187606e-06,
      "loss": 0.5031,
      "step": 6980
    },
    {
      "epoch": 9.722841225626741,
      "grad_norm": 2.102787733078003,
      "learning_rate": 1.9798180784536664e-06,
      "loss": 0.6355,
      "step": 6981
    },
    {
      "epoch": 9.724233983286908,
      "grad_norm": 2.1588542461395264,
      "learning_rate": 1.969869243888573e-06,
      "loss": 0.5202,
      "step": 6982
    },
    {
      "epoch": 9.725626740947074,
      "grad_norm": 2.134801149368286,
      "learning_rate": 1.9599204093234792e-06,
      "loss": 0.6076,
      "step": 6983
    },
    {
      "epoch": 9.727019498607243,
      "grad_norm": 2.18527889251709,
      "learning_rate": 1.9499715747583854e-06,
      "loss": 0.5973,
      "step": 6984
    },
    {
      "epoch": 9.72841225626741,
      "grad_norm": 2.2865958213806152,
      "learning_rate": 1.9400227401932916e-06,
      "loss": 0.5338,
      "step": 6985
    },
    {
      "epoch": 9.729805013927576,
      "grad_norm": 2.043187141418457,
      "learning_rate": 1.930073905628198e-06,
      "loss": 0.4877,
      "step": 6986
    },
    {
      "epoch": 9.731197771587743,
      "grad_norm": 2.2247226238250732,
      "learning_rate": 1.920125071063104e-06,
      "loss": 0.4591,
      "step": 6987
    },
    {
      "epoch": 9.732590529247911,
      "grad_norm": 1.741791009902954,
      "learning_rate": 1.9101762364980098e-06,
      "loss": 0.4618,
      "step": 6988
    },
    {
      "epoch": 9.733983286908078,
      "grad_norm": 1.886433482170105,
      "learning_rate": 1.9002274019329164e-06,
      "loss": 0.4592,
      "step": 6989
    },
    {
      "epoch": 9.735376044568245,
      "grad_norm": 1.9228953123092651,
      "learning_rate": 1.8902785673678224e-06,
      "loss": 0.4409,
      "step": 6990
    },
    {
      "epoch": 9.736768802228411,
      "grad_norm": 2.4497694969177246,
      "learning_rate": 1.8803297328027288e-06,
      "loss": 0.6366,
      "step": 6991
    },
    {
      "epoch": 9.73816155988858,
      "grad_norm": 1.7542403936386108,
      "learning_rate": 1.8703808982376347e-06,
      "loss": 0.427,
      "step": 6992
    },
    {
      "epoch": 9.739554317548746,
      "grad_norm": 1.9917913675308228,
      "learning_rate": 1.8604320636725412e-06,
      "loss": 0.4881,
      "step": 6993
    },
    {
      "epoch": 9.740947075208913,
      "grad_norm": 1.8209863901138306,
      "learning_rate": 1.8504832291074471e-06,
      "loss": 0.431,
      "step": 6994
    },
    {
      "epoch": 9.742339832869082,
      "grad_norm": 2.16464900970459,
      "learning_rate": 1.8405343945423535e-06,
      "loss": 0.5031,
      "step": 6995
    },
    {
      "epoch": 9.743732590529248,
      "grad_norm": 2.160353660583496,
      "learning_rate": 1.8305855599772597e-06,
      "loss": 0.5782,
      "step": 6996
    },
    {
      "epoch": 9.745125348189415,
      "grad_norm": 2.2408394813537598,
      "learning_rate": 1.8206367254121657e-06,
      "loss": 0.4689,
      "step": 6997
    },
    {
      "epoch": 9.746518105849582,
      "grad_norm": 1.8952913284301758,
      "learning_rate": 1.8106878908470721e-06,
      "loss": 0.4133,
      "step": 6998
    },
    {
      "epoch": 9.74791086350975,
      "grad_norm": 2.0435421466827393,
      "learning_rate": 1.800739056281978e-06,
      "loss": 0.5561,
      "step": 6999
    },
    {
      "epoch": 9.749303621169917,
      "grad_norm": 1.984863042831421,
      "learning_rate": 1.7907902217168845e-06,
      "loss": 0.4682,
      "step": 7000
    },
    {
      "epoch": 9.750696378830083,
      "grad_norm": 1.9193624258041382,
      "learning_rate": 1.7808413871517905e-06,
      "loss": 0.5653,
      "step": 7001
    },
    {
      "epoch": 9.75208913649025,
      "grad_norm": 2.372758626937866,
      "learning_rate": 1.7708925525866969e-06,
      "loss": 0.5681,
      "step": 7002
    },
    {
      "epoch": 9.753481894150418,
      "grad_norm": 2.1380016803741455,
      "learning_rate": 1.760943718021603e-06,
      "loss": 0.457,
      "step": 7003
    },
    {
      "epoch": 9.754874651810585,
      "grad_norm": 1.9594526290893555,
      "learning_rate": 1.7509948834565093e-06,
      "loss": 0.5065,
      "step": 7004
    },
    {
      "epoch": 9.756267409470752,
      "grad_norm": 2.3975584506988525,
      "learning_rate": 1.7410460488914155e-06,
