{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.97074312463429,
  "eval_steps": 500,
  "global_step": 2130,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004681100058513751,
      "grad_norm": 8.437421798706055,
      "learning_rate": 4.694835680751174e-07,
      "loss": 2.8409,
      "step": 1
    },
    {
      "epoch": 0.009362200117027502,
      "grad_norm": 8.356019973754883,
      "learning_rate": 9.389671361502348e-07,
      "loss": 2.8116,
      "step": 2
    },
    {
      "epoch": 0.014043300175541252,
      "grad_norm": 8.29000473022461,
      "learning_rate": 1.4084507042253521e-06,
      "loss": 2.8066,
      "step": 3
    },
    {
      "epoch": 0.018724400234055003,
      "grad_norm": 8.24352741241455,
      "learning_rate": 1.8779342723004696e-06,
      "loss": 2.8373,
      "step": 4
    },
    {
      "epoch": 0.023405500292568753,
      "grad_norm": 7.985064506530762,
      "learning_rate": 2.3474178403755868e-06,
      "loss": 2.8102,
      "step": 5
    },
    {
      "epoch": 0.028086600351082503,
      "grad_norm": 7.659386157989502,
      "learning_rate": 2.8169014084507042e-06,
      "loss": 2.7384,
      "step": 6
    },
    {
      "epoch": 0.032767700409596257,
      "grad_norm": 7.230035781860352,
      "learning_rate": 3.286384976525822e-06,
      "loss": 2.7179,
      "step": 7
    },
    {
      "epoch": 0.037448800468110006,
      "grad_norm": 6.8731160163879395,
      "learning_rate": 3.755868544600939e-06,
      "loss": 2.6758,
      "step": 8
    },
    {
      "epoch": 0.042129900526623756,
      "grad_norm": 6.585916519165039,
      "learning_rate": 4.225352112676056e-06,
      "loss": 2.6716,
      "step": 9
    },
    {
      "epoch": 0.046811000585137506,
      "grad_norm": 5.923017501831055,
      "learning_rate": 4.6948356807511736e-06,
      "loss": 2.5922,
      "step": 10
    },
    {
      "epoch": 0.051492100643651256,
      "grad_norm": 5.472363471984863,
      "learning_rate": 5.164319248826292e-06,
      "loss": 2.5256,
      "step": 11
    },
    {
      "epoch": 0.056173200702165006,
      "grad_norm": 5.083034515380859,
      "learning_rate": 5.6338028169014084e-06,
      "loss": 2.5569,
      "step": 12
    },
    {
      "epoch": 0.060854300760678756,
      "grad_norm": 4.812219142913818,
      "learning_rate": 6.103286384976526e-06,
      "loss": 2.5469,
      "step": 13
    },
    {
      "epoch": 0.06553540081919251,
      "grad_norm": 4.375195503234863,
      "learning_rate": 6.572769953051644e-06,
      "loss": 2.503,
      "step": 14
    },
    {
      "epoch": 0.07021650087770626,
      "grad_norm": 3.939549684524536,
      "learning_rate": 7.042253521126762e-06,
      "loss": 2.4302,
      "step": 15
    },
    {
      "epoch": 0.07489760093622001,
      "grad_norm": 3.4781973361968994,
      "learning_rate": 7.511737089201878e-06,
      "loss": 2.4014,
      "step": 16
    },
    {
      "epoch": 0.07957870099473376,
      "grad_norm": 3.1489295959472656,
      "learning_rate": 7.981220657276996e-06,
      "loss": 2.3437,
      "step": 17
    },
    {
      "epoch": 0.08425980105324751,
      "grad_norm": 2.8369641304016113,
      "learning_rate": 8.450704225352112e-06,
      "loss": 2.2593,
      "step": 18
    },
    {
      "epoch": 0.08894090111176127,
      "grad_norm": 2.472102403640747,
      "learning_rate": 8.92018779342723e-06,
      "loss": 2.2874,
      "step": 19
    },
    {
      "epoch": 0.09362200117027501,
      "grad_norm": 2.0766918659210205,
      "learning_rate": 9.389671361502347e-06,
      "loss": 2.2487,
      "step": 20
    },
    {
      "epoch": 0.09830310122878877,
      "grad_norm": 1.7876816987991333,
      "learning_rate": 9.859154929577465e-06,
      "loss": 2.2559,
      "step": 21
    },
    {
      "epoch": 0.10298420128730251,
      "grad_norm": 1.6245261430740356,
      "learning_rate": 1.0328638497652584e-05,
      "loss": 2.2229,
      "step": 22
    },
    {
      "epoch": 0.10766530134581627,
      "grad_norm": 1.4229193925857544,
      "learning_rate": 1.0798122065727701e-05,
      "loss": 2.2001,
      "step": 23
    },
    {
      "epoch": 0.11234640140433001,
      "grad_norm": 1.2889472246170044,
      "learning_rate": 1.1267605633802817e-05,
      "loss": 2.1775,
      "step": 24
    },
    {
      "epoch": 0.11702750146284377,
      "grad_norm": 1.1388767957687378,
      "learning_rate": 1.1737089201877934e-05,
      "loss": 2.1486,
      "step": 25
    },
    {
      "epoch": 0.12170860152135751,
      "grad_norm": 1.0823746919631958,
      "learning_rate": 1.2206572769953052e-05,
      "loss": 2.1248,
      "step": 26
    },
    {
      "epoch": 0.12638970157987128,
      "grad_norm": 1.0433950424194336,
      "learning_rate": 1.267605633802817e-05,
      "loss": 2.1398,
      "step": 27
    },
    {
      "epoch": 0.13107080163838503,
      "grad_norm": 1.045602798461914,
      "learning_rate": 1.3145539906103288e-05,
      "loss": 2.1176,
      "step": 28
    },
    {
      "epoch": 0.13575190169689877,
      "grad_norm": 0.9916484951972961,
      "learning_rate": 1.3615023474178404e-05,
      "loss": 2.095,
      "step": 29
    },
    {
      "epoch": 0.1404330017554125,
      "grad_norm": 0.9007664918899536,
      "learning_rate": 1.4084507042253523e-05,
      "loss": 2.1067,
      "step": 30
    },
    {
      "epoch": 0.14511410181392628,
      "grad_norm": 0.8627976775169373,
      "learning_rate": 1.4553990610328639e-05,
      "loss": 2.1083,
      "step": 31
    },
    {
      "epoch": 0.14979520187244003,
      "grad_norm": 0.8264485001564026,
      "learning_rate": 1.5023474178403756e-05,
      "loss": 2.0686,
      "step": 32
    },
    {
      "epoch": 0.15447630193095377,
      "grad_norm": 0.8104755282402039,
      "learning_rate": 1.5492957746478872e-05,
      "loss": 2.0687,
      "step": 33
    },
    {
      "epoch": 0.1591574019894675,
      "grad_norm": 0.7617929577827454,
      "learning_rate": 1.5962441314553993e-05,
      "loss": 2.019,
      "step": 34
    },
    {
      "epoch": 0.16383850204798128,
      "grad_norm": 0.7188560366630554,
      "learning_rate": 1.643192488262911e-05,
      "loss": 2.0473,
      "step": 35
    },
    {
      "epoch": 0.16851960210649503,
      "grad_norm": 0.7130621075630188,
      "learning_rate": 1.6901408450704224e-05,
      "loss": 2.0238,
      "step": 36
    },
    {
      "epoch": 0.17320070216500877,
      "grad_norm": 0.6950539350509644,
      "learning_rate": 1.7370892018779345e-05,
      "loss": 2.028,
      "step": 37
    },
    {
      "epoch": 0.17788180222352254,
      "grad_norm": 0.7059375643730164,
      "learning_rate": 1.784037558685446e-05,
      "loss": 1.9914,
      "step": 38
    },
    {
      "epoch": 0.18256290228203628,
      "grad_norm": 0.6565279960632324,
      "learning_rate": 1.830985915492958e-05,
      "loss": 1.9853,
      "step": 39
    },
    {
      "epoch": 0.18724400234055003,
      "grad_norm": 0.6930928826332092,
      "learning_rate": 1.8779342723004694e-05,
      "loss": 1.9887,
      "step": 40
    },
    {
      "epoch": 0.19192510239906377,
      "grad_norm": 0.6585078835487366,
      "learning_rate": 1.9248826291079812e-05,
      "loss": 1.9895,
      "step": 41
    },
    {
      "epoch": 0.19660620245757754,
      "grad_norm": 0.6370649933815002,
      "learning_rate": 1.971830985915493e-05,
      "loss": 1.9903,
      "step": 42
    },
    {
      "epoch": 0.20128730251609128,
      "grad_norm": 0.6276764273643494,
      "learning_rate": 2.0187793427230047e-05,
      "loss": 1.9972,
      "step": 43
    },
    {
      "epoch": 0.20596840257460503,
      "grad_norm": 0.615270733833313,
      "learning_rate": 2.0657276995305167e-05,
      "loss": 1.9474,
      "step": 44
    },
    {
      "epoch": 0.2106495026331188,
      "grad_norm": 0.62484210729599,
      "learning_rate": 2.112676056338028e-05,
      "loss": 1.9598,
      "step": 45
    },
    {
      "epoch": 0.21533060269163254,
      "grad_norm": 0.5930389165878296,
      "learning_rate": 2.1596244131455402e-05,
      "loss": 1.9726,
      "step": 46
    },
    {
      "epoch": 0.22001170275014628,
      "grad_norm": 0.6064797043800354,
      "learning_rate": 2.2065727699530516e-05,
      "loss": 1.9479,
      "step": 47
    },
    {
      "epoch": 0.22469280280866002,
      "grad_norm": 0.5708951950073242,
      "learning_rate": 2.2535211267605634e-05,
      "loss": 1.9425,
      "step": 48
    },
    {
      "epoch": 0.2293739028671738,
      "grad_norm": 0.5736585855484009,
      "learning_rate": 2.300469483568075e-05,
      "loss": 1.962,
      "step": 49
    },
    {
      "epoch": 0.23405500292568754,
      "grad_norm": 0.5813770890235901,
      "learning_rate": 2.347417840375587e-05,
      "loss": 1.9599,
      "step": 50
    },
    {
      "epoch": 0.23873610298420128,
      "grad_norm": 0.5862187147140503,
      "learning_rate": 2.3943661971830986e-05,
      "loss": 1.9415,
      "step": 51
    },
    {
      "epoch": 0.24341720304271502,
      "grad_norm": 0.5472211241722107,
      "learning_rate": 2.4413145539906104e-05,
      "loss": 1.9185,
      "step": 52
    },
    {
      "epoch": 0.2480983031012288,
      "grad_norm": 0.5598443150520325,
      "learning_rate": 2.4882629107981224e-05,
      "loss": 1.9337,
      "step": 53
    },
    {
      "epoch": 0.25277940315974257,
      "grad_norm": 0.5730856657028198,
      "learning_rate": 2.535211267605634e-05,
      "loss": 1.9126,
      "step": 54
    },
    {
      "epoch": 0.2574605032182563,
      "grad_norm": 0.5666413307189941,
      "learning_rate": 2.5821596244131456e-05,
      "loss": 1.9412,
      "step": 55
    },
    {
      "epoch": 0.26214160327677005,
      "grad_norm": 0.5389975905418396,
      "learning_rate": 2.6291079812206577e-05,
      "loss": 1.9045,
      "step": 56
    },
    {
      "epoch": 0.2668227033352838,
      "grad_norm": 0.5219833850860596,
      "learning_rate": 2.676056338028169e-05,
      "loss": 1.8992,
      "step": 57
    },
    {
      "epoch": 0.27150380339379754,
      "grad_norm": 0.5336855053901672,
      "learning_rate": 2.7230046948356808e-05,
      "loss": 1.8941,
      "step": 58
    },
    {
      "epoch": 0.2761849034523113,
      "grad_norm": 0.5400417447090149,
      "learning_rate": 2.7699530516431926e-05,
      "loss": 1.885,
      "step": 59
    },
    {
      "epoch": 0.280866003510825,
      "grad_norm": 0.5488688945770264,
      "learning_rate": 2.8169014084507046e-05,
      "loss": 1.9013,
      "step": 60
    },
    {
      "epoch": 0.28554710356933877,
      "grad_norm": 0.5290355086326599,
      "learning_rate": 2.8638497652582164e-05,
      "loss": 1.889,
      "step": 61
    },
    {
      "epoch": 0.29022820362785257,
      "grad_norm": 0.5441176295280457,
      "learning_rate": 2.9107981220657278e-05,
      "loss": 1.895,
      "step": 62
    },
    {
      "epoch": 0.2949093036863663,
      "grad_norm": 0.5422474145889282,
      "learning_rate": 2.9577464788732395e-05,
      "loss": 1.9095,
      "step": 63
    },
    {
      "epoch": 0.29959040374488005,
      "grad_norm": 0.526756763458252,
      "learning_rate": 3.0046948356807513e-05,
      "loss": 1.8849,
      "step": 64
    },
    {
      "epoch": 0.3042715038033938,
      "grad_norm": 0.5253100395202637,
      "learning_rate": 3.0516431924882634e-05,
      "loss": 1.8704,
      "step": 65
    },
    {
      "epoch": 0.30895260386190754,
      "grad_norm": 0.5327618718147278,
      "learning_rate": 3.0985915492957744e-05,
      "loss": 1.8598,
      "step": 66
    },
    {
      "epoch": 0.3136337039204213,
      "grad_norm": 0.545571506023407,
      "learning_rate": 3.1455399061032865e-05,
      "loss": 1.8789,
      "step": 67
    },
    {
      "epoch": 0.318314803978935,
      "grad_norm": 0.5246133804321289,
      "learning_rate": 3.1924882629107986e-05,
      "loss": 1.8815,
      "step": 68
    },
    {
      "epoch": 0.3229959040374488,
      "grad_norm": 0.5434203147888184,
      "learning_rate": 3.23943661971831e-05,
      "loss": 1.8822,
      "step": 69
    },
    {
      "epoch": 0.32767700409596257,
      "grad_norm": 0.5504802465438843,
      "learning_rate": 3.286384976525822e-05,
      "loss": 1.8565,
      "step": 70
    },
    {
      "epoch": 0.3323581041544763,
      "grad_norm": 0.5427384376525879,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.8748,
      "step": 71
    },
    {
      "epoch": 0.33703920421299005,
      "grad_norm": 0.538313090801239,
      "learning_rate": 3.380281690140845e-05,
      "loss": 1.8338,
      "step": 72
    },
    {
      "epoch": 0.3417203042715038,
      "grad_norm": 0.524651050567627,
      "learning_rate": 3.427230046948357e-05,
      "loss": 1.8492,
      "step": 73
    },
    {
      "epoch": 0.34640140433001754,
      "grad_norm": 0.5431910157203674,
      "learning_rate": 3.474178403755869e-05,
      "loss": 1.8543,
      "step": 74
    },
    {
      "epoch": 0.3510825043885313,
      "grad_norm": 0.5315969586372375,
      "learning_rate": 3.5211267605633805e-05,
      "loss": 1.8377,
      "step": 75
    },
    {
      "epoch": 0.3557636044470451,
      "grad_norm": 0.5872920751571655,
      "learning_rate": 3.568075117370892e-05,
      "loss": 1.8463,
      "step": 76
    },
    {
      "epoch": 0.3604447045055588,
      "grad_norm": 0.5032703280448914,
      "learning_rate": 3.615023474178404e-05,
      "loss": 1.8352,
      "step": 77
    },
    {
      "epoch": 0.36512580456407256,
      "grad_norm": 0.5578582882881165,
      "learning_rate": 3.661971830985916e-05,
      "loss": 1.8259,
      "step": 78
    },
    {
      "epoch": 0.3698069046225863,
      "grad_norm": 0.5350944995880127,
      "learning_rate": 3.7089201877934274e-05,
      "loss": 1.8459,
      "step": 79
    },
    {
      "epoch": 0.37448800468110005,
      "grad_norm": 0.5319028496742249,
      "learning_rate": 3.755868544600939e-05,
      "loss": 1.841,
      "step": 80
    },
    {
      "epoch": 0.3791691047396138,
      "grad_norm": 0.5441652536392212,
      "learning_rate": 3.802816901408451e-05,
      "loss": 1.8342,
      "step": 81
    },
    {
      "epoch": 0.38385020479812754,
      "grad_norm": 0.5238386988639832,
      "learning_rate": 3.8497652582159623e-05,
      "loss": 1.8338,
      "step": 82
    },
    {
      "epoch": 0.38853130485664134,
      "grad_norm": 0.5281409025192261,
      "learning_rate": 3.8967136150234744e-05,
      "loss": 1.8282,
      "step": 83
    },
    {
      "epoch": 0.3932124049151551,
      "grad_norm": 0.5238882899284363,
      "learning_rate": 3.943661971830986e-05,
      "loss": 1.843,
      "step": 84
    },
    {
      "epoch": 0.3978935049736688,
      "grad_norm": 0.5391219854354858,
      "learning_rate": 3.990610328638498e-05,
      "loss": 1.8056,
      "step": 85
    },
    {
      "epoch": 0.40257460503218256,
      "grad_norm": 0.5335526466369629,
      "learning_rate": 4.037558685446009e-05,
      "loss": 1.8369,
      "step": 86
    },
    {
      "epoch": 0.4072557050906963,
      "grad_norm": 0.5062191486358643,
      "learning_rate": 4.0845070422535214e-05,
      "loss": 1.7941,
      "step": 87
    },
    {
      "epoch": 0.41193680514921005,
      "grad_norm": 0.521977424621582,
      "learning_rate": 4.1314553990610335e-05,
      "loss": 1.8227,
      "step": 88
    },
    {
      "epoch": 0.4166179052077238,
      "grad_norm": 0.5376787185668945,
      "learning_rate": 4.178403755868545e-05,
      "loss": 1.817,
      "step": 89
    },
    {
      "epoch": 0.4212990052662376,
      "grad_norm": 0.5220103859901428,
      "learning_rate": 4.225352112676056e-05,
      "loss": 1.7939,
      "step": 90
    },
    {
      "epoch": 0.42598010532475133,
      "grad_norm": 0.5247204303741455,
      "learning_rate": 4.2723004694835684e-05,
      "loss": 1.8197,
      "step": 91
    },
    {
      "epoch": 0.4306612053832651,
      "grad_norm": 0.5285261273384094,
      "learning_rate": 4.3192488262910805e-05,
      "loss": 1.8184,
      "step": 92
    },
    {
      "epoch": 0.4353423054417788,
      "grad_norm": 0.532840371131897,
      "learning_rate": 4.366197183098591e-05,
      "loss": 1.8032,
      "step": 93
    },
    {
      "epoch": 0.44002340550029256,
      "grad_norm": 0.5351529121398926,
      "learning_rate": 4.413145539906103e-05,
      "loss": 1.8146,
      "step": 94
    },
    {
      "epoch": 0.4447045055588063,
      "grad_norm": 0.5198392868041992,
      "learning_rate": 4.4600938967136154e-05,
      "loss": 1.7852,
      "step": 95
    },
    {
      "epoch": 0.44938560561732005,
      "grad_norm": 0.5182085633277893,
      "learning_rate": 4.507042253521127e-05,
      "loss": 1.7841,
      "step": 96
    },
    {
      "epoch": 0.45406670567583385,
      "grad_norm": 0.5423537492752075,
      "learning_rate": 4.553990610328639e-05,
      "loss": 1.7969,
      "step": 97
    },
    {
      "epoch": 0.4587478057343476,
      "grad_norm": 0.5703156590461731,
      "learning_rate": 4.60093896713615e-05,
      "loss": 1.7823,
      "step": 98
    },
    {
      "epoch": 0.46342890579286133,
      "grad_norm": 0.5168886780738831,
      "learning_rate": 4.647887323943662e-05,
      "loss": 1.789,
      "step": 99
    },
    {
      "epoch": 0.4681100058513751,
      "grad_norm": 0.5176598429679871,
      "learning_rate": 4.694835680751174e-05,
      "loss": 1.7888,
      "step": 100
    },
    {
      "epoch": 0.4727911059098888,
      "grad_norm": 0.535539448261261,
      "learning_rate": 4.741784037558686e-05,
      "loss": 1.7912,
      "step": 101
    },
    {
      "epoch": 0.47747220596840256,
      "grad_norm": 0.5312119722366333,
      "learning_rate": 4.788732394366197e-05,
      "loss": 1.8038,
      "step": 102
    },
    {
      "epoch": 0.4821533060269163,
      "grad_norm": 0.5023992657661438,
      "learning_rate": 4.835680751173709e-05,
      "loss": 1.7891,
      "step": 103
    },
    {
      "epoch": 0.48683440608543005,
      "grad_norm": 0.5566378831863403,
      "learning_rate": 4.882629107981221e-05,
      "loss": 1.7947,
      "step": 104
    },
    {
      "epoch": 0.49151550614394385,
      "grad_norm": 0.5359525680541992,
      "learning_rate": 4.929577464788733e-05,
      "loss": 1.7786,
      "step": 105
    },
    {
      "epoch": 0.4961966062024576,
      "grad_norm": 0.535662829875946,
      "learning_rate": 4.976525821596245e-05,
      "loss": 1.7852,
      "step": 106
    },
    {
      "epoch": 0.5008777062609713,
      "grad_norm": 0.5481117367744446,
      "learning_rate": 5.023474178403756e-05,
      "loss": 1.7843,
      "step": 107
    },
    {
      "epoch": 0.5055588063194851,
      "grad_norm": 0.5280411243438721,
      "learning_rate": 5.070422535211268e-05,
      "loss": 1.7946,
      "step": 108
    },
    {
      "epoch": 0.5102399063779989,
      "grad_norm": 0.5515773296356201,
      "learning_rate": 5.117370892018779e-05,
      "loss": 1.7706,
      "step": 109
    },
    {
      "epoch": 0.5149210064365126,
      "grad_norm": 0.5569891929626465,
      "learning_rate": 5.164319248826291e-05,
      "loss": 1.755,
      "step": 110
    },
    {
      "epoch": 0.5196021064950264,
      "grad_norm": 0.5135573744773865,
      "learning_rate": 5.2112676056338026e-05,
      "loss": 1.7694,
      "step": 111
    },
    {
      "epoch": 0.5242832065535401,
      "grad_norm": 0.5315744876861572,
      "learning_rate": 5.2582159624413153e-05,
      "loss": 1.7636,
      "step": 112
    },
    {
      "epoch": 0.5289643066120538,
      "grad_norm": 0.5318157076835632,
      "learning_rate": 5.305164319248827e-05,
      "loss": 1.757,
      "step": 113
    },
    {
      "epoch": 0.5336454066705676,
      "grad_norm": 0.562716007232666,
      "learning_rate": 5.352112676056338e-05,
      "loss": 1.7723,
      "step": 114
    },
    {
      "epoch": 0.5383265067290813,
      "grad_norm": 0.5341477394104004,
      "learning_rate": 5.39906103286385e-05,
      "loss": 1.7609,
      "step": 115
    },
    {
      "epoch": 0.5430076067875951,
      "grad_norm": 0.5458403825759888,
      "learning_rate": 5.4460093896713616e-05,
      "loss": 1.7656,
      "step": 116
    },
    {
      "epoch": 0.5476887068461088,
      "grad_norm": 0.5436280965805054,
      "learning_rate": 5.492957746478874e-05,
      "loss": 1.7611,
      "step": 117
    },
    {
      "epoch": 0.5523698069046226,
      "grad_norm": 0.5393151044845581,
      "learning_rate": 5.539906103286385e-05,
      "loss": 1.7364,
      "step": 118
    },
    {
      "epoch": 0.5570509069631363,
      "grad_norm": 0.5378183126449585,
      "learning_rate": 5.5868544600938965e-05,
      "loss": 1.7419,
      "step": 119
    },
    {
      "epoch": 0.56173200702165,
      "grad_norm": 0.5988975763320923,
      "learning_rate": 5.633802816901409e-05,
      "loss": 1.7456,
      "step": 120
    },
    {
      "epoch": 0.5664131070801638,
      "grad_norm": 0.5297287702560425,
      "learning_rate": 5.68075117370892e-05,
      "loss": 1.7475,
      "step": 121
    },
    {
      "epoch": 0.5710942071386775,
      "grad_norm": 0.5265582203865051,
      "learning_rate": 5.727699530516433e-05,
      "loss": 1.7479,
      "step": 122
    },
    {
      "epoch": 0.5757753071971914,
      "grad_norm": 0.5375944972038269,
      "learning_rate": 5.774647887323944e-05,
      "loss": 1.7542,
      "step": 123
    },
    {
      "epoch": 0.5804564072557051,
      "grad_norm": 0.5334234237670898,
      "learning_rate": 5.8215962441314556e-05,
      "loss": 1.7561,
      "step": 124
    },
    {
      "epoch": 0.5851375073142189,
      "grad_norm": 0.5718224048614502,
      "learning_rate": 5.868544600938968e-05,
      "loss": 1.7557,
      "step": 125
    },
    {
      "epoch": 0.5898186073727326,
      "grad_norm": 0.5835574865341187,
      "learning_rate": 5.915492957746479e-05,
      "loss": 1.7572,
      "step": 126
    },
    {
      "epoch": 0.5944997074312464,
      "grad_norm": 0.53130042552948,
      "learning_rate": 5.9624413145539905e-05,
      "loss": 1.748,
      "step": 127
    },
    {
      "epoch": 0.5991808074897601,
      "grad_norm": 0.5725318789482117,
      "learning_rate": 6.0093896713615026e-05,
      "loss": 1.73,
      "step": 128
    },
    {
      "epoch": 0.6038619075482738,
      "grad_norm": 0.5428541898727417,
      "learning_rate": 6.056338028169014e-05,
      "loss": 1.7334,
      "step": 129
    },
    {
      "epoch": 0.6085430076067876,
      "grad_norm": 0.5751455426216125,
      "learning_rate": 6.103286384976527e-05,
      "loss": 1.7184,
      "step": 130
    },
    {
      "epoch": 0.6132241076653013,
      "grad_norm": 0.5367549061775208,
      "learning_rate": 6.150234741784038e-05,
      "loss": 1.7289,
      "step": 131
    },
    {
      "epoch": 0.6179052077238151,
      "grad_norm": 0.6348855495452881,
      "learning_rate": 6.197183098591549e-05,
      "loss": 1.7421,
      "step": 132
    },
    {
      "epoch": 0.6225863077823288,
      "grad_norm": 0.5319597721099854,
      "learning_rate": 6.244131455399061e-05,
      "loss": 1.7477,
      "step": 133
    },
    {
      "epoch": 0.6272674078408426,
      "grad_norm": 0.6505827307701111,
      "learning_rate": 6.291079812206573e-05,
      "loss": 1.7379,
      "step": 134
    },
    {
      "epoch": 0.6319485078993563,
      "grad_norm": 0.552060067653656,
      "learning_rate": 6.338028169014085e-05,
      "loss": 1.7272,
      "step": 135
    },
    {
      "epoch": 0.63662960795787,
      "grad_norm": 0.5982983708381653,
      "learning_rate": 6.384976525821597e-05,
      "loss": 1.7272,
      "step": 136
    },
    {
      "epoch": 0.6413107080163839,
      "grad_norm": 0.5840004682540894,
      "learning_rate": 6.431924882629108e-05,
      "loss": 1.748,
      "step": 137
    },
    {
      "epoch": 0.6459918080748976,
      "grad_norm": 0.5581035017967224,
      "learning_rate": 6.47887323943662e-05,
      "loss": 1.6975,
      "step": 138
    },
    {
      "epoch": 0.6506729081334114,
      "grad_norm": 0.5873554348945618,
      "learning_rate": 6.525821596244132e-05,
      "loss": 1.7248,
      "step": 139
    },
    {
      "epoch": 0.6553540081919251,
      "grad_norm": 0.5404384732246399,
      "learning_rate": 6.572769953051644e-05,
      "loss": 1.7105,
      "step": 140
    },
    {
      "epoch": 0.6600351082504389,
      "grad_norm": 0.5836492776870728,
      "learning_rate": 6.619718309859155e-05,
      "loss": 1.7131,
      "step": 141
    },
    {
      "epoch": 0.6647162083089526,
      "grad_norm": 0.556516170501709,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.7321,
      "step": 142
    },
    {
      "epoch": 0.6693973083674664,
      "grad_norm": 0.6061303019523621,
      "learning_rate": 6.713615023474179e-05,
      "loss": 1.7291,
      "step": 143
    },
    {
      "epoch": 0.6740784084259801,
      "grad_norm": 0.5574740171432495,
      "learning_rate": 6.76056338028169e-05,
      "loss": 1.7256,
      "step": 144
    },
    {
      "epoch": 0.6787595084844938,
      "grad_norm": 0.5718780159950256,
      "learning_rate": 6.807511737089203e-05,
      "loss": 1.7138,
      "step": 145
    },
    {
      "epoch": 0.6834406085430076,
      "grad_norm": 0.5740944743156433,
      "learning_rate": 6.854460093896714e-05,
      "loss": 1.7277,
      "step": 146
    },
    {
      "epoch": 0.6881217086015213,
      "grad_norm": 0.5966343283653259,
      "learning_rate": 6.901408450704226e-05,
      "loss": 1.6959,
      "step": 147
    },
    {
      "epoch": 0.6928028086600351,
      "grad_norm": 0.5656957626342773,
      "learning_rate": 6.948356807511738e-05,
      "loss": 1.7184,
      "step": 148
    },
    {
      "epoch": 0.6974839087185488,
      "grad_norm": 0.6364338994026184,
      "learning_rate": 6.995305164319249e-05,
      "loss": 1.7022,
      "step": 149
    },
    {
      "epoch": 0.7021650087770626,
      "grad_norm": 0.5378186702728271,
      "learning_rate": 7.042253521126761e-05,
      "loss": 1.731,
      "step": 150
    },
    {
      "epoch": 0.7068461088355764,
      "grad_norm": 0.5798015594482422,
      "learning_rate": 7.089201877934273e-05,
      "loss": 1.6997,
      "step": 151
    },
    {
      "epoch": 0.7115272088940902,
      "grad_norm": 0.5603360533714294,
      "learning_rate": 7.136150234741784e-05,
      "loss": 1.7121,
      "step": 152
    },
    {
      "epoch": 0.7162083089526039,
      "grad_norm": 0.5362142324447632,
      "learning_rate": 7.183098591549297e-05,
      "loss": 1.6943,
      "step": 153
    },
    {
      "epoch": 0.7208894090111176,
      "grad_norm": 0.6047829985618591,
      "learning_rate": 7.230046948356808e-05,
      "loss": 1.7229,
      "step": 154
    },
    {
      "epoch": 0.7255705090696314,
      "grad_norm": 0.5463905930519104,
      "learning_rate": 7.276995305164319e-05,
      "loss": 1.7021,
      "step": 155
    },
    {
      "epoch": 0.7302516091281451,
      "grad_norm": 0.5977831482887268,
      "learning_rate": 7.323943661971832e-05,
      "loss": 1.7043,
      "step": 156
    },
    {
      "epoch": 0.7349327091866589,
      "grad_norm": 0.5732491612434387,
      "learning_rate": 7.370892018779343e-05,
      "loss": 1.7098,
      "step": 157
    },
    {
      "epoch": 0.7396138092451726,
      "grad_norm": 0.5583239197731018,
      "learning_rate": 7.417840375586855e-05,
      "loss": 1.669,
      "step": 158
    },
    {
      "epoch": 0.7442949093036864,
      "grad_norm": 0.6422227621078491,
      "learning_rate": 7.464788732394367e-05,
      "loss": 1.6873,
      "step": 159
    },
    {
      "epoch": 0.7489760093622001,
      "grad_norm": 0.5348430275917053,
      "learning_rate": 7.511737089201878e-05,
      "loss": 1.6982,
      "step": 160
    },
    {
      "epoch": 0.7536571094207138,
      "grad_norm": 0.6607827544212341,
      "learning_rate": 7.55868544600939e-05,
      "loss": 1.708,
      "step": 161
    },
    {
      "epoch": 0.7583382094792276,
      "grad_norm": 0.5770392417907715,
      "learning_rate": 7.605633802816902e-05,
      "loss": 1.7188,
      "step": 162
    },
    {
      "epoch": 0.7630193095377413,
      "grad_norm": 0.5945390462875366,
      "learning_rate": 7.652582159624414e-05,
      "loss": 1.7049,
      "step": 163
    },
    {
      "epoch": 0.7677004095962551,
      "grad_norm": 0.6323566436767578,
      "learning_rate": 7.699530516431925e-05,
      "loss": 1.6993,
      "step": 164
    },
    {
      "epoch": 0.7723815096547688,
      "grad_norm": 0.5178484320640564,
      "learning_rate": 7.746478873239437e-05,
      "loss": 1.6599,
      "step": 165
    },
    {
      "epoch": 0.7770626097132827,
      "grad_norm": 0.6181017160415649,
      "learning_rate": 7.793427230046949e-05,
      "loss": 1.717,
      "step": 166
    },
    {
      "epoch": 0.7817437097717964,
      "grad_norm": 0.5473299026489258,
      "learning_rate": 7.840375586854461e-05,
      "loss": 1.6855,
      "step": 167
    },
    {
      "epoch": 0.7864248098303102,
      "grad_norm": 0.5608950853347778,
      "learning_rate": 7.887323943661972e-05,
      "loss": 1.6793,
      "step": 168
    },
    {
      "epoch": 0.7911059098888239,
      "grad_norm": 0.5713764429092407,
      "learning_rate": 7.934272300469484e-05,
      "loss": 1.6982,
      "step": 169
    },
    {
      "epoch": 0.7957870099473376,
      "grad_norm": 0.5279483199119568,
      "learning_rate": 7.981220657276996e-05,
      "loss": 1.7026,
      "step": 170
    },
    {
      "epoch": 0.8004681100058514,
      "grad_norm": 0.579554557800293,
      "learning_rate": 8.028169014084508e-05,
      "loss": 1.7107,
      "step": 171
    },
    {
      "epoch": 0.8051492100643651,
      "grad_norm": 0.5823881030082703,
      "learning_rate": 8.075117370892019e-05,
      "loss": 1.6994,
      "step": 172
    },
    {
      "epoch": 0.8098303101228789,
      "grad_norm": 0.5477363467216492,
      "learning_rate": 8.122065727699531e-05,
      "loss": 1.7047,
      "step": 173
    },
    {
      "epoch": 0.8145114101813926,
      "grad_norm": 0.5420286655426025,
      "learning_rate": 8.169014084507043e-05,
      "loss": 1.6781,
      "step": 174
    },
    {
      "epoch": 0.8191925102399064,
      "grad_norm": 0.558557391166687,
      "learning_rate": 8.215962441314554e-05,
      "loss": 1.6771,
      "step": 175
    },
    {
      "epoch": 0.8238736102984201,
      "grad_norm": 0.5379720330238342,
      "learning_rate": 8.262910798122067e-05,
      "loss": 1.6797,
      "step": 176
    },
    {
      "epoch": 0.8285547103569338,
      "grad_norm": 0.5545073747634888,
      "learning_rate": 8.309859154929578e-05,
      "loss": 1.6675,
      "step": 177
    },
    {
      "epoch": 0.8332358104154476,
      "grad_norm": 0.5789998769760132,
      "learning_rate": 8.35680751173709e-05,
      "loss": 1.6683,
      "step": 178
    },
    {
      "epoch": 0.8379169104739613,
      "grad_norm": 0.6201234459877014,
      "learning_rate": 8.403755868544602e-05,
      "loss": 1.6806,
      "step": 179
    },
    {
      "epoch": 0.8425980105324752,
      "grad_norm": 0.5758758783340454,
      "learning_rate": 8.450704225352113e-05,
      "loss": 1.6456,
      "step": 180
    },
    {
      "epoch": 0.8472791105909889,
      "grad_norm": 0.5807463526725769,
      "learning_rate": 8.497652582159625e-05,
      "loss": 1.6479,
      "step": 181
    },
    {
      "epoch": 0.8519602106495027,
      "grad_norm": 0.6110079884529114,
      "learning_rate": 8.544600938967137e-05,
      "loss": 1.6858,
      "step": 182
    },
    {
      "epoch": 0.8566413107080164,
      "grad_norm": 0.566091001033783,
      "learning_rate": 8.591549295774647e-05,
      "loss": 1.6743,
      "step": 183
    },
    {
      "epoch": 0.8613224107665302,
      "grad_norm": 0.6173126101493835,
      "learning_rate": 8.638497652582161e-05,
      "loss": 1.6819,
      "step": 184
    },
    {
      "epoch": 0.8660035108250439,
      "grad_norm": 0.5910197496414185,
      "learning_rate": 8.685446009389672e-05,
      "loss": 1.6901,
      "step": 185
    },
    {
      "epoch": 0.8706846108835576,
      "grad_norm": 0.6235933899879456,
      "learning_rate": 8.732394366197182e-05,
      "loss": 1.6558,
      "step": 186
    },
    {
      "epoch": 0.8753657109420714,
      "grad_norm": 0.6407201290130615,
      "learning_rate": 8.779342723004696e-05,
      "loss": 1.6538,
      "step": 187
    },
    {
      "epoch": 0.8800468110005851,
      "grad_norm": 0.5760080218315125,
      "learning_rate": 8.826291079812207e-05,
      "loss": 1.6657,
      "step": 188
    },
    {
      "epoch": 0.8847279110590989,
      "grad_norm": 0.5801392793655396,
      "learning_rate": 8.873239436619719e-05,
      "loss": 1.6783,
      "step": 189
    },
    {
      "epoch": 0.8894090111176126,
      "grad_norm": 0.6207360625267029,
      "learning_rate": 8.920187793427231e-05,
      "loss": 1.6722,
      "step": 190
    },
    {
      "epoch": 0.8940901111761264,
      "grad_norm": 0.5441955327987671,
      "learning_rate": 8.967136150234741e-05,
      "loss": 1.6895,
      "step": 191
    },
    {
      "epoch": 0.8987712112346401,
      "grad_norm": 0.6543055772781372,
      "learning_rate": 9.014084507042254e-05,
      "loss": 1.655,
      "step": 192
    },
    {
      "epoch": 0.9034523112931538,
      "grad_norm": 0.6866628527641296,
      "learning_rate": 9.061032863849766e-05,
      "loss": 1.6692,
      "step": 193
    },
    {
      "epoch": 0.9081334113516677,
      "grad_norm": 0.5562683939933777,
      "learning_rate": 9.107981220657278e-05,
      "loss": 1.6617,
      "step": 194
    },
    {
      "epoch": 0.9128145114101814,
      "grad_norm": 0.6460333466529846,
      "learning_rate": 9.15492957746479e-05,
      "loss": 1.6455,
      "step": 195
    },
    {
      "epoch": 0.9174956114686952,
      "grad_norm": 0.6548853516578674,
      "learning_rate": 9.2018779342723e-05,
      "loss": 1.6894,
      "step": 196
    },
    {
      "epoch": 0.9221767115272089,
      "grad_norm": 0.5706417560577393,
      "learning_rate": 9.248826291079813e-05,
      "loss": 1.6521,
      "step": 197
    },
    {
      "epoch": 0.9268578115857227,
      "grad_norm": 0.6393511891365051,
      "learning_rate": 9.295774647887325e-05,
      "loss": 1.6604,
      "step": 198
    },
    {
      "epoch": 0.9315389116442364,
      "grad_norm": 0.598738968372345,
      "learning_rate": 9.342723004694837e-05,
      "loss": 1.6479,
      "step": 199
    },
    {
      "epoch": 0.9362200117027502,
      "grad_norm": 0.5855012536048889,
      "learning_rate": 9.389671361502347e-05,
      "loss": 1.6348,
      "step": 200
    },
    {
      "epoch": 0.9409011117612639,
      "grad_norm": 0.5486910343170166,
      "learning_rate": 9.43661971830986e-05,
      "loss": 1.6532,
      "step": 201
    },
    {
      "epoch": 0.9455822118197776,
      "grad_norm": 0.5791751146316528,
      "learning_rate": 9.483568075117372e-05,
      "loss": 1.6438,
      "step": 202
    },
    {
      "epoch": 0.9502633118782914,
      "grad_norm": 0.6078116297721863,
      "learning_rate": 9.530516431924882e-05,
      "loss": 1.6477,
      "step": 203
    },
    {
      "epoch": 0.9549444119368051,
      "grad_norm": 0.5766721367835999,
      "learning_rate": 9.577464788732394e-05,
      "loss": 1.6264,
      "step": 204
    },
    {
      "epoch": 0.9596255119953189,
      "grad_norm": 0.5816053152084351,
      "learning_rate": 9.624413145539907e-05,
      "loss": 1.6483,
      "step": 205
    },
    {
      "epoch": 0.9643066120538326,
      "grad_norm": 0.5633952021598816,
      "learning_rate": 9.671361502347419e-05,
      "loss": 1.6346,
      "step": 206
    },
    {
      "epoch": 0.9689877121123464,
      "grad_norm": 0.592918872833252,
      "learning_rate": 9.718309859154931e-05,
      "loss": 1.6453,
      "step": 207
    },
    {
      "epoch": 0.9736688121708601,
      "grad_norm": 0.5661919116973877,
      "learning_rate": 9.765258215962441e-05,
      "loss": 1.6475,
      "step": 208
    },
    {
      "epoch": 0.978349912229374,
      "grad_norm": 0.6112465858459473,
      "learning_rate": 9.812206572769954e-05,
      "loss": 1.6551,
      "step": 209
    },
    {
      "epoch": 0.9830310122878877,
      "grad_norm": 0.5819314122200012,
      "learning_rate": 9.859154929577466e-05,
      "loss": 1.6491,
      "step": 210
    },
    {
      "epoch": 0.9877121123464014,
      "grad_norm": 0.5747060179710388,
      "learning_rate": 9.906103286384976e-05,
      "loss": 1.6388,
      "step": 211
    },
    {
      "epoch": 0.9923932124049152,
      "grad_norm": 0.5805224180221558,
      "learning_rate": 9.95305164319249e-05,
      "loss": 1.6417,
      "step": 212
    },
    {
      "epoch": 0.9970743124634289,
      "grad_norm": 0.5614432692527771,
      "learning_rate": 0.0001,
      "loss": 1.6335,
      "step": 213
    },
    {
      "epoch": 1.0017554125219426,
      "grad_norm": 2.6474854946136475,
      "learning_rate": 9.999993285782108e-05,
      "loss": 1.83,
      "step": 214
    },
    {
      "epoch": 1.0064365125804564,
      "grad_norm": 0.76060950756073,
      "learning_rate": 9.999973143146459e-05,
      "loss": 1.6414,
      "step": 215
    },
    {
      "epoch": 1.0111176126389703,
      "grad_norm": 0.6044154763221741,
      "learning_rate": 9.999939572147152e-05,
      "loss": 1.6377,
      "step": 216
    },
    {
      "epoch": 1.015798712697484,
      "grad_norm": 0.7118411064147949,
      "learning_rate": 9.99989257287435e-05,
      "loss": 1.6528,
      "step": 217
    },
    {
      "epoch": 1.0204798127559978,
      "grad_norm": 0.5965666174888611,
      "learning_rate": 9.999832145454275e-05,
      "loss": 1.6573,
      "step": 218
    },
    {
      "epoch": 1.0251609128145114,
      "grad_norm": 0.6278660297393799,
      "learning_rate": 9.99975829004922e-05,
      "loss": 1.6336,
      "step": 219
    },
    {
      "epoch": 1.0298420128730252,
      "grad_norm": 0.6511713266372681,
      "learning_rate": 9.999671006857532e-05,
      "loss": 1.6196,
      "step": 220
    },
    {
      "epoch": 1.0345231129315389,
      "grad_norm": 0.6470988392829895,
      "learning_rate": 9.999570296113632e-05,
      "loss": 1.6437,
      "step": 221
    },
    {
      "epoch": 1.0392042129900527,
      "grad_norm": 0.5942560434341431,
      "learning_rate": 9.999456158087995e-05,
      "loss": 1.6212,
      "step": 222
    },
    {
      "epoch": 1.0438853130485664,
      "grad_norm": 0.5859194993972778,
      "learning_rate": 9.999328593087158e-05,
      "loss": 1.644,
      "step": 223
    },
    {
      "epoch": 1.0485664131070802,
      "grad_norm": 0.5919376611709595,
      "learning_rate": 9.999187601453726e-05,
      "loss": 1.6231,
      "step": 224
    },
    {
      "epoch": 1.0532475131655938,
      "grad_norm": 0.5644704103469849,
      "learning_rate": 9.999033183566353e-05,
      "loss": 1.6368,
      "step": 225
    },
    {
      "epoch": 1.0579286132241077,
      "grad_norm": 0.5851888656616211,
      "learning_rate": 9.998865339839761e-05,
      "loss": 1.612,
      "step": 226
    },
    {
      "epoch": 1.0626097132826213,
      "grad_norm": 0.5965368151664734,
      "learning_rate": 9.998684070724723e-05,
      "loss": 1.6369,
      "step": 227
    },
    {
      "epoch": 1.0672908133411352,
      "grad_norm": 0.5883433222770691,
      "learning_rate": 9.998489376708073e-05,
      "loss": 1.6436,
      "step": 228
    },
    {
      "epoch": 1.0719719133996488,
      "grad_norm": 0.5617659091949463,
      "learning_rate": 9.998281258312697e-05,
      "loss": 1.6472,
      "step": 229
    },
    {
      "epoch": 1.0766530134581627,
      "grad_norm": 0.5981467366218567,
      "learning_rate": 9.998059716097536e-05,
      "loss": 1.614,
      "step": 230
    },
    {
      "epoch": 1.0813341135166765,
      "grad_norm": 0.5670073628425598,
      "learning_rate": 9.997824750657585e-05,
      "loss": 1.6469,
      "step": 231
    },
    {
      "epoch": 1.0860152135751902,
      "grad_norm": 0.5550282001495361,
      "learning_rate": 9.997576362623887e-05,
      "loss": 1.6296,
      "step": 232
    },
    {
      "epoch": 1.090696313633704,
      "grad_norm": 0.5583559274673462,
      "learning_rate": 9.99731455266353e-05,
      "loss": 1.6544,
      "step": 233
    },
    {
      "epoch": 1.0953774136922176,
      "grad_norm": 0.5706636905670166,
      "learning_rate": 9.99703932147966e-05,
      "loss": 1.6167,
      "step": 234
    },
    {
      "epoch": 1.1000585137507315,
      "grad_norm": 0.5709851980209351,
      "learning_rate": 9.996750669811459e-05,
      "loss": 1.6209,
      "step": 235
    },
    {
      "epoch": 1.1047396138092451,
      "grad_norm": 0.5738768577575684,
      "learning_rate": 9.996448598434156e-05,
      "loss": 1.6275,
      "step": 236
    },
    {
      "epoch": 1.109420713867759,
      "grad_norm": 0.5648218989372253,
      "learning_rate": 9.996133108159018e-05,
      "loss": 1.5934,
      "step": 237
    },
    {
      "epoch": 1.1141018139262726,
      "grad_norm": 0.5713233351707458,
      "learning_rate": 9.995804199833356e-05,
      "loss": 1.6167,
      "step": 238
    },
    {
      "epoch": 1.1187829139847865,
      "grad_norm": 0.6131446957588196,
      "learning_rate": 9.995461874340513e-05,
      "loss": 1.6463,
      "step": 239
    },
    {
      "epoch": 1.1234640140433,
      "grad_norm": 0.6103014349937439,
      "learning_rate": 9.995106132599869e-05,
      "loss": 1.6078,
      "step": 240
    },
    {
      "epoch": 1.128145114101814,
      "grad_norm": 0.581728994846344,
      "learning_rate": 9.994736975566834e-05,
      "loss": 1.6155,
      "step": 241
    },
    {
      "epoch": 1.1328262141603276,
      "grad_norm": 0.602294385433197,
      "learning_rate": 9.99435440423285e-05,
      "loss": 1.6319,
      "step": 242
    },
    {
      "epoch": 1.1375073142188414,
      "grad_norm": 0.6152282953262329,
      "learning_rate": 9.993958419625382e-05,
      "loss": 1.616,
      "step": 243
    },
    {
      "epoch": 1.142188414277355,
      "grad_norm": 0.6185297966003418,
      "learning_rate": 9.993549022807922e-05,
      "loss": 1.5935,
      "step": 244
    },
    {
      "epoch": 1.146869514335869,
      "grad_norm": 0.6191090941429138,
      "learning_rate": 9.993126214879982e-05,
      "loss": 1.6084,
      "step": 245
    },
    {
      "epoch": 1.1515506143943828,
      "grad_norm": 0.6081867814064026,
      "learning_rate": 9.992689996977091e-05,
      "loss": 1.6216,
      "step": 246
    },
    {
      "epoch": 1.1562317144528964,
      "grad_norm": 0.5717095136642456,
      "learning_rate": 9.992240370270795e-05,
      "loss": 1.6291,
      "step": 247
    },
    {
      "epoch": 1.1609128145114103,
      "grad_norm": 0.6211952567100525,
      "learning_rate": 9.991777335968651e-05,
      "loss": 1.606,
      "step": 248
    },
    {
      "epoch": 1.165593914569924,
      "grad_norm": 0.5580140948295593,
      "learning_rate": 9.99130089531422e-05,
      "loss": 1.6145,
      "step": 249
    },
    {
      "epoch": 1.1702750146284377,
      "grad_norm": 0.5648279786109924,
      "learning_rate": 9.990811049587077e-05,
      "loss": 1.6073,
      "step": 250
    },
    {
      "epoch": 1.1749561146869514,
      "grad_norm": 0.5821574330329895,
      "learning_rate": 9.990307800102794e-05,
      "loss": 1.6062,
      "step": 251
    },
    {
      "epoch": 1.1796372147454652,
      "grad_norm": 0.5487422943115234,
      "learning_rate": 9.98979114821294e-05,
      "loss": 1.6039,
      "step": 252
    },
    {
      "epoch": 1.1843183148039789,
      "grad_norm": 0.5623614192008972,
      "learning_rate": 9.989261095305082e-05,
      "loss": 1.5898,
      "step": 253
    },
    {
      "epoch": 1.1889994148624927,
      "grad_norm": 0.5610089898109436,
      "learning_rate": 9.988717642802775e-05,
      "loss": 1.6122,
      "step": 254
    },
    {
      "epoch": 1.1936805149210064,
      "grad_norm": 0.5645126700401306,
      "learning_rate": 9.988160792165562e-05,
      "loss": 1.6148,
      "step": 255
    },
    {
      "epoch": 1.1983616149795202,
      "grad_norm": 0.6094059348106384,
      "learning_rate": 9.98759054488897e-05,
      "loss": 1.5879,
      "step": 256
    },
    {
      "epoch": 1.203042715038034,
      "grad_norm": 0.5557142496109009,
      "learning_rate": 9.987006902504506e-05,
      "loss": 1.578,
      "step": 257
    },
    {
      "epoch": 1.2077238150965477,
      "grad_norm": 0.5883256793022156,
      "learning_rate": 9.98640986657965e-05,
      "loss": 1.6043,
      "step": 258
    },
    {
      "epoch": 1.2124049151550613,
      "grad_norm": 0.5736782550811768,
      "learning_rate": 9.985799438717853e-05,
      "loss": 1.5862,
      "step": 259
    },
    {
      "epoch": 1.2170860152135752,
      "grad_norm": 0.5847081542015076,
      "learning_rate": 9.985175620558535e-05,
      "loss": 1.5853,
      "step": 260
    },
    {
      "epoch": 1.221767115272089,
      "grad_norm": 0.5794082283973694,
      "learning_rate": 9.984538413777076e-05,
      "loss": 1.6245,
      "step": 261
    },
    {
      "epoch": 1.2264482153306027,
      "grad_norm": 0.581802248954773,
      "learning_rate": 9.983887820084812e-05,
      "loss": 1.601,
      "step": 262
    },
    {
      "epoch": 1.2311293153891165,
      "grad_norm": 0.5950344800949097,
      "learning_rate": 9.983223841229037e-05,
      "loss": 1.6066,
      "step": 263
    },
    {
      "epoch": 1.2358104154476302,
      "grad_norm": 0.6156999468803406,
      "learning_rate": 9.982546478992989e-05,
      "loss": 1.5977,
      "step": 264
    },
    {
      "epoch": 1.240491515506144,
      "grad_norm": 0.5503116250038147,
      "learning_rate": 9.98185573519585e-05,
      "loss": 1.6073,
      "step": 265
    },
    {
      "epoch": 1.2451726155646576,
      "grad_norm": 0.5789607763290405,
      "learning_rate": 9.981151611692745e-05,
      "loss": 1.6041,
      "step": 266
    },
    {
      "epoch": 1.2498537156231715,
      "grad_norm": 0.6294329762458801,
      "learning_rate": 9.980434110374724e-05,
      "loss": 1.5896,
      "step": 267
    },
    {
      "epoch": 1.2545348156816851,
      "grad_norm": 0.5368711352348328,
      "learning_rate": 9.979703233168778e-05,
      "loss": 1.5921,
      "step": 268
    },
    {
      "epoch": 1.259215915740199,
      "grad_norm": 0.548893392086029,
      "learning_rate": 9.978958982037809e-05,
      "loss": 1.5859,
      "step": 269
    },
    {
      "epoch": 1.2638970157987126,
      "grad_norm": 0.5760201811790466,
      "learning_rate": 9.978201358980645e-05,
      "loss": 1.6037,
      "step": 270
    },
    {
      "epoch": 1.2685781158572265,
      "grad_norm": 0.5687435269355774,
      "learning_rate": 9.977430366032023e-05,
      "loss": 1.5622,
      "step": 271
    },
    {
      "epoch": 1.2732592159157403,
      "grad_norm": 0.606561541557312,
      "learning_rate": 9.976646005262592e-05,
      "loss": 1.6117,
      "step": 272
    },
    {
      "epoch": 1.277940315974254,
      "grad_norm": 0.5877252817153931,
      "learning_rate": 9.975848278778897e-05,
      "loss": 1.5912,
      "step": 273
    },
    {
      "epoch": 1.2826214160327676,
      "grad_norm": 0.582199215888977,
      "learning_rate": 9.975037188723382e-05,
      "loss": 1.6029,
      "step": 274
    },
    {
      "epoch": 1.2873025160912814,
      "grad_norm": 0.5783854722976685,
      "learning_rate": 9.974212737274382e-05,
      "loss": 1.5892,
      "step": 275
    },
    {
      "epoch": 1.2919836161497953,
      "grad_norm": 0.5509589910507202,
      "learning_rate": 9.973374926646116e-05,
      "loss": 1.5857,
      "step": 276
    },
    {
      "epoch": 1.296664716208309,
      "grad_norm": 0.65985107421875,
      "learning_rate": 9.97252375908868e-05,
      "loss": 1.6011,
      "step": 277
    },
    {
      "epoch": 1.3013458162668228,
      "grad_norm": 0.5439913272857666,
      "learning_rate": 9.971659236888045e-05,
      "loss": 1.5916,
      "step": 278
    },
    {
      "epoch": 1.3060269163253364,
      "grad_norm": 0.5911054015159607,
      "learning_rate": 9.970781362366046e-05,
      "loss": 1.5833,
      "step": 279
    },
    {
      "epoch": 1.3107080163838503,
      "grad_norm": 0.5758826732635498,
      "learning_rate": 9.969890137880377e-05,
      "loss": 1.5998,
      "step": 280
    },
    {
      "epoch": 1.315389116442364,
      "grad_norm": 0.5465978980064392,
      "learning_rate": 9.968985565824596e-05,
      "loss": 1.6098,
      "step": 281
    },
    {
      "epoch": 1.3200702165008777,
      "grad_norm": 0.5745389461517334,
      "learning_rate": 9.968067648628093e-05,
      "loss": 1.593,
      "step": 282
    },
    {
      "epoch": 1.3247513165593914,
      "grad_norm": 0.543717086315155,
      "learning_rate": 9.967136388756109e-05,
      "loss": 1.5512,
      "step": 283
    },
    {
      "epoch": 1.3294324166179052,
      "grad_norm": 0.5977076292037964,
      "learning_rate": 9.966191788709716e-05,
      "loss": 1.5734,
      "step": 284
    },
    {
      "epoch": 1.3341135166764189,
      "grad_norm": 0.5420624017715454,
      "learning_rate": 9.965233851025814e-05,
      "loss": 1.5766,
      "step": 285
    },
    {
      "epoch": 1.3387946167349327,
      "grad_norm": 0.5433383584022522,
      "learning_rate": 9.964262578277127e-05,
      "loss": 1.561,
      "step": 286
    },
    {
      "epoch": 1.3434757167934466,
      "grad_norm": 0.5578792095184326,
      "learning_rate": 9.963277973072187e-05,
      "loss": 1.5546,
      "step": 287
    },
    {
      "epoch": 1.3481568168519602,
      "grad_norm": 0.585272490978241,
      "learning_rate": 9.962280038055335e-05,
      "loss": 1.5803,
      "step": 288
    },
    {
      "epoch": 1.3528379169104738,
      "grad_norm": 0.5901860594749451,
      "learning_rate": 9.961268775906715e-05,
      "loss": 1.5691,
      "step": 289
    },
    {
      "epoch": 1.3575190169689877,
      "grad_norm": 0.5460952520370483,
      "learning_rate": 9.960244189342258e-05,
      "loss": 1.5787,
      "step": 290
    },
    {
      "epoch": 1.3622001170275015,
      "grad_norm": 0.5862498879432678,
      "learning_rate": 9.959206281113684e-05,
      "loss": 1.5932,
      "step": 291
    },
    {
      "epoch": 1.3668812170860152,
      "grad_norm": 0.5581315755844116,
      "learning_rate": 9.95815505400849e-05,
      "loss": 1.5683,
      "step": 292
    },
    {
      "epoch": 1.371562317144529,
      "grad_norm": 0.541332483291626,
      "learning_rate": 9.957090510849943e-05,
      "loss": 1.5734,
      "step": 293
    },
    {
      "epoch": 1.3762434172030427,
      "grad_norm": 0.5829706192016602,
      "learning_rate": 9.956012654497074e-05,
      "loss": 1.5589,
      "step": 294
    },
    {
      "epoch": 1.3809245172615565,
      "grad_norm": 0.5810964107513428,
      "learning_rate": 9.954921487844667e-05,
      "loss": 1.5859,
      "step": 295
    },
    {
      "epoch": 1.3856056173200701,
      "grad_norm": 0.6038123965263367,
      "learning_rate": 9.953817013823252e-05,
      "loss": 1.5826,
      "step": 296
    },
    {
      "epoch": 1.390286717378584,
      "grad_norm": 0.5622268915176392,
      "learning_rate": 9.952699235399104e-05,
      "loss": 1.571,
      "step": 297
    },
    {
      "epoch": 1.3949678174370979,
      "grad_norm": 0.5940186977386475,
      "learning_rate": 9.951568155574225e-05,
      "loss": 1.5573,
      "step": 298
    },
    {
      "epoch": 1.3996489174956115,
      "grad_norm": 0.6147366166114807,
      "learning_rate": 9.950423777386341e-05,
      "loss": 1.5654,
      "step": 299
    },
    {
      "epoch": 1.4043300175541251,
      "grad_norm": 0.5400118827819824,
      "learning_rate": 9.949266103908895e-05,
      "loss": 1.5529,
      "step": 300
    },
    {
      "epoch": 1.409011117612639,
      "grad_norm": 0.6384579539299011,
      "learning_rate": 9.948095138251035e-05,
      "loss": 1.5939,
      "step": 301
    },
    {
      "epoch": 1.4136922176711528,
      "grad_norm": 0.566920816898346,
      "learning_rate": 9.946910883557607e-05,
      "loss": 1.559,
      "step": 302
    },
    {
      "epoch": 1.4183733177296665,
      "grad_norm": 0.6052601337432861,
      "learning_rate": 9.945713343009153e-05,
      "loss": 1.5554,
      "step": 303
    },
    {
      "epoch": 1.42305441778818,
      "grad_norm": 0.6517474055290222,
      "learning_rate": 9.944502519821887e-05,
      "loss": 1.579,
      "step": 304
    },
    {
      "epoch": 1.427735517846694,
      "grad_norm": 0.5794010758399963,
      "learning_rate": 9.943278417247705e-05,
      "loss": 1.5558,
      "step": 305
    },
    {
      "epoch": 1.4324166179052078,
      "grad_norm": 0.6108701229095459,
      "learning_rate": 9.942041038574162e-05,
      "loss": 1.5675,
      "step": 306
    },
    {
      "epoch": 1.4370977179637214,
      "grad_norm": 0.59806889295578,
      "learning_rate": 9.940790387124472e-05,
      "loss": 1.5565,
      "step": 307
    },
    {
      "epoch": 1.4417788180222353,
      "grad_norm": 0.5927438735961914,
      "learning_rate": 9.939526466257489e-05,
      "loss": 1.5607,
      "step": 308
    },
    {
      "epoch": 1.446459918080749,
      "grad_norm": 0.5815222263336182,
      "learning_rate": 9.938249279367714e-05,
      "loss": 1.5423,
      "step": 309
    },
    {
      "epoch": 1.4511410181392628,
      "grad_norm": 0.5558337569236755,
      "learning_rate": 9.936958829885268e-05,
      "loss": 1.5628,
      "step": 310
    },
    {
      "epoch": 1.4558221181977764,
      "grad_norm": 0.5937175750732422,
      "learning_rate": 9.935655121275898e-05,
      "loss": 1.5852,
      "step": 311
    },
    {
      "epoch": 1.4605032182562903,
      "grad_norm": 0.64537113904953,
      "learning_rate": 9.934338157040952e-05,
      "loss": 1.5671,
      "step": 312
    },
    {
      "epoch": 1.4651843183148041,
      "grad_norm": 0.5682733654975891,
      "learning_rate": 9.93300794071739e-05,
      "loss": 1.5603,
      "step": 313
    },
    {
      "epoch": 1.4698654183733177,
      "grad_norm": 0.6427823901176453,
      "learning_rate": 9.931664475877753e-05,
      "loss": 1.5602,
      "step": 314
    },
    {
      "epoch": 1.4745465184318314,
      "grad_norm": 0.5664746165275574,
      "learning_rate": 9.930307766130169e-05,
      "loss": 1.5843,
      "step": 315
    },
    {
      "epoch": 1.4792276184903452,
      "grad_norm": 0.6186694502830505,
      "learning_rate": 9.928937815118336e-05,
      "loss": 1.5918,
      "step": 316
    },
    {
      "epoch": 1.483908718548859,
      "grad_norm": 0.5681505799293518,
      "learning_rate": 9.927554626521512e-05,
      "loss": 1.5656,
      "step": 317
    },
    {
      "epoch": 1.4885898186073727,
      "grad_norm": 0.5927398204803467,
      "learning_rate": 9.926158204054511e-05,
      "loss": 1.5836,
      "step": 318
    },
    {
      "epoch": 1.4932709186658863,
      "grad_norm": 0.563877522945404,
      "learning_rate": 9.924748551467686e-05,
      "loss": 1.5628,
      "step": 319
    },
    {
      "epoch": 1.4979520187244002,
      "grad_norm": 0.5428170561790466,
      "learning_rate": 9.923325672546923e-05,
      "loss": 1.5528,
      "step": 320
    },
    {
      "epoch": 1.502633118782914,
      "grad_norm": 0.5832676291465759,
      "learning_rate": 9.921889571113628e-05,
      "loss": 1.5662,
      "step": 321
    },
    {
      "epoch": 1.5073142188414277,
      "grad_norm": 0.5446849465370178,
      "learning_rate": 9.920440251024723e-05,
      "loss": 1.5593,
      "step": 322
    },
    {
      "epoch": 1.5119953188999413,
      "grad_norm": 0.5420470237731934,
      "learning_rate": 9.918977716172628e-05,
      "loss": 1.566,
      "step": 323
    },
    {
      "epoch": 1.5166764189584554,
      "grad_norm": 0.5586629509925842,
      "learning_rate": 9.917501970485251e-05,
      "loss": 1.5522,
      "step": 324
    },
    {
      "epoch": 1.521357519016969,
      "grad_norm": 0.5785658955574036,
      "learning_rate": 9.916013017925987e-05,
      "loss": 1.5882,
      "step": 325
    },
    {
      "epoch": 1.5260386190754827,
      "grad_norm": 0.5468897223472595,
      "learning_rate": 9.914510862493693e-05,
      "loss": 1.5523,
      "step": 326
    },
    {
      "epoch": 1.5307197191339965,
      "grad_norm": 0.5480566024780273,
      "learning_rate": 9.912995508222693e-05,
      "loss": 1.5472,
      "step": 327
    },
    {
      "epoch": 1.5354008191925104,
      "grad_norm": 0.5439897179603577,
      "learning_rate": 9.911466959182751e-05,
      "loss": 1.554,
      "step": 328
    },
    {
      "epoch": 1.540081919251024,
      "grad_norm": 0.5619301795959473,
      "learning_rate": 9.909925219479072e-05,
      "loss": 1.5626,
      "step": 329
    },
    {
      "epoch": 1.5447630193095376,
      "grad_norm": 0.5582358241081238,
      "learning_rate": 9.90837029325229e-05,
      "loss": 1.5655,
      "step": 330
    },
    {
      "epoch": 1.5494441193680515,
      "grad_norm": 0.5682559609413147,
      "learning_rate": 9.906802184678445e-05,
      "loss": 1.5417,
      "step": 331
    },
    {
      "epoch": 1.5541252194265653,
      "grad_norm": 0.5619310140609741,
      "learning_rate": 9.90522089796899e-05,
      "loss": 1.5645,
      "step": 332
    },
    {
      "epoch": 1.558806319485079,
      "grad_norm": 0.551527202129364,
      "learning_rate": 9.903626437370764e-05,
      "loss": 1.5417,
      "step": 333
    },
    {
      "epoch": 1.5634874195435926,
      "grad_norm": 0.5816091895103455,
      "learning_rate": 9.90201880716599e-05,
      "loss": 1.5497,
      "step": 334
    },
    {
      "epoch": 1.5681685196021065,
      "grad_norm": 0.5371972322463989,
      "learning_rate": 9.900398011672262e-05,
      "loss": 1.5454,
      "step": 335
    },
    {
      "epoch": 1.5728496196606203,
      "grad_norm": 0.5627063512802124,
      "learning_rate": 9.898764055242526e-05,
      "loss": 1.5649,
      "step": 336
    },
    {
      "epoch": 1.577530719719134,
      "grad_norm": 0.5842807292938232,
      "learning_rate": 9.89711694226508e-05,
      "loss": 1.5496,
      "step": 337
    },
    {
      "epoch": 1.5822118197776478,
      "grad_norm": 0.5478192567825317,
      "learning_rate": 9.895456677163555e-05,
      "loss": 1.5585,
      "step": 338
    },
    {
      "epoch": 1.5868929198361617,
      "grad_norm": 0.5461902022361755,
      "learning_rate": 9.893783264396904e-05,
      "loss": 1.5429,
      "step": 339
    },
    {
      "epoch": 1.5915740198946753,
      "grad_norm": 0.5406402945518494,
      "learning_rate": 9.892096708459385e-05,
      "loss": 1.5555,
      "step": 340
    },
    {
      "epoch": 1.596255119953189,
      "grad_norm": 0.5658143162727356,
      "learning_rate": 9.890397013880566e-05,
      "loss": 1.5528,
      "step": 341
    },
    {
      "epoch": 1.6009362200117028,
      "grad_norm": 0.5685530304908752,
      "learning_rate": 9.88868418522529e-05,
      "loss": 1.5483,
      "step": 342
    },
    {
      "epoch": 1.6056173200702166,
      "grad_norm": 0.5648348331451416,
      "learning_rate": 9.886958227093682e-05,
      "loss": 1.5678,
      "step": 343
    },
    {
      "epoch": 1.6102984201287303,
      "grad_norm": 0.5785353779792786,
      "learning_rate": 9.885219144121125e-05,
      "loss": 1.5427,
      "step": 344
    },
    {
      "epoch": 1.6149795201872439,
      "grad_norm": 0.6532377004623413,
      "learning_rate": 9.883466940978252e-05,
      "loss": 1.5651,
      "step": 345
    },
    {
      "epoch": 1.6196606202457577,
      "grad_norm": 0.5949984192848206,
      "learning_rate": 9.881701622370932e-05,
      "loss": 1.5579,
      "step": 346
    },
    {
      "epoch": 1.6243417203042716,
      "grad_norm": 0.5811630487442017,
      "learning_rate": 9.879923193040256e-05,
      "loss": 1.5637,
      "step": 347
    },
    {
      "epoch": 1.6290228203627852,
      "grad_norm": 0.5769162774085999,
      "learning_rate": 9.878131657762535e-05,
      "loss": 1.5532,
      "step": 348
    },
    {
      "epoch": 1.6337039204212989,
      "grad_norm": 0.5758988261222839,
      "learning_rate": 9.876327021349267e-05,
      "loss": 1.5356,
      "step": 349
    },
    {
      "epoch": 1.6383850204798127,
      "grad_norm": 0.5641076564788818,
      "learning_rate": 9.874509288647143e-05,
      "loss": 1.5598,
      "step": 350
    },
    {
      "epoch": 1.6430661205383266,
      "grad_norm": 0.5294468402862549,
      "learning_rate": 9.872678464538021e-05,
      "loss": 1.5359,
      "step": 351
    },
    {
      "epoch": 1.6477472205968402,
      "grad_norm": 0.6214150786399841,
      "learning_rate": 9.870834553938927e-05,
      "loss": 1.5787,
      "step": 352
    },
    {
      "epoch": 1.652428320655354,
      "grad_norm": 0.5245915651321411,
      "learning_rate": 9.868977561802027e-05,
      "loss": 1.5525,
      "step": 353
    },
    {
      "epoch": 1.657109420713868,
      "grad_norm": 0.5408384799957275,
      "learning_rate": 9.867107493114617e-05,
      "loss": 1.5426,
      "step": 354
    },
    {
      "epoch": 1.6617905207723815,
      "grad_norm": 0.5473621487617493,
      "learning_rate": 9.865224352899119e-05,
      "loss": 1.5414,
      "step": 355
    },
    {
      "epoch": 1.6664716208308952,
      "grad_norm": 0.5682988166809082,
      "learning_rate": 9.86332814621306e-05,
      "loss": 1.5487,
      "step": 356
    },
    {
      "epoch": 1.671152720889409,
      "grad_norm": 0.5523189306259155,
      "learning_rate": 9.861418878149056e-05,
      "loss": 1.5656,
      "step": 357
    },
    {
      "epoch": 1.6758338209479229,
      "grad_norm": 0.5846049189567566,
      "learning_rate": 9.859496553834804e-05,
      "loss": 1.5517,
      "step": 358
    },
    {
      "epoch": 1.6805149210064365,
      "grad_norm": 0.5750234723091125,
      "learning_rate": 9.857561178433066e-05,
      "loss": 1.5317,
      "step": 359
    },
    {
      "epoch": 1.6851960210649501,
      "grad_norm": 0.5728639364242554,
      "learning_rate": 9.855612757141655e-05,
      "loss": 1.5326,
      "step": 360
    },
    {
      "epoch": 1.689877121123464,
      "grad_norm": 0.5863739252090454,
      "learning_rate": 9.85365129519342e-05,
      "loss": 1.5359,
      "step": 361
    },
    {
      "epoch": 1.6945582211819779,
      "grad_norm": 0.6063214540481567,
      "learning_rate": 9.851676797856235e-05,
      "loss": 1.5412,
      "step": 362
    },
    {
      "epoch": 1.6992393212404915,
      "grad_norm": 0.552778959274292,
      "learning_rate": 9.849689270432984e-05,
      "loss": 1.5518,
      "step": 363
    },
    {
      "epoch": 1.7039204212990051,
      "grad_norm": 0.5694995522499084,
      "learning_rate": 9.84768871826154e-05,
      "loss": 1.5323,
      "step": 364
    },
    {
      "epoch": 1.708601521357519,
      "grad_norm": 0.5607172846794128,
      "learning_rate": 9.845675146714763e-05,
      "loss": 1.5434,
      "step": 365
    },
    {
      "epoch": 1.7132826214160328,
      "grad_norm": 0.62750244140625,
      "learning_rate": 9.843648561200475e-05,
      "loss": 1.5401,
      "step": 366
    },
    {
      "epoch": 1.7179637214745465,
      "grad_norm": 0.559459924697876,
      "learning_rate": 9.841608967161451e-05,
      "loss": 1.5438,
      "step": 367
    },
    {
      "epoch": 1.7226448215330603,
      "grad_norm": 0.6135414242744446,
      "learning_rate": 9.839556370075404e-05,
      "loss": 1.5237,
      "step": 368
    },
    {
      "epoch": 1.7273259215915742,
      "grad_norm": 0.547193169593811,
      "learning_rate": 9.837490775454965e-05,
      "loss": 1.5201,
      "step": 369
    },
    {
      "epoch": 1.7320070216500878,
      "grad_norm": 0.5808367729187012,
      "learning_rate": 9.835412188847677e-05,
      "loss": 1.5461,
      "step": 370
    },
    {
      "epoch": 1.7366881217086014,
      "grad_norm": 0.5561200380325317,
      "learning_rate": 9.833320615835972e-05,
      "loss": 1.5562,
      "step": 371
    },
    {
      "epoch": 1.7413692217671153,
      "grad_norm": 0.549072265625,
      "learning_rate": 9.831216062037163e-05,
      "loss": 1.5218,
      "step": 372
    },
    {
      "epoch": 1.7460503218256291,
      "grad_norm": 0.6117122769355774,
      "learning_rate": 9.829098533103421e-05,
      "loss": 1.5527,
      "step": 373
    },
    {
      "epoch": 1.7507314218841428,
      "grad_norm": 0.5658581256866455,
      "learning_rate": 9.826968034721765e-05,
      "loss": 1.5346,
      "step": 374
    },
    {
      "epoch": 1.7554125219426564,
      "grad_norm": 0.5828503966331482,
      "learning_rate": 9.824824572614051e-05,
      "loss": 1.5191,
      "step": 375
    },
    {
      "epoch": 1.7600936220011703,
      "grad_norm": 0.5669796466827393,
      "learning_rate": 9.822668152536944e-05,
      "loss": 1.5269,
      "step": 376
    },
    {
      "epoch": 1.764774722059684,
      "grad_norm": 0.5611358284950256,
      "learning_rate": 9.820498780281917e-05,
      "loss": 1.536,
      "step": 377
    },
    {
      "epoch": 1.7694558221181977,
      "grad_norm": 0.5313979387283325,
      "learning_rate": 9.818316461675221e-05,
      "loss": 1.5189,
      "step": 378
    },
    {
      "epoch": 1.7741369221767114,
      "grad_norm": 0.5862874984741211,
      "learning_rate": 9.816121202577886e-05,
      "loss": 1.5261,
      "step": 379
    },
    {
      "epoch": 1.7788180222352252,
      "grad_norm": 0.5266857147216797,
      "learning_rate": 9.813913008885687e-05,
      "loss": 1.4967,
      "step": 380
    },
    {
      "epoch": 1.783499122293739,
      "grad_norm": 0.5802350044250488,
      "learning_rate": 9.811691886529144e-05,
      "loss": 1.5198,
      "step": 381
    },
    {
      "epoch": 1.7881802223522527,
      "grad_norm": 0.5663620829582214,
      "learning_rate": 9.809457841473496e-05,
      "loss": 1.5432,
      "step": 382
    },
    {
      "epoch": 1.7928613224107666,
      "grad_norm": 0.5301174521446228,
      "learning_rate": 9.80721087971869e-05,
      "loss": 1.5261,
      "step": 383
    },
    {
      "epoch": 1.7975424224692804,
      "grad_norm": 0.5568957924842834,
      "learning_rate": 9.804951007299359e-05,
      "loss": 1.5129,
      "step": 384
    },
    {
      "epoch": 1.802223522527794,
      "grad_norm": 0.5582797527313232,
      "learning_rate": 9.802678230284817e-05,
      "loss": 1.529,
      "step": 385
    },
    {
      "epoch": 1.8069046225863077,
      "grad_norm": 0.5716988444328308,
      "learning_rate": 9.800392554779031e-05,
      "loss": 1.5381,
      "step": 386
    },
    {
      "epoch": 1.8115857226448215,
      "grad_norm": 0.5414494276046753,
      "learning_rate": 9.798093986920608e-05,
      "loss": 1.5332,
      "step": 387
    },
    {
      "epoch": 1.8162668227033354,
      "grad_norm": 0.531887948513031,
      "learning_rate": 9.795782532882787e-05,
      "loss": 1.5408,
      "step": 388
    },
    {
      "epoch": 1.820947922761849,
      "grad_norm": 0.5692090392112732,
      "learning_rate": 9.793458198873409e-05,
      "loss": 1.5233,
      "step": 389
    },
    {
      "epoch": 1.8256290228203627,
      "grad_norm": 0.6099083423614502,
      "learning_rate": 9.791120991134904e-05,
      "loss": 1.5325,
      "step": 390
    },
    {
      "epoch": 1.8303101228788765,
      "grad_norm": 0.5257400870323181,
      "learning_rate": 9.788770915944283e-05,
      "loss": 1.5127,
      "step": 391
    },
    {
      "epoch": 1.8349912229373904,
      "grad_norm": 0.565856397151947,
      "learning_rate": 9.786407979613115e-05,
      "loss": 1.5515,
      "step": 392
    },
    {
      "epoch": 1.839672322995904,
      "grad_norm": 0.5161274671554565,
      "learning_rate": 9.784032188487506e-05,
      "loss": 1.535,
      "step": 393
    },
    {
      "epoch": 1.8443534230544176,
      "grad_norm": 0.5876962542533875,
      "learning_rate": 9.781643548948088e-05,
      "loss": 1.5211,
      "step": 394
    },
    {
      "epoch": 1.8490345231129317,
      "grad_norm": 0.5334473252296448,
      "learning_rate": 9.779242067409997e-05,
      "loss": 1.5147,
      "step": 395
    },
    {
      "epoch": 1.8537156231714453,
      "grad_norm": 0.5968698263168335,
      "learning_rate": 9.776827750322865e-05,
      "loss": 1.5136,
      "step": 396
    },
    {
      "epoch": 1.858396723229959,
      "grad_norm": 0.5792655348777771,
      "learning_rate": 9.77440060417079e-05,
      "loss": 1.5066,
      "step": 397
    },
    {
      "epoch": 1.8630778232884728,
      "grad_norm": 0.574500322341919,
      "learning_rate": 9.77196063547233e-05,
      "loss": 1.5238,
      "step": 398
    },
    {
      "epoch": 1.8677589233469867,
      "grad_norm": 0.565139889717102,
      "learning_rate": 9.769507850780474e-05,
      "loss": 1.5124,
      "step": 399
    },
    {
      "epoch": 1.8724400234055003,
      "grad_norm": 0.5808327198028564,
      "learning_rate": 9.767042256682637e-05,
      "loss": 1.5324,
      "step": 400
    },
    {
      "epoch": 1.877121123464014,
      "grad_norm": 0.5518105626106262,
      "learning_rate": 9.764563859800631e-05,
      "loss": 1.5302,
      "step": 401
    },
    {
      "epoch": 1.8818022235225278,
      "grad_norm": 0.5466589331626892,
      "learning_rate": 9.762072666790658e-05,
      "loss": 1.523,
      "step": 402
    },
    {
      "epoch": 1.8864833235810416,
      "grad_norm": 0.5499767065048218,
      "learning_rate": 9.759568684343279e-05,
      "loss": 1.5018,
      "step": 403
    },
    {
      "epoch": 1.8911644236395553,
      "grad_norm": 0.6103038787841797,
      "learning_rate": 9.757051919183411e-05,
      "loss": 1.5308,
      "step": 404
    },
    {
      "epoch": 1.895845523698069,
      "grad_norm": 0.5939445495605469,
      "learning_rate": 9.754522378070297e-05,
      "loss": 1.5091,
      "step": 405
    },
    {
      "epoch": 1.9005266237565828,
      "grad_norm": 0.5608471035957336,
      "learning_rate": 9.75198006779749e-05,
      "loss": 1.5341,
      "step": 406
    },
    {
      "epoch": 1.9052077238150966,
      "grad_norm": 0.5591056942939758,
      "learning_rate": 9.749424995192846e-05,
      "loss": 1.5248,
      "step": 407
    },
    {
      "epoch": 1.9098888238736103,
      "grad_norm": 0.5589863657951355,
      "learning_rate": 9.746857167118486e-05,
      "loss": 1.53,
      "step": 408
    },
    {
      "epoch": 1.9145699239321239,
      "grad_norm": 0.533953070640564,
      "learning_rate": 9.744276590470793e-05,
      "loss": 1.5411,
      "step": 409
    },
    {
      "epoch": 1.919251023990638,
      "grad_norm": 0.5937032699584961,
      "learning_rate": 9.741683272180391e-05,
      "loss": 1.5225,
      "step": 410
    },
    {
      "epoch": 1.9239321240491516,
      "grad_norm": 0.5463059544563293,
      "learning_rate": 9.73907721921212e-05,
      "loss": 1.5109,
      "step": 411
    },
    {
      "epoch": 1.9286132241076652,
      "grad_norm": 0.5598007440567017,
      "learning_rate": 9.736458438565023e-05,
      "loss": 1.5269,
      "step": 412
    },
    {
      "epoch": 1.933294324166179,
      "grad_norm": 0.5386488437652588,
      "learning_rate": 9.733826937272327e-05,
      "loss": 1.5212,
      "step": 413
    },
    {
      "epoch": 1.937975424224693,
      "grad_norm": 0.5409656763076782,
      "learning_rate": 9.731182722401421e-05,
      "loss": 1.5196,
      "step": 414
    },
    {
      "epoch": 1.9426565242832066,
      "grad_norm": 0.5412046313285828,
      "learning_rate": 9.728525801053836e-05,
      "loss": 1.5181,
      "step": 415
    },
    {
      "epoch": 1.9473376243417202,
      "grad_norm": 0.5616279244422913,
      "learning_rate": 9.725856180365236e-05,
      "loss": 1.5266,
      "step": 416
    },
    {
      "epoch": 1.952018724400234,
      "grad_norm": 0.5128246545791626,
      "learning_rate": 9.723173867505381e-05,
      "loss": 1.5208,
      "step": 417
    },
    {
      "epoch": 1.956699824458748,
      "grad_norm": 0.5439711213111877,
      "learning_rate": 9.720478869678132e-05,
      "loss": 1.4862,
      "step": 418
    },
    {
      "epoch": 1.9613809245172615,
      "grad_norm": 0.5718542337417603,
      "learning_rate": 9.717771194121404e-05,
      "loss": 1.5081,
      "step": 419
    },
    {
      "epoch": 1.9660620245757752,
      "grad_norm": 0.584561288356781,
      "learning_rate": 9.715050848107168e-05,
      "loss": 1.5213,
      "step": 420
    },
    {
      "epoch": 1.970743124634289,
      "grad_norm": 0.5634207725524902,
      "learning_rate": 9.712317838941423e-05,
      "loss": 1.5127,
      "step": 421
    },
    {
      "epoch": 1.9754242246928029,
      "grad_norm": 0.5634327530860901,
      "learning_rate": 9.709572173964177e-05,
      "loss": 1.5197,
      "step": 422
    },
    {
      "epoch": 1.9801053247513165,
      "grad_norm": 0.5960271954536438,
      "learning_rate": 9.706813860549424e-05,
      "loss": 1.5239,
      "step": 423
    },
    {
      "epoch": 1.9847864248098304,
      "grad_norm": 0.5209662318229675,
      "learning_rate": 9.704042906105136e-05,
      "loss": 1.5009,
      "step": 424
    },
    {
      "epoch": 1.9894675248683442,
      "grad_norm": 0.5604713559150696,
      "learning_rate": 9.701259318073226e-05,
      "loss": 1.5221,
      "step": 425
    },
    {
      "epoch": 1.9941486249268578,
      "grad_norm": 0.545125424861908,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.5277,
      "step": 426
    },
    {
      "epoch": 1.9988297249853715,
      "grad_norm": 0.5339157581329346,
      "learning_rate": 9.69565427118384e-05,
      "loss": 1.4962,
      "step": 427
    },
    {
      "epoch": 2.003510825043885,
      "grad_norm": 1.5908141136169434,
      "learning_rate": 9.692832827379768e-05,
      "loss": 1.4203,
      "step": 428
    },
    {
      "epoch": 2.008191925102399,
      "grad_norm": 0.6132961511611938,
      "learning_rate": 9.689998780094837e-05,
      "loss": 1.5247,
      "step": 429
    },
    {
      "epoch": 2.012873025160913,
      "grad_norm": 0.5658572912216187,
      "learning_rate": 9.687152136940417e-05,
      "loss": 1.5084,
      "step": 430
    },
    {
      "epoch": 2.0175541252194265,
      "grad_norm": 0.545052170753479,
      "learning_rate": 9.684292905561696e-05,
      "loss": 1.5128,
      "step": 431
    },
    {
      "epoch": 2.0222352252779405,
      "grad_norm": 0.5767326951026917,
      "learning_rate": 9.681421093637678e-05,
      "loss": 1.516,
      "step": 432
    },
    {
      "epoch": 2.026916325336454,
      "grad_norm": 0.5857481360435486,
      "learning_rate": 9.678536708881149e-05,
      "loss": 1.493,
      "step": 433
    },
    {
      "epoch": 2.031597425394968,
      "grad_norm": 0.5660700798034668,
      "learning_rate": 9.675639759038666e-05,
      "loss": 1.5246,
      "step": 434
    },
    {
      "epoch": 2.0362785254534814,
      "grad_norm": 0.5561010837554932,
      "learning_rate": 9.67273025189053e-05,
      "loss": 1.4908,
      "step": 435
    },
    {
      "epoch": 2.0409596255119955,
      "grad_norm": 0.5635765194892883,
      "learning_rate": 9.669808195250766e-05,
      "loss": 1.4967,
      "step": 436
    },
    {
      "epoch": 2.045640725570509,
      "grad_norm": 0.5964431762695312,
      "learning_rate": 9.666873596967105e-05,
      "loss": 1.5108,
      "step": 437
    },
    {
      "epoch": 2.0503218256290228,
      "grad_norm": 0.571668267250061,
      "learning_rate": 9.663926464920958e-05,
      "loss": 1.514,
      "step": 438
    },
    {
      "epoch": 2.0550029256875364,
      "grad_norm": 0.552706241607666,
      "learning_rate": 9.660966807027402e-05,
      "loss": 1.5147,
      "step": 439
    },
    {
      "epoch": 2.0596840257460505,
      "grad_norm": 0.5939638018608093,
      "learning_rate": 9.657994631235152e-05,
      "loss": 1.4984,
      "step": 440
    },
    {
      "epoch": 2.064365125804564,
      "grad_norm": 0.5550068616867065,
      "learning_rate": 9.655009945526541e-05,
      "loss": 1.5009,
      "step": 441
    },
    {
      "epoch": 2.0690462258630777,
      "grad_norm": 0.587196409702301,
      "learning_rate": 9.652012757917501e-05,
      "loss": 1.5032,
      "step": 442
    },
    {
      "epoch": 2.0737273259215914,
      "grad_norm": 0.5701729655265808,
      "learning_rate": 9.649003076457542e-05,
      "loss": 1.5315,
      "step": 443
    },
    {
      "epoch": 2.0784084259801054,
      "grad_norm": 0.5458940267562866,
      "learning_rate": 9.645980909229726e-05,
      "loss": 1.5028,
      "step": 444
    },
    {
      "epoch": 2.083089526038619,
      "grad_norm": 0.5555766224861145,
      "learning_rate": 9.642946264350648e-05,
      "loss": 1.4991,
      "step": 445
    },
    {
      "epoch": 2.0877706260971327,
      "grad_norm": 0.5843587517738342,
      "learning_rate": 9.639899149970415e-05,
      "loss": 1.5023,
      "step": 446
    },
    {
      "epoch": 2.092451726155647,
      "grad_norm": 0.5529284477233887,
      "learning_rate": 9.636839574272622e-05,
      "loss": 1.4814,
      "step": 447
    },
    {
      "epoch": 2.0971328262141604,
      "grad_norm": 0.5499287247657776,
      "learning_rate": 9.633767545474336e-05,
      "loss": 1.5233,
      "step": 448
    },
    {
      "epoch": 2.101813926272674,
      "grad_norm": 0.5526400208473206,
      "learning_rate": 9.630683071826062e-05,
      "loss": 1.5028,
      "step": 449
    },
    {
      "epoch": 2.1064950263311877,
      "grad_norm": 0.5946321487426758,
      "learning_rate": 9.627586161611732e-05,
      "loss": 1.5055,
      "step": 450
    },
    {
      "epoch": 2.1111761263897018,
      "grad_norm": 0.5409477949142456,
      "learning_rate": 9.624476823148678e-05,
      "loss": 1.4986,
      "step": 451
    },
    {
      "epoch": 2.1158572264482154,
      "grad_norm": 0.5718604922294617,
      "learning_rate": 9.62135506478761e-05,
      "loss": 1.5062,
      "step": 452
    },
    {
      "epoch": 2.120538326506729,
      "grad_norm": 0.5604239702224731,
      "learning_rate": 9.618220894912593e-05,
      "loss": 1.5252,
      "step": 453
    },
    {
      "epoch": 2.1252194265652427,
      "grad_norm": 0.5496664643287659,
      "learning_rate": 9.615074321941031e-05,
      "loss": 1.5092,
      "step": 454
    },
    {
      "epoch": 2.1299005266237567,
      "grad_norm": 0.548175036907196,
      "learning_rate": 9.61191535432363e-05,
      "loss": 1.5099,
      "step": 455
    },
    {
      "epoch": 2.1345816266822704,
      "grad_norm": 0.569646418094635,
      "learning_rate": 9.608744000544392e-05,
      "loss": 1.501,
      "step": 456
    },
    {
      "epoch": 2.139262726740784,
      "grad_norm": 0.5541195869445801,
      "learning_rate": 9.605560269120581e-05,
      "loss": 1.497,
      "step": 457
    },
    {
      "epoch": 2.1439438267992976,
      "grad_norm": 0.5321877598762512,
      "learning_rate": 9.6023641686027e-05,
      "loss": 1.4933,
      "step": 458
    },
    {
      "epoch": 2.1486249268578117,
      "grad_norm": 0.574272632598877,
      "learning_rate": 9.59915570757448e-05,
      "loss": 1.4919,
      "step": 459
    },
    {
      "epoch": 2.1533060269163253,
      "grad_norm": 0.5240908861160278,
      "learning_rate": 9.59593489465284e-05,
      "loss": 1.5041,
      "step": 460
    },
    {
      "epoch": 2.157987126974839,
      "grad_norm": 0.5542965531349182,
      "learning_rate": 9.592701738487877e-05,
      "loss": 1.4997,
      "step": 461
    },
    {
      "epoch": 2.162668227033353,
      "grad_norm": 0.562849760055542,
      "learning_rate": 9.589456247762839e-05,
      "loss": 1.5099,
      "step": 462
    },
    {
      "epoch": 2.1673493270918667,
      "grad_norm": 0.5381311178207397,
      "learning_rate": 9.586198431194095e-05,
      "loss": 1.4904,
      "step": 463
    },
    {
      "epoch": 2.1720304271503803,
      "grad_norm": 0.5613717436790466,
      "learning_rate": 9.582928297531124e-05,
      "loss": 1.4995,
      "step": 464
    },
    {
      "epoch": 2.176711527208894,
      "grad_norm": 0.529779314994812,
      "learning_rate": 9.57964585555648e-05,
      "loss": 1.4842,
      "step": 465
    },
    {
      "epoch": 2.181392627267408,
      "grad_norm": 0.5571693778038025,
      "learning_rate": 9.576351114085777e-05,
      "loss": 1.5059,
      "step": 466
    },
    {
      "epoch": 2.1860737273259216,
      "grad_norm": 0.5243085026741028,
      "learning_rate": 9.57304408196766e-05,
      "loss": 1.4824,
      "step": 467
    },
    {
      "epoch": 2.1907548273844353,
      "grad_norm": 0.55620938539505,
      "learning_rate": 9.56972476808378e-05,
      "loss": 1.4873,
      "step": 468
    },
    {
      "epoch": 2.195435927442949,
      "grad_norm": 0.5558146238327026,
      "learning_rate": 9.566393181348779e-05,
      "loss": 1.4936,
      "step": 469
    },
    {
      "epoch": 2.200117027501463,
      "grad_norm": 0.583473801612854,
      "learning_rate": 9.563049330710253e-05,
      "loss": 1.463,
      "step": 470
    },
    {
      "epoch": 2.2047981275599766,
      "grad_norm": 0.5721150636672974,
      "learning_rate": 9.559693225148743e-05,
      "loss": 1.5186,
      "step": 471
    },
    {
      "epoch": 2.2094792276184902,
      "grad_norm": 0.5524104237556458,
      "learning_rate": 9.556324873677695e-05,
      "loss": 1.4892,
      "step": 472
    },
    {
      "epoch": 2.2141603276770043,
      "grad_norm": 0.5464508533477783,
      "learning_rate": 9.55294428534345e-05,
      "loss": 1.4778,
      "step": 473
    },
    {
      "epoch": 2.218841427735518,
      "grad_norm": 0.548009991645813,
      "learning_rate": 9.549551469225208e-05,
      "loss": 1.5107,
      "step": 474
    },
    {
      "epoch": 2.2235225277940316,
      "grad_norm": 0.5196877717971802,
      "learning_rate": 9.546146434435015e-05,
      "loss": 1.5064,
      "step": 475
    },
    {
      "epoch": 2.228203627852545,
      "grad_norm": 0.553367555141449,
      "learning_rate": 9.542729190117727e-05,
      "loss": 1.499,
      "step": 476
    },
    {
      "epoch": 2.2328847279110593,
      "grad_norm": 0.5798020958900452,
      "learning_rate": 9.539299745450992e-05,
      "loss": 1.5104,
      "step": 477
    },
    {
      "epoch": 2.237565827969573,
      "grad_norm": 0.5476927757263184,
      "learning_rate": 9.535858109645228e-05,
      "loss": 1.4901,
      "step": 478
    },
    {
      "epoch": 2.2422469280280866,
      "grad_norm": 0.5290266871452332,
      "learning_rate": 9.532404291943592e-05,
      "loss": 1.4774,
      "step": 479
    },
    {
      "epoch": 2.2469280280866,
      "grad_norm": 0.5534362196922302,
      "learning_rate": 9.528938301621956e-05,
      "loss": 1.4947,
      "step": 480
    },
    {
      "epoch": 2.2516091281451143,
      "grad_norm": 0.5712639689445496,
      "learning_rate": 9.525460147988888e-05,
      "loss": 1.4757,
      "step": 481
    },
    {
      "epoch": 2.256290228203628,
      "grad_norm": 0.527601420879364,
      "learning_rate": 9.521969840385616e-05,
      "loss": 1.462,
      "step": 482
    },
    {
      "epoch": 2.2609713282621415,
      "grad_norm": 0.5285370945930481,
      "learning_rate": 9.51846738818602e-05,
      "loss": 1.4846,
      "step": 483
    },
    {
      "epoch": 2.265652428320655,
      "grad_norm": 0.5500218868255615,
      "learning_rate": 9.514952800796589e-05,
      "loss": 1.492,
      "step": 484
    },
    {
      "epoch": 2.2703335283791692,
      "grad_norm": 0.5409276485443115,
      "learning_rate": 9.511426087656401e-05,
      "loss": 1.4878,
      "step": 485
    },
    {
      "epoch": 2.275014628437683,
      "grad_norm": 0.5382949709892273,
      "learning_rate": 9.507887258237112e-05,
      "loss": 1.5157,
      "step": 486
    },
    {
      "epoch": 2.2796957284961965,
      "grad_norm": 0.5844248533248901,
      "learning_rate": 9.504336322042904e-05,
      "loss": 1.4953,
      "step": 487
    },
    {
      "epoch": 2.28437682855471,
      "grad_norm": 0.5399705767631531,
      "learning_rate": 9.500773288610484e-05,
      "loss": 1.4921,
      "step": 488
    },
    {
      "epoch": 2.289057928613224,
      "grad_norm": 0.5701977014541626,
      "learning_rate": 9.497198167509044e-05,
      "loss": 1.4958,
      "step": 489
    },
    {
      "epoch": 2.293739028671738,
      "grad_norm": 0.5427618026733398,
      "learning_rate": 9.49361096834024e-05,
      "loss": 1.4818,
      "step": 490