      "loss": 0.7089,
      "step": 7005
    },
    {
      "epoch": 9.757660167130918,
      "grad_norm": 2.2214443683624268,
      "learning_rate": 1.7310972143263217e-06,
      "loss": 0.5737,
      "step": 7006
    },
    {
      "epoch": 9.759052924791087,
      "grad_norm": 2.1189019680023193,
      "learning_rate": 1.7211483797612278e-06,
      "loss": 0.5466,
      "step": 7007
    },
    {
      "epoch": 9.760445682451254,
      "grad_norm": 2.492465019226074,
      "learning_rate": 1.7111995451961338e-06,
      "loss": 0.7129,
      "step": 7008
    },
    {
      "epoch": 9.76183844011142,
      "grad_norm": 2.2690608501434326,
      "learning_rate": 1.7012507106310402e-06,
      "loss": 0.6258,
      "step": 7009
    },
    {
      "epoch": 9.763231197771589,
      "grad_norm": 2.0491721630096436,
      "learning_rate": 1.6913018760659464e-06,
      "loss": 0.4888,
      "step": 7010
    },
    {
      "epoch": 9.764623955431755,
      "grad_norm": 2.462674379348755,
      "learning_rate": 1.6813530415008526e-06,
      "loss": 0.6891,
      "step": 7011
    },
    {
      "epoch": 9.766016713091922,
      "grad_norm": 3.877023935317993,
      "learning_rate": 1.6714042069357588e-06,
      "loss": 0.6741,
      "step": 7012
    },
    {
      "epoch": 9.767409470752089,
      "grad_norm": 2.1356163024902344,
      "learning_rate": 1.661455372370665e-06,
      "loss": 0.535,
      "step": 7013
    },
    {
      "epoch": 9.768802228412257,
      "grad_norm": 2.041917562484741,
      "learning_rate": 1.6515065378055712e-06,
      "loss": 0.5095,
      "step": 7014
    },
    {
      "epoch": 9.770194986072424,
      "grad_norm": 2.0338964462280273,
      "learning_rate": 1.6415577032404776e-06,
      "loss": 0.4833,
      "step": 7015
    },
    {
      "epoch": 9.77158774373259,
      "grad_norm": 2.070000648498535,
      "learning_rate": 1.6316088686753836e-06,
      "loss": 0.579,
      "step": 7016
    },
    {
      "epoch": 9.772980501392757,
      "grad_norm": 2.4562582969665527,
      "learning_rate": 1.62166003411029e-06,
      "loss": 0.721,
      "step": 7017
    },
    {
      "epoch": 9.774373259052926,
      "grad_norm": 2.306575059890747,
      "learning_rate": 1.611711199545196e-06,
      "loss": 0.6792,
      "step": 7018
    },
    {
      "epoch": 9.775766016713092,
      "grad_norm": 2.3740251064300537,
      "learning_rate": 1.6017623649801022e-06,
      "loss": 0.6151,
      "step": 7019
    },
    {
      "epoch": 9.777158774373259,
      "grad_norm": 2.5305862426757812,
      "learning_rate": 1.5918135304150084e-06,
      "loss": 0.6339,
      "step": 7020
    },
    {
      "epoch": 9.778551532033426,
      "grad_norm": 1.8700345754623413,
      "learning_rate": 1.5818646958499145e-06,
      "loss": 0.4044,
      "step": 7021
    },
    {
      "epoch": 9.779944289693594,
      "grad_norm": 2.096038341522217,
      "learning_rate": 1.5719158612848207e-06,
      "loss": 0.4863,
      "step": 7022
    },
    {
      "epoch": 9.78133704735376,
      "grad_norm": 2.018458127975464,
      "learning_rate": 1.561967026719727e-06,
      "loss": 0.4877,
      "step": 7023
    },
    {
      "epoch": 9.782729805013927,
      "grad_norm": 1.793351173400879,
      "learning_rate": 1.5520181921546333e-06,
      "loss": 0.5304,
      "step": 7024
    },
    {
      "epoch": 9.784122562674094,
      "grad_norm": 1.891086220741272,
      "learning_rate": 1.5420693575895393e-06,
      "loss": 0.4283,
      "step": 7025
    },
    {
      "epoch": 9.785515320334262,
      "grad_norm": 2.475612163543701,
      "learning_rate": 1.5321205230244457e-06,
      "loss": 0.5958,
      "step": 7026
    },
    {
      "epoch": 9.786908077994429,
      "grad_norm": 1.8320320844650269,
      "learning_rate": 1.5221716884593517e-06,
      "loss": 0.4097,
      "step": 7027
    },
    {
      "epoch": 9.788300835654596,
      "grad_norm": 1.9003340005874634,
      "learning_rate": 1.5122228538942579e-06,
      "loss": 0.5038,
      "step": 7028
    },
    {
      "epoch": 9.789693593314762,
      "grad_norm": 2.2090916633605957,
      "learning_rate": 1.502274019329164e-06,
      "loss": 0.6872,
      "step": 7029
    },
    {
      "epoch": 9.79108635097493,
      "grad_norm": 1.9826990365982056,
      "learning_rate": 1.4923251847640703e-06,