    },
    {
      "epoch": 2.2984201287302515,
      "grad_norm": 0.5544453263282776,
      "learning_rate": 9.49001170073817e-05,
      "loss": 1.4983,
      "step": 491
    },
    {
      "epoch": 2.3031012287887656,
      "grad_norm": 0.5501473546028137,
      "learning_rate": 9.48640037436934e-05,
      "loss": 1.4433,
      "step": 492
    },
    {
      "epoch": 2.307782328847279,
      "grad_norm": 0.5611422657966614,
      "learning_rate": 9.482776998932639e-05,
      "loss": 1.5056,
      "step": 493
    },
    {
      "epoch": 2.312463428905793,
      "grad_norm": 0.5366073250770569,
      "learning_rate": 9.479141584159324e-05,
      "loss": 1.5156,
      "step": 494
    },
    {
      "epoch": 2.3171445289643064,
      "grad_norm": 0.5340487957000732,
      "learning_rate": 9.475494139812979e-05,
      "loss": 1.4878,
      "step": 495
    },
    {
      "epoch": 2.3218256290228205,
      "grad_norm": 0.5424363017082214,
      "learning_rate": 9.471834675689499e-05,
      "loss": 1.4718,
      "step": 496
    },
    {
      "epoch": 2.326506729081334,
      "grad_norm": 0.5500884652137756,
      "learning_rate": 9.468163201617062e-05,
      "loss": 1.4989,
      "step": 497
    },
    {
      "epoch": 2.331187829139848,
      "grad_norm": 0.5656370520591736,
      "learning_rate": 9.464479727456097e-05,
      "loss": 1.494,
      "step": 498
    },
    {
      "epoch": 2.3358689291983614,
      "grad_norm": 0.5271655321121216,
      "learning_rate": 9.460784263099262e-05,
      "loss": 1.4938,
      "step": 499
    },
    {
      "epoch": 2.3405500292568755,
      "grad_norm": 0.5550000071525574,
      "learning_rate": 9.457076818471421e-05,
      "loss": 1.4722,
      "step": 500
    },
    {
      "epoch": 2.345231129315389,
      "grad_norm": 0.5688049793243408,
      "learning_rate": 9.453357403529609e-05,
      "loss": 1.5075,
      "step": 501
    },
    {
      "epoch": 2.3499122293739028,
      "grad_norm": 0.5786244869232178,
      "learning_rate": 9.449626028263011e-05,
      "loss": 1.4759,
      "step": 502
    },
    {
      "epoch": 2.354593329432417,
      "grad_norm": 0.5229558944702148,
      "learning_rate": 9.445882702692933e-05,
      "loss": 1.4841,
      "step": 503
    },
    {
      "epoch": 2.3592744294909305,
      "grad_norm": 0.530393123626709,
      "learning_rate": 9.442127436872778e-05,
      "loss": 1.4722,
      "step": 504
    },
    {
      "epoch": 2.363955529549444,
      "grad_norm": 0.5296366214752197,
      "learning_rate": 9.438360240888014e-05,
      "loss": 1.4728,
      "step": 505
    },
    {
      "epoch": 2.3686366296079577,
      "grad_norm": 0.5329553484916687,
      "learning_rate": 9.43458112485615e-05,
      "loss": 1.4928,
      "step": 506
    },
    {
      "epoch": 2.373317729666472,
      "grad_norm": 0.5391282439231873,
      "learning_rate": 9.430790098926711e-05,
      "loss": 1.4776,
      "step": 507
    },
    {
      "epoch": 2.3779988297249854,
      "grad_norm": 0.5178598761558533,
      "learning_rate": 9.426987173281208e-05,
      "loss": 1.5091,
      "step": 508
    },
    {
      "epoch": 2.382679929783499,
      "grad_norm": 0.5349472165107727,
      "learning_rate": 9.423172358133106e-05,
      "loss": 1.4791,
      "step": 509
    },
    {
      "epoch": 2.3873610298420127,
      "grad_norm": 0.5359641909599304,
      "learning_rate": 9.419345663727805e-05,
      "loss": 1.4864,
      "step": 510
    },
    {
      "epoch": 2.392042129900527,
      "grad_norm": 0.526164710521698,
      "learning_rate": 9.415507100342613e-05,
      "loss": 1.4795,
      "step": 511
    },
    {
      "epoch": 2.3967232299590404,
      "grad_norm": 0.5576329827308655,
      "learning_rate": 9.411656678286709e-05,
      "loss": 1.4974,
      "step": 512
    },
    {
      "epoch": 2.401404330017554,
      "grad_norm": 0.533876359462738,
      "learning_rate": 9.407794407901119e-05,
      "loss": 1.4956,
      "step": 513
    },
    {
      "epoch": 2.406085430076068,
      "grad_norm": 0.5780550241470337,
      "learning_rate": 9.403920299558697e-05,
      "loss": 1.4695,
      "step": 514
    },
    {
      "epoch": 2.4107665301345818,
      "grad_norm": 0.5714160799980164,
      "learning_rate": 9.400034363664083e-05,
      "loss": 1.478,
      "step": 515
    },
    {
      "epoch": 2.4154476301930954,
      "grad_norm": 0.5485972166061401,
      "learning_rate": 9.396136610653687e-05,
      "loss": 1.4872,
      "step": 516
    },
    {
      "epoch": 2.420128730251609,
      "grad_norm": 0.5578656196594238,
      "learning_rate": 9.392227050995653e-05,
      "loss": 1.4811,
      "step": 517
    },
    {
      "epoch": 2.4248098303101226,
      "grad_norm": 0.513107180595398,
      "learning_rate": 9.388305695189837e-05,
      "loss": 1.469,
      "step": 518
    },
    {
      "epoch": 2.4294909303686367,
      "grad_norm": 0.5163384079933167,
      "learning_rate": 9.384372553767771e-05,
      "loss": 1.4785,
      "step": 519
    },
    {
      "epoch": 2.4341720304271504,
      "grad_norm": 0.5341216921806335,
      "learning_rate": 9.380427637292646e-05,
      "loss": 1.4853,
      "step": 520
    },
    {
      "epoch": 2.438853130485664,
      "grad_norm": 0.5262408256530762,
      "learning_rate": 9.376470956359269e-05,
      "loss": 1.4675,
      "step": 521
    },
    {
      "epoch": 2.443534230544178,
      "grad_norm": 0.5297648310661316,
      "learning_rate": 9.372502521594052e-05,
      "loss": 1.4609,
      "step": 522
    },
    {
      "epoch": 2.4482153306026917,
      "grad_norm": 0.5424336194992065,
      "learning_rate": 9.368522343654967e-05,
      "loss": 1.5052,
      "step": 523
    },
    {
      "epoch": 2.4528964306612053,
      "grad_norm": 0.5182657837867737,
      "learning_rate": 9.364530433231524e-05,
      "loss": 1.4835,
      "step": 524
    },
    {
      "epoch": 2.457577530719719,
      "grad_norm": 0.5619286298751831,
      "learning_rate": 9.360526801044752e-05,
      "loss": 1.4511,
      "step": 525
    },
    {
      "epoch": 2.462258630778233,
      "grad_norm": 0.5724915266036987,
      "learning_rate": 9.356511457847148e-05,
      "loss": 1.4618,
      "step": 526
    },
    {
      "epoch": 2.4669397308367467,
      "grad_norm": 0.5449216365814209,
      "learning_rate": 9.352484414422674e-05,
      "loss": 1.4844,
      "step": 527
    },
    {
      "epoch": 2.4716208308952603,
      "grad_norm": 0.5833266973495483,
      "learning_rate": 9.348445681586702e-05,
      "loss": 1.4519,
      "step": 528
    },
    {
      "epoch": 2.476301930953774,
      "grad_norm": 0.5463898181915283,
      "learning_rate": 9.344395270186008e-05,
      "loss": 1.4771,
      "step": 529
    },
    {
      "epoch": 2.480983031012288,
      "grad_norm": 0.529358983039856,
      "learning_rate": 9.340333191098733e-05,
      "loss": 1.4712,
      "step": 530
    },
    {
      "epoch": 2.4856641310708016,
      "grad_norm": 0.5545377135276794,
      "learning_rate": 9.336259455234347e-05,
      "loss": 1.4848,
      "step": 531
    },
    {
      "epoch": 2.4903452311293153,
      "grad_norm": 0.5353593230247498,
      "learning_rate": 9.332174073533628e-05,
      "loss": 1.492,
      "step": 532
    },
    {
      "epoch": 2.4950263311878293,
      "grad_norm": 0.5352204442024231,
      "learning_rate": 9.328077056968639e-05,
      "loss": 1.4894,
      "step": 533
    },
    {
      "epoch": 2.499707431246343,
      "grad_norm": 0.5731208324432373,
      "learning_rate": 9.32396841654268e-05,
      "loss": 1.4775,
      "step": 534
    },
    {
      "epoch": 2.5043885313048566,
      "grad_norm": 0.5478562712669373,
      "learning_rate": 9.319848163290279e-05,
      "loss": 1.4688,
      "step": 535
    },
    {
      "epoch": 2.5090696313633702,
      "grad_norm": 0.5327141284942627,
      "learning_rate": 9.31571630827714e-05,
      "loss": 1.4791,
      "step": 536
    },
    {
      "epoch": 2.513750731421884,
      "grad_norm": 0.5497174859046936,
      "learning_rate": 9.311572862600139e-05,
      "loss": 1.4884,
      "step": 537
    },
    {
      "epoch": 2.518431831480398,
      "grad_norm": 0.5478792190551758,
      "learning_rate": 9.307417837387273e-05,
      "loss": 1.4544,
      "step": 538
    },
    {
      "epoch": 2.5231129315389116,
      "grad_norm": 0.5436558723449707,
      "learning_rate": 9.303251243797638e-05,
      "loss": 1.4773,
      "step": 539
    },
    {
      "epoch": 2.527794031597425,
      "grad_norm": 0.5304036140441895,
      "learning_rate": 9.299073093021405e-05,
      "loss": 1.4674,
      "step": 540
    },
    {
      "epoch": 2.5324751316559393,
      "grad_norm": 0.5399511456489563,
      "learning_rate": 9.294883396279774e-05,
      "loss": 1.4726,
      "step": 541
    },
    {
      "epoch": 2.537156231714453,
      "grad_norm": 0.5874165296554565,
      "learning_rate": 9.290682164824964e-05,
      "loss": 1.4801,
      "step": 542
    },
    {
      "epoch": 2.5418373317729666,
      "grad_norm": 0.54348224401474,
      "learning_rate": 9.286469409940168e-05,
      "loss": 1.4638,
      "step": 543
    },
    {
      "epoch": 2.5465184318314806,
      "grad_norm": 0.5654876232147217,
      "learning_rate": 9.282245142939528e-05,
      "loss": 1.4728,
      "step": 544
    },
    {
      "epoch": 2.5511995318899943,
      "grad_norm": 0.5578087568283081,
      "learning_rate": 9.278009375168102e-05,
      "loss": 1.4708,
      "step": 545
    },
    {
      "epoch": 2.555880631948508,
      "grad_norm": 0.5147175788879395,
      "learning_rate": 9.273762118001837e-05,
      "loss": 1.4613,
      "step": 546
    },
    {
      "epoch": 2.5605617320070215,
      "grad_norm": 0.5677844285964966,
      "learning_rate": 9.269503382847539e-05,
      "loss": 1.4805,
      "step": 547
    },
    {
      "epoch": 2.565242832065535,
      "grad_norm": 0.5775290131568909,
      "learning_rate": 9.265233181142836e-05,
      "loss": 1.4691,
      "step": 548
    },
    {
      "epoch": 2.5699239321240492,
      "grad_norm": 0.5376417636871338,
      "learning_rate": 9.260951524356157e-05,
      "loss": 1.4742,
      "step": 549
    },
    {
      "epoch": 2.574605032182563,
      "grad_norm": 0.5477780103683472,
      "learning_rate": 9.256658423986689e-05,
      "loss": 1.4852,
      "step": 550
    },
    {
      "epoch": 2.5792861322410765,
      "grad_norm": 0.5555460453033447,
      "learning_rate": 9.252353891564358e-05,
      "loss": 1.4658,
      "step": 551
    },
    {
      "epoch": 2.5839672322995906,
      "grad_norm": 0.5579097867012024,
      "learning_rate": 9.248037938649792e-05,
      "loss": 1.466,
      "step": 552
    },
    {
      "epoch": 2.588648332358104,
      "grad_norm": 0.5582410097122192,
      "learning_rate": 9.24371057683429e-05,
      "loss": 1.4521,
      "step": 553
    },
    {
      "epoch": 2.593329432416618,
      "grad_norm": 0.5532661080360413,
      "learning_rate": 9.239371817739793e-05,
      "loss": 1.4845,
      "step": 554
    },
    {
      "epoch": 2.598010532475132,
      "grad_norm": 0.5436837673187256,
      "learning_rate": 9.235021673018849e-05,
      "loss": 1.4701,
      "step": 555
    },
    {
      "epoch": 2.6026916325336455,
      "grad_norm": 0.5618695020675659,
      "learning_rate": 9.230660154354587e-05,
      "loss": 1.4644,
      "step": 556
    },
    {
      "epoch": 2.607372732592159,
      "grad_norm": 0.5521680116653442,
      "learning_rate": 9.22628727346068e-05,
      "loss": 1.4754,
      "step": 557
    },
    {
      "epoch": 2.612053832650673,
      "grad_norm": 0.5556138753890991,
      "learning_rate": 9.221903042081319e-05,
      "loss": 1.4681,
      "step": 558
    },
    {
      "epoch": 2.6167349327091864,
      "grad_norm": 0.538085401058197,
      "learning_rate": 9.21750747199118e-05,
      "loss": 1.4899,
      "step": 559
    },
    {
      "epoch": 2.6214160327677005,
      "grad_norm": 0.5571359992027283,
      "learning_rate": 9.213100574995386e-05,
      "loss": 1.4526,
      "step": 560
    },
    {
      "epoch": 2.626097132826214,
      "grad_norm": 0.5502709150314331,
      "learning_rate": 9.208682362929484e-05,
      "loss": 1.4645,
      "step": 561
    },
    {
      "epoch": 2.630778232884728,
      "grad_norm": 0.5496858358383179,
      "learning_rate": 9.204252847659411e-05,
      "loss": 1.4672,
      "step": 562
    },
    {
      "epoch": 2.635459332943242,
      "grad_norm": 0.5407517552375793,
      "learning_rate": 9.199812041081459e-05,
      "loss": 1.468,
      "step": 563
    },
    {
      "epoch": 2.6401404330017555,
      "grad_norm": 0.5413292050361633,
      "learning_rate": 9.195359955122244e-05,
      "loss": 1.4667,
      "step": 564
    },
    {
      "epoch": 2.644821533060269,
      "grad_norm": 0.5415422916412354,
      "learning_rate": 9.190896601738677e-05,
      "loss": 1.469,
      "step": 565
    },
    {
      "epoch": 2.6495026331187828,
      "grad_norm": 0.5778108239173889,
      "learning_rate": 9.186421992917928e-05,
      "loss": 1.463,
      "step": 566
    },
    {
      "epoch": 2.6541837331772964,
      "grad_norm": 0.5511682033538818,
      "learning_rate": 9.181936140677398e-05,
      "loss": 1.4663,
      "step": 567
    },
    {
      "epoch": 2.6588648332358105,
      "grad_norm": 0.5939563512802124,
      "learning_rate": 9.177439057064683e-05,
      "loss": 1.4887,
      "step": 568
    },
    {
      "epoch": 2.663545933294324,
      "grad_norm": 0.5491057634353638,
      "learning_rate": 9.17293075415754e-05,
      "loss": 1.4704,
      "step": 569
    },
    {
      "epoch": 2.6682270333528377,
      "grad_norm": 0.528983473777771,
      "learning_rate": 9.168411244063863e-05,
      "loss": 1.4699,
      "step": 570
    },
    {
      "epoch": 2.672908133411352,
      "grad_norm": 0.5451183915138245,
      "learning_rate": 9.163880538921638e-05,
      "loss": 1.4635,
      "step": 571
    },
    {
      "epoch": 2.6775892334698654,
      "grad_norm": 0.562751054763794,
      "learning_rate": 9.159338650898927e-05,
      "loss": 1.4656,
      "step": 572
    },
    {
      "epoch": 2.682270333528379,
      "grad_norm": 0.5425247550010681,
      "learning_rate": 9.15478559219382e-05,
      "loss": 1.4799,
      "step": 573
    },
    {
      "epoch": 2.686951433586893,
      "grad_norm": 0.5311916470527649,
      "learning_rate": 9.150221375034404e-05,
      "loss": 1.4587,
      "step": 574
    },
    {
      "epoch": 2.6916325336454068,
      "grad_norm": 0.5678891539573669,
      "learning_rate": 9.14564601167874e-05,
      "loss": 1.5004,
      "step": 575
    },
    {
      "epoch": 2.6963136337039204,
      "grad_norm": 0.5324888229370117,
      "learning_rate": 9.141059514414824e-05,
      "loss": 1.4589,
      "step": 576
    },
    {
      "epoch": 2.700994733762434,
      "grad_norm": 0.5441606044769287,
      "learning_rate": 9.136461895560553e-05,
      "loss": 1.4517,
      "step": 577
    },
    {
      "epoch": 2.7056758338209477,
      "grad_norm": 0.5288588404655457,
      "learning_rate": 9.131853167463692e-05,
      "loss": 1.4729,
      "step": 578
    },
    {
      "epoch": 2.7103569338794617,
      "grad_norm": 0.5603190660476685,
      "learning_rate": 9.12723334250184e-05,
      "loss": 1.4639,
      "step": 579
    },
    {
      "epoch": 2.7150380339379754,
      "grad_norm": 0.5756075382232666,
      "learning_rate": 9.122602433082408e-05,
      "loss": 1.4742,
      "step": 580
    },
    {
      "epoch": 2.719719133996489,
      "grad_norm": 0.5515328049659729,
      "learning_rate": 9.117960451642565e-05,
      "loss": 1.4475,
      "step": 581
    },
    {
      "epoch": 2.724400234055003,
      "grad_norm": 0.5273950695991516,
      "learning_rate": 9.11330741064922e-05,
      "loss": 1.4705,
      "step": 582
    },
    {
      "epoch": 2.7290813341135167,
      "grad_norm": 0.5433343052864075,
      "learning_rate": 9.108643322598991e-05,
      "loss": 1.4691,
      "step": 583
    },
    {
      "epoch": 2.7337624341720304,
      "grad_norm": 0.5376033782958984,
      "learning_rate": 9.103968200018154e-05,
      "loss": 1.4469,
      "step": 584
    },
    {
      "epoch": 2.7384435342305444,
      "grad_norm": 0.5662835240364075,
      "learning_rate": 9.09928205546263e-05,
      "loss": 1.468,
      "step": 585
    },
    {
      "epoch": 2.743124634289058,
      "grad_norm": 0.5775234699249268,
      "learning_rate": 9.094584901517933e-05,
      "loss": 1.4454,
      "step": 586
    },
    {
      "epoch": 2.7478057343475717,
      "grad_norm": 0.5586291551589966,
      "learning_rate": 9.089876750799152e-05,
      "loss": 1.4382,
      "step": 587
    },
    {
      "epoch": 2.7524868344060853,
      "grad_norm": 0.5502759218215942,
      "learning_rate": 9.085157615950906e-05,
      "loss": 1.4901,
      "step": 588
    },
    {
      "epoch": 2.757167934464599,
      "grad_norm": 0.5833450555801392,
      "learning_rate": 9.080427509647314e-05,
      "loss": 1.473,
      "step": 589
    },
    {
      "epoch": 2.761849034523113,
      "grad_norm": 0.5477181077003479,
      "learning_rate": 9.075686444591964e-05,
      "loss": 1.4784,
      "step": 590
    },
    {
      "epoch": 2.7665301345816267,
      "grad_norm": 0.5742630958557129,
      "learning_rate": 9.070934433517873e-05,
      "loss": 1.4615,
      "step": 591
    },
    {
      "epoch": 2.7712112346401403,
      "grad_norm": 0.5728650093078613,
      "learning_rate": 9.066171489187452e-05,
      "loss": 1.4766,
      "step": 592
    },
    {
      "epoch": 2.7758923346986544,
      "grad_norm": 0.5450015068054199,
      "learning_rate": 9.061397624392485e-05,
      "loss": 1.4638,
      "step": 593
    },
    {
      "epoch": 2.780573434757168,
      "grad_norm": 0.5659042000770569,
      "learning_rate": 9.056612851954077e-05,
      "loss": 1.4474,
      "step": 594
    },
    {
      "epoch": 2.7852545348156816,
      "grad_norm": 0.5580088496208191,
      "learning_rate": 9.051817184722629e-05,
      "loss": 1.4765,
      "step": 595
    },
    {
      "epoch": 2.7899356348741957,
      "grad_norm": 0.5821008086204529,
      "learning_rate": 9.047010635577805e-05,
      "loss": 1.457,
      "step": 596
    },
    {
      "epoch": 2.7946167349327093,
      "grad_norm": 0.5484012961387634,
      "learning_rate": 9.04219321742849e-05,
      "loss": 1.4658,
      "step": 597
    },
    {
      "epoch": 2.799297834991223,
      "grad_norm": 0.596698522567749,
      "learning_rate": 9.037364943212763e-05,
      "loss": 1.4584,
      "step": 598
    },
    {
      "epoch": 2.8039789350497366,
      "grad_norm": 0.5309833288192749,
      "learning_rate": 9.032525825897859e-05,
      "loss": 1.4387,
      "step": 599
    },
    {
      "epoch": 2.8086600351082502,
      "grad_norm": 0.6024175882339478,
      "learning_rate": 9.027675878480131e-05,
      "loss": 1.4539,
      "step": 600
    },
    {
      "epoch": 2.8133411351667643,
      "grad_norm": 0.5484504103660583,
      "learning_rate": 9.022815113985023e-05,
      "loss": 1.4633,
      "step": 601
    },
    {
      "epoch": 2.818022235225278,
      "grad_norm": 0.567604124546051,
      "learning_rate": 9.017943545467027e-05,
      "loss": 1.4596,
      "step": 602
    },
    {
      "epoch": 2.8227033352837916,
      "grad_norm": 0.5658008456230164,
      "learning_rate": 9.013061186009653e-05,
      "loss": 1.4869,
      "step": 603
    },
    {
      "epoch": 2.8273844353423057,
      "grad_norm": 0.5770115852355957,
      "learning_rate": 9.008168048725387e-05,
      "loss": 1.4671,
      "step": 604
    },
    {
      "epoch": 2.8320655354008193,
      "grad_norm": 0.5261843800544739,
      "learning_rate": 9.00326414675567e-05,
      "loss": 1.4276,
      "step": 605
    },
    {
      "epoch": 2.836746635459333,
      "grad_norm": 0.5513188242912292,
      "learning_rate": 8.998349493270846e-05,
      "loss": 1.4538,
      "step": 606
    },
    {
      "epoch": 2.8414277355178466,
      "grad_norm": 0.5463377237319946,
      "learning_rate": 8.993424101470136e-05,
      "loss": 1.4747,
      "step": 607
    },
    {
      "epoch": 2.84610883557636,
      "grad_norm": 0.5515256524085999,
      "learning_rate": 8.988487984581605e-05,
      "loss": 1.4463,
      "step": 608
    },
    {
      "epoch": 2.8507899356348743,
      "grad_norm": 0.5260566473007202,
      "learning_rate": 8.983541155862114e-05,
      "loss": 1.4626,
      "step": 609
    },
    {
      "epoch": 2.855471035693388,
      "grad_norm": 0.5268198251724243,
      "learning_rate": 8.9785836285973e-05,
      "loss": 1.4535,
      "step": 610
    },
    {
      "epoch": 2.8601521357519015,
      "grad_norm": 0.5588566064834595,
      "learning_rate": 8.97361541610153e-05,
      "loss": 1.4401,
      "step": 611
    },
    {
      "epoch": 2.8648332358104156,
      "grad_norm": 0.5275058746337891,
      "learning_rate": 8.968636531717868e-05,
      "loss": 1.4727,
      "step": 612
    },
    {
      "epoch": 2.8695143358689292,
      "grad_norm": 0.5546239018440247,
      "learning_rate": 8.963646988818042e-05,
      "loss": 1.482,
      "step": 613
    },
    {
      "epoch": 2.874195435927443,
      "grad_norm": 0.5342071056365967,
      "learning_rate": 8.958646800802401e-05,
      "loss": 1.4612,
      "step": 614
    },
    {
      "epoch": 2.878876535985957,
      "grad_norm": 0.5397853851318359,
      "learning_rate": 8.953635981099887e-05,
      "loss": 1.4502,
      "step": 615
    },
    {
      "epoch": 2.8835576360444706,
      "grad_norm": 0.5423009991645813,
      "learning_rate": 8.948614543167993e-05,
      "loss": 1.4689,
      "step": 616
    },
    {
      "epoch": 2.888238736102984,
      "grad_norm": 0.5220727920532227,
      "learning_rate": 8.94358250049273e-05,
      "loss": 1.4529,
      "step": 617
    },
    {
      "epoch": 2.892919836161498,
      "grad_norm": 0.5633693933486938,
      "learning_rate": 8.938539866588592e-05,
      "loss": 1.432,
      "step": 618
    },
    {
      "epoch": 2.8976009362200115,
      "grad_norm": 0.5513154864311218,
      "learning_rate": 8.933486654998515e-05,
      "loss": 1.4665,
      "step": 619
    },
    {
      "epoch": 2.9022820362785255,
      "grad_norm": 0.5335742831230164,
      "learning_rate": 8.928422879293845e-05,
      "loss": 1.4545,
      "step": 620
    },
    {
      "epoch": 2.906963136337039,
      "grad_norm": 0.556303083896637,
      "learning_rate": 8.9233485530743e-05,
      "loss": 1.4576,
      "step": 621
    },
    {
      "epoch": 2.911644236395553,
      "grad_norm": 0.5727290511131287,
      "learning_rate": 8.91826368996793e-05,
      "loss": 1.463,
      "step": 622
    },
    {
      "epoch": 2.916325336454067,
      "grad_norm": 0.5217007994651794,
      "learning_rate": 8.91316830363109e-05,
      "loss": 1.4659,
      "step": 623
    },
    {
      "epoch": 2.9210064365125805,
      "grad_norm": 0.5431603789329529,
      "learning_rate": 8.908062407748393e-05,
      "loss": 1.4581,
      "step": 624
    },
    {
      "epoch": 2.925687536571094,
      "grad_norm": 0.5315890312194824,
      "learning_rate": 8.902946016032676e-05,
      "loss": 1.4412,
      "step": 625
    },
    {
      "epoch": 2.9303686366296082,
      "grad_norm": 0.5527530908584595,
      "learning_rate": 8.897819142224967e-05,
      "loss": 1.4648,
      "step": 626
    },
    {
      "epoch": 2.935049736688122,
      "grad_norm": 0.5519826412200928,
      "learning_rate": 8.892681800094447e-05,
      "loss": 1.459,
      "step": 627
    },
    {
      "epoch": 2.9397308367466355,
      "grad_norm": 0.5363120436668396,
      "learning_rate": 8.887534003438408e-05,
      "loss": 1.4371,
      "step": 628
    },
    {
      "epoch": 2.944411936805149,
      "grad_norm": 0.5575461983680725,
      "learning_rate": 8.882375766082222e-05,
      "loss": 1.4323,
      "step": 629
    },
    {
      "epoch": 2.9490930368636628,
      "grad_norm": 0.5456847548484802,
      "learning_rate": 8.877207101879302e-05,
      "loss": 1.4561,
      "step": 630
    },
    {
      "epoch": 2.953774136922177,
      "grad_norm": 0.6018211245536804,
      "learning_rate": 8.87202802471106e-05,
      "loss": 1.4586,
      "step": 631
    },
    {
      "epoch": 2.9584552369806905,
      "grad_norm": 0.5407954454421997,
      "learning_rate": 8.86683854848688e-05,
      "loss": 1.4498,
      "step": 632
    },
    {
      "epoch": 2.963136337039204,
      "grad_norm": 0.5806236863136292,
      "learning_rate": 8.86163868714407e-05,
      "loss": 1.4716,
      "step": 633
    },
    {
      "epoch": 2.967817437097718,
      "grad_norm": 0.5460893511772156,
      "learning_rate": 8.856428454647831e-05,
      "loss": 1.4461,
      "step": 634
    },
    {
      "epoch": 2.972498537156232,
      "grad_norm": 0.5376127362251282,
      "learning_rate": 8.85120786499122e-05,
      "loss": 1.4402,
      "step": 635
    },
    {
      "epoch": 2.9771796372147454,
      "grad_norm": 0.557420015335083,
      "learning_rate": 8.845976932195103e-05,
      "loss": 1.4495,
      "step": 636
    },
    {
      "epoch": 2.981860737273259,
      "grad_norm": 0.5634372234344482,
      "learning_rate": 8.840735670308133e-05,
      "loss": 1.4728,
      "step": 637
    },
    {
      "epoch": 2.9865418373317727,
      "grad_norm": 0.5169801115989685,
      "learning_rate": 8.835484093406698e-05,
      "loss": 1.4493,
      "step": 638
    },
    {
      "epoch": 2.9912229373902868,
      "grad_norm": 0.555625319480896,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.4539,
      "step": 639
    },
    {
      "epoch": 2.9959040374488004,
      "grad_norm": 0.5352002382278442,
      "learning_rate": 8.82495005100447e-05,
      "loss": 1.4447,
      "step": 640
    },
    {
      "epoch": 3.000585137507314,
      "grad_norm": 1.3402049541473389,
      "learning_rate": 8.819667613794818e-05,
      "loss": 1.3743,
      "step": 641
    },
    {
      "epoch": 3.005266237565828,
      "grad_norm": 0.5957451462745667,
      "learning_rate": 8.81437491815291e-05,
      "loss": 1.4553,
      "step": 642
    },
    {
      "epoch": 3.0099473376243417,
      "grad_norm": 0.5585767030715942,
      "learning_rate": 8.809071978293271e-05,
      "loss": 1.4212,
      "step": 643
    },
    {
      "epoch": 3.0146284376828554,
      "grad_norm": 0.5555313229560852,
      "learning_rate": 8.803758808457939e-05,
      "loss": 1.4537,
      "step": 644
    },
    {
      "epoch": 3.019309537741369,
      "grad_norm": 0.5627725124359131,
      "learning_rate": 8.798435422916425e-05,
      "loss": 1.4445,
      "step": 645
    },
    {
      "epoch": 3.023990637799883,
      "grad_norm": 0.5429000854492188,
      "learning_rate": 8.793101835965678e-05,
      "loss": 1.4572,
      "step": 646
    },
    {
      "epoch": 3.0286717378583967,
      "grad_norm": 0.5517989993095398,
      "learning_rate": 8.787758061930043e-05,
      "loss": 1.4421,
      "step": 647
    },
    {
      "epoch": 3.0333528379169103,
      "grad_norm": 0.5385038256645203,
      "learning_rate": 8.782404115161225e-05,
      "loss": 1.4421,
      "step": 648
    },
    {
      "epoch": 3.0380339379754244,
      "grad_norm": 0.5344865322113037,
      "learning_rate": 8.777040010038251e-05,
      "loss": 1.4546,
      "step": 649
    },
    {
      "epoch": 3.042715038033938,
      "grad_norm": 0.5500159859657288,
      "learning_rate": 8.771665760967429e-05,
      "loss": 1.4635,
      "step": 650
    },
    {
      "epoch": 3.0473961380924517,
      "grad_norm": 0.5563318729400635,
      "learning_rate": 8.766281382382311e-05,
      "loss": 1.4304,
      "step": 651
    },
    {
      "epoch": 3.0520772381509653,
      "grad_norm": 0.5369390249252319,
      "learning_rate": 8.760886888743653e-05,
      "loss": 1.4385,
      "step": 652
    },
    {
      "epoch": 3.0567583382094794,
      "grad_norm": 0.5653578042984009,
      "learning_rate": 8.755482294539379e-05,
      "loss": 1.4691,
      "step": 653
    },
    {
      "epoch": 3.061439438267993,
      "grad_norm": 0.5696621537208557,
      "learning_rate": 8.750067614284534e-05,
      "loss": 1.4381,
      "step": 654
    },
    {
      "epoch": 3.0661205383265067,
      "grad_norm": 0.5326243042945862,
      "learning_rate": 8.74464286252126e-05,
      "loss": 1.4609,
      "step": 655
    },
    {
      "epoch": 3.0708016383850203,
      "grad_norm": 0.5605649352073669,
      "learning_rate": 8.73920805381874e-05,
      "loss": 1.4236,
      "step": 656
    },
    {
      "epoch": 3.0754827384435344,
      "grad_norm": 0.5605049133300781,
      "learning_rate": 8.733763202773171e-05,
      "loss": 1.4739,
      "step": 657
    },
    {
      "epoch": 3.080163838502048,
      "grad_norm": 0.5228269696235657,
      "learning_rate": 8.72830832400772e-05,
      "loss": 1.4376,
      "step": 658
    },
    {
      "epoch": 3.0848449385605616,
      "grad_norm": 0.5667708516120911,
      "learning_rate": 8.722843432172484e-05,
      "loss": 1.4552,
      "step": 659
    },
    {
      "epoch": 3.0895260386190753,
      "grad_norm": 0.5448529124259949,
      "learning_rate": 8.717368541944452e-05,
      "loss": 1.4637,
      "step": 660
    },
    {
      "epoch": 3.0942071386775893,
      "grad_norm": 0.5323814153671265,
      "learning_rate": 8.711883668027469e-05,
      "loss": 1.4236,
      "step": 661
    },
    {
      "epoch": 3.098888238736103,
      "grad_norm": 0.532503068447113,
      "learning_rate": 8.70638882515219e-05,
      "loss": 1.4307,
      "step": 662
    },
    {
      "epoch": 3.1035693387946166,
      "grad_norm": 0.5616405606269836,
      "learning_rate": 8.700884028076042e-05,
      "loss": 1.4563,
      "step": 663
    },
    {
      "epoch": 3.1082504388531307,
      "grad_norm": 0.533190906047821,
      "learning_rate": 8.695369291583188e-05,
      "loss": 1.4402,
      "step": 664
    },
    {
      "epoch": 3.1129315389116443,
      "grad_norm": 0.5488091111183167,
      "learning_rate": 8.689844630484485e-05,
      "loss": 1.4464,
      "step": 665
    },
    {
      "epoch": 3.117612638970158,
      "grad_norm": 0.5348528027534485,
      "learning_rate": 8.684310059617448e-05,
      "loss": 1.4316,
      "step": 666
    },
    {
      "epoch": 3.1222937390286716,
      "grad_norm": 0.5574296116828918,
      "learning_rate": 8.678765593846197e-05,
      "loss": 1.4475,
      "step": 667
    },
    {
      "epoch": 3.1269748390871857,
      "grad_norm": 0.5486092567443848,
      "learning_rate": 8.673211248061436e-05,
      "loss": 1.4496,
      "step": 668
    },
    {
      "epoch": 3.1316559391456993,
      "grad_norm": 0.5664199590682983,
      "learning_rate": 8.667647037180402e-05,
      "loss": 1.4527,
      "step": 669
    },
    {
      "epoch": 3.136337039204213,
      "grad_norm": 0.5176118016242981,
      "learning_rate": 8.66207297614682e-05,
      "loss": 1.4533,
      "step": 670
    },
    {
      "epoch": 3.1410181392627265,
      "grad_norm": 0.5346795916557312,
      "learning_rate": 8.656489079930878e-05,
      "loss": 1.4466,
      "step": 671
    },
    {
      "epoch": 3.1456992393212406,
      "grad_norm": 0.5439758896827698,
      "learning_rate": 8.650895363529173e-05,
      "loss": 1.464,
      "step": 672
    },
    {
      "epoch": 3.1503803393797543,
      "grad_norm": 0.5399249792098999,
      "learning_rate": 8.645291841964675e-05,
      "loss": 1.4438,
      "step": 673
    },
    {
      "epoch": 3.155061439438268,
      "grad_norm": 0.5242234468460083,
      "learning_rate": 8.639678530286693e-05,
      "loss": 1.4551,
      "step": 674
    },
    {
      "epoch": 3.159742539496782,
      "grad_norm": 0.5472708940505981,
      "learning_rate": 8.634055443570826e-05,
      "loss": 1.4368,
      "step": 675
    },
    {
      "epoch": 3.1644236395552956,
      "grad_norm": 0.5396461486816406,
      "learning_rate": 8.628422596918923e-05,
      "loss": 1.4461,
      "step": 676
    },
    {
      "epoch": 3.1691047396138092,
      "grad_norm": 0.5699884295463562,
      "learning_rate": 8.622780005459052e-05,
      "loss": 1.4359,
      "step": 677
    },
    {
      "epoch": 3.173785839672323,
      "grad_norm": 0.5235402584075928,
      "learning_rate": 8.617127684345445e-05,
      "loss": 1.4417,
      "step": 678
    },
    {
      "epoch": 3.178466939730837,
      "grad_norm": 0.5453900098800659,
      "learning_rate": 8.61146564875847e-05,
      "loss": 1.458,
      "step": 679
    },
    {
      "epoch": 3.1831480397893506,
      "grad_norm": 0.5414010286331177,
      "learning_rate": 8.60579391390458e-05,
      "loss": 1.4408,
      "step": 680
    },
    {
      "epoch": 3.187829139847864,
      "grad_norm": 0.5347760915756226,
      "learning_rate": 8.600112495016288e-05,
      "loss": 1.4272,
      "step": 681
    },
    {
      "epoch": 3.192510239906378,
      "grad_norm": 0.526759147644043,
      "learning_rate": 8.5944214073521e-05,
      "loss": 1.431,
      "step": 682
    },
    {
      "epoch": 3.197191339964892,
      "grad_norm": 0.540042519569397,
      "learning_rate": 8.588720666196499e-05,
      "loss": 1.4328,
      "step": 683
    },
    {
      "epoch": 3.2018724400234055,
      "grad_norm": 0.5147102475166321,
      "learning_rate": 8.583010286859896e-05,
      "loss": 1.4331,
      "step": 684
    },
    {
      "epoch": 3.206553540081919,
      "grad_norm": 0.5127415060997009,
      "learning_rate": 8.577290284678578e-05,
      "loss": 1.4137,
      "step": 685
    },
    {
      "epoch": 3.211234640140433,
      "grad_norm": 0.5615378618240356,
      "learning_rate": 8.571560675014687e-05,
      "loss": 1.4393,
      "step": 686
    },
    {
      "epoch": 3.215915740198947,
      "grad_norm": 0.5693033337593079,
      "learning_rate": 8.565821473256156e-05,
      "loss": 1.4372,
      "step": 687
    },
    {
      "epoch": 3.2205968402574605,
      "grad_norm": 0.5161866545677185,
      "learning_rate": 8.560072694816693e-05,
      "loss": 1.4578,
      "step": 688
    },
    {
      "epoch": 3.225277940315974,
      "grad_norm": 0.5499429106712341,
      "learning_rate": 8.554314355135711e-05,
      "loss": 1.4432,
      "step": 689
    },
    {
      "epoch": 3.2299590403744878,
      "grad_norm": 0.5418663620948792,
      "learning_rate": 8.548546469678311e-05,
      "loss": 1.4387,
      "step": 690
    },
    {
      "epoch": 3.234640140433002,
      "grad_norm": 0.5338912010192871,
      "learning_rate": 8.542769053935232e-05,
      "loss": 1.4345,
      "step": 691
    },
    {
      "epoch": 3.2393212404915155,
      "grad_norm": 0.5374161601066589,
      "learning_rate": 8.536982123422803e-05,
      "loss": 1.4377,
      "step": 692
    },
    {
      "epoch": 3.244002340550029,
      "grad_norm": 0.5633624792098999,
      "learning_rate": 8.531185693682908e-05,
      "loss": 1.4386,
      "step": 693
    },
    {
      "epoch": 3.248683440608543,
      "grad_norm": 0.5362398624420166,
      "learning_rate": 8.525379780282947e-05,
      "loss": 1.4442,
      "step": 694
    },
    {
      "epoch": 3.253364540667057,
      "grad_norm": 0.5384619832038879,
      "learning_rate": 8.519564398815783e-05,
      "loss": 1.4351,
      "step": 695
    },
    {
      "epoch": 3.2580456407255705,
      "grad_norm": 0.5073276162147522,
      "learning_rate": 8.513739564899715e-05,
      "loss": 1.4234,
      "step": 696
    },
    {
      "epoch": 3.262726740784084,
      "grad_norm": 0.5378078818321228,
      "learning_rate": 8.507905294178422e-05,
      "loss": 1.4497,
      "step": 697
    },
    {
      "epoch": 3.267407840842598,
      "grad_norm": 0.5337597727775574,
      "learning_rate": 8.502061602320933e-05,
      "loss": 1.4409,
      "step": 698
    },
    {
      "epoch": 3.272088940901112,
      "grad_norm": 0.5651869177818298,
      "learning_rate": 8.496208505021571e-05,
      "loss": 1.4611,
      "step": 699
    },
    {
      "epoch": 3.2767700409596254,
      "grad_norm": 0.5655667185783386,
      "learning_rate": 8.490346017999929e-05,
      "loss": 1.4335,
      "step": 700
    },
    {
      "epoch": 3.281451141018139,
      "grad_norm": 0.548124372959137,
      "learning_rate": 8.484474157000812e-05,
      "loss": 1.45,
      "step": 701
    },
    {
      "epoch": 3.286132241076653,
      "grad_norm": 0.5557027459144592,
      "learning_rate": 8.4785929377942e-05,
      "loss": 1.459,
      "step": 702
    },
    {
      "epoch": 3.2908133411351668,
      "grad_norm": 0.5695424675941467,
      "learning_rate": 8.47270237617521e-05,
      "loss": 1.4587,
      "step": 703
    },
    {
      "epoch": 3.2954944411936804,
      "grad_norm": 0.5496786832809448,
      "learning_rate": 8.466802487964047e-05,
      "loss": 1.4273,
      "step": 704
    },
    {
      "epoch": 3.3001755412521945,
      "grad_norm": 0.5440911054611206,
      "learning_rate": 8.460893289005965e-05,
      "loss": 1.4467,
      "step": 705
    },
    {
      "epoch": 3.304856641310708,
      "grad_norm": 0.5482668876647949,
      "learning_rate": 8.454974795171222e-05,
      "loss": 1.4618,
      "step": 706
    },
    {
      "epoch": 3.3095377413692217,
      "grad_norm": 0.562824010848999,
      "learning_rate": 8.449047022355043e-05,
      "loss": 1.4451,
      "step": 707
    },
    {
      "epoch": 3.3142188414277354,
      "grad_norm": 0.5314816832542419,
      "learning_rate": 8.443109986477573e-05,
      "loss": 1.4324,
      "step": 708
    },
    {
      "epoch": 3.3188999414862494,
      "grad_norm": 0.5433753728866577,
      "learning_rate": 8.437163703483827e-05,
      "loss": 1.4271,
      "step": 709
    },
    {
      "epoch": 3.323581041544763,
      "grad_norm": 0.5342699885368347,
      "learning_rate": 8.43120818934367e-05,
      "loss": 1.447,
      "step": 710
    },
    {
      "epoch": 3.3282621416032767,
      "grad_norm": 0.5455060005187988,
      "learning_rate": 8.425243460051739e-05,
      "loss": 1.4147,
      "step": 711
    },
    {
      "epoch": 3.3329432416617903,
      "grad_norm": 0.5278387665748596,
      "learning_rate": 8.419269531627439e-05,
      "loss": 1.4347,
      "step": 712
    },
    {
      "epoch": 3.3376243417203044,
      "grad_norm": 0.5744187235832214,
      "learning_rate": 8.41328642011487e-05,
      "loss": 1.4277,
      "step": 713
    },
    {
      "epoch": 3.342305441778818,
      "grad_norm": 0.5402573347091675,
      "learning_rate": 8.407294141582797e-05,
      "loss": 1.4347,
      "step": 714
    },
    {
      "epoch": 3.3469865418373317,
      "grad_norm": 0.5673272609710693,
      "learning_rate": 8.401292712124609e-05,
      "loss": 1.4444,
      "step": 715
    },
    {
      "epoch": 3.3516676418958458,
      "grad_norm": 0.5388854742050171,
      "learning_rate": 8.395282147858264e-05,
      "loss": 1.4518,
      "step": 716
    },
    {
      "epoch": 3.3563487419543594,
      "grad_norm": 0.5453377962112427,
      "learning_rate": 8.389262464926256e-05,
      "loss": 1.4257,
      "step": 717
    },
    {
      "epoch": 3.361029842012873,
      "grad_norm": 0.5486723184585571,
      "learning_rate": 8.383233679495577e-05,
      "loss": 1.4255,
      "step": 718
    },
    {
      "epoch": 3.3657109420713867,
      "grad_norm": 0.5446637272834778,
      "learning_rate": 8.377195807757652e-05,
      "loss": 1.4526,
      "step": 719
    },
    {
      "epoch": 3.3703920421299003,
      "grad_norm": 0.5255379676818848,
      "learning_rate": 8.371148865928319e-05,
      "loss": 1.4246,
      "step": 720
    },
    {
      "epoch": 3.3750731421884144,
      "grad_norm": 0.5421748757362366,
      "learning_rate": 8.36509287024777e-05,
      "loss": 1.4381,
      "step": 721
    },
    {
      "epoch": 3.379754242246928,
      "grad_norm": 0.5305628180503845,
      "learning_rate": 8.359027836980517e-05,
      "loss": 1.4356,
      "step": 722
    },
    {
      "epoch": 3.3844353423054416,
      "grad_norm": 0.5197787880897522,
      "learning_rate": 8.352953782415341e-05,
      "loss": 1.4422,
      "step": 723
    },
    {
      "epoch": 3.3891164423639557,
      "grad_norm": 0.521529495716095,
      "learning_rate": 8.346870722865251e-05,
      "loss": 1.429,
      "step": 724
    },
    {
      "epoch": 3.3937975424224693,
      "grad_norm": 0.5614930987358093,
      "learning_rate": 8.340778674667444e-05,
      "loss": 1.4318,
      "step": 725
    },
    {
      "epoch": 3.398478642480983,
      "grad_norm": 0.5214779376983643,
      "learning_rate": 8.334677654183254e-05,
      "loss": 1.3981,
      "step": 726
    },
    {
      "epoch": 3.4031597425394966,
      "grad_norm": 0.5232903957366943,
      "learning_rate": 8.328567677798115e-05,
      "loss": 1.4362,
      "step": 727
    },
    {
      "epoch": 3.4078408425980107,
      "grad_norm": 0.5370655655860901,
      "learning_rate": 8.322448761921511e-05,
      "loss": 1.4426,
      "step": 728
    },
    {
      "epoch": 3.4125219426565243,
      "grad_norm": 0.5535497069358826,
      "learning_rate": 8.316320922986935e-05,
      "loss": 1.4286,
      "step": 729
    },
    {
      "epoch": 3.417203042715038,
      "grad_norm": 0.5056213736534119,
      "learning_rate": 8.310184177451847e-05,
      "loss": 1.4402,
      "step": 730
    },
    {
      "epoch": 3.4218841427735516,
      "grad_norm": 0.5447074174880981,
      "learning_rate": 8.304038541797625e-05,
      "loss": 1.4167,
      "step": 731
    },
    {
      "epoch": 3.4265652428320656,
      "grad_norm": 0.5250815749168396,
      "learning_rate": 8.297884032529523e-05,
      "loss": 1.414,
      "step": 732
    },
    {
      "epoch": 3.4312463428905793,
      "grad_norm": 0.5571607351303101,
      "learning_rate": 8.29172066617663e-05,
      "loss": 1.4144,
      "step": 733
    },
    {
      "epoch": 3.435927442949093,
      "grad_norm": 0.551026463508606,
      "learning_rate": 8.285548459291817e-05,
      "loss": 1.4486,
      "step": 734
    },
    {
      "epoch": 3.440608543007607,
      "grad_norm": 0.5348817110061646,
      "learning_rate": 8.279367428451702e-05,
      "loss": 1.4424,
      "step": 735
    },
    {
      "epoch": 3.4452896430661206,
      "grad_norm": 0.5166352987289429,
      "learning_rate": 8.2731775902566e-05,
      "loss": 1.4447,
      "step": 736
    },
    {
      "epoch": 3.4499707431246343,
      "grad_norm": 0.5199401378631592,
      "learning_rate": 8.266978961330479e-05,
      "loss": 1.4445,
      "step": 737
    },
    {
      "epoch": 3.454651843183148,
      "grad_norm": 0.5190863609313965,
      "learning_rate": 8.260771558320919e-05,
      "loss": 1.437,
      "step": 738
    },
    {
      "epoch": 3.459332943241662,
      "grad_norm": 0.5204188227653503,
      "learning_rate": 8.254555397899061e-05,
      "loss": 1.4279,
      "step": 739
    },
    {
      "epoch": 3.4640140433001756,
      "grad_norm": 0.5144184827804565,
      "learning_rate": 8.248330496759568e-05,
      "loss": 1.4376,
      "step": 740
    },
    {
      "epoch": 3.4686951433586892,
      "grad_norm": 0.5174996256828308,
      "learning_rate": 8.242096871620578e-05,
      "loss": 1.4067,
      "step": 741
    },
    {
      "epoch": 3.473376243417203,
      "grad_norm": 0.5053052306175232,
      "learning_rate": 8.235854539223654e-05,
      "loss": 1.4425,
      "step": 742
    },
    {
      "epoch": 3.478057343475717,
      "grad_norm": 0.5285958051681519,
      "learning_rate": 8.229603516333753e-05,
      "loss": 1.4215,
      "step": 743
    },
    {
      "epoch": 3.4827384435342306,
      "grad_norm": 0.5122373104095459,
      "learning_rate": 8.223343819739164e-05,
      "loss": 1.4413,
      "step": 744
    },
    {
      "epoch": 3.487419543592744,
      "grad_norm": 0.5227556228637695,
      "learning_rate": 8.217075466251477e-05,
      "loss": 1.4256,
      "step": 745
    },
    {
      "epoch": 3.4921006436512583,
      "grad_norm": 0.5254419445991516,
      "learning_rate": 8.210798472705522e-05,
      "loss": 1.4248,
      "step": 746
    },
    {
      "epoch": 3.496781743709772,
      "grad_norm": 0.5361796617507935,
      "learning_rate": 8.204512855959345e-05,
      "loss": 1.4356,
      "step": 747
    },
    {
      "epoch": 3.5014628437682855,
      "grad_norm": 0.5470665693283081,
      "learning_rate": 8.198218632894145e-05,
      "loss": 1.4229,
      "step": 748
    },
    {
      "epoch": 3.506143943826799,
      "grad_norm": 0.5319539904594421,
      "learning_rate": 8.191915820414237e-05,
      "loss": 1.4288,
      "step": 749
    },
    {
      "epoch": 3.510825043885313,
      "grad_norm": 0.5314258337020874,
      "learning_rate": 8.185604435447002e-05,
      "loss": 1.4321,
      "step": 750
    },
    {
      "epoch": 3.515506143943827,
      "grad_norm": 0.55251544713974,
      "learning_rate": 8.179284494942846e-05,
      "loss": 1.435,
      "step": 751
    },
    {
      "epoch": 3.5201872440023405,
      "grad_norm": 0.550113320350647,
      "learning_rate": 8.172956015875151e-05,
      "loss": 1.4424,
      "step": 752
    },
    {
      "epoch": 3.524868344060854,
      "grad_norm": 0.5319443345069885,
      "learning_rate": 8.166619015240236e-05,
      "loss": 1.4439,
      "step": 753
    },
    {
      "epoch": 3.529549444119368,
      "grad_norm": 0.5305471420288086,
      "learning_rate": 8.160273510057296e-05,
      "loss": 1.4407,
      "step": 754
    },
    {
      "epoch": 3.534230544177882,
      "grad_norm": 0.5263088345527649,
      "learning_rate": 8.15391951736838e-05,
      "loss": 1.4359,
      "step": 755
    },
    {
      "epoch": 3.5389116442363955,
      "grad_norm": 0.5275413990020752,
      "learning_rate": 8.147557054238317e-05,
      "loss": 1.4003,
      "step": 756
    },
    {
      "epoch": 3.5435927442949096,
      "grad_norm": 0.5371294617652893,
      "learning_rate": 8.141186137754698e-05,
      "loss": 1.4161,
      "step": 757
    },
    {
      "epoch": 3.548273844353423,
      "grad_norm": 0.4912378191947937,
      "learning_rate": 8.13480678502781e-05,
      "loss": 1.4107,
      "step": 758
    },
    {
      "epoch": 3.552954944411937,
      "grad_norm": 0.5293853878974915,
      "learning_rate": 8.128419013190598e-05,
      "loss": 1.3992,
      "step": 759
    },
    {
      "epoch": 3.5576360444704505,
      "grad_norm": 0.5455768704414368,
      "learning_rate": 8.12202283939862e-05,
      "loss": 1.4105,
      "step": 760
    },
    {
      "epoch": 3.562317144528964,
      "grad_norm": 0.5033174157142639,
      "learning_rate": 8.115618280829997e-05,
      "loss": 1.4323,
      "step": 761
    },
    {
      "epoch": 3.566998244587478,
      "grad_norm": 0.5291778445243835,
      "learning_rate": 8.109205354685368e-05,
      "loss": 1.4055,
      "step": 762
    },
    {
      "epoch": 3.571679344645992,
      "grad_norm": 0.5420578718185425,
      "learning_rate": 8.102784078187851e-05,
      "loss": 1.4305,
      "step": 763
    },
    {
      "epoch": 3.5763604447045054,
      "grad_norm": 0.5107989311218262,
      "learning_rate": 8.096354468582981e-05,
      "loss": 1.4081,
      "step": 764
    },
    {
      "epoch": 3.5810415447630195,
      "grad_norm": 0.5385276675224304,
      "learning_rate": 8.089916543138681e-05,
      "loss": 1.4376,
      "step": 765
    },
    {
      "epoch": 3.585722644821533,
      "grad_norm": 0.5747690796852112,
      "learning_rate": 8.083470319145203e-05,
      "loss": 1.4271,
      "step": 766
    },
    {
      "epoch": 3.5904037448800468,
      "grad_norm": 0.5204617977142334,
      "learning_rate": 8.077015813915088e-05,
      "loss": 1.4334,
      "step": 767
    },
    {
      "epoch": 3.5950848449385604,
      "grad_norm": 0.560840904712677,
      "learning_rate": 8.070553044783119e-05,
      "loss": 1.4072,
      "step": 768
    },
    {
      "epoch": 3.5997659449970745,
      "grad_norm": 0.5681152939796448,
      "learning_rate": 8.064082029106271e-05,
      "loss": 1.4163,
      "step": 769
    },
    {
      "epoch": 3.604447045055588,
      "grad_norm": 0.5297346711158752,
      "learning_rate": 8.05760278426367e-05,
      "loss": 1.4132,
      "step": 770
    },
    {
      "epoch": 3.6091281451141017,
      "grad_norm": 0.5584424734115601,
      "learning_rate": 8.051115327656538e-05,
      "loss": 1.4256,
      "step": 771
    },
    {
      "epoch": 3.6138092451726154,
      "grad_norm": 0.5776448249816895,
      "learning_rate": 8.044619676708153e-05,
      "loss": 1.437,
      "step": 772
    },
    {
      "epoch": 3.6184903452311294,
      "grad_norm": 0.523693859577179,
      "learning_rate": 8.038115848863805e-05,
      "loss": 1.4335,
      "step": 773
    },
    {
      "epoch": 3.623171445289643,
      "grad_norm": 0.52964186668396,
      "learning_rate": 8.031603861590737e-05,
      "loss": 1.4266,
      "step": 774
    },
    {
      "epoch": 3.6278525453481567,
      "grad_norm": 0.5207391977310181,
      "learning_rate": 8.025083732378114e-05,
      "loss": 1.4182,
      "step": 775
    },
    {
      "epoch": 3.632533645406671,
      "grad_norm": 0.5130608677864075,
      "learning_rate": 8.01855547873696e-05,
      "loss": 1.4123,
      "step": 776
    },
    {
      "epoch": 3.6372147454651844,
      "grad_norm": 0.5629760026931763,
      "learning_rate": 8.012019118200123e-05,
      "loss": 1.4324,
      "step": 777
    },
    {
      "epoch": 3.641895845523698,
      "grad_norm": 0.5352134108543396,
      "learning_rate": 8.005474668322221e-05,
      "loss": 1.4471,
      "step": 778
    },
    {
      "epoch": 3.6465769455822117,
      "grad_norm": 0.5482569932937622,
      "learning_rate": 7.998922146679602e-05,
      "loss": 1.4192,
      "step": 779
    },
    {
      "epoch": 3.6512580456407253,
      "grad_norm": 0.5770877599716187,
      "learning_rate": 7.992361570870288e-05,
      "loss": 1.4312,
      "step": 780
    },
    {
      "epoch": 3.6559391456992394,
      "grad_norm": 0.540606677532196,
      "learning_rate": 7.985792958513931e-05,
      "loss": 1.441,
      "step": 781
    },
    {
      "epoch": 3.660620245757753,
      "grad_norm": 0.5749906301498413,
      "learning_rate": 7.979216327251772e-05,
      "loss": 1.4313,
      "step": 782
    },
    {
      "epoch": 3.6653013458162667,
      "grad_norm": 0.5137084126472473,
      "learning_rate": 7.972631694746583e-05,
      "loss": 1.4219,
      "step": 783
    },
    {
      "epoch": 3.6699824458747807,
      "grad_norm": 0.5164451003074646,
      "learning_rate": 7.966039078682627e-05,
      "loss": 1.4088,
      "step": 784
    },
    {
      "epoch": 3.6746635459332944,
      "grad_norm": 0.5294907093048096,
      "learning_rate": 7.959438496765611e-05,
      "loss": 1.4178,
      "step": 785
    },
    {
      "epoch": 3.679344645991808,
      "grad_norm": 0.5210172533988953,
      "learning_rate": 7.95282996672263e-05,
      "loss": 1.4223,
      "step": 786
    },
    {
      "epoch": 3.684025746050322,
      "grad_norm": 0.5051605105400085,
      "learning_rate": 7.946213506302129e-05,
      "loss": 1.4262,
      "step": 787
    },
    {
      "epoch": 3.6887068461088357,
      "grad_norm": 0.52712082862854,
      "learning_rate": 7.939589133273853e-05,
      "loss": 1.4094,
      "step": 788
    },
    {
      "epoch": 3.6933879461673493,
      "grad_norm": 0.5296499133110046,
      "learning_rate": 7.932956865428791e-05,
      "loss": 1.4226,
      "step": 789
    },
    {
      "epoch": 3.698069046225863,
      "grad_norm": 0.5116936564445496,
      "learning_rate": 7.926316720579144e-05,
      "loss": 1.4022,
      "step": 790
    },
    {
      "epoch": 3.7027501462843766,
      "grad_norm": 0.5399641394615173,
      "learning_rate": 7.919668716558262e-05,
      "loss": 1.423,
      "step": 791
    },
    {
      "epoch": 3.7074312463428907,
      "grad_norm": 0.5201460123062134,
      "learning_rate": 7.913012871220604e-05,
      "loss": 1.4254,
      "step": 792
    },
    {
      "epoch": 3.7121123464014043,
      "grad_norm": 0.5204175114631653,
      "learning_rate": 7.906349202441687e-05,
      "loss": 1.4126,
      "step": 793
    },
    {
      "epoch": 3.716793446459918,
      "grad_norm": 0.5481407046318054,
      "learning_rate": 7.899677728118043e-05,
      "loss": 1.418,
      "step": 794
    },
    {
      "epoch": 3.721474546518432,
      "grad_norm": 0.5074529051780701,
      "learning_rate": 7.892998466167165e-05,
      "loss": 1.3903,
      "step": 795
    },
    {
      "epoch": 3.7261556465769456,
      "grad_norm": 0.5236367583274841,
      "learning_rate": 7.886311434527459e-05,
      "loss": 1.4446,
      "step": 796
    },
    {
      "epoch": 3.7308367466354593,
      "grad_norm": 0.5233679413795471,
      "learning_rate": 7.879616651158201e-05,
      "loss": 1.3958,
      "step": 797
    },
    {
      "epoch": 3.7355178466939734,
      "grad_norm": 0.5118744969367981,
      "learning_rate": 7.872914134039484e-05,
      "loss": 1.4242,
      "step": 798
    },
    {
      "epoch": 3.740198946752487,
      "grad_norm": 0.5215593576431274,
      "learning_rate": 7.866203901172175e-05,
      "loss": 1.4231,
      "step": 799
    },
    {
      "epoch": 3.7448800468110006,
      "grad_norm": 0.5329627990722656,
      "learning_rate": 7.859485970577856e-05,
      "loss": 1.431,
      "step": 800
    },
    {
      "epoch": 3.7495611468695142,
      "grad_norm": 0.5146481990814209,
      "learning_rate": 7.85276036029879e-05,
      "loss": 1.392,
      "step": 801
    },
    {
      "epoch": 3.754242246928028,
      "grad_norm": 0.5273050665855408,
      "learning_rate": 7.846027088397862e-05,
      "loss": 1.4377,
      "step": 802
    },
    {
      "epoch": 3.758923346986542,
      "grad_norm": 0.5356155633926392,
      "learning_rate": 7.83928617295853e-05,
      "loss": 1.4286,
      "step": 803
    },
    {
      "epoch": 3.7636044470450556,
      "grad_norm": 0.5126620531082153,
      "learning_rate": 7.83253763208479e-05,
      "loss": 1.4101,
      "step": 804
    },
    {
      "epoch": 3.768285547103569,
      "grad_norm": 0.5486011505126953,
      "learning_rate": 7.825781483901109e-05,
      "loss": 1.4076,
      "step": 805
    },
    {
      "epoch": 3.7729666471620833,
      "grad_norm": 0.5352131128311157,
      "learning_rate": 7.819017746552387e-05,
      "loss": 1.3961,
      "step": 806
    },
    {
      "epoch": 3.777647747220597,
      "grad_norm": 0.5462915301322937,
      "learning_rate": 7.812246438203904e-05,
      "loss": 1.4247,
      "step": 807
    },
    {
      "epoch": 3.7823288472791106,
      "grad_norm": 0.5541925430297852,
      "learning_rate": 7.80546757704128e-05,
      "loss": 1.4295,
      "step": 808
    },
    {
      "epoch": 3.787009947337624,
      "grad_norm": 0.5532597303390503,
      "learning_rate": 7.798681181270413e-05,
      "loss": 1.413,
      "step": 809
    },
    {
      "epoch": 3.791691047396138,
      "grad_norm": 0.5388747453689575,
      "learning_rate": 7.791887269117442e-05,
      "loss": 1.4304,
      "step": 810
    },
    {
      "epoch": 3.796372147454652,
      "grad_norm": 0.5408781170845032,
      "learning_rate": 7.785085858828685e-05,
      "loss": 1.4126,
      "step": 811
    },
    {
      "epoch": 3.8010532475131655,
      "grad_norm": 0.5439730882644653,
      "learning_rate": 7.778276968670606e-05,
      "loss": 1.4035,
      "step": 812
    },
    {
      "epoch": 3.805734347571679,
      "grad_norm": 0.5271381735801697,
      "learning_rate": 7.77146061692975e-05,
      "loss": 1.4185,
      "step": 813
    },
    {
      "epoch": 3.8104154476301932,
      "grad_norm": 0.5220599174499512,
      "learning_rate": 7.764636821912711e-05,
      "loss": 1.3992,
      "step": 814
    },
    {
      "epoch": 3.815096547688707,
      "grad_norm": 0.5409356951713562,
      "learning_rate": 7.757805601946064e-05,
      "loss": 1.4282,
      "step": 815
    },
    {
      "epoch": 3.8197776477472205,
      "grad_norm": 0.5270222425460815,
      "learning_rate": 7.750966975376328e-05,
      "loss": 1.43,
      "step": 816
    },
    {
      "epoch": 3.8244587478057346,
      "grad_norm": 0.5513463616371155,
      "learning_rate": 7.744120960569918e-05,
      "loss": 1.4232,
      "step": 817
    },
    {
      "epoch": 3.829139847864248,
      "grad_norm": 0.5306671857833862,
      "learning_rate": 7.737267575913083e-05,
      "loss": 1.4234,
      "step": 818
    },
    {
      "epoch": 3.833820947922762,
      "grad_norm": 0.519351601600647,
      "learning_rate": 7.730406839811874e-05,
      "loss": 1.4073,
      "step": 819
    },
    {
      "epoch": 3.8385020479812755,
      "grad_norm": 0.5287462472915649,
      "learning_rate": 7.723538770692081e-05,
      "loss": 1.4187,
      "step": 820
    },
    {
      "epoch": 3.843183148039789,
      "grad_norm": 0.5309276580810547,
      "learning_rate": 7.71666338699919e-05,
      "loss": 1.4321,
      "step": 821
    },
    {
      "epoch": 3.847864248098303,
      "grad_norm": 0.5215145945549011,
      "learning_rate": 7.709780707198328e-05,
      "loss": 1.4212,
      "step": 822
    },
    {
      "epoch": 3.852545348156817,
      "grad_norm": 0.5737372636795044,
      "learning_rate": 7.70289074977422e-05,
      "loss": 1.4051,
      "step": 823
    },
    {
      "epoch": 3.8572264482153304,
      "grad_norm": 0.5519363880157471,
      "learning_rate": 7.69599353323114e-05,
      "loss": 1.4322,
      "step": 824
    },
    {
      "epoch": 3.8619075482738445,
      "grad_norm": 0.5450279712677002,
      "learning_rate": 7.68908907609285e-05,
      "loss": 1.4188,
      "step": 825
    },
    {
      "epoch": 3.866588648332358,
      "grad_norm": 0.6025644540786743,
      "learning_rate": 7.682177396902566e-05,
      "loss": 1.4205,
      "step": 826
    },
    {
      "epoch": 3.871269748390872,
      "grad_norm": 0.5163719654083252,
      "learning_rate": 7.67525851422289e-05,
      "loss": 1.43,
      "step": 827
    },
    {
      "epoch": 3.875950848449386,
      "grad_norm": 0.5643473863601685,
      "learning_rate": 7.668332446635779e-05,
      "loss": 1.4233,
      "step": 828
    },
    {
      "epoch": 3.8806319485078995,
      "grad_norm": 0.5531340837478638,
      "learning_rate": 7.661399212742487e-05,
      "loss": 1.4196,
      "step": 829
    },
    {
      "epoch": 3.885313048566413,
      "grad_norm": 0.5392515063285828,
      "learning_rate": 7.654458831163506e-05,
      "loss": 1.4164,
      "step": 830
    },
    {
      "epoch": 3.8899941486249268,
      "grad_norm": 0.5888575911521912,
      "learning_rate": 7.647511320538536e-05,
      "loss": 1.419,
      "step": 831
    },
    {
      "epoch": 3.8946752486834404,
      "grad_norm": 0.542989194393158,
      "learning_rate": 7.64055669952641e-05,
      "loss": 1.3953,
      "step": 832
    },
    {
      "epoch": 3.8993563487419545,
      "grad_norm": 0.5783752202987671,
      "learning_rate": 7.633594986805069e-05,
      "loss": 1.3971,
      "step": 833
    },
    {
      "epoch": 3.904037448800468,
      "grad_norm": 0.5455073118209839,
      "learning_rate": 7.626626201071494e-05,
      "loss": 1.4088,
      "step": 834
    },
    {
      "epoch": 3.9087185488589817,
      "grad_norm": 0.6045266389846802,
      "learning_rate": 7.619650361041663e-05,
      "loss": 1.4383,
      "step": 835
    },
    {
      "epoch": 3.913399648917496,
      "grad_norm": 0.5085153579711914,
      "learning_rate": 7.612667485450502e-05,
      "loss": 1.4218,
      "step": 836
    },
    {
      "epoch": 3.9180807489760094,
      "grad_norm": 0.5967952013015747,
      "learning_rate": 7.605677593051827e-05,
      "loss": 1.4266,
      "step": 837
    },
    {
      "epoch": 3.922761849034523,
      "grad_norm": 0.560104250907898,
      "learning_rate": 7.598680702618304e-05,
      "loss": 1.4135,
      "step": 838
    },
    {
      "epoch": 3.9274429490930367,
      "grad_norm": 0.5937385559082031,
      "learning_rate": 7.591676832941394e-05,
      "loss": 1.4419,
      "step": 839
    },
    {
      "epoch": 3.9321240491515503,
      "grad_norm": 0.616173505783081,
      "learning_rate": 7.584666002831296e-05,
      "loss": 1.4358,
      "step": 840
    },
    {
      "epoch": 3.9368051492100644,
      "grad_norm": 0.5356134176254272,
      "learning_rate": 7.577648231116907e-05,
      "loss": 1.3979,
      "step": 841
    },
    {
      "epoch": 3.941486249268578,
      "grad_norm": 0.5758895874023438,
      "learning_rate": 7.570623536645768e-05,
      "loss": 1.4052,
      "step": 842
    },
    {
      "epoch": 3.9461673493270917,
      "grad_norm": 0.5662379860877991,
      "learning_rate": 7.563591938284011e-05,
      "loss": 1.418,
      "step": 843
    },
    {
      "epoch": 3.9508484493856058,
      "grad_norm": 0.5756869912147522,
      "learning_rate": 7.556553454916308e-05,
      "loss": 1.3976,
      "step": 844
    },
    {
      "epoch": 3.9555295494441194,
      "grad_norm": 0.5514070391654968,
      "learning_rate": 7.549508105445825e-05,
      "loss": 1.422,
      "step": 845
    },
    {
      "epoch": 3.960210649502633,
      "grad_norm": 0.5394649505615234,
      "learning_rate": 7.542455908794163e-05,
      "loss": 1.407,
      "step": 846
    },
    {
      "epoch": 3.964891749561147,
      "grad_norm": 0.5769184827804565,
      "learning_rate": 7.535396883901322e-05,
      "loss": 1.415,
      "step": 847
    },
    {
      "epoch": 3.9695728496196607,
      "grad_norm": 0.5303329229354858,
      "learning_rate": 7.528331049725627e-05,
      "loss": 1.4326,
      "step": 848
    },
    {
      "epoch": 3.9742539496781744,
      "grad_norm": 0.5458217859268188,
      "learning_rate": 7.521258425243705e-05,
      "loss": 1.4167,
      "step": 849
    },
    {
      "epoch": 3.978935049736688,
      "grad_norm": 0.5328347086906433,
      "learning_rate": 7.514179029450409e-05,
      "loss": 1.4191,
      "step": 850
    },
    {
      "epoch": 3.9836161497952016,
      "grad_norm": 0.525581955909729,
      "learning_rate": 7.507092881358782e-05,
      "loss": 1.4093,
      "step": 851
    },
    {
      "epoch": 3.9882972498537157,
      "grad_norm": 0.5150876641273499,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.3958,
      "step": 852
    },
    {
      "epoch": 3.9929783499122293,
      "grad_norm": 0.5269944071769714,
      "learning_rate": 7.492900404423326e-05,
      "loss": 1.4066,
      "step": 853
    },
    {
      "epoch": 3.997659449970743,
      "grad_norm": 0.5274633765220642,
      "learning_rate": 7.48579411369605e-05,
      "loss": 1.4296,
      "step": 854
    },
    {
      "epoch": 4.002340550029257,
      "grad_norm": 2.507246971130371,
      "learning_rate": 7.478681146903448e-05,
      "loss": 1.4841,
      "step": 855
    },
    {
      "epoch": 4.00702165008777,
      "grad_norm": 0.5649014115333557,
      "learning_rate": 7.471561523148722e-05,
      "loss": 1.4352,
      "step": 856
    },
    {
      "epoch": 4.011702750146284,
      "grad_norm": 0.5550701022148132,
      "learning_rate": 7.464435261552956e-05,
      "loss": 1.4092,
      "step": 857
    },
    {
      "epoch": 4.016383850204798,
      "grad_norm": 0.5513316988945007,
      "learning_rate": 7.457302381255054e-05,
      "loss": 1.4058,
      "step": 858
    },
    {
      "epoch": 4.021064950263312,
      "grad_norm": 0.5378918647766113,
      "learning_rate": 7.450162901411708e-05,
      "loss": 1.4232,
      "step": 859
    },
    {
      "epoch": 4.025746050321826,
      "grad_norm": 0.5452824234962463,
      "learning_rate": 7.443016841197323e-05,
      "loss": 1.4016,
      "step": 860
    },
    {
      "epoch": 4.03042715038034,
      "grad_norm": 0.5211923122406006,
      "learning_rate": 7.435864219803983e-05,
      "loss": 1.4091,
      "step": 861
    },
    {
      "epoch": 4.035108250438853,
      "grad_norm": 0.5528005361557007,
      "learning_rate": 7.42870505644139e-05,
      "loss": 1.409,
      "step": 862
    },
    {
      "epoch": 4.039789350497367,
      "grad_norm": 0.536733090877533,
      "learning_rate": 7.421539370336819e-05,
      "loss": 1.4136,
      "step": 863
    },
    {
      "epoch": 4.044470450555881,
      "grad_norm": 0.5282482504844666,
      "learning_rate": 7.414367180735058e-05,
      "loss": 1.3946,
      "step": 864
    },
    {
      "epoch": 4.049151550614394,
      "grad_norm": 0.5195043087005615,
      "learning_rate": 7.407188506898367e-05,
      "loss": 1.4206,
      "step": 865
    },
    {
      "epoch": 4.053832650672908,
      "grad_norm": 0.5188748240470886,
      "learning_rate": 7.400003368106419e-05,
      "loss": 1.4144,
      "step": 866
    },
    {
      "epoch": 4.0585137507314215,
      "grad_norm": 0.5505629181861877,
      "learning_rate": 7.392811783656246e-05,
      "loss": 1.4355,
      "step": 867
    },
    {
      "epoch": 4.063194850789936,
      "grad_norm": 0.5169565677642822,
      "learning_rate": 7.385613772862197e-05,
      "loss": 1.4153,
      "step": 868
    },
    {
      "epoch": 4.06787595084845,
      "grad_norm": 0.5217868685722351,
      "learning_rate": 7.378409355055874e-05,
      "loss": 1.4018,
      "step": 869
    },
    {
      "epoch": 4.072557050906963,
      "grad_norm": 0.5039651393890381,
      "learning_rate": 7.371198549586091e-05,
      "loss": 1.3759,
      "step": 870
    },
    {
      "epoch": 4.077238150965477,
      "grad_norm": 0.5173787474632263,
      "learning_rate": 7.363981375818817e-05,
      "loss": 1.4033,
      "step": 871
    },
    {
      "epoch": 4.081919251023991,
      "grad_norm": 0.5146439075469971,
      "learning_rate": 7.35675785313712e-05,
      "loss": 1.4211,
      "step": 872
    },
    {
      "epoch": 4.086600351082504,
      "grad_norm": 0.5264257192611694,
      "learning_rate": 7.349528000941126e-05,
      "loss": 1.3901,
      "step": 873
    },
    {
      "epoch": 4.091281451141018,
      "grad_norm": 0.5119170546531677,
      "learning_rate": 7.342291838647952e-05,
      "loss": 1.4119,
      "step": 874
    },
    {
      "epoch": 4.0959625511995315,
      "grad_norm": 0.5248039960861206,
      "learning_rate": 7.335049385691669e-05,
      "loss": 1.3874,
      "step": 875
    },
    {
      "epoch": 4.1006436512580455,
      "grad_norm": 0.5059455633163452,
      "learning_rate": 7.327800661523238e-05,
      "loss": 1.4059,
      "step": 876
    },
    {
      "epoch": 4.10532475131656,
      "grad_norm": 0.5265378952026367,
      "learning_rate": 7.320545685610467e-05,
      "loss": 1.4202,
      "step": 877
    },
    {
      "epoch": 4.110005851375073,
      "grad_norm": 0.5115812420845032,
      "learning_rate": 7.31328447743795e-05,
      "loss": 1.4021,
      "step": 878
    },
    {
      "epoch": 4.114686951433587,
      "grad_norm": 0.5200651288032532,
      "learning_rate": 7.306017056507017e-05,
      "loss": 1.4136,
      "step": 879
    },
    {
      "epoch": 4.119368051492101,
      "grad_norm": 0.5201098322868347,
      "learning_rate": 7.298743442335693e-05,
      "loss": 1.4063,
      "step": 880
    },
    {
      "epoch": 4.124049151550614,
      "grad_norm": 0.5152908563613892,
      "learning_rate": 7.291463654458629e-05,
      "loss": 1.4263,
      "step": 881
    },
    {
      "epoch": 4.128730251609128,
      "grad_norm": 0.5363911390304565,
      "learning_rate": 7.284177712427056e-05,
      "loss": 1.4095,
      "step": 882
    },
    {
      "epoch": 4.133411351667642,
      "grad_norm": 0.5588401556015015,
      "learning_rate": 7.276885635808734e-05,
      "loss": 1.3952,
      "step": 883
    },
    {
      "epoch": 4.1380924517261555,
      "grad_norm": 0.50605708360672,
      "learning_rate": 7.269587444187902e-05,
      "loss": 1.3897,
      "step": 884
    },
    {
      "epoch": 4.1427735517846696,
      "grad_norm": 0.5475851893424988,
      "learning_rate": 7.262283157165219e-05,
      "loss": 1.3952,
      "step": 885
    },
    {
      "epoch": 4.147454651843183,
      "grad_norm": 0.5669999122619629,
      "learning_rate": 7.254972794357713e-05,
      "loss": 1.4031,
      "step": 886
    },
    {
      "epoch": 4.152135751901697,
      "grad_norm": 0.5178671479225159,
      "learning_rate": 7.247656375398734e-05,
      "loss": 1.3768,
      "step": 887
    },
    {
      "epoch": 4.156816851960211,
      "grad_norm": 0.5260223746299744,
      "learning_rate": 7.240333919937893e-05,
      "loss": 1.4019,
      "step": 888
    },
    {
      "epoch": 4.161497952018724,
      "grad_norm": 0.5484073162078857,
      "learning_rate": 7.233005447641015e-05,
      "loss": 1.3978,
      "step": 889
    },
    {
      "epoch": 4.166179052077238,
      "grad_norm": 0.5305057764053345,
      "learning_rate": 7.225670978190085e-05,
      "loss": 1.4178,
      "step": 890
    },
    {
      "epoch": 4.170860152135752,
      "grad_norm": 0.5260404348373413,
      "learning_rate": 7.21833053128319e-05,
      "loss": 1.4109,
      "step": 891
    },
    {
      "epoch": 4.175541252194265,
      "grad_norm": 0.5581408739089966,
      "learning_rate": 7.210984126634476e-05,
      "loss": 1.4049,
      "step": 892
    },
    {
      "epoch": 4.1802223522527795,
      "grad_norm": 0.5089554786682129,
      "learning_rate": 7.20363178397409e-05,
      "loss": 1.3986,
      "step": 893
    },
    {
      "epoch": 4.184903452311294,
      "grad_norm": 0.5193794965744019,
      "learning_rate": 7.19627352304812e-05,
      "loss": 1.4088,
      "step": 894
    },
    {
      "epoch": 4.189584552369807,
      "grad_norm": 0.5327683091163635,
      "learning_rate": 7.188909363618557e-05,
      "loss": 1.4,
      "step": 895
    },
    {
      "epoch": 4.194265652428321,
      "grad_norm": 0.5190610885620117,
      "learning_rate": 7.181539325463227e-05,
      "loss": 1.4081,
      "step": 896
    },
    {
      "epoch": 4.198946752486834,
      "grad_norm": 0.5354302525520325,
      "learning_rate": 7.174163428375748e-05,
      "loss": 1.4022,
      "step": 897
    },
    {
      "epoch": 4.203627852545348,
      "grad_norm": 0.5408794283866882,
      "learning_rate": 7.16678169216547e-05,
      "loss": 1.4132,
      "step": 898
    },
    {
      "epoch": 4.208308952603862,
      "grad_norm": 0.5260702967643738,
      "learning_rate": 7.159394136657427e-05,
      "loss": 1.4131,
      "step": 899
    },
    {
      "epoch": 4.212990052662375,
      "grad_norm": 0.546550452709198,
      "learning_rate": 7.152000781692286e-05,
      "loss": 1.3992,
      "step": 900
    },
    {
      "epoch": 4.217671152720889,
      "grad_norm": 0.516927182674408,
      "learning_rate": 7.144601647126281e-05,
      "loss": 1.3858,
      "step": 901
    },
    {
      "epoch": 4.2223522527794035,
      "grad_norm": 0.5527111887931824,
      "learning_rate": 7.137196752831177e-05,
      "loss": 1.3951,
      "step": 902
    },
    {
      "epoch": 4.227033352837917,
      "grad_norm": 0.5520824790000916,
      "learning_rate": 7.1297861186942e-05,
      "loss": 1.3779,
      "step": 903
    },
    {
      "epoch": 4.231714452896431,
      "grad_norm": 0.5436320304870605,
      "learning_rate": 7.122369764617993e-05,
      "loss": 1.4142,
      "step": 904
    },
    {
      "epoch": 4.236395552954944,
      "grad_norm": 0.5542235374450684,
      "learning_rate": 7.114947710520569e-05,
      "loss": 1.4066,
      "step": 905
    },
    {
      "epoch": 4.241076653013458,
      "grad_norm": 0.5395337343215942,
      "learning_rate": 7.10751997633524e-05,
      "loss": 1.403,
      "step": 906
    },
    {
      "epoch": 4.245757753071972,
      "grad_norm": 0.5479816198348999,
      "learning_rate": 7.100086582010576e-05,
      "loss": 1.3946,
      "step": 907
    },
    {
      "epoch": 4.250438853130485,
      "grad_norm": 0.5550580620765686,
      "learning_rate": 7.092647547510351e-05,
      "loss": 1.3902,
      "step": 908
    },
    {
      "epoch": 4.255119953188999,
      "grad_norm": 0.5406416058540344,
      "learning_rate": 7.08520289281348e-05,
      "loss": 1.4019,
      "step": 909
    },
    {
      "epoch": 4.2598010532475135,
      "grad_norm": 0.5209553837776184,
      "learning_rate": 7.07775263791398e-05,
      "loss": 1.4078,
      "step": 910
    },
    {
      "epoch": 4.264482153306027,
      "grad_norm": 0.5705069303512573,
      "learning_rate": 7.070296802820907e-05,
      "loss": 1.3928,
      "step": 911
    },
    {
      "epoch": 4.269163253364541,
      "grad_norm": 0.5689321756362915,
      "learning_rate": 7.062835407558294e-05,
      "loss": 1.3885,
      "step": 912
    },
    {
      "epoch": 4.273844353423055,
      "grad_norm": 0.5145436525344849,
      "learning_rate": 7.055368472165123e-05,
      "loss": 1.4017,
      "step": 913
    },
    {
      "epoch": 4.278525453481568,
      "grad_norm": 0.5727246403694153,
      "learning_rate": 7.047896016695239e-05,
      "loss": 1.3992,
      "step": 914
    },
    {
      "epoch": 4.283206553540082,
      "grad_norm": 0.5670337677001953,
      "learning_rate": 7.040418061217325e-05,
      "loss": 1.4002,
      "step": 915
    },
    {
      "epoch": 4.287887653598595,
      "grad_norm": 0.5520856380462646,
      "learning_rate": 7.032934625814829e-05,
      "loss": 1.4064,
      "step": 916
    },
    {
      "epoch": 4.292568753657109,
      "grad_norm": 0.556670069694519,
      "learning_rate": 7.025445730585915e-05,
      "loss": 1.4072,
      "step": 917
    },
    {
      "epoch": 4.297249853715623,
      "grad_norm": 0.5678969025611877,
      "learning_rate": 7.017951395643417e-05,
      "loss": 1.424,
      "step": 918
    },
    {
      "epoch": 4.301930953774137,
      "grad_norm": 0.5321788787841797,
      "learning_rate": 7.010451641114768e-05,
      "loss": 1.3822,
      "step": 919
    },
    {
      "epoch": 4.306612053832651,
      "grad_norm": 0.5413317084312439,
      "learning_rate": 7.002946487141967e-05,
      "loss": 1.4016,
      "step": 920
    },
    {
      "epoch": 4.311293153891165,
      "grad_norm": 0.5802444219589233,
      "learning_rate": 6.995435953881509e-05,
      "loss": 1.4047,
      "step": 921
    },
    {
      "epoch": 4.315974253949678,
      "grad_norm": 0.5204755067825317,
      "learning_rate": 6.987920061504336e-05,
      "loss": 1.3945,
      "step": 922
    },
    {
      "epoch": 4.320655354008192,
      "grad_norm": 0.5474327802658081,
      "learning_rate": 6.980398830195785e-05,
      "loss": 1.3974,
      "step": 923
    },
    {
      "epoch": 4.325336454066706,
      "grad_norm": 0.5491933226585388,
      "learning_rate": 6.972872280155528e-05,
      "loss": 1.4182,
      "step": 924
    },
    {
      "epoch": 4.330017554125219,
      "grad_norm": 0.5143939256668091,
      "learning_rate": 6.965340431597525e-05,
      "loss": 1.4069,
      "step": 925
    },
    {
      "epoch": 4.334698654183733,
      "grad_norm": 0.5000368356704712,
      "learning_rate": 6.957803304749966e-05,
      "loss": 1.3999,
      "step": 926
    },
    {
      "epoch": 4.3393797542422465,
      "grad_norm": 0.5126700401306152,
      "learning_rate": 6.950260919855213e-05,
      "loss": 1.3851,
      "step": 927
    },
    {
      "epoch": 4.344060854300761,
      "grad_norm": 0.5102867484092712,
      "learning_rate": 6.942713297169755e-05,
      "loss": 1.3897,
      "step": 928
    },
    {
      "epoch": 4.348741954359275,
      "grad_norm": 0.5007814764976501,
      "learning_rate": 6.935160456964144e-05,
      "loss": 1.3965,
      "step": 929
    },
    {
      "epoch": 4.353423054417788,
      "grad_norm": 0.5208529829978943,
      "learning_rate": 6.927602419522947e-05,
      "loss": 1.3983,
      "step": 930
    },
    {
      "epoch": 4.358104154476302,
      "grad_norm": 0.5251065492630005,
      "learning_rate": 6.920039205144686e-05,
      "loss": 1.4218,
      "step": 931
    },
    {
      "epoch": 4.362785254534816,
      "grad_norm": 0.49510329961776733,
      "learning_rate": 6.912470834141791e-05,
      "loss": 1.3946,
      "step": 932
    },
    {
      "epoch": 4.367466354593329,
      "grad_norm": 0.5090747475624084,
      "learning_rate": 6.904897326840537e-05,
      "loss": 1.3917,
      "step": 933
    },
    {
      "epoch": 4.372147454651843,
      "grad_norm": 0.49734053015708923,
      "learning_rate": 6.897318703580995e-05,
      "loss": 1.3687,
      "step": 934
    },
    {
      "epoch": 4.376828554710357,
      "grad_norm": 0.5525650978088379,
      "learning_rate": 6.889734984716982e-05,
      "loss": 1.4092,
      "step": 935
    },
    {
      "epoch": 4.3815096547688706,
      "grad_norm": 0.5270445942878723,
      "learning_rate": 6.882146190615984e-05,
      "loss": 1.4065,
      "step": 936
    },
    {
      "epoch": 4.386190754827385,
      "grad_norm": 0.5141691565513611,
      "learning_rate": 6.874552341659137e-05,
      "loss": 1.4111,
      "step": 937
    },
    {
      "epoch": 4.390871854885898,
      "grad_norm": 0.5064695477485657,
      "learning_rate": 6.86695345824114e-05,
      "loss": 1.387,
      "step": 938
    },
    {
      "epoch": 4.395552954944412,
      "grad_norm": 0.5093940496444702,
      "learning_rate": 6.859349560770213e-05,
      "loss": 1.4198,
      "step": 939
    },
    {
      "epoch": 4.400234055002926,
      "grad_norm": 0.5096136927604675,
      "learning_rate": 6.851740669668055e-05,
      "loss": 1.3782,
      "step": 940
    },
    {
      "epoch": 4.404915155061439,
      "grad_norm": 0.5299738049507141,
      "learning_rate": 6.844126805369758e-05,
      "loss": 1.3926,
      "step": 941
    },
    {
      "epoch": 4.409596255119953,
      "grad_norm": 0.5203460454940796,
      "learning_rate": 6.836507988323784e-05,
      "loss": 1.3865,
      "step": 942
    },
    {
      "epoch": 4.414277355178467,
      "grad_norm": 0.5185226202011108,
      "learning_rate": 6.828884238991892e-05,
      "loss": 1.4046,
      "step": 943
    },
    {
      "epoch": 4.4189584552369805,
      "grad_norm": 0.5208016037940979,
      "learning_rate": 6.821255577849086e-05,
      "loss": 1.4046,
      "step": 944
    },
    {
      "epoch": 4.423639555295495,
      "grad_norm": 0.5160471200942993,
      "learning_rate": 6.813622025383565e-05,
      "loss": 1.4017,
      "step": 945
    },
    {
      "epoch": 4.428320655354009,
      "grad_norm": 0.5242071151733398,
      "learning_rate": 6.805983602096661e-05,
      "loss": 1.3965,
      "step": 946
    },
    {
      "epoch": 4.433001755412522,
      "grad_norm": 0.5100763440132141,
      "learning_rate": 6.798340328502792e-05,
      "loss": 1.4058,
      "step": 947
    },
    {
      "epoch": 4.437682855471036,
      "grad_norm": 0.5151635408401489,
      "learning_rate": 6.790692225129398e-05,
      "loss": 1.4159,
      "step": 948
    },
    {
      "epoch": 4.442363955529549,
      "grad_norm": 0.5144315958023071,
      "learning_rate": 6.783039312516889e-05,
      "loss": 1.3947,
      "step": 949
    },
    {
      "epoch": 4.447045055588063,
      "grad_norm": 0.5269882678985596,
      "learning_rate": 6.7753816112186e-05,
      "loss": 1.383,
      "step": 950
    },
    {
      "epoch": 4.451726155646577,
      "grad_norm": 0.5096060037612915,
      "learning_rate": 6.767719141800717e-05,
      "loss": 1.3778,
      "step": 951
    },
    {
      "epoch": 4.45640725570509,
      "grad_norm": 0.4936077892780304,
      "learning_rate": 6.760051924842239e-05,
      "loss": 1.3645,
      "step": 952
    },
    {
      "epoch": 4.4610883557636045,
      "grad_norm": 0.5201056003570557,
      "learning_rate": 6.752379980934909e-05,
      "loss": 1.3968,
      "step": 953
    },
    {
      "epoch": 4.465769455822119,
      "grad_norm": 0.516392707824707,
      "learning_rate": 6.744703330683169e-05,
      "loss": 1.3793,
      "step": 954
    },
    {
      "epoch": 4.470450555880632,
      "grad_norm": 0.505644679069519,
      "learning_rate": 6.7370219947041e-05,
      "loss": 1.3902,
      "step": 955
    },
    {
      "epoch": 4.475131655939146,
      "grad_norm": 0.4903636574745178,
      "learning_rate": 6.729335993627368e-05,
      "loss": 1.3741,
      "step": 956
    },
    {
      "epoch": 4.479812755997659,
      "grad_norm": 0.5225103497505188,
      "learning_rate": 6.721645348095167e-05,
      "loss": 1.3859,
      "step": 957
    },
    {
      "epoch": 4.484493856056173,
      "grad_norm": 0.5010210275650024,
      "learning_rate": 6.713950078762165e-05,
      "loss": 1.4044,
      "step": 958
    },
    {
      "epoch": 4.489174956114687,
      "grad_norm": 0.5460252165794373,
      "learning_rate": 6.706250206295449e-05,
      "loss": 1.3896,
      "step": 959
    },
    {
      "epoch": 4.4938560561732,
      "grad_norm": 0.5131853222846985,
      "learning_rate": 6.698545751374465e-05,
      "loss": 1.3853,
      "step": 960
    },
    {
      "epoch": 4.4985371562317145,
      "grad_norm": 0.5487529039382935,
      "learning_rate": 6.69083673469097e-05,
      "loss": 1.401,
      "step": 961
    },
    {
      "epoch": 4.5032182562902285,
      "grad_norm": 0.5241919159889221,
      "learning_rate": 6.683123176948973e-05,
      "loss": 1.3839,
      "step": 962
    },
    {
      "epoch": 4.507899356348742,
      "grad_norm": 0.5286917686462402,
      "learning_rate": 6.675405098864675e-05,
      "loss": 1.3946,
      "step": 963
    },
    {
      "epoch": 4.512580456407256,
      "grad_norm": 0.5300504565238953,
      "learning_rate": 6.667682521166419e-05,
      "loss": 1.4107,
      "step": 964
    },
    {
      "epoch": 4.517261556465769,
      "grad_norm": 0.5106096267700195,
      "learning_rate": 6.659955464594632e-05,
      "loss": 1.3983,
      "step": 965
    },
    {
      "epoch": 4.521942656524283,
      "grad_norm": 0.5417435169219971,
      "learning_rate": 6.652223949901774e-05,
      "loss": 1.4005,
      "step": 966
    },
    {
      "epoch": 4.526623756582797,
      "grad_norm": 0.5418683290481567,
      "learning_rate": 6.644487997852272e-05,
      "loss": 1.4127,
      "step": 967
    },
    {
      "epoch": 4.53130485664131,
      "grad_norm": 0.5197861790657043,
      "learning_rate": 6.636747629222475e-05,
      "loss": 1.391,
      "step": 968
    },
    {
      "epoch": 4.535985956699824,
      "grad_norm": 0.5314134359359741,
      "learning_rate": 6.629002864800589e-05,
      "loss": 1.3912,
      "step": 969
    },
    {
      "epoch": 4.5406670567583385,
      "grad_norm": 0.526654839515686,
      "learning_rate": 6.621253725386628e-05,
      "loss": 1.3949,
      "step": 970
    },
    {
      "epoch": 4.545348156816852,
      "grad_norm": 0.527079701423645,
      "learning_rate": 6.61350023179236e-05,
      "loss": 1.4042,
      "step": 971
    },
    {
      "epoch": 4.550029256875366,
      "grad_norm": 0.5492599606513977,
      "learning_rate": 6.605742404841241e-05,
      "loss": 1.4006,
      "step": 972
    },
    {
      "epoch": 4.55471035693388,
      "grad_norm": 0.522839367389679,
      "learning_rate": 6.597980265368367e-05,
      "loss": 1.3986,
      "step": 973
    },
    {
      "epoch": 4.559391456992393,
      "grad_norm": 0.5506901741027832,
      "learning_rate": 6.590213834220415e-05,
      "loss": 1.3812,
      "step": 974
    },
    {
      "epoch": 4.564072557050907,
      "grad_norm": 0.5584288835525513,
      "learning_rate": 6.582443132255592e-05,
      "loss": 1.4006,
      "step": 975
    },
    {
      "epoch": 4.56875365710942,
      "grad_norm": 0.5198907256126404,
      "learning_rate": 6.574668180343571e-05,
      "loss": 1.4016,
      "step": 976
    },
    {
      "epoch": 4.573434757167934,
      "grad_norm": 0.5312773585319519,
      "learning_rate": 6.566888999365441e-05,
      "loss": 1.3853,
      "step": 977
    },
    {
      "epoch": 4.578115857226448,
      "grad_norm": 0.5397941470146179,
      "learning_rate": 6.559105610213649e-05,
      "loss": 1.376,
      "step": 978
    },
    {
      "epoch": 4.582796957284962,
      "grad_norm": 0.515097439289093,
      "learning_rate": 6.551318033791942e-05,
      "loss": 1.4004,
      "step": 979
    },
    {
      "epoch": 4.587478057343476,
      "grad_norm": 0.554069459438324,
      "learning_rate": 6.543526291015315e-05,
      "loss": 1.3937,
      "step": 980
    },
    {
      "epoch": 4.59215915740199,
      "grad_norm": 0.5135395526885986,
      "learning_rate": 6.53573040280995e-05,
      "loss": 1.3848,
      "step": 981
    },
    {
      "epoch": 4.596840257460503,
      "grad_norm": 0.554758608341217,
      "learning_rate": 6.527930390113166e-05,
      "loss": 1.3784,
      "step": 982