      "loss": 0.5234,
      "step": 7030
    },
    {
      "epoch": 9.792479108635098,
      "grad_norm": 1.7873624563217163,
      "learning_rate": 1.4823763501989767e-06,
      "loss": 0.4322,
      "step": 7031
    },
    {
      "epoch": 9.793871866295264,
      "grad_norm": 2.042003870010376,
      "learning_rate": 1.4724275156338827e-06,
      "loss": 0.5007,
      "step": 7032
    },
    {
      "epoch": 9.795264623955433,
      "grad_norm": 2.289184093475342,
      "learning_rate": 1.462478681068789e-06,
      "loss": 0.605,
      "step": 7033
    },
    {
      "epoch": 9.7966573816156,
      "grad_norm": 3.1260499954223633,
      "learning_rate": 1.452529846503695e-06,
      "loss": 0.6609,
      "step": 7034
    },
    {
      "epoch": 9.798050139275766,
      "grad_norm": 1.9191718101501465,
      "learning_rate": 1.4425810119386014e-06,
      "loss": 0.4787,
      "step": 7035
    },
    {
      "epoch": 9.799442896935933,
      "grad_norm": 2.3396522998809814,
      "learning_rate": 1.4326321773735074e-06,
      "loss": 0.6021,
      "step": 7036
    },
    {
      "epoch": 9.800835654596101,
      "grad_norm": 1.95595121383667,
      "learning_rate": 1.4226833428084138e-06,
      "loss": 0.386,
      "step": 7037
    },
    {
      "epoch": 9.802228412256268,
      "grad_norm": 2.068617820739746,
      "learning_rate": 1.41273450824332e-06,
      "loss": 0.6495,
      "step": 7038
    },
    {
      "epoch": 9.803621169916434,
      "grad_norm": 2.383984088897705,
      "learning_rate": 1.402785673678226e-06,
      "loss": 0.6752,
      "step": 7039
    },
    {
      "epoch": 9.805013927576601,
      "grad_norm": 2.521237373352051,
      "learning_rate": 1.3928368391131324e-06,
      "loss": 0.6401,
      "step": 7040
    },
    {
      "epoch": 9.80640668523677,
      "grad_norm": 1.938478946685791,
      "learning_rate": 1.3828880045480384e-06,
      "loss": 0.5717,
      "step": 7041
    },
    {
      "epoch": 9.807799442896936,
      "grad_norm": 1.7377694845199585,
      "learning_rate": 1.3729391699829448e-06,
      "loss": 0.4244,
      "step": 7042
    },
    {
      "epoch": 9.809192200557103,
      "grad_norm": 2.3117964267730713,
      "learning_rate": 1.3629903354178508e-06,
      "loss": 0.6161,
      "step": 7043
    },
    {
      "epoch": 9.81058495821727,
      "grad_norm": 2.130361795425415,
      "learning_rate": 1.3530415008527572e-06,
      "loss": 0.5192,
      "step": 7044
    },
    {
      "epoch": 9.811977715877438,
      "grad_norm": 2.2317090034484863,
      "learning_rate": 1.3430926662876634e-06,
      "loss": 0.4929,
      "step": 7045
    },
    {
      "epoch": 9.813370473537605,
      "grad_norm": 2.0268118381500244,
      "learning_rate": 1.3331438317225696e-06,
      "loss": 0.6043,
      "step": 7046
    },
    {
      "epoch": 9.814763231197771,
      "grad_norm": 2.0953783988952637,
      "learning_rate": 1.3231949971574758e-06,
      "loss": 0.5248,
      "step": 7047
    },
    {
      "epoch": 9.816155988857938,
      "grad_norm": 1.854250431060791,
      "learning_rate": 1.3132461625923817e-06,
      "loss": 0.5196,
      "step": 7048
    },
    {
      "epoch": 9.817548746518106,
      "grad_norm": 2.054757595062256,
      "learning_rate": 1.3032973280272881e-06,
      "loss": 0.5887,
      "step": 7049
    },
    {
      "epoch": 9.818941504178273,
      "grad_norm": 2.184720516204834,
      "learning_rate": 1.2933484934621941e-06,
      "loss": 0.5577,
      "step": 7050
    },
    {
      "epoch": 9.82033426183844,
      "grad_norm": 2.0001440048217773,
      "learning_rate": 1.2833996588971005e-06,
      "loss": 0.4692,
      "step": 7051
    },
    {
      "epoch": 9.821727019498606,
      "grad_norm": 1.9419692754745483,
      "learning_rate": 1.2734508243320067e-06,
      "loss": 0.5137,
      "step": 7052
    },
    {
      "epoch": 9.823119777158775,
      "grad_norm": 2.0789315700531006,
      "learning_rate": 1.263501989766913e-06,
      "loss": 0.4364,
      "step": 7053
    },
    {
      "epoch": 9.824512534818941,
      "grad_norm": 1.8646198511123657,
      "learning_rate": 1.2535531552018191e-06,
      "loss": 0.4988,
      "step": 7054
    },
    {
      "epoch": 9.825905292479108,
      "grad_norm": 1.80806565284729,
      "learning_rate": 1.2436043206367253e-06,
      "loss": 0.4913,