    },
    {
      "epoch": 4.601521357519017,
      "grad_norm": 0.5125584006309509,
      "learning_rate": 6.520126273873356e-05,
      "loss": 1.3807,
      "step": 983
    },
    {
      "epoch": 4.606202457577531,
      "grad_norm": 0.5136418342590332,
      "learning_rate": 6.512318075049935e-05,
      "loss": 1.3824,
      "step": 984
    },
    {
      "epoch": 4.610883557636044,
      "grad_norm": 0.5604468584060669,
      "learning_rate": 6.50450581461328e-05,
      "loss": 1.3738,
      "step": 985
    },
    {
      "epoch": 4.615564657694558,
      "grad_norm": 0.5250749588012695,
      "learning_rate": 6.496689513544682e-05,
      "loss": 1.4121,
      "step": 986
    },
    {
      "epoch": 4.620245757753072,
      "grad_norm": 0.5173577666282654,
      "learning_rate": 6.488869192836278e-05,
      "loss": 1.3659,
      "step": 987
    },
    {
      "epoch": 4.624926857811586,
      "grad_norm": 0.5003069639205933,
      "learning_rate": 6.481044873491004e-05,
      "loss": 1.3967,
      "step": 988
    },
    {
      "epoch": 4.6296079578701,
      "grad_norm": 0.49318012595176697,
      "learning_rate": 6.473216576522535e-05,
      "loss": 1.3986,
      "step": 989
    },
    {
      "epoch": 4.634289057928613,
      "grad_norm": 0.5480040907859802,
      "learning_rate": 6.465384322955224e-05,
      "loss": 1.404,
      "step": 990
    },
    {
      "epoch": 4.638970157987127,
      "grad_norm": 0.49322742223739624,
      "learning_rate": 6.457548133824057e-05,
      "loss": 1.3786,
      "step": 991
    },
    {
      "epoch": 4.643651258045641,
      "grad_norm": 0.5618907809257507,
      "learning_rate": 6.449708030174587e-05,
      "loss": 1.3984,
      "step": 992
    },
    {
      "epoch": 4.648332358104154,
      "grad_norm": 0.5069143772125244,
      "learning_rate": 6.441864033062879e-05,
      "loss": 1.3675,
      "step": 993
    },
    {
      "epoch": 4.653013458162668,
      "grad_norm": 0.5381274223327637,
      "learning_rate": 6.434016163555452e-05,
      "loss": 1.3848,
      "step": 994
    },
    {
      "epoch": 4.657694558221182,
      "grad_norm": 0.5342992544174194,
      "learning_rate": 6.426164442729232e-05,
      "loss": 1.3804,
      "step": 995
    },
    {
      "epoch": 4.662375658279696,
      "grad_norm": 0.5064785480499268,
      "learning_rate": 6.418308891671484e-05,
      "loss": 1.3686,
      "step": 996
    },
    {
      "epoch": 4.66705675833821,
      "grad_norm": 0.5631519556045532,
      "learning_rate": 6.410449531479761e-05,
      "loss": 1.3819,
      "step": 997
    },
    {
      "epoch": 4.671737858396723,
      "grad_norm": 0.5275139808654785,
      "learning_rate": 6.402586383261844e-05,
      "loss": 1.3877,
      "step": 998
    },
    {
      "epoch": 4.676418958455237,
      "grad_norm": 0.5353507995605469,
      "learning_rate": 6.394719468135691e-05,
      "loss": 1.388,
      "step": 999
    },
    {
      "epoch": 4.681100058513751,
      "grad_norm": 0.543814480304718,
      "learning_rate": 6.386848807229373e-05,
      "loss": 1.4066,
      "step": 1000
    },
    {
      "epoch": 4.685781158572264,
      "grad_norm": 0.5282537937164307,
      "learning_rate": 6.378974421681027e-05,
      "loss": 1.3901,
      "step": 1001
    },
    {
      "epoch": 4.690462258630778,
      "grad_norm": 0.5356928110122681,
      "learning_rate": 6.371096332638784e-05,
      "loss": 1.3953,
      "step": 1002
    },
    {
      "epoch": 4.695143358689292,
      "grad_norm": 0.503050684928894,
      "learning_rate": 6.36321456126073e-05,
      "loss": 1.3777,
      "step": 1003
    },
    {
      "epoch": 4.6998244587478055,
      "grad_norm": 0.5379858016967773,
      "learning_rate": 6.355329128714834e-05,
      "loss": 1.3895,
      "step": 1004
    },
    {
      "epoch": 4.70450555880632,
      "grad_norm": 0.5181742310523987,
      "learning_rate": 6.347440056178904e-05,
      "loss": 1.3797,
      "step": 1005
    },
    {
      "epoch": 4.709186658864834,
      "grad_norm": 0.5005363821983337,
      "learning_rate": 6.339547364840522e-05,
      "loss": 1.3904,
      "step": 1006
    },
    {
      "epoch": 4.713867758923347,
      "grad_norm": 0.5222419500350952,
      "learning_rate": 6.331651075896983e-05,
      "loss": 1.3625,
      "step": 1007
    },
    {
      "epoch": 4.718548858981861,
      "grad_norm": 0.5192439556121826,
      "learning_rate": 6.323751210555252e-05,
      "loss": 1.3799,
      "step": 1008
    },
    {
      "epoch": 4.723229959040374,
      "grad_norm": 0.5143327116966248,
      "learning_rate": 6.315847790031895e-05,
      "loss": 1.3915,
      "step": 1009
    },
    {
      "epoch": 4.727911059098888,
      "grad_norm": 0.5154076814651489,
      "learning_rate": 6.307940835553027e-05,
      "loss": 1.3921,
      "step": 1010
    },
    {
      "epoch": 4.732592159157402,
      "grad_norm": 0.5128686428070068,
      "learning_rate": 6.300030368354255e-05,
      "loss": 1.3899,
      "step": 1011
    },
    {
      "epoch": 4.7372732592159155,
      "grad_norm": 0.5078786015510559,
      "learning_rate": 6.292116409680617e-05,
      "loss": 1.3807,
      "step": 1012
    },
    {
      "epoch": 4.7419543592744295,
      "grad_norm": 0.5336173176765442,
      "learning_rate": 6.284198980786533e-05,
      "loss": 1.4099,
      "step": 1013
    },
    {
      "epoch": 4.746635459332944,
      "grad_norm": 0.5144036412239075,
      "learning_rate": 6.276278102935739e-05,
      "loss": 1.3845,
      "step": 1014
    },
    {
      "epoch": 4.751316559391457,
      "grad_norm": 0.5357654094696045,
      "learning_rate": 6.268353797401234e-05,
      "loss": 1.373,
      "step": 1015
    },
    {
      "epoch": 4.755997659449971,
      "grad_norm": 0.5412711501121521,
      "learning_rate": 6.260426085465225e-05,
      "loss": 1.3678,
      "step": 1016
    },
    {
      "epoch": 4.760678759508485,
      "grad_norm": 0.5578483939170837,
      "learning_rate": 6.252494988419066e-05,
      "loss": 1.3851,
      "step": 1017
    },
    {
      "epoch": 4.765359859566998,
      "grad_norm": 0.5388118028640747,
      "learning_rate": 6.244560527563199e-05,
      "loss": 1.3954,
      "step": 1018
    },
    {
      "epoch": 4.770040959625512,
      "grad_norm": 0.5390235781669617,
      "learning_rate": 6.23662272420711e-05,
      "loss": 1.3814,
      "step": 1019
    },
    {
      "epoch": 4.774722059684025,
      "grad_norm": 0.5449773073196411,
      "learning_rate": 6.228681599669248e-05,
      "loss": 1.3758,
      "step": 1020
    },
    {
      "epoch": 4.7794031597425395,
      "grad_norm": 0.515874445438385,
      "learning_rate": 6.220737175276996e-05,
      "loss": 1.362,
      "step": 1021
    },
    {
      "epoch": 4.784084259801054,
      "grad_norm": 0.5363708138465881,
      "learning_rate": 6.212789472366591e-05,
      "loss": 1.4039,
      "step": 1022
    },
    {
      "epoch": 4.788765359859567,
      "grad_norm": 0.5405462980270386,
      "learning_rate": 6.204838512283072e-05,
      "loss": 1.3834,
      "step": 1023
    },
    {
      "epoch": 4.793446459918081,
      "grad_norm": 0.5273584723472595,
      "learning_rate": 6.196884316380237e-05,
      "loss": 1.369,
      "step": 1024
    },
    {
      "epoch": 4.798127559976594,
      "grad_norm": 0.5312931537628174,
      "learning_rate": 6.188926906020562e-05,
      "loss": 1.3913,
      "step": 1025
    },
    {
      "epoch": 4.802808660035108,
      "grad_norm": 0.5499268174171448,
      "learning_rate": 6.180966302575166e-05,
      "loss": 1.4118,
      "step": 1026
    },
    {
      "epoch": 4.807489760093622,
      "grad_norm": 0.5172853469848633,
      "learning_rate": 6.173002527423737e-05,
      "loss": 1.3869,
      "step": 1027
    },
    {
      "epoch": 4.812170860152136,
      "grad_norm": 0.5686773657798767,
      "learning_rate": 6.165035601954485e-05,
      "loss": 1.3715,
      "step": 1028
    },
    {
      "epoch": 4.816851960210649,
      "grad_norm": 0.5325746536254883,
      "learning_rate": 6.157065547564079e-05,
      "loss": 1.3785,
      "step": 1029
    },
    {
      "epoch": 4.8215330602691635,
      "grad_norm": 0.5127576589584351,
      "learning_rate": 6.149092385657591e-05,
      "loss": 1.3906,
      "step": 1030
    },
    {
      "epoch": 4.826214160327677,
      "grad_norm": 0.5501847267150879,
      "learning_rate": 6.14111613764844e-05,
      "loss": 1.4056,
      "step": 1031
    },
    {
      "epoch": 4.830895260386191,
      "grad_norm": 0.5234781503677368,
      "learning_rate": 6.133136824958335e-05,
      "loss": 1.4034,
      "step": 1032
    },
    {
      "epoch": 4.835576360444705,
      "grad_norm": 0.5253031849861145,
      "learning_rate": 6.125154469017209e-05,
      "loss": 1.3904,
      "step": 1033
    },
    {
      "epoch": 4.840257460503218,
      "grad_norm": 0.5461185574531555,
      "learning_rate": 6.117169091263177e-05,
      "loss": 1.3921,
      "step": 1034
    },
    {
      "epoch": 4.844938560561732,
      "grad_norm": 0.5113133788108826,
      "learning_rate": 6.109180713142465e-05,
      "loss": 1.3875,
      "step": 1035
    },
    {
      "epoch": 4.849619660620245,
      "grad_norm": 0.5434524416923523,
      "learning_rate": 6.1011893561093535e-05,
      "loss": 1.412,
      "step": 1036
    },
    {
      "epoch": 4.854300760678759,
      "grad_norm": 0.5480051636695862,
      "learning_rate": 6.093195041626132e-05,
      "loss": 1.4039,
      "step": 1037
    },
    {
      "epoch": 4.8589818607372735,
      "grad_norm": 0.520328938961029,
      "learning_rate": 6.085197791163028e-05,
      "loss": 1.3992,
      "step": 1038
    },
    {
      "epoch": 4.863662960795787,
      "grad_norm": 0.5311024188995361,
      "learning_rate": 6.077197626198151e-05,
      "loss": 1.3981,
      "step": 1039
    },
    {
      "epoch": 4.868344060854301,
      "grad_norm": 0.5565276741981506,
      "learning_rate": 6.0691945682174434e-05,
      "loss": 1.3742,
      "step": 1040
    },
    {
      "epoch": 4.873025160912815,
      "grad_norm": 0.518947958946228,
      "learning_rate": 6.0611886387146156e-05,
      "loss": 1.3944,
      "step": 1041
    },
    {
      "epoch": 4.877706260971328,
      "grad_norm": 0.5651046633720398,
      "learning_rate": 6.0531798591910896e-05,
      "loss": 1.4093,
      "step": 1042
    },
    {
      "epoch": 4.882387361029842,
      "grad_norm": 0.5247631072998047,
      "learning_rate": 6.0451682511559416e-05,
      "loss": 1.3807,
      "step": 1043
    },
    {
      "epoch": 4.887068461088356,
      "grad_norm": 0.5134245753288269,
      "learning_rate": 6.037153836125844e-05,
      "loss": 1.383,
      "step": 1044
    },
    {
      "epoch": 4.891749561146869,
      "grad_norm": 0.5301898717880249,
      "learning_rate": 6.029136635625008e-05,
      "loss": 1.3791,
      "step": 1045
    },
    {
      "epoch": 4.896430661205383,
      "grad_norm": 0.5449047684669495,
      "learning_rate": 6.0211166711851264e-05,
      "loss": 1.3943,
      "step": 1046
    },
    {
      "epoch": 4.901111761263897,
      "grad_norm": 0.5147010684013367,
      "learning_rate": 6.013093964345314e-05,
      "loss": 1.3803,
      "step": 1047
    },
    {
      "epoch": 4.905792861322411,
      "grad_norm": 0.5083584785461426,
      "learning_rate": 6.0050685366520545e-05,
      "loss": 1.3966,
      "step": 1048
    },
    {
      "epoch": 4.910473961380925,
      "grad_norm": 0.5392999053001404,
      "learning_rate": 5.997040409659132e-05,
      "loss": 1.3822,
      "step": 1049
    },
    {
      "epoch": 4.915155061439438,
      "grad_norm": 0.5468650460243225,
      "learning_rate": 5.989009604927587e-05,
      "loss": 1.3845,
      "step": 1050
    },
    {
      "epoch": 4.919836161497952,
      "grad_norm": 0.5140033960342407,
      "learning_rate": 5.980976144025647e-05,
      "loss": 1.3736,
      "step": 1051
    },
    {
      "epoch": 4.924517261556466,
      "grad_norm": 0.5236037969589233,
      "learning_rate": 5.972940048528675e-05,
      "loss": 1.3911,
      "step": 1052
    },
    {
      "epoch": 4.929198361614979,
      "grad_norm": 0.5266885161399841,
      "learning_rate": 5.9649013400191104e-05,
      "loss": 1.3906,
      "step": 1053
    },
    {
      "epoch": 4.933879461673493,
      "grad_norm": 0.5133062601089478,
      "learning_rate": 5.956860040086407e-05,
      "loss": 1.3672,
      "step": 1054
    },
    {
      "epoch": 4.938560561732007,
      "grad_norm": 0.5198325514793396,
      "learning_rate": 5.948816170326985e-05,
      "loss": 1.3813,
      "step": 1055
    },
    {
      "epoch": 4.943241661790521,
      "grad_norm": 0.530920684337616,
      "learning_rate": 5.9407697523441576e-05,
      "loss": 1.3997,
      "step": 1056
    },
    {
      "epoch": 4.947922761849035,
      "grad_norm": 0.5203038454055786,
      "learning_rate": 5.932720807748089e-05,
      "loss": 1.3741,
      "step": 1057
    },
    {
      "epoch": 4.952603861907548,
      "grad_norm": 0.5104234218597412,
      "learning_rate": 5.924669358155727e-05,
      "loss": 1.3938,
      "step": 1058
    },
    {
      "epoch": 4.957284961966062,
      "grad_norm": 0.5079901814460754,
      "learning_rate": 5.916615425190743e-05,
      "loss": 1.371,
      "step": 1059
    },
    {
      "epoch": 4.961966062024576,
      "grad_norm": 0.5349999070167542,
      "learning_rate": 5.908559030483484e-05,
      "loss": 1.3715,
      "step": 1060
    },
    {
      "epoch": 4.966647162083089,
      "grad_norm": 0.5009754300117493,
      "learning_rate": 5.900500195670905e-05,
      "loss": 1.3818,
      "step": 1061
    },
    {
      "epoch": 4.971328262141603,
      "grad_norm": 0.4894314706325531,
      "learning_rate": 5.8924389423965155e-05,
      "loss": 1.379,
      "step": 1062
    },
    {
      "epoch": 4.976009362200117,
      "grad_norm": 0.5114794969558716,
      "learning_rate": 5.884375292310319e-05,
      "loss": 1.37,
      "step": 1063
    },
    {
      "epoch": 4.9806904622586305,
      "grad_norm": 0.5092068314552307,
      "learning_rate": 5.8763092670687584e-05,
      "loss": 1.4138,
      "step": 1064
    },
    {
      "epoch": 4.985371562317145,
      "grad_norm": 0.511785089969635,
      "learning_rate": 5.868240888334653e-05,
      "loss": 1.3846,
      "step": 1065
    },
    {
      "epoch": 4.990052662375659,
      "grad_norm": 0.508854329586029,
      "learning_rate": 5.860170177777144e-05,
      "loss": 1.3671,
      "step": 1066
    },
    {
      "epoch": 4.994733762434172,
      "grad_norm": 0.5381656885147095,
      "learning_rate": 5.852097157071635e-05,
      "loss": 1.3861,
      "step": 1067
    },
    {
      "epoch": 4.999414862492686,
      "grad_norm": 0.5062696933746338,
      "learning_rate": 5.844021847899734e-05,
      "loss": 1.3698,
      "step": 1068
    },
    {
      "epoch": 5.004095962551199,
      "grad_norm": 1.824023723602295,
      "learning_rate": 5.8359442719491975e-05,
      "loss": 1.3047,
      "step": 1069
    },
    {
      "epoch": 5.008777062609713,
      "grad_norm": 0.5977228879928589,
      "learning_rate": 5.8278644509138645e-05,
      "loss": 1.3578,
      "step": 1070
    },
    {
      "epoch": 5.013458162668227,
      "grad_norm": 0.5300642848014832,
      "learning_rate": 5.8197824064936066e-05,
      "loss": 1.3888,
      "step": 1071
    },
    {
      "epoch": 5.0181392627267405,
      "grad_norm": 0.5731930136680603,
      "learning_rate": 5.81169816039427e-05,
      "loss": 1.3759,
      "step": 1072
    },
    {
      "epoch": 5.022820362785255,
      "grad_norm": 0.5253560543060303,
      "learning_rate": 5.8036117343276065e-05,
      "loss": 1.361,
      "step": 1073
    },
    {
      "epoch": 5.027501462843769,
      "grad_norm": 0.5331753492355347,
      "learning_rate": 5.7955231500112296e-05,
      "loss": 1.4112,
      "step": 1074
    },
    {
      "epoch": 5.032182562902282,
      "grad_norm": 0.536668062210083,
      "learning_rate": 5.7874324291685446e-05,
      "loss": 1.4051,
      "step": 1075
    },
    {
      "epoch": 5.036863662960796,
      "grad_norm": 0.5135036110877991,
      "learning_rate": 5.7793395935286964e-05,
      "loss": 1.3882,
      "step": 1076
    },
    {
      "epoch": 5.04154476301931,
      "grad_norm": 0.5347151756286621,
      "learning_rate": 5.771244664826512e-05,
      "loss": 1.3693,
      "step": 1077
    },
    {
      "epoch": 5.046225863077823,
      "grad_norm": 0.5390860438346863,
      "learning_rate": 5.763147664802435e-05,
      "loss": 1.3811,
      "step": 1078
    },
    {
      "epoch": 5.050906963136337,
      "grad_norm": 0.5213562250137329,
      "learning_rate": 5.7550486152024774e-05,
      "loss": 1.3884,
      "step": 1079
    },
    {
      "epoch": 5.05558806319485,
      "grad_norm": 0.5514521598815918,
      "learning_rate": 5.74694753777815e-05,
      "loss": 1.3939,
      "step": 1080
    },
    {
      "epoch": 5.0602691632533645,
      "grad_norm": 0.5222517251968384,
      "learning_rate": 5.738844454286411e-05,
      "loss": 1.3776,
      "step": 1081
    },
    {
      "epoch": 5.064950263311879,
      "grad_norm": 0.5084861516952515,
      "learning_rate": 5.730739386489612e-05,
      "loss": 1.3786,
      "step": 1082
    },
    {
      "epoch": 5.069631363370392,
      "grad_norm": 0.5144921541213989,
      "learning_rate": 5.7226323561554276e-05,
      "loss": 1.3698,
      "step": 1083
    },
    {
      "epoch": 5.074312463428906,
      "grad_norm": 0.5156840085983276,
      "learning_rate": 5.714523385056802e-05,
      "loss": 1.3798,
      "step": 1084
    },
    {
      "epoch": 5.07899356348742,
      "grad_norm": 0.5120629668235779,
      "learning_rate": 5.706412494971901e-05,
      "loss": 1.3657,
      "step": 1085
    },
    {
      "epoch": 5.083674663545933,
      "grad_norm": 0.5259561538696289,
      "learning_rate": 5.6982997076840306e-05,
      "loss": 1.3836,
      "step": 1086
    },
    {
      "epoch": 5.088355763604447,
      "grad_norm": 0.5179267525672913,
      "learning_rate": 5.690185044981606e-05,
      "loss": 1.4128,
      "step": 1087
    },
    {
      "epoch": 5.09303686366296,
      "grad_norm": 0.4868166148662567,
      "learning_rate": 5.68206852865807e-05,
      "loss": 1.3634,
      "step": 1088
    },
    {
      "epoch": 5.0977179637214745,
      "grad_norm": 0.5132505297660828,
      "learning_rate": 5.673950180511845e-05,
      "loss": 1.3701,
      "step": 1089
    },
    {
      "epoch": 5.1023990637799885,
      "grad_norm": 0.5255224108695984,
      "learning_rate": 5.665830022346277e-05,
      "loss": 1.3893,
      "step": 1090
    },
    {
      "epoch": 5.107080163838502,
      "grad_norm": 0.5020806789398193,
      "learning_rate": 5.657708075969567e-05,
      "loss": 1.3687,
      "step": 1091
    },
    {
      "epoch": 5.111761263897016,
      "grad_norm": 0.4931749701499939,
      "learning_rate": 5.649584363194724e-05,
      "loss": 1.3766,
      "step": 1092
    },
    {
      "epoch": 5.11644236395553,
      "grad_norm": 0.49240025877952576,
      "learning_rate": 5.6414589058395004e-05,
      "loss": 1.3917,
      "step": 1093
    },
    {
      "epoch": 5.121123464014043,
      "grad_norm": 0.5174960494041443,
      "learning_rate": 5.6333317257263305e-05,
      "loss": 1.3842,
      "step": 1094
    },
    {
      "epoch": 5.125804564072557,
      "grad_norm": 0.5175259709358215,
      "learning_rate": 5.62520284468228e-05,
      "loss": 1.3828,
      "step": 1095
    },
    {
      "epoch": 5.130485664131071,
      "grad_norm": 0.5083780884742737,
      "learning_rate": 5.6170722845389765e-05,
      "loss": 1.3807,
      "step": 1096
    },
    {
      "epoch": 5.135166764189584,
      "grad_norm": 0.4956238865852356,
      "learning_rate": 5.6089400671325644e-05,
      "loss": 1.3676,
      "step": 1097
    },
    {
      "epoch": 5.1398478642480985,
      "grad_norm": 0.5338079929351807,
      "learning_rate": 5.6008062143036346e-05,
      "loss": 1.4095,
      "step": 1098
    },
    {
      "epoch": 5.144528964306612,
      "grad_norm": 0.5186750888824463,
      "learning_rate": 5.592670747897171e-05,
      "loss": 1.3557,
      "step": 1099
    },
    {
      "epoch": 5.149210064365126,
      "grad_norm": 0.505437970161438,
      "learning_rate": 5.5845336897624915e-05,
      "loss": 1.3948,
      "step": 1100
    },
    {
      "epoch": 5.15389116442364,
      "grad_norm": 0.4789367914199829,
      "learning_rate": 5.576395061753187e-05,
      "loss": 1.3748,
      "step": 1101
    },
    {
      "epoch": 5.158572264482153,
      "grad_norm": 0.5059368014335632,
      "learning_rate": 5.568254885727068e-05,
      "loss": 1.3792,
      "step": 1102
    },
    {
      "epoch": 5.163253364540667,
      "grad_norm": 0.5207875370979309,
      "learning_rate": 5.5601131835461003e-05,
      "loss": 1.3598,
      "step": 1103
    },
    {
      "epoch": 5.167934464599181,
      "grad_norm": 0.5086791515350342,
      "learning_rate": 5.55196997707635e-05,
      "loss": 1.3763,
      "step": 1104
    },
    {
      "epoch": 5.172615564657694,
      "grad_norm": 0.5212629437446594,
      "learning_rate": 5.5438252881879194e-05,
      "loss": 1.3568,
      "step": 1105
    },
    {
      "epoch": 5.177296664716208,
      "grad_norm": 0.5509768724441528,
      "learning_rate": 5.535679138754897e-05,
      "loss": 1.3703,
      "step": 1106
    },
    {
      "epoch": 5.181977764774722,
      "grad_norm": 0.5098535418510437,
      "learning_rate": 5.5275315506552906e-05,
      "loss": 1.3798,
      "step": 1107
    },
    {
      "epoch": 5.186658864833236,
      "grad_norm": 0.5017245411872864,
      "learning_rate": 5.5193825457709745e-05,
      "loss": 1.3775,
      "step": 1108
    },
    {
      "epoch": 5.19133996489175,
      "grad_norm": 0.523624062538147,
      "learning_rate": 5.511232145987626e-05,
      "loss": 1.3714,
      "step": 1109
    },
    {
      "epoch": 5.196021064950263,
      "grad_norm": 0.5008803009986877,
      "learning_rate": 5.5030803731946665e-05,
      "loss": 1.3593,
      "step": 1110
    },
    {
      "epoch": 5.200702165008777,
      "grad_norm": 0.5039325952529907,
      "learning_rate": 5.494927249285211e-05,
      "loss": 1.3629,
      "step": 1111
    },
    {
      "epoch": 5.205383265067291,
      "grad_norm": 0.4887968897819519,
      "learning_rate": 5.486772796155999e-05,
      "loss": 1.3852,
      "step": 1112
    },
    {
      "epoch": 5.210064365125804,
      "grad_norm": 0.5224624276161194,
      "learning_rate": 5.478617035707337e-05,
      "loss": 1.3957,
      "step": 1113
    },
    {
      "epoch": 5.214745465184318,
      "grad_norm": 0.5084558129310608,
      "learning_rate": 5.470459989843052e-05,
      "loss": 1.3731,
      "step": 1114
    },
    {
      "epoch": 5.219426565242832,
      "grad_norm": 0.48302987217903137,
      "learning_rate": 5.462301680470412e-05,
      "loss": 1.3728,
      "step": 1115
    },
    {
      "epoch": 5.224107665301346,
      "grad_norm": 0.5186018943786621,
      "learning_rate": 5.4541421295000864e-05,
      "loss": 1.3822,
      "step": 1116
    },
    {
      "epoch": 5.22878876535986,
      "grad_norm": 0.5260018110275269,
      "learning_rate": 5.445981358846077e-05,
      "loss": 1.3599,
      "step": 1117
    },
    {
      "epoch": 5.233469865418373,
      "grad_norm": 0.5030722618103027,
      "learning_rate": 5.437819390425659e-05,
      "loss": 1.3905,
      "step": 1118
    },
    {
      "epoch": 5.238150965476887,
      "grad_norm": 0.5315456986427307,
      "learning_rate": 5.429656246159327e-05,
      "loss": 1.3853,
      "step": 1119
    },
    {
      "epoch": 5.242832065535401,
      "grad_norm": 0.49964120984077454,
      "learning_rate": 5.4214919479707325e-05,
      "loss": 1.3709,
      "step": 1120
    },
    {
      "epoch": 5.247513165593914,
      "grad_norm": 0.5210468769073486,
      "learning_rate": 5.4133265177866256e-05,
      "loss": 1.3792,
      "step": 1121
    },
    {
      "epoch": 5.252194265652428,
      "grad_norm": 0.5345668792724609,
      "learning_rate": 5.4051599775368004e-05,
      "loss": 1.3763,
      "step": 1122
    },
    {
      "epoch": 5.256875365710942,
      "grad_norm": 0.5057262778282166,
      "learning_rate": 5.396992349154024e-05,
      "loss": 1.355,
      "step": 1123
    },
    {
      "epoch": 5.261556465769456,
      "grad_norm": 0.512408971786499,
      "learning_rate": 5.3888236545739955e-05,
      "loss": 1.3738,
      "step": 1124
    },
    {
      "epoch": 5.26623756582797,
      "grad_norm": 0.5097825527191162,
      "learning_rate": 5.380653915735272e-05,
      "loss": 1.388,
      "step": 1125
    },
    {
      "epoch": 5.270918665886484,
      "grad_norm": 0.5049353241920471,
      "learning_rate": 5.372483154579213e-05,
      "loss": 1.361,
      "step": 1126
    },
    {
      "epoch": 5.275599765944997,
      "grad_norm": 0.5072355270385742,
      "learning_rate": 5.364311393049931e-05,
      "loss": 1.3837,
      "step": 1127
    },
    {
      "epoch": 5.280280866003511,
      "grad_norm": 0.5134193897247314,
      "learning_rate": 5.356138653094219e-05,
      "loss": 1.3833,
      "step": 1128
    },
    {
      "epoch": 5.284961966062024,
      "grad_norm": 0.48993170261383057,
      "learning_rate": 5.347964956661502e-05,
      "loss": 1.399,
      "step": 1129
    },
    {
      "epoch": 5.289643066120538,
      "grad_norm": 0.5153975486755371,
      "learning_rate": 5.3397903257037685e-05,
      "loss": 1.3627,
      "step": 1130
    },
    {
      "epoch": 5.294324166179052,
      "grad_norm": 0.504741370677948,
      "learning_rate": 5.3316147821755205e-05,
      "loss": 1.3756,
      "step": 1131
    },
    {
      "epoch": 5.2990052662375655,
      "grad_norm": 0.502116322517395,
      "learning_rate": 5.3234383480337104e-05,
      "loss": 1.3584,
      "step": 1132
    },
    {
      "epoch": 5.30368636629608,
      "grad_norm": 0.5001406669616699,
      "learning_rate": 5.315261045237684e-05,
      "loss": 1.3828,
      "step": 1133
    },
    {
      "epoch": 5.308367466354594,
      "grad_norm": 0.49901527166366577,
      "learning_rate": 5.3070828957491156e-05,
      "loss": 1.3853,
      "step": 1134
    },
    {
      "epoch": 5.313048566413107,
      "grad_norm": 0.5057284832000732,
      "learning_rate": 5.29890392153196e-05,
      "loss": 1.3496,
      "step": 1135
    },
    {
      "epoch": 5.317729666471621,
      "grad_norm": 0.5117406845092773,
      "learning_rate": 5.290724144552379e-05,
      "loss": 1.3709,
      "step": 1136
    },
    {
      "epoch": 5.322410766530135,
      "grad_norm": 0.5053237080574036,
      "learning_rate": 5.2825435867786975e-05,
      "loss": 1.366,
      "step": 1137
    },
    {
      "epoch": 5.327091866588648,
      "grad_norm": 0.5771435499191284,
      "learning_rate": 5.2743622701813344e-05,
      "loss": 1.3787,
      "step": 1138
    },
    {
      "epoch": 5.331772966647162,
      "grad_norm": 0.5015900135040283,
      "learning_rate": 5.2661802167327445e-05,
      "loss": 1.3677,
      "step": 1139
    },
    {
      "epoch": 5.3364540667056755,
      "grad_norm": 0.4910103380680084,
      "learning_rate": 5.2579974484073655e-05,
      "loss": 1.357,
      "step": 1140
    },
    {
      "epoch": 5.3411351667641895,
      "grad_norm": 0.5642368793487549,
      "learning_rate": 5.249813987181553e-05,
      "loss": 1.3786,
      "step": 1141
    },
    {
      "epoch": 5.345816266822704,
      "grad_norm": 0.519067645072937,
      "learning_rate": 5.241629855033523e-05,
      "loss": 1.3667,
      "step": 1142
    },
    {
      "epoch": 5.350497366881217,
      "grad_norm": 0.483230859041214,
      "learning_rate": 5.233445073943295e-05,
      "loss": 1.366,
      "step": 1143
    },
    {
      "epoch": 5.355178466939731,
      "grad_norm": 0.494340717792511,
      "learning_rate": 5.225259665892629e-05,
      "loss": 1.3661,
      "step": 1144
    },
    {
      "epoch": 5.359859566998245,
      "grad_norm": 0.5203878879547119,
      "learning_rate": 5.217073652864972e-05,
      "loss": 1.3753,
      "step": 1145
    },
    {
      "epoch": 5.364540667056758,
      "grad_norm": 0.4912766218185425,
      "learning_rate": 5.208887056845394e-05,
      "loss": 1.3673,
      "step": 1146
    },
    {
      "epoch": 5.369221767115272,
      "grad_norm": 0.5085307359695435,
      "learning_rate": 5.200699899820527e-05,
      "loss": 1.3987,
      "step": 1147
    },
    {
      "epoch": 5.373902867173786,
      "grad_norm": 0.5408493876457214,
      "learning_rate": 5.192512203778519e-05,
      "loss": 1.362,
      "step": 1148
    },
    {
      "epoch": 5.3785839672322995,
      "grad_norm": 0.5014744997024536,
      "learning_rate": 5.184323990708958e-05,
      "loss": 1.3545,
      "step": 1149
    },
    {
      "epoch": 5.3832650672908136,
      "grad_norm": 0.5128848552703857,
      "learning_rate": 5.176135282602821e-05,
      "loss": 1.3659,
      "step": 1150
    },
    {
      "epoch": 5.387946167349327,
      "grad_norm": 0.49814894795417786,
      "learning_rate": 5.167946101452419e-05,
      "loss": 1.3796,
      "step": 1151
    },
    {
      "epoch": 5.392627267407841,
      "grad_norm": 0.4915131628513336,
      "learning_rate": 5.159756469251327e-05,
      "loss": 1.3845,
      "step": 1152
    },
    {
      "epoch": 5.397308367466355,
      "grad_norm": 0.48490697145462036,
      "learning_rate": 5.151566407994339e-05,
      "loss": 1.3686,
      "step": 1153
    },
    {
      "epoch": 5.401989467524868,
      "grad_norm": 0.5091562867164612,
      "learning_rate": 5.143375939677396e-05,
      "loss": 1.365,
      "step": 1154
    },
    {
      "epoch": 5.406670567583382,
      "grad_norm": 0.5359136462211609,
      "learning_rate": 5.1351850862975315e-05,
      "loss": 1.363,
      "step": 1155
    },
    {
      "epoch": 5.411351667641896,
      "grad_norm": 0.4946351945400238,
      "learning_rate": 5.126993869852819e-05,
      "loss": 1.3613,
      "step": 1156
    },
    {
      "epoch": 5.416032767700409,
      "grad_norm": 0.5049864649772644,
      "learning_rate": 5.118802312342299e-05,
      "loss": 1.3847,
      "step": 1157
    },
    {
      "epoch": 5.4207138677589235,
      "grad_norm": 0.506483256816864,
      "learning_rate": 5.110610435765934e-05,
      "loss": 1.3732,
      "step": 1158
    },
    {
      "epoch": 5.425394967817437,
      "grad_norm": 0.5119485855102539,
      "learning_rate": 5.1024182621245455e-05,
      "loss": 1.3457,
      "step": 1159
    },
    {
      "epoch": 5.430076067875951,
      "grad_norm": 0.507080614566803,
      "learning_rate": 5.094225813419743e-05,
      "loss": 1.3595,
      "step": 1160
    },
    {
      "epoch": 5.434757167934465,
      "grad_norm": 0.49319717288017273,
      "learning_rate": 5.086033111653884e-05,
      "loss": 1.3653,
      "step": 1161
    },
    {
      "epoch": 5.439438267992978,
      "grad_norm": 0.5073138475418091,
      "learning_rate": 5.0778401788300026e-05,
      "loss": 1.3739,
      "step": 1162
    },
    {
      "epoch": 5.444119368051492,
      "grad_norm": 0.4962432086467743,
      "learning_rate": 5.069647036951751e-05,
      "loss": 1.3528,
      "step": 1163
    },
    {
      "epoch": 5.448800468110006,
      "grad_norm": 0.48607367277145386,
      "learning_rate": 5.0614537080233484e-05,
      "loss": 1.346,
      "step": 1164
    },
    {
      "epoch": 5.453481568168519,
      "grad_norm": 0.5067170858383179,
      "learning_rate": 5.05326021404951e-05,
      "loss": 1.3729,
      "step": 1165
    },
    {
      "epoch": 5.458162668227033,
      "grad_norm": 0.5128630995750427,
      "learning_rate": 5.045066577035401e-05,
      "loss": 1.3561,
      "step": 1166
    },
    {
      "epoch": 5.4628437682855475,
      "grad_norm": 0.508293867111206,
      "learning_rate": 5.036872818986562e-05,
      "loss": 1.3688,
      "step": 1167
    },
    {
      "epoch": 5.467524868344061,
      "grad_norm": 0.5079975128173828,
      "learning_rate": 5.028678961908868e-05,
      "loss": 1.3581,
      "step": 1168
    },
    {
      "epoch": 5.472205968402575,
      "grad_norm": 0.517815113067627,
      "learning_rate": 5.020485027808455e-05,
      "loss": 1.37,
      "step": 1169
    },
    {
      "epoch": 5.476887068461088,
      "grad_norm": 0.5143315196037292,
      "learning_rate": 5.0122910386916656e-05,
      "loss": 1.3897,
      "step": 1170
    },
    {
      "epoch": 5.481568168519602,
      "grad_norm": 0.5109765529632568,
      "learning_rate": 5.0040970165649906e-05,
      "loss": 1.3633,
      "step": 1171
    },
    {
      "epoch": 5.486249268578116,
      "grad_norm": 0.5113378167152405,
      "learning_rate": 4.995902983435012e-05,
      "loss": 1.3613,
      "step": 1172
    },
    {
      "epoch": 5.490930368636629,
      "grad_norm": 0.4970690608024597,
      "learning_rate": 4.9877089613083356e-05,
      "loss": 1.3629,
      "step": 1173
    },
    {
      "epoch": 5.495611468695143,
      "grad_norm": 0.5230845212936401,
      "learning_rate": 4.9795149721915476e-05,
      "loss": 1.383,
      "step": 1174
    },
    {
      "epoch": 5.5002925687536575,
      "grad_norm": 0.5186367034912109,
      "learning_rate": 4.971321038091133e-05,
      "loss": 1.3685,
      "step": 1175
    },
    {
      "epoch": 5.504973668812171,
      "grad_norm": 0.5045866966247559,
      "learning_rate": 4.963127181013438e-05,
      "loss": 1.3856,
      "step": 1176
    },
    {
      "epoch": 5.509654768870685,
      "grad_norm": 0.49818751215934753,
      "learning_rate": 4.954933422964602e-05,
      "loss": 1.3566,
      "step": 1177
    },
    {
      "epoch": 5.514335868929198,
      "grad_norm": 0.5398290753364563,
      "learning_rate": 4.94673978595049e-05,
      "loss": 1.376,
      "step": 1178
    },
    {
      "epoch": 5.519016968987712,
      "grad_norm": 0.5045533180236816,
      "learning_rate": 4.938546291976654e-05,
      "loss": 1.3568,
      "step": 1179
    },
    {
      "epoch": 5.523698069046226,
      "grad_norm": 0.5707494020462036,
      "learning_rate": 4.9303529630482495e-05,
      "loss": 1.3664,
      "step": 1180
    },
    {
      "epoch": 5.528379169104739,
      "grad_norm": 0.505876362323761,
      "learning_rate": 4.922159821169999e-05,
      "loss": 1.3668,
      "step": 1181
    },
    {
      "epoch": 5.533060269163253,
      "grad_norm": 0.5047862529754639,
      "learning_rate": 4.913966888346118e-05,
      "loss": 1.3598,
      "step": 1182
    },
    {
      "epoch": 5.537741369221767,
      "grad_norm": 0.4833430349826813,
      "learning_rate": 4.905774186580259e-05,
      "loss": 1.3625,
      "step": 1183
    },
    {
      "epoch": 5.542422469280281,
      "grad_norm": 0.5005712509155273,
      "learning_rate": 4.8975817378754576e-05,
      "loss": 1.3525,
      "step": 1184
    },
    {
      "epoch": 5.547103569338795,
      "grad_norm": 0.4941693842411041,
      "learning_rate": 4.889389564234066e-05,
      "loss": 1.3613,
      "step": 1185
    },
    {
      "epoch": 5.551784669397309,
      "grad_norm": 0.5212202072143555,
      "learning_rate": 4.881197687657702e-05,
      "loss": 1.38,
      "step": 1186
    },
    {
      "epoch": 5.556465769455822,
      "grad_norm": 0.4857902526855469,
      "learning_rate": 4.8730061301471836e-05,
      "loss": 1.378,
      "step": 1187
    },
    {
      "epoch": 5.561146869514336,
      "grad_norm": 0.4982302486896515,
      "learning_rate": 4.864814913702469e-05,
      "loss": 1.3625,
      "step": 1188
    },
    {
      "epoch": 5.565827969572849,
      "grad_norm": 0.5094555020332336,
      "learning_rate": 4.856624060322606e-05,
      "loss": 1.3751,
      "step": 1189
    },
    {
      "epoch": 5.570509069631363,
      "grad_norm": 0.4969262480735779,
      "learning_rate": 4.8484335920056615e-05,
      "loss": 1.346,
      "step": 1190
    },
    {
      "epoch": 5.575190169689877,
      "grad_norm": 0.5096610188484192,
      "learning_rate": 4.8402435307486734e-05,
      "loss": 1.3752,
      "step": 1191
    },
    {
      "epoch": 5.5798712697483905,
      "grad_norm": 0.5154902338981628,
      "learning_rate": 4.832053898547583e-05,
      "loss": 1.3772,
      "step": 1192
    },
    {
      "epoch": 5.584552369806905,
      "grad_norm": 0.5006143450737,
      "learning_rate": 4.82386471739718e-05,
      "loss": 1.3897,
      "step": 1193
    },
    {
      "epoch": 5.589233469865419,
      "grad_norm": 0.5159085392951965,
      "learning_rate": 4.8156760092910444e-05,
      "loss": 1.387,
      "step": 1194
    },
    {
      "epoch": 5.593914569923932,
      "grad_norm": 0.4952409565448761,
      "learning_rate": 4.807487796221482e-05,
      "loss": 1.3462,
      "step": 1195
    },
    {
      "epoch": 5.598595669982446,
      "grad_norm": 0.49595585465431213,
      "learning_rate": 4.799300100179473e-05,
      "loss": 1.3811,
      "step": 1196
    },
    {
      "epoch": 5.60327677004096,
      "grad_norm": 0.49851763248443604,
      "learning_rate": 4.7911129431546084e-05,
      "loss": 1.3673,
      "step": 1197
    },
    {
      "epoch": 5.607957870099473,
      "grad_norm": 0.5071251392364502,
      "learning_rate": 4.782926347135029e-05,
      "loss": 1.3756,
      "step": 1198
    },
    {
      "epoch": 5.612638970157987,
      "grad_norm": 0.49200981855392456,
      "learning_rate": 4.7747403341073725e-05,
      "loss": 1.3631,
      "step": 1199
    },
    {
      "epoch": 5.6173200702165005,
      "grad_norm": 0.47780096530914307,
      "learning_rate": 4.766554926056707e-05,
      "loss": 1.3533,
      "step": 1200
    },
    {
      "epoch": 5.622001170275015,
      "grad_norm": 0.5060575008392334,
      "learning_rate": 4.758370144966477e-05,
      "loss": 1.4004,
      "step": 1201
    },
    {
      "epoch": 5.626682270333529,
      "grad_norm": 0.5059685707092285,
      "learning_rate": 4.750186012818448e-05,
      "loss": 1.3682,
      "step": 1202
    },
    {
      "epoch": 5.631363370392042,
      "grad_norm": 0.4853475093841553,
      "learning_rate": 4.742002551592635e-05,
      "loss": 1.3636,
      "step": 1203
    },
    {
      "epoch": 5.636044470450556,
      "grad_norm": 0.5116673707962036,
      "learning_rate": 4.7338197832672566e-05,
      "loss": 1.3723,
      "step": 1204
    },
    {
      "epoch": 5.64072557050907,
      "grad_norm": 0.5065430998802185,
      "learning_rate": 4.725637729818667e-05,
      "loss": 1.3578,
      "step": 1205
    },
    {
      "epoch": 5.645406670567583,
      "grad_norm": 0.4856509864330292,
      "learning_rate": 4.717456413221302e-05,
      "loss": 1.3414,
      "step": 1206
    },
    {
      "epoch": 5.650087770626097,
      "grad_norm": 0.49742671847343445,
      "learning_rate": 4.709275855447621e-05,
      "loss": 1.3629,
      "step": 1207
    },
    {
      "epoch": 5.654768870684611,
      "grad_norm": 0.5032638311386108,
      "learning_rate": 4.701096078468041e-05,
      "loss": 1.354,
      "step": 1208
    },
    {
      "epoch": 5.6594499707431245,
      "grad_norm": 0.490828275680542,
      "learning_rate": 4.6929171042508855e-05,
      "loss": 1.3569,
      "step": 1209
    },
    {
      "epoch": 5.664131070801639,
      "grad_norm": 0.5146189332008362,
      "learning_rate": 4.684738954762316e-05,
      "loss": 1.3846,
      "step": 1210
    },
    {
      "epoch": 5.668812170860152,
      "grad_norm": 0.4796738028526306,
      "learning_rate": 4.6765616519662894e-05,
      "loss": 1.3576,
      "step": 1211
    },
    {
      "epoch": 5.673493270918666,
      "grad_norm": 0.5143107771873474,
      "learning_rate": 4.668385217824482e-05,
      "loss": 1.365,
      "step": 1212
    },
    {
      "epoch": 5.67817437097718,
      "grad_norm": 0.5042698979377747,
      "learning_rate": 4.660209674296233e-05,
      "loss": 1.3561,
      "step": 1213
    },
    {
      "epoch": 5.682855471035693,
      "grad_norm": 0.4811539351940155,
      "learning_rate": 4.652035043338501e-05,
      "loss": 1.3464,
      "step": 1214
    },
    {
      "epoch": 5.687536571094207,
      "grad_norm": 0.5001482963562012,
      "learning_rate": 4.643861346905781e-05,
      "loss": 1.3742,
      "step": 1215
    },
    {
      "epoch": 5.692217671152721,
      "grad_norm": 0.5193770527839661,
      "learning_rate": 4.6356886069500685e-05,
      "loss": 1.3502,
      "step": 1216
    },
    {
      "epoch": 5.6968987712112344,
      "grad_norm": 0.5073791742324829,
      "learning_rate": 4.6275168454207885e-05,
      "loss": 1.3572,
      "step": 1217
    },
    {
      "epoch": 5.7015798712697485,
      "grad_norm": 0.4926114082336426,
      "learning_rate": 4.619346084264729e-05,
      "loss": 1.3439,
      "step": 1218
    },
    {
      "epoch": 5.706260971328263,
      "grad_norm": 0.5074840784072876,
      "learning_rate": 4.611176345426006e-05,
      "loss": 1.386,
      "step": 1219
    },
    {
      "epoch": 5.710942071386776,
      "grad_norm": 0.5240839719772339,
      "learning_rate": 4.6030076508459776e-05,
      "loss": 1.3681,
      "step": 1220
    },
    {
      "epoch": 5.71562317144529,
      "grad_norm": 0.46889132261276245,
      "learning_rate": 4.5948400224632e-05,
      "loss": 1.3777,
      "step": 1221
    },
    {
      "epoch": 5.720304271503803,
      "grad_norm": 0.5044253468513489,
      "learning_rate": 4.5866734822133755e-05,
      "loss": 1.3609,
      "step": 1222
    },
    {
      "epoch": 5.724985371562317,
      "grad_norm": 0.5392898321151733,
      "learning_rate": 4.5785080520292687e-05,
      "loss": 1.3505,
      "step": 1223
    },
    {
      "epoch": 5.729666471620831,
      "grad_norm": 0.4860037863254547,
      "learning_rate": 4.570343753840675e-05,
      "loss": 1.3765,
      "step": 1224
    },
    {
      "epoch": 5.734347571679344,
      "grad_norm": 0.49485546350479126,
      "learning_rate": 4.562180609574343e-05,
      "loss": 1.378,
      "step": 1225
    },
    {
      "epoch": 5.7390286717378585,
      "grad_norm": 0.4981268048286438,
      "learning_rate": 4.5540186411539234e-05,
      "loss": 1.3644,
      "step": 1226
    },
    {
      "epoch": 5.743709771796372,
      "grad_norm": 0.5137907862663269,