      "step": 7055
    },
    {
      "epoch": 9.827298050139277,
      "grad_norm": 2.497965097427368,
      "learning_rate": 1.2336554860716315e-06,
      "loss": 0.5724,
      "step": 7056
    },
    {
      "epoch": 9.828690807799443,
      "grad_norm": 1.9820994138717651,
      "learning_rate": 1.223706651506538e-06,
      "loss": 0.5237,
      "step": 7057
    },
    {
      "epoch": 9.83008356545961,
      "grad_norm": 1.5877418518066406,
      "learning_rate": 1.2137578169414439e-06,
      "loss": 0.3922,
      "step": 7058
    },
    {
      "epoch": 9.831476323119777,
      "grad_norm": 2.035332441329956,
      "learning_rate": 1.20380898237635e-06,
      "loss": 0.5689,
      "step": 7059
    },
    {
      "epoch": 9.832869080779945,
      "grad_norm": 2.17326283454895,
      "learning_rate": 1.1938601478112563e-06,
      "loss": 0.6454,
      "step": 7060
    },
    {
      "epoch": 9.834261838440112,
      "grad_norm": 2.0205471515655518,
      "learning_rate": 1.1839113132461625e-06,
      "loss": 0.5385,
      "step": 7061
    },
    {
      "epoch": 9.835654596100278,
      "grad_norm": 1.927617073059082,
      "learning_rate": 1.1739624786810686e-06,
      "loss": 0.4974,
      "step": 7062
    },
    {
      "epoch": 9.837047353760445,
      "grad_norm": 1.6703886985778809,
      "learning_rate": 1.1640136441159748e-06,
      "loss": 0.3911,
      "step": 7063
    },
    {
      "epoch": 9.838440111420613,
      "grad_norm": 2.4799304008483887,
      "learning_rate": 1.1540648095508812e-06,
      "loss": 0.6664,
      "step": 7064
    },
    {
      "epoch": 9.83983286908078,
      "grad_norm": 2.192898988723755,
      "learning_rate": 1.1441159749857872e-06,
      "loss": 0.6836,
      "step": 7065
    },
    {
      "epoch": 9.841225626740947,
      "grad_norm": 1.9271363019943237,
      "learning_rate": 1.1341671404206936e-06,
      "loss": 0.4509,
      "step": 7066
    },
    {
      "epoch": 9.842618384401113,
      "grad_norm": 1.6498823165893555,
      "learning_rate": 1.1242183058555996e-06,
      "loss": 0.406,
      "step": 7067
    },
    {
      "epoch": 9.844011142061282,
      "grad_norm": 2.2960798740386963,
      "learning_rate": 1.114269471290506e-06,
      "loss": 0.7303,
      "step": 7068
    },
    {
      "epoch": 9.845403899721449,
      "grad_norm": 2.159357786178589,
      "learning_rate": 1.104320636725412e-06,
      "loss": 0.6743,
      "step": 7069
    },
    {
      "epoch": 9.846796657381615,
      "grad_norm": 1.9666335582733154,
      "learning_rate": 1.0943718021603182e-06,
      "loss": 0.4534,
      "step": 7070
    },
    {
      "epoch": 9.848189415041782,
      "grad_norm": 1.9147229194641113,
      "learning_rate": 1.0844229675952246e-06,
      "loss": 0.5186,
      "step": 7071
    },
    {
      "epoch": 9.84958217270195,
      "grad_norm": 1.8351507186889648,
      "learning_rate": 1.0744741330301308e-06,
      "loss": 0.4674,
      "step": 7072
    },
    {
      "epoch": 9.850974930362117,
      "grad_norm": 2.234679937362671,
      "learning_rate": 1.064525298465037e-06,
      "loss": 0.6809,
      "step": 7073
    },
    {
      "epoch": 9.852367688022284,
      "grad_norm": 1.74303138256073,
      "learning_rate": 1.054576463899943e-06,
      "loss": 0.3896,
      "step": 7074
    },
    {
      "epoch": 9.85376044568245,
      "grad_norm": 2.025094747543335,
      "learning_rate": 1.0446276293348492e-06,
      "loss": 0.5635,
      "step": 7075
    },
    {
      "epoch": 9.855153203342619,
      "grad_norm": 2.3703227043151855,
      "learning_rate": 1.0346787947697553e-06,
      "loss": 0.6656,
      "step": 7076
    },
    {
      "epoch": 9.856545961002785,
      "grad_norm": 2.1646673679351807,
      "learning_rate": 1.0247299602046615e-06,
      "loss": 0.5863,
      "step": 7077
    },
    {
      "epoch": 9.857938718662952,
      "grad_norm": 2.322664737701416,
      "learning_rate": 1.014781125639568e-06,
      "loss": 0.552,
      "step": 7078
    },
    {
      "epoch": 9.85933147632312,
      "grad_norm": 2.183500051498413,
      "learning_rate": 1.0048322910744741e-06,
      "loss": 0.576,
      "step": 7079
    },
    {
      "epoch": 9.860724233983287,
      "grad_norm": 3.6958634853363037,
      "learning_rate": 9.948834565093803e-07,
      "loss": 0.4973,
      "step": 7080
    },
    {
      "epoch": 9.862116991643454,
      "grad_norm": 2.184685468673706,
      "learning_rate": 9.849346219442865e-07,
      "loss": 0.5928,
      "step": 7081
    },
    {
      "epoch": 9.86350974930362,
      "grad_norm": 1.7897632122039795,
      "learning_rate": 9.749857873791927e-07,
      "loss": 0.4739,
      "step": 7082
    },
    {
      "epoch": 9.864902506963789,
      "grad_norm": 2.104130744934082,
      "learning_rate": 9.65036952814099e-07,
      "loss": 0.504,
      "step": 7083
    },
    {
      "epoch": 9.866295264623956,
      "grad_norm": 1.6344671249389648,
      "learning_rate": 9.550881182490049e-07,
      "loss": 0.4088,
      "step": 7084
    },
    {
      "epoch": 9.867688022284122,
      "grad_norm": 2.351177215576172,
      "learning_rate": 9.451392836839112e-07,
      "loss": 0.3861,
      "step": 7085
    },
    {
      "epoch": 9.869080779944289,
      "grad_norm": 2.248156785964966,
      "learning_rate": 9.351904491188174e-07,
      "loss": 0.6107,
      "step": 7086
    },
    {
      "epoch": 9.870473537604457,
      "grad_norm": 2.3252768516540527,
      "learning_rate": 9.252416145537236e-07,
      "loss": 0.6035,
      "step": 7087
    },
    {
      "epoch": 9.871866295264624,
      "grad_norm": 1.9758137464523315,
      "learning_rate": 9.152927799886299e-07,
      "loss": 0.4369,
      "step": 7088
    },
    {
      "epoch": 9.87325905292479,
      "grad_norm": 2.075624465942383,
      "learning_rate": 9.053439454235361e-07,
      "loss": 0.5758,
      "step": 7089
    },
    {
      "epoch": 9.874651810584957,
      "grad_norm": 1.892717719078064,
      "learning_rate": 8.953951108584422e-07,
      "loss": 0.4483,
      "step": 7090
    },
    {
      "epoch": 9.876044568245126,
      "grad_norm": 2.014420747756958,
      "learning_rate": 8.854462762933484e-07,
      "loss": 0.6405,
      "step": 7091
    },
    {
      "epoch": 9.877437325905293,
      "grad_norm": 1.8224351406097412,
      "learning_rate": 8.754974417282546e-07,
      "loss": 0.4362,
      "step": 7092
    },
    {
      "epoch": 9.87883008356546,
      "grad_norm": 2.152827024459839,
      "learning_rate": 8.655486071631608e-07,
      "loss": 0.4982,
      "step": 7093
    },
    {
      "epoch": 9.880222841225628,
      "grad_norm": 1.9619979858398438,
      "learning_rate": 8.555997725980669e-07,
      "loss": 0.4952,
      "step": 7094
    },
    {
      "epoch": 9.881615598885794,
      "grad_norm": 2.4509429931640625,
      "learning_rate": 8.456509380329732e-07,
      "loss": 0.6591,
      "step": 7095
    },
    {
      "epoch": 9.883008356545961,
      "grad_norm": 2.236889362335205,
      "learning_rate": 8.357021034678794e-07,
      "loss": 0.6118,
      "step": 7096
    },
    {
      "epoch": 9.884401114206128,
      "grad_norm": 2.0628082752227783,
      "learning_rate": 8.257532689027856e-07,
      "loss": 0.5942,
      "step": 7097
    },
    {
      "epoch": 9.885793871866294,
      "grad_norm": 2.272329092025757,
      "learning_rate": 8.158044343376918e-07,
      "loss": 0.6116,
      "step": 7098
    },
    {
      "epoch": 9.887186629526463,
      "grad_norm": 2.2354860305786133,
      "learning_rate": 8.05855599772598e-07,
      "loss": 0.5908,
      "step": 7099
    },
    {
      "epoch": 9.88857938718663,
      "grad_norm": 1.988637089729309,
      "learning_rate": 7.959067652075042e-07,
      "loss": 0.4926,
      "step": 7100
    },
    {
      "epoch": 9.889972144846796,
      "grad_norm": 2.5095486640930176,
      "learning_rate": 7.859579306424104e-07,
      "loss": 0.5934,
      "step": 7101
    },
    {
      "epoch": 9.891364902506965,
      "grad_norm": 2.0553159713745117,
      "learning_rate": 7.760090960773167e-07,
      "loss": 0.5687,
      "step": 7102
    },
    {
      "epoch": 9.892757660167131,
      "grad_norm": 2.137578248977661,
      "learning_rate": 7.660602615122229e-07,
      "loss": 0.5213,
      "step": 7103
    },
    {
      "epoch": 9.894150417827298,
      "grad_norm": 2.0234615802764893,
      "learning_rate": 7.561114269471289e-07,
      "loss": 0.4872,
      "step": 7104
    },
    {
      "epoch": 9.895543175487465,
      "grad_norm": 1.8893370628356934,
      "learning_rate": 7.461625923820351e-07,
      "loss": 0.6072,
      "step": 7105
    },
    {