      "learning_rate": 4.545857870499914e-05,
      "loss": 1.3722,
      "step": 1227
    },
    {
      "epoch": 5.748390871854886,
      "grad_norm": 0.5029009580612183,
      "learning_rate": 4.537698319529588e-05,
      "loss": 1.351,
      "step": 1228
    },
    {
      "epoch": 5.7530719719134,
      "grad_norm": 0.4825573265552521,
      "learning_rate": 4.52954001015695e-05,
      "loss": 1.3512,
      "step": 1229
    },
    {
      "epoch": 5.757753071971914,
      "grad_norm": 0.5327299237251282,
      "learning_rate": 4.521382964292663e-05,
      "loss": 1.3836,
      "step": 1230
    },
    {
      "epoch": 5.762434172030427,
      "grad_norm": 0.49493491649627686,
      "learning_rate": 4.513227203844003e-05,
      "loss": 1.353,
      "step": 1231
    },
    {
      "epoch": 5.767115272088941,
      "grad_norm": 0.4785345196723938,
      "learning_rate": 4.5050727507147905e-05,
      "loss": 1.362,
      "step": 1232
    },
    {
      "epoch": 5.771796372147454,
      "grad_norm": 0.5051293969154358,
      "learning_rate": 4.496919626805334e-05,
      "loss": 1.3773,
      "step": 1233
    },
    {
      "epoch": 5.776477472205968,
      "grad_norm": 0.48809775710105896,
      "learning_rate": 4.488767854012376e-05,
      "loss": 1.3907,
      "step": 1234
    },
    {
      "epoch": 5.7811585722644825,
      "grad_norm": 0.5047741532325745,
      "learning_rate": 4.4806174542290266e-05,
      "loss": 1.3612,
      "step": 1235
    },
    {
      "epoch": 5.785839672322996,
      "grad_norm": 0.48427894711494446,
      "learning_rate": 4.472468449344709e-05,
      "loss": 1.3631,
      "step": 1236
    },
    {
      "epoch": 5.79052077238151,
      "grad_norm": 0.49840694665908813,
      "learning_rate": 4.4643208612451043e-05,
      "loss": 1.3762,
      "step": 1237
    },
    {
      "epoch": 5.795201872440023,
      "grad_norm": 0.5097300410270691,
      "learning_rate": 4.456174711812082e-05,
      "loss": 1.3545,
      "step": 1238
    },
    {
      "epoch": 5.799882972498537,
      "grad_norm": 0.5060076117515564,
      "learning_rate": 4.448030022923652e-05,
      "loss": 1.3726,
      "step": 1239
    },
    {
      "epoch": 5.804564072557051,
      "grad_norm": 0.4817342460155487,
      "learning_rate": 4.4398868164539e-05,
      "loss": 1.3556,
      "step": 1240
    },
    {
      "epoch": 5.809245172615564,
      "grad_norm": 0.48569947481155396,
      "learning_rate": 4.4317451142729324e-05,
      "loss": 1.3296,
      "step": 1241
    },
    {
      "epoch": 5.813926272674078,
      "grad_norm": 0.5207067131996155,
      "learning_rate": 4.423604938246815e-05,
      "loss": 1.3719,
      "step": 1242
    },
    {
      "epoch": 5.818607372732592,
      "grad_norm": 0.496518075466156,
      "learning_rate": 4.4154663102375104e-05,
      "loss": 1.3624,
      "step": 1243
    },
    {
      "epoch": 5.823288472791106,
      "grad_norm": 0.49017754197120667,
      "learning_rate": 4.407329252102831e-05,
      "loss": 1.3624,
      "step": 1244
    },
    {
      "epoch": 5.82796957284962,
      "grad_norm": 0.491558700799942,
      "learning_rate": 4.399193785696366e-05,
      "loss": 1.3518,
      "step": 1245
    },
    {
      "epoch": 5.832650672908134,
      "grad_norm": 0.48818331956863403,
      "learning_rate": 4.3910599328674354e-05,
      "loss": 1.3409,
      "step": 1246
    },
    {
      "epoch": 5.837331772966647,
      "grad_norm": 0.4837343096733093,
      "learning_rate": 4.3829277154610246e-05,
      "loss": 1.3607,
      "step": 1247
    },
    {
      "epoch": 5.842012873025161,
      "grad_norm": 0.49654173851013184,
      "learning_rate": 4.374797155317722e-05,
      "loss": 1.35,
      "step": 1248
    },
    {
      "epoch": 5.846693973083674,
      "grad_norm": 0.49282020330429077,
      "learning_rate": 4.36666827427367e-05,
      "loss": 1.3615,
      "step": 1249
    },
    {
      "epoch": 5.851375073142188,
      "grad_norm": 0.48395588994026184,
      "learning_rate": 4.3585410941605e-05,
      "loss": 1.364,
      "step": 1250
    },
    {
      "epoch": 5.856056173200702,
      "grad_norm": 0.4863456189632416,
      "learning_rate": 4.3504156368052765e-05,
      "loss": 1.3643,
      "step": 1251
    },
    {
      "epoch": 5.860737273259216,
      "grad_norm": 0.4858078062534332,
      "learning_rate": 4.3422919240304344e-05,
      "loss": 1.3664,
      "step": 1252
    },
    {
      "epoch": 5.86541837331773,
      "grad_norm": 0.49061718583106995,
      "learning_rate": 4.334169977653725e-05,
      "loss": 1.3704,
      "step": 1253
    },
    {
      "epoch": 5.870099473376244,
      "grad_norm": 0.4844505488872528,
      "learning_rate": 4.326049819488157e-05,
      "loss": 1.3413,
      "step": 1254
    },
    {
      "epoch": 5.874780573434757,
      "grad_norm": 0.4844004213809967,
      "learning_rate": 4.3179314713419314e-05,
      "loss": 1.356,
      "step": 1255
    },
    {
      "epoch": 5.879461673493271,
      "grad_norm": 0.48914822936058044,
      "learning_rate": 4.3098149550183936e-05,
      "loss": 1.3519,
      "step": 1256
    },
    {
      "epoch": 5.884142773551785,
      "grad_norm": 0.4801008105278015,
      "learning_rate": 4.30170029231597e-05,
      "loss": 1.3625,
      "step": 1257
    },
    {
      "epoch": 5.888823873610298,
      "grad_norm": 0.4835020899772644,
      "learning_rate": 4.2935875050281e-05,
      "loss": 1.3461,
      "step": 1258
    },
    {
      "epoch": 5.893504973668812,
      "grad_norm": 0.5038294196128845,
      "learning_rate": 4.285476614943199e-05,
      "loss": 1.3209,
      "step": 1259
    },
    {
      "epoch": 5.8981860737273255,
      "grad_norm": 0.4965617060661316,
      "learning_rate": 4.277367643844574e-05,
      "loss": 1.3918,
      "step": 1260
    },
    {
      "epoch": 5.90286717378584,
      "grad_norm": 0.4906851351261139,
      "learning_rate": 4.2692606135103884e-05,
      "loss": 1.3669,
      "step": 1261
    },
    {
      "epoch": 5.907548273844354,
      "grad_norm": 0.48122575879096985,
      "learning_rate": 4.26115554571359e-05,
      "loss": 1.3443,
      "step": 1262
    },
    {
      "epoch": 5.912229373902867,
      "grad_norm": 0.48655739426612854,
      "learning_rate": 4.253052462221851e-05,
      "loss": 1.3694,
      "step": 1263
    },
    {
      "epoch": 5.916910473961381,
      "grad_norm": 0.4788222312927246,
      "learning_rate": 4.244951384797525e-05,
      "loss": 1.3731,
      "step": 1264
    },
    {
      "epoch": 5.921591574019895,
      "grad_norm": 0.5071664452552795,
      "learning_rate": 4.236852335197566e-05,
      "loss": 1.3895,
      "step": 1265
    },
    {
      "epoch": 5.926272674078408,
      "grad_norm": 0.49297115206718445,
      "learning_rate": 4.228755335173488e-05,
      "loss": 1.362,
      "step": 1266
    },
    {
      "epoch": 5.930953774136922,
      "grad_norm": 0.49331197142601013,
      "learning_rate": 4.2206604064713054e-05,
      "loss": 1.3823,
      "step": 1267
    },
    {
      "epoch": 5.935634874195436,
      "grad_norm": 0.4724307060241699,
      "learning_rate": 4.212567570831457e-05,
      "loss": 1.3621,
      "step": 1268
    },
    {
      "epoch": 5.9403159742539495,
      "grad_norm": 0.4933937191963196,
      "learning_rate": 4.204476849988773e-05,
      "loss": 1.3606,
      "step": 1269
    },
    {
      "epoch": 5.944997074312464,
      "grad_norm": 0.4997156858444214,
      "learning_rate": 4.1963882656723954e-05,
      "loss": 1.3673,
      "step": 1270
    },
    {
      "epoch": 5.949678174370977,
      "grad_norm": 0.49473080039024353,
      "learning_rate": 4.1883018396057307e-05,
      "loss": 1.3366,
      "step": 1271
    },
    {
      "epoch": 5.954359274429491,
      "grad_norm": 0.5106769800186157,
      "learning_rate": 4.180217593506394e-05,
      "loss": 1.3577,
      "step": 1272
    },
    {
      "epoch": 5.959040374488005,
      "grad_norm": 0.4847675561904907,
      "learning_rate": 4.1721355490861367e-05,
      "loss": 1.3707,
      "step": 1273
    },
    {
      "epoch": 5.963721474546518,
      "grad_norm": 0.5004474520683289,
      "learning_rate": 4.1640557280508044e-05,
      "loss": 1.3667,
      "step": 1274
    },
    {
      "epoch": 5.968402574605032,
      "grad_norm": 0.5049918293952942,
      "learning_rate": 4.1559781521002664e-05,
      "loss": 1.3726,
      "step": 1275
    },
    {
      "epoch": 5.973083674663546,
      "grad_norm": 0.49814268946647644,
      "learning_rate": 4.147902842928366e-05,
      "loss": 1.3621,
      "step": 1276
    },
    {
      "epoch": 5.9777647747220595,
      "grad_norm": 0.5119578838348389,
      "learning_rate": 4.139829822222858e-05,
      "loss": 1.3846,
      "step": 1277
    },
    {
      "epoch": 5.9824458747805735,
      "grad_norm": 0.4856437146663666,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.341,
      "step": 1278
    },
    {
      "epoch": 5.987126974839088,
      "grad_norm": 0.5015390515327454,
      "learning_rate": 4.1236907329312435e-05,
      "loss": 1.3391,
      "step": 1279
    },
    {
      "epoch": 5.991808074897601,
      "grad_norm": 0.5185198187828064,
      "learning_rate": 4.115624707689683e-05,
      "loss": 1.3637,
      "step": 1280
    },
    {
      "epoch": 5.996489174956115,
      "grad_norm": 0.5117372274398804,
      "learning_rate": 4.107561057603486e-05,
      "loss": 1.3667,
      "step": 1281
    },
    {
      "epoch": 6.001170275014628,
      "grad_norm": 1.6847134828567505,
      "learning_rate": 4.0994998043290964e-05,
      "loss": 1.3563,
      "step": 1282
    },
    {
      "epoch": 6.005851375073142,
      "grad_norm": 0.5279262065887451,
      "learning_rate": 4.0914409695165165e-05,
      "loss": 1.3405,
      "step": 1283
    },
    {
      "epoch": 6.010532475131656,
      "grad_norm": 0.5033664703369141,
      "learning_rate": 4.0833845748092586e-05,
      "loss": 1.3468,
      "step": 1284
    },
    {
      "epoch": 6.015213575190169,
      "grad_norm": 0.5114166736602783,
      "learning_rate": 4.075330641844275e-05,
      "loss": 1.3727,
      "step": 1285
    },
    {
      "epoch": 6.0198946752486835,
      "grad_norm": 0.5349748134613037,
      "learning_rate": 4.067279192251911e-05,
      "loss": 1.3489,
      "step": 1286
    },
    {
      "epoch": 6.024575775307198,
      "grad_norm": 0.5186452269554138,
      "learning_rate": 4.059230247655843e-05,
      "loss": 1.3473,
      "step": 1287
    },
    {
      "epoch": 6.029256875365711,
      "grad_norm": 0.5082773566246033,
      "learning_rate": 4.051183829673016e-05,
      "loss": 1.3607,
      "step": 1288
    },
    {
      "epoch": 6.033937975424225,
      "grad_norm": 0.5214784741401672,
      "learning_rate": 4.043139959913593e-05,
      "loss": 1.3429,
      "step": 1289
    },
    {
      "epoch": 6.038619075482738,
      "grad_norm": 0.5377727746963501,
      "learning_rate": 4.035098659980891e-05,
      "loss": 1.3523,
      "step": 1290
    },
    {
      "epoch": 6.043300175541252,
      "grad_norm": 0.48703083395957947,
      "learning_rate": 4.0270599514713256e-05,
      "loss": 1.3616,
      "step": 1291
    },
    {
      "epoch": 6.047981275599766,
      "grad_norm": 0.5291474461555481,
      "learning_rate": 4.019023855974354e-05,
      "loss": 1.3612,
      "step": 1292
    },
    {
      "epoch": 6.052662375658279,
      "grad_norm": 0.541254460811615,
      "learning_rate": 4.0109903950724134e-05,
      "loss": 1.3449,
      "step": 1293
    },
    {
      "epoch": 6.057343475716793,
      "grad_norm": 0.49508151412010193,
      "learning_rate": 4.002959590340869e-05,
      "loss": 1.3784,
      "step": 1294
    },
    {
      "epoch": 6.0620245757753075,
      "grad_norm": 0.4925527572631836,
      "learning_rate": 3.994931463347946e-05,
      "loss": 1.3626,
      "step": 1295
    },
    {
      "epoch": 6.066705675833821,
      "grad_norm": 0.522050142288208,
      "learning_rate": 3.9869060356546855e-05,
      "loss": 1.3445,
      "step": 1296
    },
    {
      "epoch": 6.071386775892335,
      "grad_norm": 0.498424232006073,
      "learning_rate": 3.978883328814875e-05,
      "loss": 1.3489,
      "step": 1297
    },
    {
      "epoch": 6.076067875950849,
      "grad_norm": 0.4769817590713501,
      "learning_rate": 3.9708633643749924e-05,
      "loss": 1.354,
      "step": 1298
    },
    {
      "epoch": 6.080748976009362,
      "grad_norm": 0.4856712818145752,
      "learning_rate": 3.9628461638741584e-05,
      "loss": 1.3503,
      "step": 1299
    },
    {
      "epoch": 6.085430076067876,
      "grad_norm": 0.48528093099594116,
      "learning_rate": 3.954831748844059e-05,
      "loss": 1.3485,
      "step": 1300
    },
    {
      "epoch": 6.090111176126389,
      "grad_norm": 0.489759236574173,
      "learning_rate": 3.94682014080891e-05,
      "loss": 1.3547,
      "step": 1301
    },
    {
      "epoch": 6.094792276184903,
      "grad_norm": 0.5019160509109497,
      "learning_rate": 3.9388113612853855e-05,
      "loss": 1.3499,
      "step": 1302
    },
    {
      "epoch": 6.0994733762434175,
      "grad_norm": 0.49624067544937134,
      "learning_rate": 3.9308054317825564e-05,
      "loss": 1.3506,
      "step": 1303
    },
    {
      "epoch": 6.104154476301931,
      "grad_norm": 0.5070812702178955,
      "learning_rate": 3.9228023738018513e-05,
      "loss": 1.3611,
      "step": 1304
    },
    {
      "epoch": 6.108835576360445,
      "grad_norm": 0.4832218885421753,
      "learning_rate": 3.914802208836973e-05,
      "loss": 1.3443,
      "step": 1305
    },
    {
      "epoch": 6.113516676418959,
      "grad_norm": 0.4949527382850647,
      "learning_rate": 3.906804958373868e-05,
      "loss": 1.3631,
      "step": 1306
    },
    {
      "epoch": 6.118197776477472,
      "grad_norm": 0.4915694296360016,
      "learning_rate": 3.898810643890648e-05,
      "loss": 1.351,
      "step": 1307
    },
    {
      "epoch": 6.122878876535986,
      "grad_norm": 0.5031571984291077,
      "learning_rate": 3.8908192868575366e-05,
      "loss": 1.347,
      "step": 1308
    },
    {
      "epoch": 6.1275599765945,
      "grad_norm": 0.4988706707954407,
      "learning_rate": 3.882830908736825e-05,
      "loss": 1.3601,
      "step": 1309
    },
    {
      "epoch": 6.132241076653013,
      "grad_norm": 0.49566569924354553,
      "learning_rate": 3.874845530982792e-05,
      "loss": 1.3587,
      "step": 1310
    },
    {
      "epoch": 6.136922176711527,
      "grad_norm": 0.4733045995235443,
      "learning_rate": 3.866863175041666e-05,
      "loss": 1.3419,
      "step": 1311
    },
    {
      "epoch": 6.141603276770041,
      "grad_norm": 0.48297199606895447,
      "learning_rate": 3.858883862351561e-05,
      "loss": 1.3464,
      "step": 1312
    },
    {
      "epoch": 6.146284376828555,
      "grad_norm": 0.4759235084056854,
      "learning_rate": 3.85090761434241e-05,
      "loss": 1.3634,
      "step": 1313
    },
    {
      "epoch": 6.150965476887069,
      "grad_norm": 0.4884612560272217,
      "learning_rate": 3.842934452435923e-05,
      "loss": 1.3608,
      "step": 1314
    },
    {
      "epoch": 6.155646576945582,
      "grad_norm": 0.48609334230422974,
      "learning_rate": 3.8349643980455165e-05,
      "loss": 1.3485,
      "step": 1315
    },
    {
      "epoch": 6.160327677004096,
      "grad_norm": 0.48751768469810486,
      "learning_rate": 3.8269974725762624e-05,
      "loss": 1.3482,
      "step": 1316
    },
    {
      "epoch": 6.16500877706261,
      "grad_norm": 0.49484318494796753,
      "learning_rate": 3.8190336974248356e-05,
      "loss": 1.3699,
      "step": 1317
    },
    {
      "epoch": 6.169689877121123,
      "grad_norm": 0.5006123781204224,
      "learning_rate": 3.811073093979439e-05,
      "loss": 1.3687,
      "step": 1318
    },
    {
      "epoch": 6.174370977179637,
      "grad_norm": 0.4835350215435028,
      "learning_rate": 3.803115683619766e-05,
      "loss": 1.3669,
      "step": 1319
    },
    {
      "epoch": 6.1790520772381505,
      "grad_norm": 0.482292115688324,
      "learning_rate": 3.7951614877169284e-05,
      "loss": 1.3527,
      "step": 1320
    },
    {
      "epoch": 6.183733177296665,
      "grad_norm": 0.5049312114715576,
      "learning_rate": 3.787210527633411e-05,
      "loss": 1.3698,
      "step": 1321
    },
    {
      "epoch": 6.188414277355179,
      "grad_norm": 0.5008333921432495,
      "learning_rate": 3.779262824723005e-05,
      "loss": 1.3593,
      "step": 1322
    },
    {
      "epoch": 6.193095377413692,
      "grad_norm": 0.4848458170890808,
      "learning_rate": 3.771318400330752e-05,
      "loss": 1.3598,
      "step": 1323
    },
    {
      "epoch": 6.197776477472206,
      "grad_norm": 0.4954104721546173,
      "learning_rate": 3.763377275792893e-05,
      "loss": 1.3414,
      "step": 1324
    },
    {
      "epoch": 6.20245757753072,
      "grad_norm": 0.47016122937202454,
      "learning_rate": 3.755439472436801e-05,
      "loss": 1.3665,
      "step": 1325
    },
    {
      "epoch": 6.207138677589233,
      "grad_norm": 0.5091321468353271,
      "learning_rate": 3.747505011580936e-05,
      "loss": 1.3651,
      "step": 1326
    },
    {
      "epoch": 6.211819777647747,
      "grad_norm": 0.4931752383708954,
      "learning_rate": 3.739573914534776e-05,
      "loss": 1.3484,
      "step": 1327
    },
    {
      "epoch": 6.216500877706261,
      "grad_norm": 0.49978458881378174,
      "learning_rate": 3.7316462025987666e-05,
      "loss": 1.3557,
      "step": 1328
    },
    {
      "epoch": 6.2211819777647746,
      "grad_norm": 0.5076282024383545,
      "learning_rate": 3.7237218970642626e-05,
      "loss": 1.368,
      "step": 1329
    },
    {
      "epoch": 6.225863077823289,
      "grad_norm": 0.5032605528831482,
      "learning_rate": 3.715801019213467e-05,
      "loss": 1.3485,
      "step": 1330
    },
    {
      "epoch": 6.230544177881802,
      "grad_norm": 0.515126645565033,
      "learning_rate": 3.7078835903193824e-05,
      "loss": 1.3915,
      "step": 1331
    },
    {
      "epoch": 6.235225277940316,
      "grad_norm": 0.5021912455558777,
      "learning_rate": 3.6999696316457465e-05,
      "loss": 1.366,
      "step": 1332
    },
    {
      "epoch": 6.23990637799883,
      "grad_norm": 0.4850209057331085,
      "learning_rate": 3.6920591644469736e-05,
      "loss": 1.3302,
      "step": 1333
    },
    {
      "epoch": 6.244587478057343,
      "grad_norm": 0.4773012697696686,
      "learning_rate": 3.684152209968107e-05,
      "loss": 1.3373,
      "step": 1334
    },
    {
      "epoch": 6.249268578115857,
      "grad_norm": 0.486314058303833,
      "learning_rate": 3.67624878944475e-05,
      "loss": 1.3466,
      "step": 1335
    },
    {
      "epoch": 6.253949678174371,
      "grad_norm": 0.500917375087738,
      "learning_rate": 3.6683489241030176e-05,
      "loss": 1.3574,
      "step": 1336
    },
    {
      "epoch": 6.2586307782328845,
      "grad_norm": 0.4872746765613556,
      "learning_rate": 3.66045263515948e-05,
      "loss": 1.3672,
      "step": 1337
    },
    {
      "epoch": 6.263311878291399,
      "grad_norm": 0.48777884244918823,
      "learning_rate": 3.6525599438210956e-05,
      "loss": 1.3494,
      "step": 1338
    },
    {
      "epoch": 6.267992978349913,
      "grad_norm": 0.48457857966423035,
      "learning_rate": 3.6446708712851666e-05,
      "loss": 1.3252,
      "step": 1339
    },
    {
      "epoch": 6.272674078408426,
      "grad_norm": 0.49213770031929016,
      "learning_rate": 3.6367854387392716e-05,
      "loss": 1.3592,
      "step": 1340
    },
    {
      "epoch": 6.27735517846694,
      "grad_norm": 0.4870537221431732,
      "learning_rate": 3.628903667361216e-05,
      "loss": 1.3518,
      "step": 1341
    },
    {
      "epoch": 6.282036278525453,
      "grad_norm": 0.48960310220718384,
      "learning_rate": 3.621025578318975e-05,
      "loss": 1.3404,
      "step": 1342
    },
    {
      "epoch": 6.286717378583967,
      "grad_norm": 0.48424863815307617,
      "learning_rate": 3.6131511927706264e-05,
      "loss": 1.3386,
      "step": 1343
    },
    {
      "epoch": 6.291398478642481,
      "grad_norm": 0.5022472739219666,
      "learning_rate": 3.605280531864311e-05,
      "loss": 1.3462,
      "step": 1344
    },
    {
      "epoch": 6.296079578700994,
      "grad_norm": 0.483844131231308,
      "learning_rate": 3.597413616738157e-05,
      "loss": 1.3318,
      "step": 1345
    },
    {
      "epoch": 6.3007606787595085,
      "grad_norm": 0.48008736968040466,
      "learning_rate": 3.589550468520242e-05,
      "loss": 1.3731,
      "step": 1346
    },
    {
      "epoch": 6.305441778818023,
      "grad_norm": 0.48773452639579773,
      "learning_rate": 3.581691108328517e-05,
      "loss": 1.3305,
      "step": 1347
    },
    {
      "epoch": 6.310122878876536,
      "grad_norm": 0.4856274724006653,
      "learning_rate": 3.573835557270768e-05,
      "loss": 1.3369,
      "step": 1348
    },
    {
      "epoch": 6.31480397893505,
      "grad_norm": 0.5108739137649536,
      "learning_rate": 3.5659838364445505e-05,
      "loss": 1.3618,
      "step": 1349
    },
    {
      "epoch": 6.319485078993564,
      "grad_norm": 0.47577667236328125,
      "learning_rate": 3.558135966937123e-05,
      "loss": 1.3557,
      "step": 1350
    },
    {
      "epoch": 6.324166179052077,
      "grad_norm": 0.49220508337020874,
      "learning_rate": 3.550291969825415e-05,
      "loss": 1.3505,
      "step": 1351
    },
    {
      "epoch": 6.328847279110591,
      "grad_norm": 0.47751569747924805,
      "learning_rate": 3.542451866175944e-05,
      "loss": 1.3495,
      "step": 1352
    },
    {
      "epoch": 6.333528379169104,
      "grad_norm": 0.47069454193115234,
      "learning_rate": 3.5346156770447755e-05,
      "loss": 1.3436,
      "step": 1353
    },
    {
      "epoch": 6.3382094792276185,
      "grad_norm": 0.49512651562690735,
      "learning_rate": 3.526783423477468e-05,
      "loss": 1.3386,
      "step": 1354
    },
    {
      "epoch": 6.3428905792861325,
      "grad_norm": 0.48573729395866394,
      "learning_rate": 3.518955126508997e-05,
      "loss": 1.3515,
      "step": 1355
    },
    {
      "epoch": 6.347571679344646,
      "grad_norm": 0.478813111782074,
      "learning_rate": 3.511130807163724e-05,
      "loss": 1.3693,
      "step": 1356
    },
    {
      "epoch": 6.35225277940316,
      "grad_norm": 0.48378655314445496,
      "learning_rate": 3.503310486455319e-05,
      "loss": 1.3762,
      "step": 1357
    },
    {
      "epoch": 6.356933879461674,
      "grad_norm": 0.4784218966960907,
      "learning_rate": 3.495494185386721e-05,
      "loss": 1.3584,
      "step": 1358
    },
    {
      "epoch": 6.361614979520187,
      "grad_norm": 0.4655744135379791,
      "learning_rate": 3.487681924950067e-05,
      "loss": 1.3628,
      "step": 1359
    },
    {
      "epoch": 6.366296079578701,
      "grad_norm": 0.4663156270980835,
      "learning_rate": 3.4798737261266456e-05,
      "loss": 1.363,
      "step": 1360
    },
    {
      "epoch": 6.370977179637215,
      "grad_norm": 0.48525822162628174,
      "learning_rate": 3.472069609886835e-05,
      "loss": 1.3482,
      "step": 1361
    },
    {
      "epoch": 6.375658279695728,
      "grad_norm": 0.4797033667564392,
      "learning_rate": 3.464269597190051e-05,
      "loss": 1.3476,
      "step": 1362
    },
    {
      "epoch": 6.3803393797542425,
      "grad_norm": 0.48956650495529175,
      "learning_rate": 3.4564737089846864e-05,
      "loss": 1.3509,
      "step": 1363
    },
    {
      "epoch": 6.385020479812756,
      "grad_norm": 0.48538145422935486,
      "learning_rate": 3.4486819662080595e-05,
      "loss": 1.3626,
      "step": 1364
    },
    {
      "epoch": 6.38970157987127,
      "grad_norm": 0.4890855550765991,
      "learning_rate": 3.440894389786352e-05,
      "loss": 1.3428,
      "step": 1365
    },
    {
      "epoch": 6.394382679929784,
      "grad_norm": 0.4814460873603821,
      "learning_rate": 3.43311100063456e-05,
      "loss": 1.3474,
      "step": 1366
    },
    {
      "epoch": 6.399063779988297,
      "grad_norm": 0.48000338673591614,
      "learning_rate": 3.42533181965643e-05,
      "loss": 1.3375,
      "step": 1367
    },
    {
      "epoch": 6.403744880046811,
      "grad_norm": 0.46856164932250977,
      "learning_rate": 3.4175568677444094e-05,
      "loss": 1.3479,
      "step": 1368
    },
    {
      "epoch": 6.408425980105325,
      "grad_norm": 0.4865092933177948,
      "learning_rate": 3.4097861657795864e-05,
      "loss": 1.3561,
      "step": 1369
    },
    {
      "epoch": 6.413107080163838,
      "grad_norm": 0.4856918752193451,
      "learning_rate": 3.4020197346316354e-05,
      "loss": 1.3588,
      "step": 1370
    },
    {
      "epoch": 6.417788180222352,
      "grad_norm": 0.4821678400039673,
      "learning_rate": 3.394257595158761e-05,
      "loss": 1.3443,
      "step": 1371
    },
    {
      "epoch": 6.422469280280866,
      "grad_norm": 0.4700441360473633,
      "learning_rate": 3.386499768207642e-05,
      "loss": 1.3538,
      "step": 1372
    },
    {
      "epoch": 6.42715038033938,
      "grad_norm": 0.48163434863090515,
      "learning_rate": 3.378746274613372e-05,
      "loss": 1.3491,
      "step": 1373
    },
    {
      "epoch": 6.431831480397894,
      "grad_norm": 0.47138711810112,
      "learning_rate": 3.370997135199413e-05,
      "loss": 1.3282,
      "step": 1374
    },
    {
      "epoch": 6.436512580456407,
      "grad_norm": 0.4872971475124359,
      "learning_rate": 3.3632523707775266e-05,
      "loss": 1.3651,
      "step": 1375
    },
    {
      "epoch": 6.441193680514921,
      "grad_norm": 0.4702310860157013,
      "learning_rate": 3.355512002147729e-05,
      "loss": 1.3585,
      "step": 1376
    },
    {
      "epoch": 6.445874780573435,
      "grad_norm": 0.47843244671821594,
      "learning_rate": 3.347776050098226e-05,
      "loss": 1.3806,
      "step": 1377
    },
    {
      "epoch": 6.450555880631948,
      "grad_norm": 0.47986504435539246,
      "learning_rate": 3.340044535405367e-05,
      "loss": 1.3363,
      "step": 1378
    },
    {
      "epoch": 6.455236980690462,
      "grad_norm": 0.4765663743019104,
      "learning_rate": 3.332317478833582e-05,
      "loss": 1.3649,
      "step": 1379
    },
    {
      "epoch": 6.4599180807489756,
      "grad_norm": 0.47521644830703735,
      "learning_rate": 3.3245949011353264e-05,
      "loss": 1.341,
      "step": 1380
    },
    {
      "epoch": 6.46459918080749,
      "grad_norm": 0.48216161131858826,
      "learning_rate": 3.3168768230510285e-05,
      "loss": 1.3613,
      "step": 1381
    },
    {
      "epoch": 6.469280280866004,
      "grad_norm": 0.47645995020866394,
      "learning_rate": 3.30916326530903e-05,
      "loss": 1.3461,
      "step": 1382
    },
    {
      "epoch": 6.473961380924517,
      "grad_norm": 0.4928266406059265,
      "learning_rate": 3.301454248625536e-05,
      "loss": 1.3645,
      "step": 1383
    },
    {
      "epoch": 6.478642480983031,
      "grad_norm": 0.48274198174476624,
      "learning_rate": 3.293749793704554e-05,
      "loss": 1.3732,
      "step": 1384
    },
    {
      "epoch": 6.483323581041545,
      "grad_norm": 0.48078644275665283,
      "learning_rate": 3.286049921237836e-05,
      "loss": 1.3342,
      "step": 1385
    },
    {
      "epoch": 6.488004681100058,
      "grad_norm": 0.4662472903728485,
      "learning_rate": 3.278354651904836e-05,
      "loss": 1.3518,
      "step": 1386
    },
    {
      "epoch": 6.492685781158572,
      "grad_norm": 0.4929116368293762,
      "learning_rate": 3.270664006372633e-05,
      "loss": 1.3618,
      "step": 1387
    },
    {
      "epoch": 6.497366881217086,
      "grad_norm": 0.4652104079723358,
      "learning_rate": 3.262978005295901e-05,
      "loss": 1.3246,
      "step": 1388
    },
    {
      "epoch": 6.5020479812756,
      "grad_norm": 0.4658075273036957,
      "learning_rate": 3.255296669316834e-05,
      "loss": 1.3454,
      "step": 1389
    },
    {
      "epoch": 6.506729081334114,
      "grad_norm": 0.4817008674144745,
      "learning_rate": 3.247620019065092e-05,
      "loss": 1.3584,
      "step": 1390
    },
    {
      "epoch": 6.511410181392627,
      "grad_norm": 0.4935586154460907,
      "learning_rate": 3.239948075157764e-05,
      "loss": 1.3621,
      "step": 1391
    },
    {
      "epoch": 6.516091281451141,
      "grad_norm": 0.4779127240180969,
      "learning_rate": 3.2322808581992825e-05,
      "loss": 1.3482,
      "step": 1392
    },
    {
      "epoch": 6.520772381509655,
      "grad_norm": 0.48167383670806885,
      "learning_rate": 3.2246183887813995e-05,
      "loss": 1.3475,
      "step": 1393
    },
    {
      "epoch": 6.525453481568168,
      "grad_norm": 0.4878900647163391,
      "learning_rate": 3.216960687483112e-05,
      "loss": 1.345,
      "step": 1394
    },
    {
      "epoch": 6.530134581626682,
      "grad_norm": 0.4895574748516083,
      "learning_rate": 3.209307774870603e-05,
      "loss": 1.3432,
      "step": 1395
    },
    {
      "epoch": 6.534815681685196,
      "grad_norm": 0.4896577298641205,
      "learning_rate": 3.2016596714972105e-05,
      "loss": 1.3488,
      "step": 1396
    },
    {
      "epoch": 6.5394967817437095,
      "grad_norm": 0.4758218824863434,
      "learning_rate": 3.19401639790334e-05,
      "loss": 1.3542,
      "step": 1397
    },
    {
      "epoch": 6.544177881802224,
      "grad_norm": 0.48330479860305786,
      "learning_rate": 3.1863779746164355e-05,
      "loss": 1.3713,
      "step": 1398
    },
    {
      "epoch": 6.548858981860738,
      "grad_norm": 0.4900497794151306,
      "learning_rate": 3.178744422150915e-05,
      "loss": 1.3462,
      "step": 1399
    },
    {
      "epoch": 6.553540081919251,
      "grad_norm": 0.4635152518749237,
      "learning_rate": 3.1711157610081096e-05,
      "loss": 1.3494,
      "step": 1400
    },
    {
      "epoch": 6.558221181977765,
      "grad_norm": 0.481869101524353,
      "learning_rate": 3.1634920116762176e-05,
      "loss": 1.362,
      "step": 1401
    },
    {
      "epoch": 6.562902282036278,
      "grad_norm": 0.49324488639831543,
      "learning_rate": 3.1558731946302435e-05,
      "loss": 1.3497,
      "step": 1402
    },
    {
      "epoch": 6.567583382094792,
      "grad_norm": 0.48508989810943604,
      "learning_rate": 3.148259330331945e-05,
      "loss": 1.3583,
      "step": 1403
    },
    {
      "epoch": 6.572264482153306,
      "grad_norm": 0.48708096146583557,
      "learning_rate": 3.1406504392297864e-05,
      "loss": 1.3572,
      "step": 1404
    },
    {
      "epoch": 6.5769455822118195,
      "grad_norm": 0.5007489919662476,
      "learning_rate": 3.133046541758862e-05,
      "loss": 1.353,
      "step": 1405
    },
    {
      "epoch": 6.5816266822703335,
      "grad_norm": 0.46692708134651184,
      "learning_rate": 3.1254476583408645e-05,
      "loss": 1.3486,
      "step": 1406
    },
    {
      "epoch": 6.586307782328848,
      "grad_norm": 0.45835116505622864,
      "learning_rate": 3.117853809384016e-05,
      "loss": 1.324,
      "step": 1407
    },
    {
      "epoch": 6.590988882387361,
      "grad_norm": 0.49309778213500977,
      "learning_rate": 3.11026501528302e-05,
      "loss": 1.3584,
      "step": 1408
    },
    {
      "epoch": 6.595669982445875,
      "grad_norm": 0.48946860432624817,
      "learning_rate": 3.1026812964190044e-05,
      "loss": 1.3495,
      "step": 1409
    },
    {
      "epoch": 6.600351082504389,
      "grad_norm": 0.501943051815033,
      "learning_rate": 3.0951026731594635e-05,
      "loss": 1.3578,
      "step": 1410
    },
    {
      "epoch": 6.605032182562902,
      "grad_norm": 0.48318126797676086,
      "learning_rate": 3.0875291658582104e-05,
      "loss": 1.3469,
      "step": 1411
    },
    {
      "epoch": 6.609713282621416,
      "grad_norm": 0.4796597957611084,
      "learning_rate": 3.079960794855315e-05,
      "loss": 1.3554,
      "step": 1412
    },
    {
      "epoch": 6.614394382679929,
      "grad_norm": 0.48338282108306885,
      "learning_rate": 3.072397580477054e-05,
      "loss": 1.3621,
      "step": 1413
    },
    {
      "epoch": 6.6190754827384435,
      "grad_norm": 0.47976669669151306,
      "learning_rate": 3.0648395430358564e-05,
      "loss": 1.3513,
      "step": 1414
    },
    {
      "epoch": 6.623756582796958,
      "grad_norm": 0.4931601881980896,
      "learning_rate": 3.057286702830246e-05,
      "loss": 1.3551,
      "step": 1415
    },
    {
      "epoch": 6.628437682855471,
      "grad_norm": 0.48094189167022705,
      "learning_rate": 3.0497390801447883e-05,
      "loss": 1.3491,
      "step": 1416
    },
    {
      "epoch": 6.633118782913985,
      "grad_norm": 0.48256900906562805,
      "learning_rate": 3.0421966952500358e-05,
      "loss": 1.3491,
      "step": 1417
    },
    {
      "epoch": 6.637799882972499,
      "grad_norm": 0.5100786685943604,
      "learning_rate": 3.0346595684024755e-05,
      "loss": 1.3603,
      "step": 1418
    },
    {
      "epoch": 6.642480983031012,
      "grad_norm": 0.492291122674942,
      "learning_rate": 3.0271277198444735e-05,
      "loss": 1.3553,
      "step": 1419
    },
    {
      "epoch": 6.647162083089526,
      "grad_norm": 0.47800853848457336,
      "learning_rate": 3.019601169804216e-05,
      "loss": 1.3443,
      "step": 1420
    },
    {
      "epoch": 6.65184318314804,
      "grad_norm": 0.4766044020652771,
      "learning_rate": 3.0120799384956642e-05,
      "loss": 1.3411,
      "step": 1421
    },
    {
      "epoch": 6.656524283206553,
      "grad_norm": 0.46603795886039734,
      "learning_rate": 3.0045640461184915e-05,
      "loss": 1.337,
      "step": 1422
    },
    {
      "epoch": 6.6612053832650675,
      "grad_norm": 0.5020335912704468,
      "learning_rate": 2.9970535128580336e-05,
      "loss": 1.3384,
      "step": 1423
    },
    {
      "epoch": 6.665886483323581,
      "grad_norm": 0.4919477701187134,
      "learning_rate": 2.9895483588852337e-05,
      "loss": 1.347,
      "step": 1424
    },
    {
      "epoch": 6.670567583382095,
      "grad_norm": 0.48846378922462463,
      "learning_rate": 2.9820486043565854e-05,
      "loss": 1.3286,
      "step": 1425
    },
    {
      "epoch": 6.675248683440609,
      "grad_norm": 0.4913786053657532,
      "learning_rate": 2.974554269414086e-05,
      "loss": 1.356,
      "step": 1426
    },
    {
      "epoch": 6.679929783499122,
      "grad_norm": 0.47736066579818726,
      "learning_rate": 2.9670653741851728e-05,
      "loss": 1.3642,
      "step": 1427
    },
    {
      "epoch": 6.684610883557636,
      "grad_norm": 0.4830758273601532,
      "learning_rate": 2.959581938782675e-05,
      "loss": 1.345,
      "step": 1428
    },
    {
      "epoch": 6.68929198361615,
      "grad_norm": 0.4828513562679291,
      "learning_rate": 2.9521039833047614e-05,
      "loss": 1.3532,
      "step": 1429
    },
    {
      "epoch": 6.693973083674663,
      "grad_norm": 0.4844878017902374,
      "learning_rate": 2.944631527834879e-05,
      "loss": 1.3532,
      "step": 1430
    },
    {
      "epoch": 6.6986541837331774,
      "grad_norm": 0.4727339744567871,
      "learning_rate": 2.937164592441708e-05,
      "loss": 1.3393,
      "step": 1431
    },
    {
      "epoch": 6.7033352837916915,
      "grad_norm": 0.4912458062171936,
      "learning_rate": 2.9297031971790955e-05,
      "loss": 1.3598,
      "step": 1432
    },
    {
      "epoch": 6.708016383850205,
      "grad_norm": 0.48164111375808716,
      "learning_rate": 2.9222473620860198e-05,
      "loss": 1.3473,
      "step": 1433
    },
    {
      "epoch": 6.712697483908719,
      "grad_norm": 0.48862725496292114,
      "learning_rate": 2.9147971071865216e-05,
      "loss": 1.3608,
      "step": 1434
    },
    {
      "epoch": 6.717378583967232,
      "grad_norm": 0.49266523122787476,
      "learning_rate": 2.9073524524896513e-05,
      "loss": 1.3581,
      "step": 1435
    },
    {
      "epoch": 6.722059684025746,
      "grad_norm": 0.4824602007865906,
      "learning_rate": 2.8999134179894244e-05,
      "loss": 1.3519,
      "step": 1436
    },
    {
      "epoch": 6.72674078408426,
      "grad_norm": 0.48006853461265564,
      "learning_rate": 2.8924800236647597e-05,
      "loss": 1.3383,
      "step": 1437
    },
    {
      "epoch": 6.731421884142773,
      "grad_norm": 0.48835617303848267,
      "learning_rate": 2.8850522894794307e-05,
      "loss": 1.3685,
      "step": 1438
    },
    {
      "epoch": 6.736102984201287,
      "grad_norm": 0.49173474311828613,
      "learning_rate": 2.877630235382007e-05,
      "loss": 1.3541,
      "step": 1439
    },
    {
      "epoch": 6.740784084259801,
      "grad_norm": 0.4922891855239868,
      "learning_rate": 2.870213881305802e-05,
      "loss": 1.3433,
      "step": 1440
    },
    {
      "epoch": 6.745465184318315,
      "grad_norm": 0.4828687012195587,
      "learning_rate": 2.862803247168825e-05,
      "loss": 1.3671,
      "step": 1441
    },
    {
      "epoch": 6.750146284376829,
      "grad_norm": 0.4786439836025238,
      "learning_rate": 2.855398352873718e-05,
      "loss": 1.337,
      "step": 1442
    },
    {
      "epoch": 6.754827384435342,
      "grad_norm": 0.5012813806533813,
      "learning_rate": 2.8479992183077142e-05,
      "loss": 1.3564,
      "step": 1443
    },
    {
      "epoch": 6.759508484493856,
      "grad_norm": 0.49249881505966187,
      "learning_rate": 2.8406058633425737e-05,
      "loss": 1.3373,
      "step": 1444
    },
    {
      "epoch": 6.76418958455237,
      "grad_norm": 0.47396138310432434,
      "learning_rate": 2.833218307834532e-05,
      "loss": 1.3354,
      "step": 1445
    },
    {
      "epoch": 6.768870684610883,
      "grad_norm": 0.47091707587242126,
      "learning_rate": 2.825836571624254e-05,
      "loss": 1.3607,
      "step": 1446
    },
    {
      "epoch": 6.773551784669397,
      "grad_norm": 0.47440093755722046,
      "learning_rate": 2.818460674536773e-05,
      "loss": 1.3427,
      "step": 1447
    },
    {
      "epoch": 6.778232884727911,
      "grad_norm": 0.4885600507259369,
      "learning_rate": 2.8110906363814426e-05,
      "loss": 1.3415,
      "step": 1448
    },
    {
      "epoch": 6.782913984786425,
      "grad_norm": 0.4825218915939331,
      "learning_rate": 2.8037264769518802e-05,
      "loss": 1.3577,
      "step": 1449
    },
    {
      "epoch": 6.787595084844939,
      "grad_norm": 0.5013914704322815,
      "learning_rate": 2.7963682160259107e-05,
      "loss": 1.3404,
      "step": 1450
    },
    {
      "epoch": 6.792276184903452,
      "grad_norm": 0.4755021035671234,
      "learning_rate": 2.7890158733655237e-05,
      "loss": 1.3505,
      "step": 1451
    },
    {
      "epoch": 6.796957284961966,
      "grad_norm": 0.48797109723091125,
      "learning_rate": 2.781669468716811e-05,
      "loss": 1.328,
      "step": 1452
    },
    {
      "epoch": 6.80163838502048,
      "grad_norm": 0.48389190435409546,
      "learning_rate": 2.774329021809916e-05,
      "loss": 1.3503,
      "step": 1453
    },
    {
      "epoch": 6.806319485078993,
      "grad_norm": 0.48728588223457336,
      "learning_rate": 2.766994552358987e-05,
      "loss": 1.3429,
      "step": 1454
    },
    {
      "epoch": 6.811000585137507,
      "grad_norm": 0.48784175515174866,
      "learning_rate": 2.7596660800621078e-05,
      "loss": 1.3491,
      "step": 1455
    },
    {
      "epoch": 6.815681685196021,
      "grad_norm": 0.48684337735176086,
      "learning_rate": 2.7523436246012686e-05,
      "loss": 1.3521,
      "step": 1456
    },
    {
      "epoch": 6.8203627852545345,
      "grad_norm": 0.4770592749118805,
      "learning_rate": 2.7450272056422864e-05,
      "loss": 1.357,
      "step": 1457
    },
    {
      "epoch": 6.825043885313049,
      "grad_norm": 0.47391945123672485,
      "learning_rate": 2.737716842834781e-05,
      "loss": 1.3396,
      "step": 1458
    },
    {
      "epoch": 6.829724985371563,
      "grad_norm": 0.5123258829116821,
      "learning_rate": 2.730412555812099e-05,
      "loss": 1.3551,
      "step": 1459
    },
    {
      "epoch": 6.834406085430076,
      "grad_norm": 0.47993412613868713,
      "learning_rate": 2.7231143641912672e-05,
      "loss": 1.3518,
      "step": 1460
    },
    {
      "epoch": 6.83908718548859,
      "grad_norm": 0.4489990472793579,
      "learning_rate": 2.7158222875729472e-05,
      "loss": 1.3236,
      "step": 1461
    },
    {
      "epoch": 6.843768285547103,
      "grad_norm": 0.4816996455192566,
      "learning_rate": 2.708536345541371e-05,
      "loss": 1.335,
      "step": 1462
    },
    {
      "epoch": 6.848449385605617,
      "grad_norm": 0.48490139842033386,
      "learning_rate": 2.7012565576643056e-05,
      "loss": 1.3511,
      "step": 1463
    },
    {
      "epoch": 6.853130485664131,
      "grad_norm": 0.47340846061706543,
      "learning_rate": 2.693982943492983e-05,
      "loss": 1.3428,
      "step": 1464
    },
    {
      "epoch": 6.8578115857226445,
      "grad_norm": 0.4921325743198395,
      "learning_rate": 2.6867155225620523e-05,
      "loss": 1.3538,
      "step": 1465
    },
    {
      "epoch": 6.862492685781159,
      "grad_norm": 0.4751735329627991,
      "learning_rate": 2.6794543143895356e-05,
      "loss": 1.3534,
      "step": 1466
    },
    {
      "epoch": 6.867173785839673,
      "grad_norm": 0.4749985635280609,
      "learning_rate": 2.672199338476763e-05,
      "loss": 1.3496,
      "step": 1467
    },
    {
      "epoch": 6.871854885898186,
      "grad_norm": 0.4956899881362915,
      "learning_rate": 2.6649506143083302e-05,
      "loss": 1.355,
      "step": 1468
    },
    {
      "epoch": 6.8765359859567,
      "grad_norm": 0.48751288652420044,
      "learning_rate": 2.6577081613520487e-05,
      "loss": 1.3493,
      "step": 1469
    },
    {
      "epoch": 6.881217086015214,
      "grad_norm": 0.4699368476867676,
      "learning_rate": 2.650471999058875e-05,
      "loss": 1.3395,
      "step": 1470
    },
    {
      "epoch": 6.885898186073727,