      "epoch": 9.896935933147633,
      "grad_norm": 2.027456283569336,
      "learning_rate": 7.362137578169413e-07,
      "loss": 0.5366,
      "step": 7106
    },
    {
      "epoch": 9.8983286908078,
      "grad_norm": 2.1616761684417725,
      "learning_rate": 7.262649232518475e-07,
      "loss": 0.5211,
      "step": 7107
    },
    {
      "epoch": 9.899721448467966,
      "grad_norm": 2.145045518875122,
      "learning_rate": 7.163160886867537e-07,
      "loss": 0.5437,
      "step": 7108
    },
    {
      "epoch": 9.901114206128133,
      "grad_norm": 1.9581626653671265,
      "learning_rate": 7.0636725412166e-07,
      "loss": 0.5937,
      "step": 7109
    },
    {
      "epoch": 9.902506963788301,
      "grad_norm": 2.170980453491211,
      "learning_rate": 6.964184195565662e-07,
      "loss": 0.7115,
      "step": 7110
    },
    {
      "epoch": 9.903899721448468,
      "grad_norm": 2.0341475009918213,
      "learning_rate": 6.864695849914724e-07,
      "loss": 0.572,
      "step": 7111
    },
    {
      "epoch": 9.905292479108635,
      "grad_norm": 2.295051097869873,
      "learning_rate": 6.765207504263786e-07,
      "loss": 0.5601,
      "step": 7112
    },
    {
      "epoch": 9.906685236768801,
      "grad_norm": 1.8335386514663696,
      "learning_rate": 6.665719158612848e-07,
      "loss": 0.3945,
      "step": 7113
    },
    {
      "epoch": 9.90807799442897,
      "grad_norm": 2.219296932220459,
      "learning_rate": 6.566230812961909e-07,
      "loss": 0.6245,
      "step": 7114
    },
    {
      "epoch": 9.909470752089137,
      "grad_norm": 2.0192878246307373,
      "learning_rate": 6.466742467310971e-07,
      "loss": 0.5099,
      "step": 7115
    },
    {
      "epoch": 9.910863509749303,
      "grad_norm": 2.1819090843200684,
      "learning_rate": 6.367254121660034e-07,
      "loss": 0.5059,
      "step": 7116
    },
    {
      "epoch": 9.912256267409472,
      "grad_norm": 1.8525947332382202,
      "learning_rate": 6.267765776009096e-07,
      "loss": 0.4052,
      "step": 7117
    },
    {
      "epoch": 9.913649025069638,
      "grad_norm": 2.503136396408081,
      "learning_rate": 6.168277430358157e-07,
      "loss": 0.6114,
      "step": 7118
    },
    {
      "epoch": 9.915041782729805,
      "grad_norm": 2.1122305393218994,
      "learning_rate": 6.068789084707219e-07,
      "loss": 0.5232,
      "step": 7119
    },
    {
      "epoch": 9.916434540389972,
      "grad_norm": 1.8866273164749146,
      "learning_rate": 5.969300739056281e-07,
      "loss": 0.5607,
      "step": 7120
    },
    {
      "epoch": 9.91782729805014,
      "grad_norm": 2.3245203495025635,
      "learning_rate": 5.869812393405343e-07,
      "loss": 0.5957,
      "step": 7121
    },
    {
      "epoch": 9.919220055710307,
      "grad_norm": 1.9134774208068848,
      "learning_rate": 5.770324047754406e-07,
      "loss": 0.5709,
      "step": 7122
    },
    {
      "epoch": 9.920612813370473,
      "grad_norm": 1.9723302125930786,
      "learning_rate": 5.670835702103468e-07,
      "loss": 0.4957,
      "step": 7123
    },
    {
      "epoch": 9.92200557103064,
      "grad_norm": 2.3764803409576416,
      "learning_rate": 5.57134735645253e-07,
      "loss": 0.6448,
      "step": 7124
    },
    {
      "epoch": 9.923398328690809,
      "grad_norm": 1.8192933797836304,
      "learning_rate": 5.471859010801591e-07,
      "loss": 0.4076,
      "step": 7125
    },
    {
      "epoch": 9.924791086350975,
      "grad_norm": 2.030700445175171,
      "learning_rate": 5.372370665150654e-07,
      "loss": 0.5361,
      "step": 7126
    },
    {
      "epoch": 9.926183844011142,
      "grad_norm": 1.692780613899231,
      "learning_rate": 5.272882319499715e-07,
      "loss": 0.4395,
      "step": 7127
    },
    {
      "epoch": 9.927576601671309,
      "grad_norm": 1.9694042205810547,
      "learning_rate": 5.173393973848777e-07,
      "loss": 0.4873,
      "step": 7128
    },
    {
      "epoch": 9.928969359331477,
      "grad_norm": 2.3899731636047363,
      "learning_rate": 5.07390562819784e-07,
      "loss": 0.6094,
      "step": 7129
    },
    {
      "epoch": 9.930362116991644,
      "grad_norm": 1.9247006177902222,
      "learning_rate": 4.974417282546902e-07,
      "loss": 0.5094,
      "step": 7130
    },
    {
      "epoch": 9.93175487465181,
      "grad_norm": 2.264085531234741,
      "learning_rate": 4.874928936895964e-07,
      "loss": 0.5281,
      "step": 7131
    },
    {
      "epoch": 9.933147632311977,
      "grad_norm": 1.9155348539352417,
      "learning_rate": 4.775440591245024e-07,
      "loss": 0.43,
      "step": 7132
    },
    {
      "epoch": 9.934540389972145,
      "grad_norm": 2.153841972351074,
      "learning_rate": 4.675952245594087e-07,
      "loss": 0.5799,
      "step": 7133
    },
    {
      "epoch": 9.935933147632312,
      "grad_norm": 2.1322433948516846,
      "learning_rate": 4.5764638999431493e-07,
      "loss": 0.7184,
      "step": 7134
    },
    {
      "epoch": 9.937325905292479,
      "grad_norm": 2.050550699234009,
      "learning_rate": 4.476975554292211e-07,
      "loss": 0.5051,
      "step": 7135
    },
    {
      "epoch": 9.938718662952645,
      "grad_norm": 1.484911561012268,
      "learning_rate": 4.377487208641273e-07,
      "loss": 0.3215,
      "step": 7136
    },
    {
      "epoch": 9.940111420612814,
      "grad_norm": 1.9908186197280884,
      "learning_rate": 4.2779988629903346e-07,
      "loss": 0.4299,
      "step": 7137
    },
    {
      "epoch": 9.94150417827298,
      "grad_norm": 2.2484664916992188,
      "learning_rate": 4.178510517339397e-07,
      "loss": 0.5732,
      "step": 7138
    },
    {
      "epoch": 9.942896935933147,
      "grad_norm": 2.211859703063965,
      "learning_rate": 4.079022171688459e-07,
      "loss": 0.7317,
      "step": 7139
    },
    {
      "epoch": 9.944289693593316,
      "grad_norm": 1.795798897743225,
      "learning_rate": 3.979533826037521e-07,
      "loss": 0.4229,
      "step": 7140
    },
    {
      "epoch": 9.945682451253482,
      "grad_norm": 1.943630576133728,
      "learning_rate": 3.8800454803865833e-07,
      "loss": 0.4857,
      "step": 7141
    },
    {
      "epoch": 9.947075208913649,
      "grad_norm": 2.133408784866333,
      "learning_rate": 3.7805571347356447e-07,
      "loss": 0.5843,
      "step": 7142
    },
    {
      "epoch": 9.948467966573816,
      "grad_norm": 1.9784224033355713,
      "learning_rate": 3.6810687890847067e-07,
      "loss": 0.4794,
      "step": 7143
    },
    {
      "epoch": 9.949860724233984,
      "grad_norm": 2.6341850757598877,
      "learning_rate": 3.5815804434337686e-07,
      "loss": 0.7599,
      "step": 7144
    },
    {
      "epoch": 9.95125348189415,
      "grad_norm": 2.1995270252227783,
      "learning_rate": 3.482092097782831e-07,
      "loss": 0.5679,
      "step": 7145
    },
    {
      "epoch": 9.952646239554317,
      "grad_norm": 1.9202507734298706,
      "learning_rate": 3.382603752131893e-07,
      "loss": 0.4951,
      "step": 7146
    },
    {
      "epoch": 9.954038997214484,
      "grad_norm": 2.0263235569000244,
      "learning_rate": 3.2831154064809544e-07,
      "loss": 0.5948,
      "step": 7147
    },
    {
      "epoch": 9.955431754874652,
      "grad_norm": 2.282881021499634,
      "learning_rate": 3.183627060830017e-07,
      "loss": 0.5656,
      "step": 7148
    },
    {
      "epoch": 9.95682451253482,
      "grad_norm": 2.2149147987365723,
      "learning_rate": 3.0841387151790787e-07,
      "loss": 0.718,
      "step": 7149
    },
    {
      "epoch": 9.958217270194986,
      "grad_norm": 1.9350368976593018,
      "learning_rate": 2.9846503695281407e-07,
      "loss": 0.5345,
      "step": 7150
    },
    {
      "epoch": 9.959610027855152,
      "grad_norm": 1.8354272842407227,
      "learning_rate": 2.885162023877203e-07,
      "loss": 0.5589,
      "step": 7151
    },
    {
      "epoch": 9.961002785515321,
      "grad_norm": 1.8528400659561157,
      "learning_rate": 2.785673678226265e-07,
      "loss": 0.526,
      "step": 7152
    },
    {
      "epoch": 9.962395543175488,
      "grad_norm": 2.0718677043914795,
      "learning_rate": 2.686185332575327e-07,
      "loss": 0.6267,
      "step": 7153
    },
    {
      "epoch": 9.963788300835654,
      "grad_norm": 2.057283878326416,
      "learning_rate": 2.5866969869243884e-07,
      "loss": 0.5607,
      "step": 7154
    },
    {
      "epoch": 9.965181058495821,
      "grad_norm": 1.8653982877731323,
      "learning_rate": 2.487208641273451e-07,
      "loss": 0.5396,
      "step": 7155
    },
    {
      "epoch": 9.96657381615599,
      "grad_norm": 2.0835800170898438,