      "grad_norm": 0.484819233417511,
      "learning_rate": 2.643242146862881e-05,
      "loss": 1.3556,
      "step": 1471
    },
    {
      "epoch": 6.890579286132241,
      "grad_norm": 0.48918503522872925,
      "learning_rate": 2.6360186241811847e-05,
      "loss": 1.3558,
      "step": 1472
    },
    {
      "epoch": 6.895260386190754,
      "grad_norm": 0.4830945134162903,
      "learning_rate": 2.6288014504139104e-05,
      "loss": 1.3523,
      "step": 1473
    },
    {
      "epoch": 6.8999414862492685,
      "grad_norm": 0.46876898407936096,
      "learning_rate": 2.6215906449441273e-05,
      "loss": 1.3617,
      "step": 1474
    },
    {
      "epoch": 6.904622586307783,
      "grad_norm": 0.49341315031051636,
      "learning_rate": 2.6143862271378052e-05,
      "loss": 1.3386,
      "step": 1475
    },
    {
      "epoch": 6.909303686366296,
      "grad_norm": 0.4862746298313141,
      "learning_rate": 2.607188216343756e-05,
      "loss": 1.3538,
      "step": 1476
    },
    {
      "epoch": 6.91398478642481,
      "grad_norm": 0.4845553934574127,
      "learning_rate": 2.5999966318935832e-05,
      "loss": 1.3513,
      "step": 1477
    },
    {
      "epoch": 6.918665886483324,
      "grad_norm": 0.472523033618927,
      "learning_rate": 2.5928114931016333e-05,
      "loss": 1.3533,
      "step": 1478
    },
    {
      "epoch": 6.923346986541837,
      "grad_norm": 0.4786662757396698,
      "learning_rate": 2.5856328192649426e-05,
      "loss": 1.3594,
      "step": 1479
    },
    {
      "epoch": 6.928028086600351,
      "grad_norm": 0.4927366077899933,
      "learning_rate": 2.578460629663182e-05,
      "loss": 1.3549,
      "step": 1480
    },
    {
      "epoch": 6.932709186658865,
      "grad_norm": 0.4627937972545624,
      "learning_rate": 2.571294943558612e-05,
      "loss": 1.3319,
      "step": 1481
    },
    {
      "epoch": 6.9373902867173785,
      "grad_norm": 0.4778825044631958,
      "learning_rate": 2.5641357801960187e-05,
      "loss": 1.3407,
      "step": 1482
    },
    {
      "epoch": 6.9420713867758925,
      "grad_norm": 0.4876656234264374,
      "learning_rate": 2.556983158802677e-05,
      "loss": 1.3611,
      "step": 1483
    },
    {
      "epoch": 6.946752486834406,
      "grad_norm": 0.5010808110237122,
      "learning_rate": 2.5498370985882926e-05,
      "loss": 1.3416,
      "step": 1484
    },
    {
      "epoch": 6.95143358689292,
      "grad_norm": 0.47430136799812317,
      "learning_rate": 2.542697618744945e-05,
      "loss": 1.3295,
      "step": 1485
    },
    {
      "epoch": 6.956114686951434,
      "grad_norm": 0.49215957522392273,
      "learning_rate": 2.535564738447047e-05,
      "loss": 1.3425,
      "step": 1486
    },
    {
      "epoch": 6.960795787009947,
      "grad_norm": 0.4640463590621948,
      "learning_rate": 2.528438476851279e-05,
      "loss": 1.3567,
      "step": 1487
    },
    {
      "epoch": 6.965476887068461,
      "grad_norm": 0.4978911578655243,
      "learning_rate": 2.5213188530965525e-05,
      "loss": 1.3512,
      "step": 1488
    },
    {
      "epoch": 6.970157987126975,
      "grad_norm": 0.4663708806037903,
      "learning_rate": 2.5142058863039498e-05,
      "loss": 1.3495,
      "step": 1489
    },
    {
      "epoch": 6.974839087185488,
      "grad_norm": 0.48394128680229187,
      "learning_rate": 2.507099595576674e-05,
      "loss": 1.3511,
      "step": 1490
    },
    {
      "epoch": 6.9795201872440025,
      "grad_norm": 0.49097976088523865,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.33,
      "step": 1491
    },
    {
      "epoch": 6.9842012873025165,
      "grad_norm": 0.47492849826812744,
      "learning_rate": 2.4929071186412194e-05,
      "loss": 1.3337,
      "step": 1492
    },
    {
      "epoch": 6.98888238736103,
      "grad_norm": 0.48000726103782654,
      "learning_rate": 2.485820970549592e-05,
      "loss": 1.3532,
      "step": 1493
    },
    {
      "epoch": 6.993563487419544,
      "grad_norm": 0.4719773232936859,
      "learning_rate": 2.4787415747562958e-05,
      "loss": 1.3438,
      "step": 1494
    },
    {
      "epoch": 6.998244587478057,
      "grad_norm": 0.4728144407272339,
      "learning_rate": 2.4716689502743724e-05,
      "loss": 1.3398,
      "step": 1495
    },
    {
      "epoch": 7.002925687536571,
      "grad_norm": 1.8404959440231323,
      "learning_rate": 2.4646031160986803e-05,
      "loss": 1.2867,
      "step": 1496
    },
    {
      "epoch": 7.007606787595085,
      "grad_norm": 0.4895849823951721,
      "learning_rate": 2.4575440912058374e-05,
      "loss": 1.3655,
      "step": 1497
    },
    {
      "epoch": 7.012287887653598,
      "grad_norm": 0.5031997561454773,
      "learning_rate": 2.450491894554176e-05,
      "loss": 1.3445,
      "step": 1498
    },
    {
      "epoch": 7.016968987712112,
      "grad_norm": 0.4948151111602783,
      "learning_rate": 2.4434465450836917e-05,
      "loss": 1.3692,
      "step": 1499
    },
    {
      "epoch": 7.0216500877706265,
      "grad_norm": 0.47969523072242737,
      "learning_rate": 2.4364080617159886e-05,
      "loss": 1.3499,
      "step": 1500
    },
    {
      "epoch": 7.02633118782914,
      "grad_norm": 0.4878917932510376,
      "learning_rate": 2.429376463354232e-05,
      "loss": 1.3335,
      "step": 1501
    },
    {
      "epoch": 7.031012287887654,
      "grad_norm": 0.4990362524986267,
      "learning_rate": 2.4223517688830933e-05,
      "loss": 1.3581,
      "step": 1502
    },
    {
      "epoch": 7.035693387946167,
      "grad_norm": 0.4753541648387909,
      "learning_rate": 2.415333997168705e-05,
      "loss": 1.3447,
      "step": 1503
    },
    {
      "epoch": 7.040374488004681,
      "grad_norm": 0.4919619858264923,
      "learning_rate": 2.4083231670586082e-05,
      "loss": 1.3306,
      "step": 1504
    },
    {
      "epoch": 7.045055588063195,
      "grad_norm": 0.46794456243515015,
      "learning_rate": 2.4013192973816944e-05,
      "loss": 1.3356,
      "step": 1505
    },
    {
      "epoch": 7.049736688121708,
      "grad_norm": 0.46644270420074463,
      "learning_rate": 2.3943224069481734e-05,
      "loss": 1.3382,
      "step": 1506
    },
    {
      "epoch": 7.054417788180222,
      "grad_norm": 0.4738415479660034,
      "learning_rate": 2.387332514549499e-05,
      "loss": 1.3437,
      "step": 1507
    },
    {
      "epoch": 7.059098888238736,
      "grad_norm": 0.4804893136024475,
      "learning_rate": 2.3803496389583364e-05,
      "loss": 1.364,
      "step": 1508
    },
    {
      "epoch": 7.06377998829725,
      "grad_norm": 0.4757007360458374,
      "learning_rate": 2.373373798928507e-05,
      "loss": 1.3311,
      "step": 1509
    },
    {
      "epoch": 7.068461088355764,
      "grad_norm": 0.4703696370124817,
      "learning_rate": 2.36640501319493e-05,
      "loss": 1.3387,
      "step": 1510
    },
    {
      "epoch": 7.073142188414278,
      "grad_norm": 0.4751465916633606,
      "learning_rate": 2.3594433004735906e-05,
      "loss": 1.3535,
      "step": 1511
    },
    {
      "epoch": 7.077823288472791,
      "grad_norm": 0.4735172986984253,
      "learning_rate": 2.3524886794614653e-05,
      "loss": 1.3424,
      "step": 1512
    },
    {
      "epoch": 7.082504388531305,
      "grad_norm": 0.48852401971817017,
      "learning_rate": 2.3455411688364925e-05,
      "loss": 1.339,
      "step": 1513
    },
    {
      "epoch": 7.087185488589818,
      "grad_norm": 0.47311729192733765,
      "learning_rate": 2.3386007872575143e-05,
      "loss": 1.342,
      "step": 1514
    },
    {
      "epoch": 7.091866588648332,
      "grad_norm": 0.466234028339386,
      "learning_rate": 2.3316675533642214e-05,
      "loss": 1.3322,
      "step": 1515
    },
    {
      "epoch": 7.096547688706846,
      "grad_norm": 0.46330374479293823,
      "learning_rate": 2.3247414857771115e-05,
      "loss": 1.3402,
      "step": 1516
    },
    {
      "epoch": 7.10122878876536,
      "grad_norm": 0.48585692048072815,
      "learning_rate": 2.317822603097436e-05,
      "loss": 1.3419,
      "step": 1517
    },
    {
      "epoch": 7.105909888823874,
      "grad_norm": 0.4767219126224518,
      "learning_rate": 2.3109109239071492e-05,
      "loss": 1.355,
      "step": 1518
    },
    {
      "epoch": 7.110590988882388,
      "grad_norm": 0.4704568088054657,
      "learning_rate": 2.304006466768861e-05,
      "loss": 1.3285,
      "step": 1519
    },
    {
      "epoch": 7.115272088940901,
      "grad_norm": 0.4667569100856781,
      "learning_rate": 2.29710925022578e-05,
      "loss": 1.3553,
      "step": 1520
    },
    {
      "epoch": 7.119953188999415,
      "grad_norm": 0.4830743074417114,
      "learning_rate": 2.2902192928016734e-05,
      "loss": 1.3438,
      "step": 1521
    },
    {
      "epoch": 7.124634289057928,
      "grad_norm": 0.4988841712474823,
      "learning_rate": 2.2833366130008117e-05,
      "loss": 1.3273,
      "step": 1522
    },
    {
      "epoch": 7.129315389116442,
      "grad_norm": 0.4610385596752167,
      "learning_rate": 2.276461229307921e-05,
      "loss": 1.3582,
      "step": 1523
    },
    {
      "epoch": 7.133996489174956,
      "grad_norm": 0.4775535762310028,
      "learning_rate": 2.2695931601881276e-05,
      "loss": 1.3579,
      "step": 1524
    },
    {
      "epoch": 7.1386775892334695,
      "grad_norm": 0.4786890745162964,
      "learning_rate": 2.262732424086918e-05,
      "loss": 1.3439,
      "step": 1525
    },
    {
      "epoch": 7.143358689291984,
      "grad_norm": 0.45839619636535645,
      "learning_rate": 2.2558790394300838e-05,
      "loss": 1.3358,
      "step": 1526
    },
    {
      "epoch": 7.148039789350498,
      "grad_norm": 0.47785043716430664,
      "learning_rate": 2.249033024623672e-05,
      "loss": 1.3269,
      "step": 1527
    },
    {
      "epoch": 7.152720889409011,
      "grad_norm": 0.4758484363555908,
      "learning_rate": 2.2421943980539384e-05,
      "loss": 1.3318,
      "step": 1528
    },
    {
      "epoch": 7.157401989467525,
      "grad_norm": 0.48065611720085144,
      "learning_rate": 2.23536317808729e-05,
      "loss": 1.3469,
      "step": 1529
    },
    {
      "epoch": 7.162083089526039,
      "grad_norm": 0.45679306983947754,
      "learning_rate": 2.22853938307025e-05,
      "loss": 1.3517,
      "step": 1530
    },
    {
      "epoch": 7.166764189584552,
      "grad_norm": 0.4631989002227783,
      "learning_rate": 2.2217230313293953e-05,
      "loss": 1.3512,
      "step": 1531
    },
    {
      "epoch": 7.171445289643066,
      "grad_norm": 0.47275522351264954,
      "learning_rate": 2.2149141411713154e-05,
      "loss": 1.3471,
      "step": 1532
    },
    {
      "epoch": 7.1761263897015795,
      "grad_norm": 0.48173314332962036,
      "learning_rate": 2.2081127308825612e-05,
      "loss": 1.3281,
      "step": 1533
    },
    {
      "epoch": 7.1808074897600935,
      "grad_norm": 0.47281208634376526,
      "learning_rate": 2.201318818729588e-05,
      "loss": 1.3564,
      "step": 1534
    },
    {
      "epoch": 7.185488589818608,
      "grad_norm": 0.45339658856391907,
      "learning_rate": 2.1945324229587217e-05,
      "loss": 1.3358,
      "step": 1535
    },
    {
      "epoch": 7.190169689877121,
      "grad_norm": 0.4726531505584717,
      "learning_rate": 2.187753561796097e-05,
      "loss": 1.3333,
      "step": 1536
    },
    {
      "epoch": 7.194850789935635,
      "grad_norm": 0.46895188093185425,
      "learning_rate": 2.180982253447615e-05,
      "loss": 1.3366,
      "step": 1537
    },
    {
      "epoch": 7.199531889994149,
      "grad_norm": 0.4762912094593048,
      "learning_rate": 2.174218516098893e-05,
      "loss": 1.3633,
      "step": 1538
    },
    {
      "epoch": 7.204212990052662,
      "grad_norm": 0.48002806305885315,
      "learning_rate": 2.1674623679152105e-05,
      "loss": 1.3578,
      "step": 1539
    },
    {
      "epoch": 7.208894090111176,
      "grad_norm": 0.47279906272888184,
      "learning_rate": 2.1607138270414695e-05,
      "loss": 1.3504,
      "step": 1540
    },
    {
      "epoch": 7.21357519016969,
      "grad_norm": 0.45677801966667175,
      "learning_rate": 2.153972911602139e-05,
      "loss": 1.3378,
      "step": 1541
    },
    {
      "epoch": 7.2182562902282035,
      "grad_norm": 0.4792294502258301,
      "learning_rate": 2.14723963970121e-05,
      "loss": 1.3568,
      "step": 1542
    },
    {
      "epoch": 7.2229373902867176,
      "grad_norm": 0.49012744426727295,
      "learning_rate": 2.1405140294221453e-05,
      "loss": 1.3102,
      "step": 1543
    },
    {
      "epoch": 7.227618490345231,
      "grad_norm": 0.47090527415275574,
      "learning_rate": 2.1337960988278265e-05,
      "loss": 1.3445,
      "step": 1544
    },
    {
      "epoch": 7.232299590403745,
      "grad_norm": 0.46279802918434143,
      "learning_rate": 2.1270858659605158e-05,
      "loss": 1.3388,
      "step": 1545
    },
    {
      "epoch": 7.236980690462259,
      "grad_norm": 0.4582454562187195,
      "learning_rate": 2.1203833488418014e-05,
      "loss": 1.3283,
      "step": 1546
    },
    {
      "epoch": 7.241661790520772,
      "grad_norm": 0.46114638447761536,
      "learning_rate": 2.1136885654725412e-05,
      "loss": 1.3387,
      "step": 1547
    },
    {
      "epoch": 7.246342890579286,
      "grad_norm": 0.47302740812301636,
      "learning_rate": 2.107001533832837e-05,
      "loss": 1.3528,
      "step": 1548
    },
    {
      "epoch": 7.2510239906378,
      "grad_norm": 0.45719102025032043,
      "learning_rate": 2.1003222718819575e-05,
      "loss": 1.3551,
      "step": 1549
    },
    {
      "epoch": 7.255705090696313,
      "grad_norm": 0.4898391366004944,
      "learning_rate": 2.093650797558313e-05,
      "loss": 1.342,
      "step": 1550
    },
    {
      "epoch": 7.2603861907548275,
      "grad_norm": 0.4674202799797058,
      "learning_rate": 2.0869871287793985e-05,
      "loss": 1.3365,
      "step": 1551
    },
    {
      "epoch": 7.265067290813342,
      "grad_norm": 0.45634809136390686,
      "learning_rate": 2.080331283441738e-05,
      "loss": 1.3345,
      "step": 1552
    },
    {
      "epoch": 7.269748390871855,
      "grad_norm": 0.4628802239894867,
      "learning_rate": 2.0736832794208573e-05,
      "loss": 1.3492,
      "step": 1553
    },
    {
      "epoch": 7.274429490930369,
      "grad_norm": 0.45460382103919983,
      "learning_rate": 2.0670431345712092e-05,
      "loss": 1.3191,
      "step": 1554
    },
    {
      "epoch": 7.279110590988882,
      "grad_norm": 0.4698864817619324,
      "learning_rate": 2.0604108667261483e-05,
      "loss": 1.3391,
      "step": 1555
    },
    {
      "epoch": 7.283791691047396,
      "grad_norm": 0.4649961292743683,
      "learning_rate": 2.0537864936978722e-05,
      "loss": 1.3407,
      "step": 1556
    },
    {
      "epoch": 7.28847279110591,
      "grad_norm": 0.4733186960220337,
      "learning_rate": 2.0471700332773715e-05,
      "loss": 1.3436,
      "step": 1557
    },
    {
      "epoch": 7.293153891164423,
      "grad_norm": 0.4652688503265381,
      "learning_rate": 2.04056150323439e-05,
      "loss": 1.3421,
      "step": 1558
    },
    {
      "epoch": 7.297834991222937,
      "grad_norm": 0.47101065516471863,
      "learning_rate": 2.0339609213173726e-05,
      "loss": 1.3424,
      "step": 1559
    },
    {
      "epoch": 7.3025160912814515,
      "grad_norm": 0.47479957342147827,
      "learning_rate": 2.0273683052534175e-05,
      "loss": 1.3319,
      "step": 1560
    },
    {
      "epoch": 7.307197191339965,
      "grad_norm": 0.4701590836048126,
      "learning_rate": 2.02078367274823e-05,
      "loss": 1.3557,
      "step": 1561
    },
    {
      "epoch": 7.311878291398479,
      "grad_norm": 0.46992096304893494,
      "learning_rate": 2.0142070414860704e-05,
      "loss": 1.3161,
      "step": 1562
    },
    {
      "epoch": 7.316559391456993,
      "grad_norm": 0.4736483693122864,
      "learning_rate": 2.0076384291297134e-05,
      "loss": 1.3502,
      "step": 1563
    },
    {
      "epoch": 7.321240491515506,
      "grad_norm": 0.478202760219574,
      "learning_rate": 2.0010778533203982e-05,
      "loss": 1.3343,
      "step": 1564
    },
    {
      "epoch": 7.32592159157402,
      "grad_norm": 0.48485836386680603,
      "learning_rate": 1.9945253316777785e-05,
      "loss": 1.3362,
      "step": 1565
    },
    {
      "epoch": 7.330602691632533,
      "grad_norm": 0.47226443886756897,
      "learning_rate": 1.987980881799879e-05,
      "loss": 1.3349,
      "step": 1566
    },
    {
      "epoch": 7.335283791691047,
      "grad_norm": 0.47480282187461853,
      "learning_rate": 1.981444521263041e-05,
      "loss": 1.3419,
      "step": 1567
    },
    {
      "epoch": 7.3399648917495615,
      "grad_norm": 0.4802916944026947,
      "learning_rate": 1.974916267621887e-05,
      "loss": 1.343,
      "step": 1568
    },
    {
      "epoch": 7.344645991808075,
      "grad_norm": 0.4808948338031769,
      "learning_rate": 1.9683961384092626e-05,
      "loss": 1.3161,
      "step": 1569
    },
    {
      "epoch": 7.349327091866589,
      "grad_norm": 0.47402530908584595,
      "learning_rate": 1.9618841511361957e-05,
      "loss": 1.3419,
      "step": 1570
    },
    {
      "epoch": 7.354008191925103,
      "grad_norm": 0.46208834648132324,
      "learning_rate": 1.9553803232918482e-05,
      "loss": 1.333,
      "step": 1571
    },
    {
      "epoch": 7.358689291983616,
      "grad_norm": 0.4791894257068634,
      "learning_rate": 1.9488846723434646e-05,
      "loss": 1.3353,
      "step": 1572
    },
    {
      "epoch": 7.36337039204213,
      "grad_norm": 0.4807760417461395,
      "learning_rate": 1.9423972157363318e-05,
      "loss": 1.3629,
      "step": 1573
    },
    {
      "epoch": 7.368051492100643,
      "grad_norm": 0.49156269431114197,
      "learning_rate": 1.935917970893729e-05,
      "loss": 1.3448,
      "step": 1574
    },
    {
      "epoch": 7.372732592159157,
      "grad_norm": 0.46572548151016235,
      "learning_rate": 1.9294469552168813e-05,
      "loss": 1.3447,
      "step": 1575
    },
    {
      "epoch": 7.377413692217671,
      "grad_norm": 0.4842897057533264,
      "learning_rate": 1.9229841860849134e-05,
      "loss": 1.326,
      "step": 1576
    },
    {
      "epoch": 7.382094792276185,
      "grad_norm": 0.46771901845932007,
      "learning_rate": 1.916529680854799e-05,
      "loss": 1.3459,
      "step": 1577
    },
    {
      "epoch": 7.386775892334699,
      "grad_norm": 0.4856923520565033,
      "learning_rate": 1.910083456861321e-05,
      "loss": 1.3448,
      "step": 1578
    },
    {
      "epoch": 7.391456992393213,
      "grad_norm": 0.4657568335533142,
      "learning_rate": 1.90364553141702e-05,
      "loss": 1.3459,
      "step": 1579
    },
    {
      "epoch": 7.396138092451726,
      "grad_norm": 0.4637135863304138,
      "learning_rate": 1.89721592181215e-05,
      "loss": 1.3061,
      "step": 1580
    },
    {
      "epoch": 7.40081919251024,
      "grad_norm": 0.4781709909439087,
      "learning_rate": 1.890794645314633e-05,
      "loss": 1.3539,
      "step": 1581
    },
    {
      "epoch": 7.405500292568753,
      "grad_norm": 0.460147500038147,
      "learning_rate": 1.8843817191700047e-05,
      "loss": 1.3321,
      "step": 1582
    },
    {
      "epoch": 7.410181392627267,
      "grad_norm": 0.4599703252315521,
      "learning_rate": 1.8779771606013808e-05,
      "loss": 1.3311,
      "step": 1583
    },
    {
      "epoch": 7.414862492685781,
      "grad_norm": 0.46007290482521057,
      "learning_rate": 1.871580986809402e-05,
      "loss": 1.339,
      "step": 1584
    },
    {
      "epoch": 7.4195435927442945,
      "grad_norm": 0.4541688561439514,
      "learning_rate": 1.8651932149721897e-05,
      "loss": 1.3403,
      "step": 1585
    },
    {
      "epoch": 7.424224692802809,
      "grad_norm": 0.4677172899246216,
      "learning_rate": 1.8588138622453027e-05,
      "loss": 1.3507,
      "step": 1586
    },
    {
      "epoch": 7.428905792861323,
      "grad_norm": 0.472500741481781,
      "learning_rate": 1.8524429457616832e-05,
      "loss": 1.3553,
      "step": 1587
    },
    {
      "epoch": 7.433586892919836,
      "grad_norm": 0.4556261897087097,
      "learning_rate": 1.846080482631623e-05,
      "loss": 1.3387,
      "step": 1588
    },
    {
      "epoch": 7.43826799297835,
      "grad_norm": 0.4568939208984375,
      "learning_rate": 1.839726489942703e-05,
      "loss": 1.3512,
      "step": 1589
    },
    {
      "epoch": 7.442949093036864,
      "grad_norm": 0.46659862995147705,
      "learning_rate": 1.8333809847597642e-05,
      "loss": 1.3337,
      "step": 1590
    },
    {
      "epoch": 7.447630193095377,
      "grad_norm": 0.4842297434806824,
      "learning_rate": 1.8270439841248492e-05,
      "loss": 1.3537,
      "step": 1591
    },
    {
      "epoch": 7.452311293153891,
      "grad_norm": 0.4623686373233795,
      "learning_rate": 1.820715505057155e-05,
      "loss": 1.3189,
      "step": 1592
    },
    {
      "epoch": 7.4569923932124045,
      "grad_norm": 0.45595666766166687,
      "learning_rate": 1.8143955645530007e-05,
      "loss": 1.3282,
      "step": 1593
    },
    {
      "epoch": 7.4616734932709186,
      "grad_norm": 0.4729546904563904,
      "learning_rate": 1.8080841795857633e-05,
      "loss": 1.3528,
      "step": 1594
    },
    {
      "epoch": 7.466354593329433,
      "grad_norm": 0.47479796409606934,
      "learning_rate": 1.8017813671058547e-05,
      "loss": 1.3553,
      "step": 1595
    },
    {
      "epoch": 7.471035693387946,
      "grad_norm": 0.48458370566368103,
      "learning_rate": 1.7954871440406562e-05,
      "loss": 1.3396,
      "step": 1596
    },
    {
      "epoch": 7.47571679344646,
      "grad_norm": 0.4696284830570221,
      "learning_rate": 1.789201527294479e-05,
      "loss": 1.3267,
      "step": 1597
    },
    {
      "epoch": 7.480397893504974,
      "grad_norm": 0.45866280794143677,
      "learning_rate": 1.7829245337485267e-05,
      "loss": 1.3247,
      "step": 1598
    },
    {
      "epoch": 7.485078993563487,
      "grad_norm": 0.46521177887916565,
      "learning_rate": 1.7766561802608373e-05,
      "loss": 1.3272,
      "step": 1599
    },
    {
      "epoch": 7.489760093622001,
      "grad_norm": 0.4628099799156189,
      "learning_rate": 1.7703964836662462e-05,
      "loss": 1.3311,
      "step": 1600
    },
    {
      "epoch": 7.494441193680515,
      "grad_norm": 0.47476711869239807,
      "learning_rate": 1.7641454607763464e-05,
      "loss": 1.3509,
      "step": 1601
    },
    {
      "epoch": 7.4991222937390285,
      "grad_norm": 0.45366212725639343,
      "learning_rate": 1.7579031283794236e-05,
      "loss": 1.3437,
      "step": 1602
    },
    {
      "epoch": 7.503803393797543,
      "grad_norm": 0.45894238352775574,
      "learning_rate": 1.751669503240434e-05,
      "loss": 1.3451,
      "step": 1603
    },
    {
      "epoch": 7.508484493856056,
      "grad_norm": 0.49631863832473755,
      "learning_rate": 1.7454446021009408e-05,
      "loss": 1.3531,
      "step": 1604
    },
    {
      "epoch": 7.51316559391457,
      "grad_norm": 0.4589841067790985,
      "learning_rate": 1.739228441679081e-05,
      "loss": 1.3305,
      "step": 1605
    },
    {
      "epoch": 7.517846693973084,
      "grad_norm": 0.4836769104003906,
      "learning_rate": 1.7330210386695216e-05,
      "loss": 1.3564,
      "step": 1606
    },
    {
      "epoch": 7.522527794031597,
      "grad_norm": 0.46482783555984497,
      "learning_rate": 1.726822409743401e-05,
      "loss": 1.3357,
      "step": 1607
    },
    {
      "epoch": 7.527208894090111,
      "grad_norm": 0.4549084007740021,
      "learning_rate": 1.7206325715483e-05,
      "loss": 1.3272,
      "step": 1608
    },
    {
      "epoch": 7.531889994148625,
      "grad_norm": 0.45632073283195496,
      "learning_rate": 1.7144515407081845e-05,
      "loss": 1.3506,
      "step": 1609
    },
    {
      "epoch": 7.536571094207138,
      "grad_norm": 0.4481995105743408,
      "learning_rate": 1.7082793338233716e-05,
      "loss": 1.3287,
      "step": 1610
    },
    {
      "epoch": 7.5412521942656525,
      "grad_norm": 0.46668148040771484,
      "learning_rate": 1.702115967470477e-05,
      "loss": 1.3391,
      "step": 1611
    },
    {
      "epoch": 7.545933294324167,
      "grad_norm": 0.4636210501194,
      "learning_rate": 1.6959614582023752e-05,
      "loss": 1.3302,
      "step": 1612
    },
    {
      "epoch": 7.55061439438268,
      "grad_norm": 0.4625440239906311,
      "learning_rate": 1.6898158225481548e-05,
      "loss": 1.3625,
      "step": 1613
    },
    {
      "epoch": 7.555295494441194,
      "grad_norm": 0.46369972825050354,
      "learning_rate": 1.683679077013066e-05,
      "loss": 1.3406,
      "step": 1614
    },
    {
      "epoch": 7.559976594499707,
      "grad_norm": 0.46081992983818054,
      "learning_rate": 1.6775512380784903e-05,
      "loss": 1.3516,
      "step": 1615
    },
    {
      "epoch": 7.564657694558221,
      "grad_norm": 0.4579129219055176,
      "learning_rate": 1.6714323222018858e-05,
      "loss": 1.335,
      "step": 1616
    },
    {
      "epoch": 7.569338794616735,
      "grad_norm": 0.47029703855514526,
      "learning_rate": 1.665322345816746e-05,
      "loss": 1.345,
      "step": 1617
    },
    {
      "epoch": 7.574019894675248,
      "grad_norm": 0.45141252875328064,
      "learning_rate": 1.6592213253325578e-05,
      "loss": 1.3341,
      "step": 1618
    },
    {
      "epoch": 7.5787009947337625,
      "grad_norm": 0.44940832257270813,
      "learning_rate": 1.65312927713475e-05,
      "loss": 1.3287,
      "step": 1619
    },
    {
      "epoch": 7.5833820947922765,
      "grad_norm": 0.45708319544792175,
      "learning_rate": 1.647046217584661e-05,
      "loss": 1.3354,
      "step": 1620
    },
    {
      "epoch": 7.58806319485079,
      "grad_norm": 0.4675891399383545,
      "learning_rate": 1.640972163019484e-05,
      "loss": 1.3383,
      "step": 1621
    },
    {
      "epoch": 7.592744294909304,
      "grad_norm": 0.4550844132900238,
      "learning_rate": 1.63490712975223e-05,
      "loss": 1.3535,
      "step": 1622
    },
    {
      "epoch": 7.597425394967818,
      "grad_norm": 0.46031132340431213,
      "learning_rate": 1.6288511340716833e-05,
      "loss": 1.3197,
      "step": 1623
    },
    {
      "epoch": 7.602106495026331,
      "grad_norm": 0.46827811002731323,
      "learning_rate": 1.6228041922423498e-05,
      "loss": 1.3385,
      "step": 1624
    },
    {
      "epoch": 7.606787595084845,
      "grad_norm": 0.4582699239253998,
      "learning_rate": 1.6167663205044247e-05,
      "loss": 1.341,
      "step": 1625
    },
    {
      "epoch": 7.611468695143358,
      "grad_norm": 0.46605154871940613,
      "learning_rate": 1.6107375350737437e-05,
      "loss": 1.342,
      "step": 1626
    },
    {
      "epoch": 7.616149795201872,
      "grad_norm": 0.46738141775131226,
      "learning_rate": 1.6047178521417376e-05,
      "loss": 1.3368,
      "step": 1627
    },
    {
      "epoch": 7.6208308952603865,
      "grad_norm": 0.47278207540512085,
      "learning_rate": 1.5987072878753933e-05,
      "loss": 1.3454,
      "step": 1628
    },
    {
      "epoch": 7.6255119953189,
      "grad_norm": 0.46338996291160583,
      "learning_rate": 1.5927058584172034e-05,
      "loss": 1.3399,
      "step": 1629
    },
    {
      "epoch": 7.630193095377414,
      "grad_norm": 0.4683593511581421,
      "learning_rate": 1.5867135798851306e-05,
      "loss": 1.341,
      "step": 1630
    },
    {
      "epoch": 7.634874195435928,
      "grad_norm": 0.46051323413848877,
      "learning_rate": 1.580730468372561e-05,
      "loss": 1.3181,
      "step": 1631
    },
    {
      "epoch": 7.639555295494441,
      "grad_norm": 0.4732109010219574,
      "learning_rate": 1.5747565399482604e-05,
      "loss": 1.3449,
      "step": 1632
    },
    {
      "epoch": 7.644236395552955,
      "grad_norm": 0.4626827836036682,
      "learning_rate": 1.5687918106563326e-05,
      "loss": 1.3269,
      "step": 1633
    },
    {
      "epoch": 7.648917495611469,
      "grad_norm": 0.4615435004234314,
      "learning_rate": 1.5628362965161724e-05,
      "loss": 1.3449,
      "step": 1634
    },
    {
      "epoch": 7.653598595669982,
      "grad_norm": 0.46386954188346863,
      "learning_rate": 1.556890013522428e-05,
      "loss": 1.3274,
      "step": 1635
    },
    {
      "epoch": 7.658279695728496,
      "grad_norm": 0.4549250304698944,
      "learning_rate": 1.5509529776449577e-05,
      "loss": 1.3291,
      "step": 1636
    },
    {
      "epoch": 7.66296079578701,
      "grad_norm": 0.4708329737186432,
      "learning_rate": 1.545025204828777e-05,
      "loss": 1.3441,
      "step": 1637
    },
    {
      "epoch": 7.667641895845524,
      "grad_norm": 0.4606434404850006,
      "learning_rate": 1.5391067109940366e-05,
      "loss": 1.3346,
      "step": 1638
    },
    {
      "epoch": 7.672322995904038,
      "grad_norm": 0.4673614501953125,
      "learning_rate": 1.5331975120359536e-05,
      "loss": 1.3247,
      "step": 1639
    },
    {
      "epoch": 7.677004095962551,
      "grad_norm": 0.45525097846984863,
      "learning_rate": 1.5272976238247903e-05,
      "loss": 1.3584,
      "step": 1640
    },
    {
      "epoch": 7.681685196021065,
      "grad_norm": 0.4707421660423279,
      "learning_rate": 1.5214070622058013e-05,
      "loss": 1.3507,
      "step": 1641
    },
    {
      "epoch": 7.686366296079578,
      "grad_norm": 0.46960994601249695,
      "learning_rate": 1.5155258429991882e-05,
      "loss": 1.342,
      "step": 1642
    },
    {
      "epoch": 7.691047396138092,
      "grad_norm": 0.46895959973335266,
      "learning_rate": 1.5096539820000721e-05,
      "loss": 1.3518,
      "step": 1643
    },
    {
      "epoch": 7.695728496196606,
      "grad_norm": 0.47289028763771057,
      "learning_rate": 1.5037914949784299e-05,
      "loss": 1.3365,
      "step": 1644
    },
    {
      "epoch": 7.7004095962551204,
      "grad_norm": 0.46617165207862854,
      "learning_rate": 1.4979383976790695e-05,
      "loss": 1.34,
      "step": 1645
    },
    {
      "epoch": 7.705090696313634,
      "grad_norm": 0.45962589979171753,
      "learning_rate": 1.49209470582158e-05,
      "loss": 1.3341,
      "step": 1646
    },
    {
      "epoch": 7.709771796372148,
      "grad_norm": 0.4578133523464203,
      "learning_rate": 1.4862604351002867e-05,
      "loss": 1.3283,
      "step": 1647
    },
    {
      "epoch": 7.714452896430661,
      "grad_norm": 0.4768761098384857,
      "learning_rate": 1.480435601184218e-05,
      "loss": 1.3328,
      "step": 1648
    },
    {
      "epoch": 7.719133996489175,
      "grad_norm": 0.45739585161209106,
      "learning_rate": 1.4746202197170544e-05,
      "loss": 1.319,
      "step": 1649
    },
    {
      "epoch": 7.723815096547689,
      "grad_norm": 0.4618678092956543,
      "learning_rate": 1.4688143063170923e-05,
      "loss": 1.3188,
      "step": 1650
    },
    {
      "epoch": 7.728496196606202,
      "grad_norm": 0.4686064124107361,
      "learning_rate": 1.463017876577199e-05,
      "loss": 1.3439,
      "step": 1651
    },
    {
      "epoch": 7.733177296664716,
      "grad_norm": 0.47376856207847595,
      "learning_rate": 1.4572309460647693e-05,
      "loss": 1.3228,
      "step": 1652
    },
    {
      "epoch": 7.7378583967232295,
      "grad_norm": 0.4711912274360657,
      "learning_rate": 1.4514535303216892e-05,
      "loss": 1.3317,
      "step": 1653
    },
    {
      "epoch": 7.742539496781744,
      "grad_norm": 0.45707976818084717,
      "learning_rate": 1.4456856448642903e-05,
      "loss": 1.3257,
      "step": 1654
    },
    {
      "epoch": 7.747220596840258,
      "grad_norm": 0.4613284170627594,
      "learning_rate": 1.4399273051833084e-05,
      "loss": 1.3464,
      "step": 1655
    },
    {
      "epoch": 7.751901696898771,
      "grad_norm": 0.4727132022380829,
      "learning_rate": 1.434178526743844e-05,
      "loss": 1.3484,
      "step": 1656
    },
    {
      "epoch": 7.756582796957285,
      "grad_norm": 0.4638311564922333,
      "learning_rate": 1.4284393249853145e-05,
      "loss": 1.3378,
      "step": 1657
    },
    {
      "epoch": 7.761263897015799,
      "grad_norm": 0.4656262695789337,
      "learning_rate": 1.422709715321422e-05,
      "loss": 1.3348,
      "step": 1658
    },
    {
      "epoch": 7.765944997074312,
      "grad_norm": 0.47884249687194824,
      "learning_rate": 1.416989713140105e-05,
      "loss": 1.3641,
      "step": 1659
    },
    {
      "epoch": 7.770626097132826,
      "grad_norm": 0.4640868604183197,
      "learning_rate": 1.4112793338035002e-05,
      "loss": 1.2996,
      "step": 1660
    },
    {
      "epoch": 7.77530719719134,
      "grad_norm": 0.4635297656059265,
      "learning_rate": 1.4055785926479021e-05,
      "loss": 1.3456,
      "step": 1661
    },
    {
      "epoch": 7.7799882972498535,
      "grad_norm": 0.4484289586544037,
      "learning_rate": 1.399887504983714e-05,
      "loss": 1.333,
      "step": 1662
    },
    {
      "epoch": 7.784669397308368,
      "grad_norm": 0.4604371190071106,
      "learning_rate": 1.3942060860954192e-05,
      "loss": 1.3154,
      "step": 1663
    },
    {
      "epoch": 7.789350497366881,
      "grad_norm": 0.45666757225990295,
      "learning_rate": 1.388534351241531e-05,
      "loss": 1.3377,
      "step": 1664
    },
    {
      "epoch": 7.794031597425395,
      "grad_norm": 0.4564230442047119,
      "learning_rate": 1.3828723156545553e-05,
      "loss": 1.3194,
      "step": 1665
    },
    {
      "epoch": 7.798712697483909,
      "grad_norm": 0.4641699492931366,
      "learning_rate": 1.3772199945409497e-05,
      "loss": 1.348,
      "step": 1666
    },
    {
      "epoch": 7.803393797542422,
      "grad_norm": 0.4646790027618408,
      "learning_rate": 1.3715774030810773e-05,
      "loss": 1.3318,
      "step": 1667
    },
    {
      "epoch": 7.808074897600936,
      "grad_norm": 0.46432945132255554,
      "learning_rate": 1.3659445564291757e-05,
      "loss": 1.3358,
      "step": 1668
    },
    {
      "epoch": 7.81275599765945,
      "grad_norm": 0.46811240911483765,
      "learning_rate": 1.3603214697133071e-05,
      "loss": 1.3521,
      "step": 1669
    },
    {
      "epoch": 7.8174370977179635,
      "grad_norm": 0.4617379307746887,
      "learning_rate": 1.3547081580353249e-05,
      "loss": 1.3268,
      "step": 1670
    },
    {
      "epoch": 7.8221181977764775,
      "grad_norm": 0.45402026176452637,
      "learning_rate": 1.3491046364708293e-05,
      "loss": 1.3137,
      "step": 1671
    },
    {
      "epoch": 7.826799297834992,
      "grad_norm": 0.45435112714767456,
      "learning_rate": 1.3435109200691232e-05,
      "loss": 1.3199,
      "step": 1672
    },
    {
      "epoch": 7.831480397893505,
      "grad_norm": 0.4637871980667114,
      "learning_rate": 1.3379270238531798e-05,
      "loss": 1.3306,
      "step": 1673
    },
    {
      "epoch": 7.836161497952019,
      "grad_norm": 0.4626833498477936,
      "learning_rate": 1.3323529628195985e-05,
      "loss": 1.3166,
      "step": 1674
    },
    {
      "epoch": 7.840842598010532,
      "grad_norm": 0.46381741762161255,
      "learning_rate": 1.3267887519385624e-05,
      "loss": 1.3437,
      "step": 1675
    },
    {
      "epoch": 7.845523698069046,
      "grad_norm": 0.4669346511363983,
      "learning_rate": 1.3212344061538034e-05,
      "loss": 1.3285,
      "step": 1676
    },
    {
      "epoch": 7.85020479812756,
      "grad_norm": 0.45426100492477417,
      "learning_rate": 1.3156899403825534e-05,
      "loss": 1.3491,
      "step": 1677
    },
    {
      "epoch": 7.854885898186073,
      "grad_norm": 0.476627916097641,
      "learning_rate": 1.3101553695155156e-05,
      "loss": 1.3265,
      "step": 1678
    },
    {
      "epoch": 7.8595669982445875,
      "grad_norm": 0.4632202386856079,
      "learning_rate": 1.3046307084168119e-05,
      "loss": 1.3334,
      "step": 1679
    },
    {
      "epoch": 7.864248098303102,
      "grad_norm": 0.4600007236003876,
      "learning_rate": 1.2991159719239582e-05,
      "loss": 1.3446,
      "step": 1680
    },
    {
      "epoch": 7.868929198361615,
      "grad_norm": 0.4647494852542877,
      "learning_rate": 1.2936111748478114e-05,
      "loss": 1.3327,
      "step": 1681
    },
    {
      "epoch": 7.873610298420129,
      "grad_norm": 0.4659038484096527,
      "learning_rate": 1.2881163319725303e-05,
      "loss": 1.3418,
      "step": 1682
    },
    {
      "epoch": 7.878291398478643,
      "grad_norm": 0.4638218879699707,
      "learning_rate": 1.2826314580555487e-05,
      "loss": 1.3293,
      "step": 1683
    },
    {
      "epoch": 7.882972498537156,
      "grad_norm": 0.467208594083786,
      "learning_rate": 1.277156567827516e-05,
      "loss": 1.3503,
      "step": 1684
    },
    {
      "epoch": 7.88765359859567,
      "grad_norm": 0.45110321044921875,
      "learning_rate": 1.2716916759922797e-05,
      "loss": 1.3304,
      "step": 1685
    },
    {
      "epoch": 7.892334698654183,
      "grad_norm": 0.45999690890312195,
      "learning_rate": 1.2662367972268297e-05,
      "loss": 1.3333,
      "step": 1686
    },
    {
      "epoch": 7.897015798712697,
      "grad_norm": 0.46465185284614563,
      "learning_rate": 1.2607919461812606e-05,
      "loss": 1.3489,
      "step": 1687
    },
    {
      "epoch": 7.9016968987712115,
      "grad_norm": 0.47059544920921326,
      "learning_rate": 1.255357137478742e-05,
      "loss": 1.3456,
      "step": 1688
    },
    {
      "epoch": 7.906377998829725,
      "grad_norm": 0.4738016724586487,
      "learning_rate": 1.2499323857154671e-05,
      "loss": 1.3287,
      "step": 1689
    },
    {
      "epoch": 7.911059098888239,
      "grad_norm": 0.47977060079574585,
      "learning_rate": 1.2445177054606216e-05,
      "loss": 1.3477,
      "step": 1690
    },
    {
      "epoch": 7.915740198946753,
      "grad_norm": 0.47547101974487305,
      "learning_rate": 1.239113111256347e-05,
      "loss": 1.3249,
      "step": 1691
    },
    {
      "epoch": 7.920421299005266,
      "grad_norm": 0.45657679438591003,
      "learning_rate": 1.2337186176176891e-05,
      "loss": 1.3387,
      "step": 1692
    },
    {
      "epoch": 7.92510239906378,
      "grad_norm": 0.4775516390800476,
      "learning_rate": 1.2283342390325724e-05,
      "loss": 1.344,
      "step": 1693
    },
    {
      "epoch": 7.929783499122294,
      "grad_norm": 0.46691226959228516,
      "learning_rate": 1.22295998996175e-05,
      "loss": 1.3408,
      "step": 1694
    },
    {
      "epoch": 7.934464599180807,
      "grad_norm": 0.4656122624874115,
      "learning_rate": 1.2175958848387765e-05,
      "loss": 1.3272,
      "step": 1695
    },
    {
      "epoch": 7.9391456992393215,
      "grad_norm": 0.4623662531375885,
      "learning_rate": 1.2122419380699585e-05,
      "loss": 1.3214,
      "step": 1696
    },
    {
      "epoch": 7.943826799297835,
      "grad_norm": 0.46682363748550415,
      "learning_rate": 1.2068981640343229e-05,
      "loss": 1.3343,
      "step": 1697
    },
    {
      "epoch": 7.948507899356349,
      "grad_norm": 0.4599725902080536,
      "learning_rate": 1.2015645770835764e-05,
      "loss": 1.3421,
      "step": 1698
    },
    {
      "epoch": 7.953188999414863,
      "grad_norm": 0.46313902735710144,
      "learning_rate": 1.1962411915420618e-05,
      "loss": 1.322,
      "step": 1699
    },
    {
      "epoch": 7.957870099473376,
      "grad_norm": 0.44506651163101196,
      "learning_rate": 1.1909280217067292e-05,
      "loss": 1.328,
      "step": 1700
    },
    {
      "epoch": 7.96255119953189,
      "grad_norm": 0.4529295563697815,
      "learning_rate": 1.1856250818470905e-05,
      "loss": 1.317,
      "step": 1701
    },
    {
      "epoch": 7.967232299590404,
      "grad_norm": 0.4700178802013397,
      "learning_rate": 1.180332386205183e-05,
      "loss": 1.3333,
      "step": 1702
    },
    {
      "epoch": 7.971913399648917,
      "grad_norm": 0.4718801975250244,
      "learning_rate": 1.1750499489955325e-05,
      "loss": 1.3283,
      "step": 1703
    },
    {
      "epoch": 7.976594499707431,
      "grad_norm": 0.4550124406814575,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.3333,
      "step": 1704
    },
    {
      "epoch": 7.9812755997659455,
      "grad_norm": 0.4479660391807556,
      "learning_rate": 1.164515906593303e-05,
      "loss": 1.3239,
      "step": 1705
    },
    {
      "epoch": 7.985956699824459,
      "grad_norm": 0.45768308639526367,
      "learning_rate": 1.1592643296918676e-05,
      "loss": 1.336,
      "step": 1706
    },
    {
      "epoch": 7.990637799882973,
      "grad_norm": 0.463931679725647,
      "learning_rate": 1.1540230678048968e-05,
      "loss": 1.3118,
      "step": 1707
    },
    {
      "epoch": 7.995318899941486,
      "grad_norm": 0.4714531898498535,
      "learning_rate": 1.148792135008782e-05,
      "loss": 1.3385,
      "step": 1708
    },
    {
      "epoch": 8.0,
      "grad_norm": 1.5786850452423096,
      "learning_rate": 1.1435715453521695e-05,
      "loss": 1.2702,
      "step": 1709
    },
    {
      "epoch": 8.004681100058514,
      "grad_norm": 0.4669012129306793,
      "learning_rate": 1.1383613128559306e-05,
      "loss": 1.3242,
      "step": 1710
    },
    {
      "epoch": 8.009362200117028,
      "grad_norm": 0.47372370958328247,
      "learning_rate": 1.1331614515131207e-05,
      "loss": 1.3339,
      "step": 1711
    },
    {
      "epoch": 8.01404330017554,
      "grad_norm": 0.45693066716194153,
      "learning_rate": 1.1279719752889401e-05,
      "loss": 1.3129,
      "step": 1712
    },
    {
      "epoch": 8.018724400234055,
      "grad_norm": 0.47375747561454773,
      "learning_rate": 1.1227928981207003e-05,
      "loss": 1.3499,
      "step": 1713
    },
    {
      "epoch": 8.023405500292569,
      "grad_norm": 0.4803107976913452,
      "learning_rate": 1.1176242339177789e-05,
      "loss": 1.3435,