      "learning_rate": 2.387720295622512e-07,
      "loss": 0.5667,
      "step": 7156
    },
    {
      "epoch": 9.967966573816156,
      "grad_norm": 2.089766025543213,
      "learning_rate": 2.2882319499715747e-07,
      "loss": 0.5627,
      "step": 7157
    },
    {
      "epoch": 9.969359331476323,
      "grad_norm": 1.9209058284759521,
      "learning_rate": 2.1887436043206366e-07,
      "loss": 0.451,
      "step": 7158
    },
    {
      "epoch": 9.97075208913649,
      "grad_norm": 1.976711392402649,
      "learning_rate": 2.0892552586696985e-07,
      "loss": 0.4475,
      "step": 7159
    },
    {
      "epoch": 9.972144846796658,
      "grad_norm": 2.1920390129089355,
      "learning_rate": 1.9897669130187604e-07,
      "loss": 0.5494,
      "step": 7160
    },
    {
      "epoch": 9.973537604456824,
      "grad_norm": 2.1984899044036865,
      "learning_rate": 1.8902785673678224e-07,
      "loss": 0.6537,
      "step": 7161
    },
    {
      "epoch": 9.974930362116991,
      "grad_norm": 2.0709049701690674,
      "learning_rate": 1.7907902217168843e-07,
      "loss": 0.5861,
      "step": 7162
    },
    {
      "epoch": 9.97632311977716,
      "grad_norm": 2.4112558364868164,
      "learning_rate": 1.6913018760659465e-07,
      "loss": 0.5579,
      "step": 7163
    },
    {
      "epoch": 9.977715877437326,
      "grad_norm": 2.3421363830566406,
      "learning_rate": 1.5918135304150084e-07,
      "loss": 0.5993,
      "step": 7164
    },
    {
      "epoch": 9.979108635097493,
      "grad_norm": 1.897717833518982,
      "learning_rate": 1.4923251847640703e-07,
      "loss": 0.4771,
      "step": 7165
    },
    {
      "epoch": 9.98050139275766,
      "grad_norm": 1.9442927837371826,
      "learning_rate": 1.3928368391131325e-07,
      "loss": 0.4929,
      "step": 7166
    },
    {
      "epoch": 9.981894150417828,
      "grad_norm": 2.2649590969085693,
      "learning_rate": 1.2933484934621942e-07,
      "loss": 0.5187,
      "step": 7167
    },
    {
      "epoch": 9.983286908077995,
      "grad_norm": 2.1614112854003906,
      "learning_rate": 1.193860147811256e-07,
      "loss": 0.6393,
      "step": 7168
    },
    {
      "epoch": 9.984679665738161,
      "grad_norm": 2.230652332305908,
      "learning_rate": 1.0943718021603183e-07,
      "loss": 0.5468,
      "step": 7169
    },
    {
      "epoch": 9.986072423398328,
      "grad_norm": 1.8186545372009277,
      "learning_rate": 9.948834565093802e-08,
      "loss": 0.3882,
      "step": 7170
    },
    {
      "epoch": 9.987465181058496,
      "grad_norm": 1.8561375141143799,
      "learning_rate": 8.953951108584421e-08,
      "loss": 0.5552,
      "step": 7171
    },
    {
      "epoch": 9.988857938718663,
      "grad_norm": 2.0926353931427,
      "learning_rate": 7.959067652075042e-08,
      "loss": 0.5638,
      "step": 7172
    },
    {
      "epoch": 9.99025069637883,
      "grad_norm": 2.2047741413116455,
      "learning_rate": 6.964184195565663e-08,
      "loss": 0.6578,
      "step": 7173
    },
    {
      "epoch": 9.991643454038996,
      "grad_norm": 1.7672758102416992,
      "learning_rate": 5.96930073905628e-08,
      "loss": 0.399,
      "step": 7174
    },
    {
      "epoch": 9.993036211699165,
      "grad_norm": 2.1503844261169434,
      "learning_rate": 4.974417282546901e-08,
      "loss": 0.5326,
      "step": 7175
    },
    {
      "epoch": 9.994428969359332,
      "grad_norm": 2.4471487998962402,
      "learning_rate": 3.979533826037521e-08,
      "loss": 0.7769,
      "step": 7176
    },
    {
      "epoch": 9.995821727019498,
      "grad_norm": 1.9396417140960693,
      "learning_rate": 2.98465036952814e-08,
      "loss": 0.409,
      "step": 7177
    },
    {
      "epoch": 9.997214484679667,
      "grad_norm": 1.933851718902588,
      "learning_rate": 1.9897669130187605e-08,
      "loss": 0.5888,
      "step": 7178
    },
    {
      "epoch": 9.998607242339833,
      "grad_norm": 2.037083148956299,
      "learning_rate": 9.948834565093803e-09,
      "loss": 0.5062,
      "step": 7179
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.248157501220703,
      "learning_rate": 0.0,
      "loss": 0.6042,
      "step": 7180
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.941285959873331e+17,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}