      "step": 1714
    },
    {
      "epoch": 8.028086600351083,
      "grad_norm": 0.4778917133808136,
      "learning_rate": 1.1124659965615925e-05,
      "loss": 1.3598,
      "step": 1715
    },
    {
      "epoch": 8.032767700409597,
      "grad_norm": 0.4473143219947815,
      "learning_rate": 1.1073181999055538e-05,
      "loss": 1.3386,
      "step": 1716
    },
    {
      "epoch": 8.03744880046811,
      "grad_norm": 0.46125826239585876,
      "learning_rate": 1.1021808577750332e-05,
      "loss": 1.3061,
      "step": 1717
    },
    {
      "epoch": 8.042129900526623,
      "grad_norm": 0.46183252334594727,
      "learning_rate": 1.0970539839673255e-05,
      "loss": 1.3459,
      "step": 1718
    },
    {
      "epoch": 8.046811000585137,
      "grad_norm": 0.4538736343383789,
      "learning_rate": 1.0919375922516084e-05,
      "loss": 1.327,
      "step": 1719
    },
    {
      "epoch": 8.051492100643651,
      "grad_norm": 0.4843650460243225,
      "learning_rate": 1.0868316963689113e-05,
      "loss": 1.3228,
      "step": 1720
    },
    {
      "epoch": 8.056173200702165,
      "grad_norm": 0.4713257849216461,
      "learning_rate": 1.0817363100320693e-05,
      "loss": 1.3251,
      "step": 1721
    },
    {
      "epoch": 8.06085430076068,
      "grad_norm": 0.46485134959220886,
      "learning_rate": 1.0766514469257006e-05,
      "loss": 1.3323,
      "step": 1722
    },
    {
      "epoch": 8.065535400819192,
      "grad_norm": 0.4672873914241791,
      "learning_rate": 1.0715771207061558e-05,
      "loss": 1.3446,
      "step": 1723
    },
    {
      "epoch": 8.070216500877706,
      "grad_norm": 0.46149742603302,
      "learning_rate": 1.0665133450014858e-05,
      "loss": 1.3405,
      "step": 1724
    },
    {
      "epoch": 8.07489760093622,
      "grad_norm": 0.4756418764591217,
      "learning_rate": 1.0614601334114099e-05,
      "loss": 1.3565,
      "step": 1725
    },
    {
      "epoch": 8.079578700994734,
      "grad_norm": 0.4548105299472809,
      "learning_rate": 1.05641749950727e-05,
      "loss": 1.3311,
      "step": 1726
    },
    {
      "epoch": 8.084259801053248,
      "grad_norm": 0.4672495126724243,
      "learning_rate": 1.0513854568320075e-05,
      "loss": 1.3208,
      "step": 1727
    },
    {
      "epoch": 8.088940901111762,
      "grad_norm": 0.4649643003940582,
      "learning_rate": 1.0463640189001145e-05,
      "loss": 1.3411,
      "step": 1728
    },
    {
      "epoch": 8.093622001170274,
      "grad_norm": 0.4581691324710846,
      "learning_rate": 1.0413531991975995e-05,
      "loss": 1.3272,
      "step": 1729
    },
    {
      "epoch": 8.098303101228788,
      "grad_norm": 0.4784697890281677,
      "learning_rate": 1.0363530111819591e-05,
      "loss": 1.3027,
      "step": 1730
    },
    {
      "epoch": 8.102984201287303,
      "grad_norm": 0.4707012474536896,
      "learning_rate": 1.0313634682821321e-05,
      "loss": 1.3103,
      "step": 1731
    },
    {
      "epoch": 8.107665301345817,
      "grad_norm": 0.4484752416610718,
      "learning_rate": 1.0263845838984692e-05,
      "loss": 1.346,
      "step": 1732
    },
    {
      "epoch": 8.11234640140433,
      "grad_norm": 0.458938330411911,
      "learning_rate": 1.021416371402701e-05,
      "loss": 1.3482,
      "step": 1733
    },
    {
      "epoch": 8.117027501462843,
      "grad_norm": 0.4532400071620941,
      "learning_rate": 1.016458844137887e-05,
      "loss": 1.3382,
      "step": 1734
    },
    {
      "epoch": 8.121708601521357,
      "grad_norm": 0.4518270194530487,
      "learning_rate": 1.0115120154183976e-05,
      "loss": 1.3185,
      "step": 1735
    },
    {
      "epoch": 8.126389701579871,
      "grad_norm": 0.45866602659225464,
      "learning_rate": 1.0065758985298651e-05,
      "loss": 1.355,
      "step": 1736
    },
    {
      "epoch": 8.131070801638385,
      "grad_norm": 0.45490261912345886,
      "learning_rate": 1.0016505067291543e-05,
      "loss": 1.3502,
      "step": 1737
    },
    {
      "epoch": 8.1357519016969,
      "grad_norm": 0.45964813232421875,
      "learning_rate": 9.967358532443305e-06,
      "loss": 1.3151,
      "step": 1738
    },
    {
      "epoch": 8.140433001755412,
      "grad_norm": 0.4557470679283142,
      "learning_rate": 9.918319512746128e-06,
      "loss": 1.308,
      "step": 1739
    },
    {
      "epoch": 8.145114101813926,
      "grad_norm": 0.45845234394073486,
      "learning_rate": 9.869388139903496e-06,
      "loss": 1.3277,
      "step": 1740
    },
    {
      "epoch": 8.14979520187244,
      "grad_norm": 0.4491667151451111,
      "learning_rate": 9.820564545329736e-06,
      "loss": 1.3318,
      "step": 1741
    },
    {
      "epoch": 8.154476301930954,
      "grad_norm": 0.4657641053199768,
      "learning_rate": 9.771848860149774e-06,
      "loss": 1.3297,
      "step": 1742
    },
    {
      "epoch": 8.159157401989468,
      "grad_norm": 0.4560374915599823,
      "learning_rate": 9.723241215198692e-06,
      "loss": 1.3404,
      "step": 1743
    },
    {
      "epoch": 8.163838502047982,
      "grad_norm": 0.4573279619216919,
      "learning_rate": 9.674741741021425e-06,
      "loss": 1.3341,
      "step": 1744
    },
    {
      "epoch": 8.168519602106494,
      "grad_norm": 0.4648374915122986,
      "learning_rate": 9.626350567872389e-06,
      "loss": 1.337,
      "step": 1745
    },
    {
      "epoch": 8.173200702165008,
      "grad_norm": 0.45301535725593567,
      "learning_rate": 9.578067825715126e-06,
      "loss": 1.3112,
      "step": 1746
    },
    {
      "epoch": 8.177881802223522,
      "grad_norm": 0.46872368454933167,
      "learning_rate": 9.529893644221976e-06,
      "loss": 1.3534,
      "step": 1747
    },
    {
      "epoch": 8.182562902282037,
      "grad_norm": 0.44515344500541687,
      "learning_rate": 9.481828152773725e-06,
      "loss": 1.3264,
      "step": 1748
    },
    {
      "epoch": 8.18724400234055,
      "grad_norm": 0.46138012409210205,
      "learning_rate": 9.433871480459245e-06,
      "loss": 1.3399,
      "step": 1749
    },
    {
      "epoch": 8.191925102399063,
      "grad_norm": 0.448217511177063,
      "learning_rate": 9.386023756075169e-06,
      "loss": 1.3195,
      "step": 1750
    },
    {
      "epoch": 8.196606202457577,
      "grad_norm": 0.4604373872280121,
      "learning_rate": 9.338285108125488e-06,
      "loss": 1.3468,
      "step": 1751
    },
    {
      "epoch": 8.201287302516091,
      "grad_norm": 0.45444729924201965,
      "learning_rate": 9.290655664821296e-06,
      "loss": 1.3239,
      "step": 1752
    },
    {
      "epoch": 8.205968402574605,
      "grad_norm": 0.46766898036003113,
      "learning_rate": 9.243135554080367e-06,
      "loss": 1.308,
      "step": 1753
    },
    {
      "epoch": 8.21064950263312,
      "grad_norm": 0.462932288646698,
      "learning_rate": 9.19572490352686e-06,
      "loss": 1.3439,
      "step": 1754
    },
    {
      "epoch": 8.215330602691633,
      "grad_norm": 0.4593763053417206,
      "learning_rate": 9.148423840490954e-06,
      "loss": 1.315,
      "step": 1755
    },
    {
      "epoch": 8.220011702750146,
      "grad_norm": 0.4483340084552765,
      "learning_rate": 9.101232492008487e-06,
      "loss": 1.3394,
      "step": 1756
    },
    {
      "epoch": 8.22469280280866,
      "grad_norm": 0.4449338912963867,
      "learning_rate": 9.054150984820675e-06,
      "loss": 1.3198,
      "step": 1757
    },
    {
      "epoch": 8.229373902867174,
      "grad_norm": 0.4492490887641907,
      "learning_rate": 9.007179445373715e-06,
      "loss": 1.3308,
      "step": 1758
    },
    {
      "epoch": 8.234055002925688,
      "grad_norm": 0.45008864998817444,
      "learning_rate": 8.960317999818458e-06,
      "loss": 1.3189,
      "step": 1759
    },
    {
      "epoch": 8.238736102984202,
      "grad_norm": 0.45013290643692017,
      "learning_rate": 8.913566774010107e-06,
      "loss": 1.3102,
      "step": 1760
    },
    {
      "epoch": 8.243417203042714,
      "grad_norm": 0.44994527101516724,
      "learning_rate": 8.866925893507805e-06,
      "loss": 1.3277,
      "step": 1761
    },
    {
      "epoch": 8.248098303101228,
      "grad_norm": 0.46309390664100647,
      "learning_rate": 8.820395483574368e-06,
      "loss": 1.3304,
      "step": 1762
    },
    {
      "epoch": 8.252779403159742,
      "grad_norm": 0.4659731388092041,
      "learning_rate": 8.773975669175938e-06,
      "loss": 1.3284,
      "step": 1763
    },
    {
      "epoch": 8.257460503218256,
      "grad_norm": 0.4545668959617615,
      "learning_rate": 8.727666574981597e-06,
      "loss": 1.3347,
      "step": 1764
    },
    {
      "epoch": 8.26214160327677,
      "grad_norm": 0.4606211483478546,
      "learning_rate": 8.681468325363106e-06,
      "loss": 1.3235,
      "step": 1765
    },
    {
      "epoch": 8.266822703335285,
      "grad_norm": 0.4521343410015106,
      "learning_rate": 8.635381044394486e-06,
      "loss": 1.3262,
      "step": 1766
    },
    {
      "epoch": 8.271503803393797,
      "grad_norm": 0.45838385820388794,
      "learning_rate": 8.589404855851762e-06,
      "loss": 1.3327,
      "step": 1767
    },
    {
      "epoch": 8.276184903452311,
      "grad_norm": 0.45257940888404846,
      "learning_rate": 8.543539883212615e-06,
      "loss": 1.3334,
      "step": 1768
    },
    {
      "epoch": 8.280866003510825,
      "grad_norm": 0.4488631784915924,
      "learning_rate": 8.497786249655965e-06,
      "loss": 1.3299,
      "step": 1769
    },
    {
      "epoch": 8.285547103569339,
      "grad_norm": 0.45357194542884827,
      "learning_rate": 8.452144078061818e-06,
      "loss": 1.3388,
      "step": 1770
    },
    {
      "epoch": 8.290228203627853,
      "grad_norm": 0.4556363821029663,
      "learning_rate": 8.406613491010723e-06,
      "loss": 1.3151,
      "step": 1771
    },
    {
      "epoch": 8.294909303686365,
      "grad_norm": 0.44982144236564636,
      "learning_rate": 8.361194610783612e-06,
      "loss": 1.3407,
      "step": 1772
    },
    {
      "epoch": 8.29959040374488,
      "grad_norm": 0.4660349488258362,
      "learning_rate": 8.31588755936139e-06,
      "loss": 1.3342,
      "step": 1773
    },
    {
      "epoch": 8.304271503803394,
      "grad_norm": 0.45148947834968567,
      "learning_rate": 8.2706924584246e-06,
      "loss": 1.3419,
      "step": 1774
    },
    {
      "epoch": 8.308952603861908,
      "grad_norm": 0.4520352780818939,
      "learning_rate": 8.225609429353187e-06,
      "loss": 1.3123,
      "step": 1775
    },
    {
      "epoch": 8.313633703920422,
      "grad_norm": 0.4512474238872528,
      "learning_rate": 8.18063859322602e-06,
      "loss": 1.3502,
      "step": 1776
    },
    {
      "epoch": 8.318314803978936,
      "grad_norm": 0.4652462899684906,
      "learning_rate": 8.13578007082072e-06,
      "loss": 1.3465,
      "step": 1777
    },
    {
      "epoch": 8.322995904037448,
      "grad_norm": 0.44895434379577637,
      "learning_rate": 8.091033982613245e-06,
      "loss": 1.3161,
      "step": 1778
    },
    {
      "epoch": 8.327677004095962,
      "grad_norm": 0.453987717628479,
      "learning_rate": 8.046400448777574e-06,
      "loss": 1.3343,
      "step": 1779
    },
    {
      "epoch": 8.332358104154476,
      "grad_norm": 0.4530025124549866,
      "learning_rate": 8.001879589185423e-06,
      "loss": 1.3368,
      "step": 1780
    },
    {
      "epoch": 8.33703920421299,
      "grad_norm": 0.45067164301872253,
      "learning_rate": 7.957471523405895e-06,
      "loss": 1.3291,
      "step": 1781
    },
    {
      "epoch": 8.341720304271504,
      "grad_norm": 0.44522637128829956,
      "learning_rate": 7.913176370705167e-06,
      "loss": 1.3405,
      "step": 1782
    },
    {
      "epoch": 8.346401404330017,
      "grad_norm": 0.4582405388355255,
      "learning_rate": 7.868994250046163e-06,
      "loss": 1.3564,
      "step": 1783
    },
    {
      "epoch": 8.35108250438853,
      "grad_norm": 0.455962210893631,
      "learning_rate": 7.824925280088219e-06,
      "loss": 1.3292,
      "step": 1784
    },
    {
      "epoch": 8.355763604447045,
      "grad_norm": 0.45344966650009155,
      "learning_rate": 7.780969579186814e-06,
      "loss": 1.3404,
      "step": 1785
    },
    {
      "epoch": 8.360444704505559,
      "grad_norm": 0.451633483171463,
      "learning_rate": 7.737127265393206e-06,
      "loss": 1.3244,
      "step": 1786
    },
    {
      "epoch": 8.365125804564073,
      "grad_norm": 0.46836763620376587,
      "learning_rate": 7.693398456454142e-06,
      "loss": 1.3558,
      "step": 1787
    },
    {
      "epoch": 8.369806904622587,
      "grad_norm": 0.4535449147224426,
      "learning_rate": 7.649783269811523e-06,
      "loss": 1.3454,
      "step": 1788
    },
    {
      "epoch": 8.3744880046811,
      "grad_norm": 0.44840291142463684,
      "learning_rate": 7.606281822602079e-06,
      "loss": 1.3136,
      "step": 1789
    },
    {
      "epoch": 8.379169104739614,
      "grad_norm": 0.4514668583869934,
      "learning_rate": 7.562894231657103e-06,
      "loss": 1.3278,
      "step": 1790
    },
    {
      "epoch": 8.383850204798128,
      "grad_norm": 0.4571569561958313,
      "learning_rate": 7.519620613502082e-06,
      "loss": 1.3273,
      "step": 1791
    },
    {
      "epoch": 8.388531304856642,
      "grad_norm": 0.45401352643966675,
      "learning_rate": 7.476461084356423e-06,
      "loss": 1.3273,
      "step": 1792
    },
    {
      "epoch": 8.393212404915156,
      "grad_norm": 0.45479846000671387,
      "learning_rate": 7.433415760133128e-06,
      "loss": 1.3231,
      "step": 1793
    },
    {
      "epoch": 8.397893504973668,
      "grad_norm": 0.45161619782447815,
      "learning_rate": 7.390484756438448e-06,
      "loss": 1.3104,
      "step": 1794
    },
    {
      "epoch": 8.402574605032182,
      "grad_norm": 0.4587078392505646,
      "learning_rate": 7.347668188571643e-06,
      "loss": 1.3049,
      "step": 1795
    },
    {
      "epoch": 8.407255705090696,
      "grad_norm": 0.4491417407989502,
      "learning_rate": 7.304966171524619e-06,
      "loss": 1.3255,
      "step": 1796
    },
    {
      "epoch": 8.41193680514921,
      "grad_norm": 0.46726784110069275,
      "learning_rate": 7.26237881998163e-06,
      "loss": 1.3332,
      "step": 1797
    },
    {
      "epoch": 8.416617905207724,
      "grad_norm": 0.45452451705932617,
      "learning_rate": 7.219906248318998e-06,
      "loss": 1.3464,
      "step": 1798
    },
    {
      "epoch": 8.421299005266238,
      "grad_norm": 0.44723427295684814,
      "learning_rate": 7.177548570604731e-06,
      "loss": 1.3075,
      "step": 1799
    },
    {
      "epoch": 8.42598010532475,
      "grad_norm": 0.4556967616081238,
      "learning_rate": 7.135305900598321e-06,
      "loss": 1.3382,
      "step": 1800
    },
    {
      "epoch": 8.430661205383265,
      "grad_norm": 0.4641600549221039,
      "learning_rate": 7.093178351750357e-06,
      "loss": 1.3465,
      "step": 1801
    },
    {
      "epoch": 8.435342305441779,
      "grad_norm": 0.4607059955596924,
      "learning_rate": 7.051166037202261e-06,
      "loss": 1.3453,
      "step": 1802
    },
    {
      "epoch": 8.440023405500293,
      "grad_norm": 0.4600765109062195,
      "learning_rate": 7.009269069785973e-06,
      "loss": 1.3375,
      "step": 1803
    },
    {
      "epoch": 8.444704505558807,
      "grad_norm": 0.467617928981781,
      "learning_rate": 6.967487562023622e-06,
      "loss": 1.3491,
      "step": 1804
    },
    {
      "epoch": 8.44938560561732,
      "grad_norm": 0.4564569592475891,
      "learning_rate": 6.925821626127282e-06,
      "loss": 1.3369,
      "step": 1805
    },
    {
      "epoch": 8.454066705675833,
      "grad_norm": 0.44401881098747253,
      "learning_rate": 6.8842713739986075e-06,
      "loss": 1.3317,
      "step": 1806
    },
    {
      "epoch": 8.458747805734347,
      "grad_norm": 0.4608115255832672,
      "learning_rate": 6.84283691722859e-06,
      "loss": 1.3376,
      "step": 1807
    },
    {
      "epoch": 8.463428905792862,
      "grad_norm": 0.44086068868637085,
      "learning_rate": 6.801518367097226e-06,
      "loss": 1.3354,
      "step": 1808
    },
    {
      "epoch": 8.468110005851376,
      "grad_norm": 0.45129072666168213,
      "learning_rate": 6.760315834573194e-06,
      "loss": 1.3327,
      "step": 1809
    },
    {
      "epoch": 8.472791105909888,
      "grad_norm": 0.4539918303489685,
      "learning_rate": 6.719229430313623e-06,
      "loss": 1.3101,
      "step": 1810
    },
    {
      "epoch": 8.477472205968402,
      "grad_norm": 0.45194685459136963,
      "learning_rate": 6.678259264663711e-06,
      "loss": 1.3517,
      "step": 1811
    },
    {
      "epoch": 8.482153306026916,
      "grad_norm": 0.45311862230300903,
      "learning_rate": 6.637405447656541e-06,
      "loss": 1.3246,
      "step": 1812
    },
    {
      "epoch": 8.48683440608543,
      "grad_norm": 0.44423723220825195,
      "learning_rate": 6.5966680890126865e-06,
      "loss": 1.3205,
      "step": 1813
    },
    {
      "epoch": 8.491515506143944,
      "grad_norm": 0.45881083607673645,
      "learning_rate": 6.556047298139917e-06,
      "loss": 1.3379,
      "step": 1814
    },
    {
      "epoch": 8.496196606202458,
      "grad_norm": 0.4573395550251007,
      "learning_rate": 6.515543184132999e-06,
      "loss": 1.3407,
      "step": 1815
    },
    {
      "epoch": 8.50087770626097,
      "grad_norm": 0.4606288969516754,
      "learning_rate": 6.475155855773279e-06,
      "loss": 1.3165,
      "step": 1816
    },
    {
      "epoch": 8.505558806319485,
      "grad_norm": 0.464618444442749,
      "learning_rate": 6.434885421528513e-06,
      "loss": 1.3339,
      "step": 1817
    },
    {
      "epoch": 8.510239906377999,
      "grad_norm": 0.45734116435050964,
      "learning_rate": 6.394731989552494e-06,
      "loss": 1.3336,
      "step": 1818
    },
    {
      "epoch": 8.514921006436513,
      "grad_norm": 0.44989585876464844,
      "learning_rate": 6.354695667684757e-06,
      "loss": 1.3429,
      "step": 1819
    },
    {
      "epoch": 8.519602106495027,
      "grad_norm": 0.44147491455078125,
      "learning_rate": 6.314776563450353e-06,
      "loss": 1.3215,
      "step": 1820
    },
    {
      "epoch": 8.524283206553541,
      "grad_norm": 0.448146790266037,
      "learning_rate": 6.274974784059495e-06,
      "loss": 1.3327,
      "step": 1821
    },
    {
      "epoch": 8.528964306612053,
      "grad_norm": 0.4638636112213135,
      "learning_rate": 6.235290436407304e-06,
      "loss": 1.3297,
      "step": 1822
    },
    {
      "epoch": 8.533645406670567,
      "grad_norm": 0.4473375082015991,
      "learning_rate": 6.195723627073552e-06,
      "loss": 1.3237,
      "step": 1823
    },
    {
      "epoch": 8.538326506729081,
      "grad_norm": 0.44932791590690613,
      "learning_rate": 6.156274462322292e-06,
      "loss": 1.3407,
      "step": 1824
    },
    {
      "epoch": 8.543007606787596,
      "grad_norm": 0.47016915678977966,
      "learning_rate": 6.1169430481016475e-06,
      "loss": 1.3339,
      "step": 1825
    },
    {
      "epoch": 8.54768870684611,
      "grad_norm": 0.44749605655670166,
      "learning_rate": 6.077729490043477e-06,
      "loss": 1.3432,
      "step": 1826
    },
    {
      "epoch": 8.552369806904622,
      "grad_norm": 0.4496007263660431,
      "learning_rate": 6.038633893463125e-06,
      "loss": 1.3372,
      "step": 1827
    },
    {
      "epoch": 8.557050906963136,
      "grad_norm": 0.4475160539150238,
      "learning_rate": 5.999656363359174e-06,
      "loss": 1.3329,
      "step": 1828
    },
    {
      "epoch": 8.56173200702165,
      "grad_norm": 0.4565197825431824,
      "learning_rate": 5.960797004413038e-06,
      "loss": 1.3349,
      "step": 1829
    },
    {
      "epoch": 8.566413107080164,
      "grad_norm": 0.4637001156806946,
      "learning_rate": 5.922055920988817e-06,
      "loss": 1.3355,
      "step": 1830
    },
    {
      "epoch": 8.571094207138678,
      "grad_norm": 0.4525499939918518,
      "learning_rate": 5.883433217132927e-06,
      "loss": 1.3355,
      "step": 1831
    },
    {
      "epoch": 8.57577530719719,
      "grad_norm": 0.45388099551200867,
      "learning_rate": 5.844928996573873e-06,
      "loss": 1.3177,
      "step": 1832
    },
    {
      "epoch": 8.580456407255705,
      "grad_norm": 0.45121920108795166,
      "learning_rate": 5.806543362721945e-06,
      "loss": 1.3092,
      "step": 1833
    },
    {
      "epoch": 8.585137507314219,
      "grad_norm": 0.4436412453651428,
      "learning_rate": 5.768276418668955e-06,
      "loss": 1.3184,
      "step": 1834
    },
    {
      "epoch": 8.589818607372733,
      "grad_norm": 0.44664260745048523,
      "learning_rate": 5.730128267187939e-06,
      "loss": 1.3334,
      "step": 1835
    },
    {
      "epoch": 8.594499707431247,
      "grad_norm": 0.4499116539955139,
      "learning_rate": 5.692099010732893e-06,
      "loss": 1.3212,
      "step": 1836
    },
    {
      "epoch": 8.59918080748976,
      "grad_norm": 0.4458361566066742,
      "learning_rate": 5.654188751438505e-06,
      "loss": 1.3272,
      "step": 1837
    },
    {
      "epoch": 8.603861907548273,
      "grad_norm": 0.4464578330516815,
      "learning_rate": 5.616397591119871e-06,
      "loss": 1.3164,
      "step": 1838
    },
    {
      "epoch": 8.608543007606787,
      "grad_norm": 0.4481458067893982,
      "learning_rate": 5.5787256312722245e-06,
      "loss": 1.3402,
      "step": 1839
    },
    {
      "epoch": 8.613224107665301,
      "grad_norm": 0.4532814025878906,
      "learning_rate": 5.5411729730706795e-06,
      "loss": 1.3215,
      "step": 1840
    },
    {
      "epoch": 8.617905207723815,
      "grad_norm": 0.44967561960220337,
      "learning_rate": 5.503739717369899e-06,
      "loss": 1.3291,
      "step": 1841
    },
    {
      "epoch": 8.62258630778233,
      "grad_norm": 0.4423333406448364,
      "learning_rate": 5.466425964703914e-06,
      "loss": 1.3176,
      "step": 1842
    },
    {
      "epoch": 8.627267407840842,
      "grad_norm": 0.45483851432800293,
      "learning_rate": 5.42923181528579e-06,
      "loss": 1.337,
      "step": 1843
    },
    {
      "epoch": 8.631948507899356,
      "grad_norm": 0.44831711053848267,
      "learning_rate": 5.392157369007378e-06,
      "loss": 1.3405,
      "step": 1844
    },
    {
      "epoch": 8.63662960795787,
      "grad_norm": 0.4486755430698395,
      "learning_rate": 5.355202725439046e-06,
      "loss": 1.3194,
      "step": 1845
    },
    {
      "epoch": 8.641310708016384,
      "grad_norm": 0.44130879640579224,
      "learning_rate": 5.318367983829392e-06,
      "loss": 1.3231,
      "step": 1846
    },
    {
      "epoch": 8.645991808074898,
      "grad_norm": 0.4567999839782715,
      "learning_rate": 5.281653243105011e-06,
      "loss": 1.351,
      "step": 1847
    },
    {
      "epoch": 8.650672908133412,
      "grad_norm": 0.4551226496696472,
      "learning_rate": 5.245058601870223e-06,
      "loss": 1.333,
      "step": 1848
    },
    {
      "epoch": 8.655354008191924,
      "grad_norm": 0.4408561885356903,
      "learning_rate": 5.2085841584067765e-06,
      "loss": 1.3289,
      "step": 1849
    },
    {
      "epoch": 8.660035108250439,
      "grad_norm": 0.4413094222545624,
      "learning_rate": 5.172230010673623e-06,
      "loss": 1.3271,
      "step": 1850
    },
    {
      "epoch": 8.664716208308953,
      "grad_norm": 0.44439423084259033,
      "learning_rate": 5.135996256306619e-06,
      "loss": 1.335,
      "step": 1851
    },
    {
      "epoch": 8.669397308367467,
      "grad_norm": 0.44356393814086914,
      "learning_rate": 5.099882992618299e-06,
      "loss": 1.3437,
      "step": 1852
    },
    {
      "epoch": 8.67407840842598,
      "grad_norm": 0.46065422892570496,
      "learning_rate": 5.063890316597592e-06,
      "loss": 1.3389,
      "step": 1853
    },
    {
      "epoch": 8.678759508484493,
      "grad_norm": 0.4485087990760803,
      "learning_rate": 5.028018324909562e-06,
      "loss": 1.3382,
      "step": 1854
    },
    {
      "epoch": 8.683440608543007,
      "grad_norm": 0.44895005226135254,
      "learning_rate": 4.992267113895172e-06,
      "loss": 1.3381,
      "step": 1855
    },
    {
      "epoch": 8.688121708601521,
      "grad_norm": 0.4537106454372406,
      "learning_rate": 4.956636779570972e-06,
      "loss": 1.3266,
      "step": 1856
    },
    {
      "epoch": 8.692802808660035,
      "grad_norm": 0.46100160479545593,
      "learning_rate": 4.921127417628896e-06,
      "loss": 1.3495,
      "step": 1857
    },
    {
      "epoch": 8.69748390871855,
      "grad_norm": 0.4382075071334839,
      "learning_rate": 4.885739123435984e-06,
      "loss": 1.3377,
      "step": 1858
    },
    {
      "epoch": 8.702165008777062,
      "grad_norm": 0.4552956223487854,
      "learning_rate": 4.850471992034128e-06,
      "loss": 1.3324,
      "step": 1859
    },
    {
      "epoch": 8.706846108835576,
      "grad_norm": 0.4699152112007141,
      "learning_rate": 4.8153261181398125e-06,
      "loss": 1.3317,
      "step": 1860
    },
    {
      "epoch": 8.71152720889409,
      "grad_norm": 0.46199172735214233,
      "learning_rate": 4.780301596143843e-06,
      "loss": 1.3389,
      "step": 1861
    },
    {
      "epoch": 8.716208308952604,
      "grad_norm": 0.44489288330078125,
      "learning_rate": 4.745398520111138e-06,
      "loss": 1.3299,
      "step": 1862
    },
    {
      "epoch": 8.720889409011118,
      "grad_norm": 0.4437996745109558,
      "learning_rate": 4.7106169837804525e-06,
      "loss": 1.3249,
      "step": 1863
    },
    {
      "epoch": 8.725570509069632,
      "grad_norm": 0.44859573245048523,
      "learning_rate": 4.675957080564087e-06,
      "loss": 1.3419,
      "step": 1864
    },
    {
      "epoch": 8.730251609128144,
      "grad_norm": 0.44944876432418823,
      "learning_rate": 4.641418903547723e-06,
      "loss": 1.3171,
      "step": 1865
    },
    {
      "epoch": 8.734932709186658,
      "grad_norm": 0.4558664858341217,
      "learning_rate": 4.60700254549008e-06,
      "loss": 1.3359,
      "step": 1866
    },
    {
      "epoch": 8.739613809245173,
      "grad_norm": 0.4526291787624359,
      "learning_rate": 4.5727080988227365e-06,
      "loss": 1.3089,
      "step": 1867
    },
    {
      "epoch": 8.744294909303687,
      "grad_norm": 0.4543936550617218,
      "learning_rate": 4.5385356556498616e-06,
      "loss": 1.3268,
      "step": 1868
    },
    {
      "epoch": 8.7489760093622,
      "grad_norm": 0.44007402658462524,
      "learning_rate": 4.504485307747913e-06,
      "loss": 1.3264,
      "step": 1869
    },
    {
      "epoch": 8.753657109420715,
      "grad_norm": 0.4477139115333557,
      "learning_rate": 4.470557146565507e-06,
      "loss": 1.3395,
      "step": 1870
    },
    {
      "epoch": 8.758338209479227,
      "grad_norm": 0.4843866527080536,
      "learning_rate": 4.436751263223055e-06,
      "loss": 1.3241,
      "step": 1871
    },
    {
      "epoch": 8.763019309537741,
      "grad_norm": 0.45234960317611694,
      "learning_rate": 4.40306774851259e-06,
      "loss": 1.3434,
      "step": 1872
    },
    {
      "epoch": 8.767700409596255,
      "grad_norm": 0.4426249563694,
      "learning_rate": 4.369506692897479e-06,
      "loss": 1.3105,
      "step": 1873
    },
    {
      "epoch": 8.77238150965477,
      "grad_norm": 0.44732552766799927,
      "learning_rate": 4.3360681865122325e-06,
      "loss": 1.3296,
      "step": 1874
    },
    {
      "epoch": 8.777062609713283,
      "grad_norm": 0.4477425813674927,
      "learning_rate": 4.302752319162212e-06,
      "loss": 1.364,
      "step": 1875
    },
    {
      "epoch": 8.781743709771796,
      "grad_norm": 0.4506896734237671,
      "learning_rate": 4.269559180323418e-06,
      "loss": 1.3366,
      "step": 1876
    },
    {
      "epoch": 8.78642480983031,
      "grad_norm": 0.44847944378852844,
      "learning_rate": 4.23648885914224e-06,
      "loss": 1.3518,
      "step": 1877
    },
    {
      "epoch": 8.791105909888824,
      "grad_norm": 0.442094087600708,
      "learning_rate": 4.20354144443521e-06,
      "loss": 1.3154,
      "step": 1878
    },
    {
      "epoch": 8.795787009947338,
      "grad_norm": 0.4490923285484314,
      "learning_rate": 4.170717024688775e-06,
      "loss": 1.3304,
      "step": 1879
    },
    {
      "epoch": 8.800468110005852,
      "grad_norm": 0.4516635537147522,
      "learning_rate": 4.1380156880590545e-06,
      "loss": 1.325,
      "step": 1880
    },
    {
      "epoch": 8.805149210064364,
      "grad_norm": 0.4438028931617737,
      "learning_rate": 4.105437522371619e-06,
      "loss": 1.3166,
      "step": 1881
    },
    {
      "epoch": 8.809830310122878,
      "grad_norm": 0.45484694838523865,
      "learning_rate": 4.072982615121235e-06,
      "loss": 1.3366,
      "step": 1882
    },
    {
      "epoch": 8.814511410181392,
      "grad_norm": 0.4515603482723236,
      "learning_rate": 4.040651053471605e-06,
      "loss": 1.337,
      "step": 1883
    },
    {
      "epoch": 8.819192510239906,
      "grad_norm": 0.44962701201438904,
      "learning_rate": 4.008442924255207e-06,
      "loss": 1.3385,
      "step": 1884
    },
    {
      "epoch": 8.82387361029842,
      "grad_norm": 0.4372497797012329,
      "learning_rate": 3.976358313973e-06,
      "loss": 1.3294,
      "step": 1885
    },
    {
      "epoch": 8.828554710356935,
      "grad_norm": 0.4433598518371582,
      "learning_rate": 3.944397308794201e-06,
      "loss": 1.3325,
      "step": 1886
    },
    {
      "epoch": 8.833235810415447,
      "grad_norm": 0.4615612030029297,
      "learning_rate": 3.912559994556086e-06,
      "loss": 1.3363,
      "step": 1887
    },
    {
      "epoch": 8.837916910473961,
      "grad_norm": 0.4531205892562866,
      "learning_rate": 3.880846456763704e-06,
      "loss": 1.3281,
      "step": 1888
    },
    {
      "epoch": 8.842598010532475,
      "grad_norm": 0.453538715839386,
      "learning_rate": 3.849256780589705e-06,
      "loss": 1.3542,
      "step": 1889
    },
    {
      "epoch": 8.84727911059099,
      "grad_norm": 0.44648537039756775,
      "learning_rate": 3.81779105087407e-06,
      "loss": 1.3061,
      "step": 1890
    },
    {
      "epoch": 8.851960210649503,
      "grad_norm": 0.4507054090499878,
      "learning_rate": 3.7864493521239164e-06,
      "loss": 1.3451,
      "step": 1891
    },
    {
      "epoch": 8.856641310708017,
      "grad_norm": 0.4552081823348999,
      "learning_rate": 3.75523176851324e-06,
      "loss": 1.3292,
      "step": 1892
    },
    {
      "epoch": 8.86132241076653,
      "grad_norm": 0.4661564826965332,
      "learning_rate": 3.724138383882697e-06,
      "loss": 1.3446,
      "step": 1893
    },
    {
      "epoch": 8.866003510825044,
      "grad_norm": 0.4496918022632599,
      "learning_rate": 3.6931692817393894e-06,
      "loss": 1.3254,
      "step": 1894
    },
    {
      "epoch": 8.870684610883558,
      "grad_norm": 0.45760682225227356,
      "learning_rate": 3.6623245452566455e-06,
      "loss": 1.3331,
      "step": 1895
    },
    {
      "epoch": 8.875365710942072,
      "grad_norm": 0.4535819888114929,
      "learning_rate": 3.631604257273774e-06,
      "loss": 1.3239,
      "step": 1896
    },
    {
      "epoch": 8.880046811000586,
      "grad_norm": 0.44883638620376587,
      "learning_rate": 3.601008500295866e-06,
      "loss": 1.346,
      "step": 1897
    },
    {
      "epoch": 8.884727911059098,
      "grad_norm": 0.4418095350265503,
      "learning_rate": 3.5705373564935364e-06,
      "loss": 1.3128,
      "step": 1898
    },
    {
      "epoch": 8.889409011117612,
      "grad_norm": 0.4573911428451538,
      "learning_rate": 3.5401909077027508e-06,
      "loss": 1.32,
      "step": 1899
    },
    {
      "epoch": 8.894090111176126,
      "grad_norm": 0.44246190786361694,
      "learning_rate": 3.5099692354245818e-06,
      "loss": 1.3316,
      "step": 1900
    },
    {
      "epoch": 8.89877121123464,
      "grad_norm": 0.4418052136898041,
      "learning_rate": 3.4798724208249866e-06,
      "loss": 1.3308,
      "step": 1901
    },
    {
      "epoch": 8.903452311293155,
      "grad_norm": 0.44418641924858093,
      "learning_rate": 3.449900544734602e-06,
      "loss": 1.338,
      "step": 1902
    },
    {
      "epoch": 8.908133411351667,
      "grad_norm": 0.4427943825721741,
      "learning_rate": 3.420053687648489e-06,
      "loss": 1.3242,
      "step": 1903
    },
    {
      "epoch": 8.91281451141018,
      "grad_norm": 0.44212570786476135,
      "learning_rate": 3.390331929725982e-06,
      "loss": 1.3209,
      "step": 1904
    },
    {
      "epoch": 8.917495611468695,
      "grad_norm": 0.4524332284927368,
      "learning_rate": 3.3607353507904283e-06,
      "loss": 1.3122,
      "step": 1905
    },
    {
      "epoch": 8.922176711527209,
      "grad_norm": 0.4490658640861511,
      "learning_rate": 3.3312640303289566e-06,
      "loss": 1.3247,
      "step": 1906
    },
    {
      "epoch": 8.926857811585723,
      "grad_norm": 0.44071638584136963,
      "learning_rate": 3.3019180474923527e-06,
      "loss": 1.336,
      "step": 1907
    },
    {
      "epoch": 8.931538911644237,
      "grad_norm": 0.4507541060447693,
      "learning_rate": 3.2726974810947107e-06,
      "loss": 1.3551,
      "step": 1908
    },
    {
      "epoch": 8.93622001170275,
      "grad_norm": 0.443243145942688,
      "learning_rate": 3.24360240961335e-06,
      "loss": 1.3267,
      "step": 1909
    },
    {
      "epoch": 8.940901111761264,
      "grad_norm": 0.46107861399650574,
      "learning_rate": 3.2146329111885256e-06,
      "loss": 1.3407,
      "step": 1910
    },
    {
      "epoch": 8.945582211819778,
      "grad_norm": 0.44938021898269653,
      "learning_rate": 3.18578906362324e-06,
      "loss": 1.3299,
      "step": 1911
    },
    {
      "epoch": 8.950263311878292,
      "grad_norm": 0.4389936923980713,
      "learning_rate": 3.157070944383056e-06,
      "loss": 1.3161,
      "step": 1912
    },
    {
      "epoch": 8.954944411936806,
      "grad_norm": 0.45983871817588806,
      "learning_rate": 3.1284786305958436e-06,
      "loss": 1.3426,
      "step": 1913
    },
    {
      "epoch": 8.959625511995318,
      "grad_norm": 0.4543863832950592,
      "learning_rate": 3.1000121990516274e-06,
      "loss": 1.3435,
      "step": 1914
    },
    {
      "epoch": 8.964306612053832,
      "grad_norm": 0.4446238577365875,
      "learning_rate": 3.0716717262023353e-06,
      "loss": 1.3197,
      "step": 1915
    },
    {
      "epoch": 8.968987712112346,
      "grad_norm": 0.4496222138404846,
      "learning_rate": 3.0434572881615997e-06,
      "loss": 1.3434,
      "step": 1916
    },
    {
      "epoch": 8.97366881217086,
      "grad_norm": 0.44096076488494873,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.3389,
      "step": 1917
    },
    {
      "epoch": 8.978349912229374,
      "grad_norm": 0.44681215286254883,
      "learning_rate": 2.987406819267746e-06,
      "loss": 1.3135,
      "step": 1918
    },
    {
      "epoch": 8.983031012287888,
      "grad_norm": 0.45022034645080566,
      "learning_rate": 2.959570938948647e-06,
      "loss": 1.3247,
      "step": 1919
    },
    {
      "epoch": 8.9877121123464,
      "grad_norm": 0.4587979316711426,
      "learning_rate": 2.931861394505764e-06,
      "loss": 1.3403,
      "step": 1920
    },
    {
      "epoch": 8.992393212404915,
      "grad_norm": 0.44257092475891113,
      "learning_rate": 2.9042782603582507e-06,
      "loss": 1.3504,
      "step": 1921
    },
    {
      "epoch": 8.997074312463429,
      "grad_norm": 0.4537746012210846,
      "learning_rate": 2.876821610585784e-06,
      "loss": 1.3282,
      "step": 1922
    },
    {
      "epoch": 9.001755412521943,
      "grad_norm": 1.9876453876495361,
      "learning_rate": 2.8494915189283324e-06,
      "loss": 1.3033,
      "step": 1923
    },
    {
      "epoch": 9.006436512580457,
      "grad_norm": 0.44968175888061523,
      "learning_rate": 2.8222880587859713e-06,
      "loss": 1.3326,
      "step": 1924
    },
    {
      "epoch": 9.01111761263897,
      "grad_norm": 0.4442802667617798,
      "learning_rate": 2.795211303218698e-06,
      "loss": 1.3317,
      "step": 1925
    },
    {
      "epoch": 9.015798712697483,
      "grad_norm": 0.45361271500587463,
      "learning_rate": 2.7682613249461863e-06,
      "loss": 1.3406,
      "step": 1926
    },
    {
      "epoch": 9.020479812755998,
      "grad_norm": 0.4399416744709015,
      "learning_rate": 2.7414381963476564e-06,
      "loss": 1.316,
      "step": 1927
    },
    {
      "epoch": 9.025160912814512,
      "grad_norm": 0.4478228986263275,
      "learning_rate": 2.7147419894616387e-06,
      "loss": 1.3169,
      "step": 1928
    },
    {
      "epoch": 9.029842012873026,
      "grad_norm": 0.4422329366207123,
      "learning_rate": 2.688172775985792e-06,
      "loss": 1.3357,
      "step": 1929
    },
    {
      "epoch": 9.03452311293154,
      "grad_norm": 0.44650009274482727,
      "learning_rate": 2.661730627276726e-06,
      "loss": 1.3251,
      "step": 1930
    },
    {
      "epoch": 9.039204212990052,
      "grad_norm": 0.444793164730072,
      "learning_rate": 2.6354156143497622e-06,
      "loss": 1.3069,
      "step": 1931
    },
    {
      "epoch": 9.043885313048566,
      "grad_norm": 0.4586767852306366,
      "learning_rate": 2.6092278078788e-06,
      "loss": 1.3298,
      "step": 1932
    },
    {
      "epoch": 9.04856641310708,
      "grad_norm": 0.43380340933799744,
      "learning_rate": 2.5831672781960913e-06,
      "loss": 1.3245,
      "step": 1933
    },
    {
      "epoch": 9.053247513165594,
      "grad_norm": 0.4407908320426941,
      "learning_rate": 2.557234095292066e-06,
      "loss": 1.329,
      "step": 1934
    },
    {
      "epoch": 9.057928613224108,
      "grad_norm": 0.4531652629375458,
      "learning_rate": 2.531428328815155e-06,
      "loss": 1.3414,
      "step": 1935
    },
    {
      "epoch": 9.06260971328262,
      "grad_norm": 0.44452014565467834,
      "learning_rate": 2.505750048071548e-06,
      "loss": 1.333,
      "step": 1936
    },
    {
      "epoch": 9.067290813341135,
      "grad_norm": 0.445748507976532,
      "learning_rate": 2.4801993220250953e-06,
      "loss": 1.314,
      "step": 1937
    },
    {
      "epoch": 9.071971913399649,
      "grad_norm": 0.4494490921497345,
      "learning_rate": 2.4547762192970392e-06,
      "loss": 1.3561,
      "step": 1938
    },
    {
      "epoch": 9.076653013458163,
      "grad_norm": 0.4584897756576538,
      "learning_rate": 2.4294808081658906e-06,
      "loss": 1.333,
      "step": 1939
    },
    {
      "epoch": 9.081334113516677,
      "grad_norm": 0.45643746852874756,
      "learning_rate": 2.404313156567217e-06,
      "loss": 1.3133,
      "step": 1940
    },
    {
      "epoch": 9.086015213575191,
      "grad_norm": 0.4478752017021179,
      "learning_rate": 2.3792733320934346e-06,
      "loss": 1.3533,
      "step": 1941
    },
    {
      "epoch": 9.090696313633703,
      "grad_norm": 0.44859951734542847,
      "learning_rate": 2.3543614019936987e-06,
      "loss": 1.324,
      "step": 1942
    },
    {
      "epoch": 9.095377413692217,
      "grad_norm": 0.4443413317203522,
      "learning_rate": 2.3295774331736408e-06,
      "loss": 1.3408,
      "step": 1943
    },
    {
      "epoch": 9.100058513750731,
      "grad_norm": 0.45216187834739685,
      "learning_rate": 2.304921492195261e-06,
      "loss": 1.3462,
      "step": 1944
    },
    {
      "epoch": 9.104739613809246,
      "grad_norm": 0.45232582092285156,
      "learning_rate": 2.280393645276713e-06,
      "loss": 1.3386,
      "step": 1945
    },
    {
      "epoch": 9.10942071386776,
      "grad_norm": 0.4486156702041626,
      "learning_rate": 2.2559939582920963e-06,
      "loss": 1.3437,
      "step": 1946
    },
    {
      "epoch": 9.114101813926272,
      "grad_norm": 0.44333913922309875,
      "learning_rate": 2.2317224967713603e-06,
      "loss": 1.3391,
      "step": 1947
    },
    {
      "epoch": 9.118782913984786,
      "grad_norm": 0.44496697187423706,
      "learning_rate": 2.207579325900033e-06,
      "loss": 1.2984,
      "step": 1948
    },
    {
      "epoch": 9.1234640140433,
      "grad_norm": 0.44474706053733826,
      "learning_rate": 2.183564510519137e-06,
      "loss": 1.3285,
      "step": 1949
    },
    {
      "epoch": 9.128145114101814,
      "grad_norm": 0.44082364439964294,
      "learning_rate": 2.1596781151249524e-06,
      "loss": 1.3319,
      "step": 1950
    },
    {
      "epoch": 9.132826214160328,
      "grad_norm": 0.4394497275352478,
      "learning_rate": 2.1359202038688575e-06,
      "loss": 1.31,
      "step": 1951
    },
    {
      "epoch": 9.13750731421884,
      "grad_norm": 0.44066622853279114,
      "learning_rate": 2.112290840557174e-06,
      "loss": 1.3268,
      "step": 1952
    },
    {
      "epoch": 9.142188414277355,
      "grad_norm": 0.4529281258583069,
      "learning_rate": 2.088790088650977e-06,
      "loss": 1.3423,
      "step": 1953
    },
    {
      "epoch": 9.146869514335869,
      "grad_norm": 0.4531400799751282,
      "learning_rate": 2.065418011265924e-06,
      "loss": 1.3273,
      "step": 1954
    },
    {
      "epoch": 9.151550614394383,
      "grad_norm": 0.4527547359466553,
      "learning_rate": 2.042174671172126e-06,
      "loss": 1.3125,
      "step": 1955
    },
    {
      "epoch": 9.156231714452897,
      "grad_norm": 0.4387628138065338,
      "learning_rate": 2.0190601307939094e-06,
      "loss": 1.3394,
      "step": 1956
    },
    {
      "epoch": 9.160912814511411,
      "grad_norm": 0.4485406279563904,
      "learning_rate": 1.9960744522097007e-06,
      "loss": 1.3379,
      "step": 1957
    },
    {
      "epoch": 9.165593914569923,
      "grad_norm": 0.4442361891269684,
      "learning_rate": 1.973217697151836e-06,
      "loss": 1.3386,
      "step": 1958
    },
    {
      "epoch": 9.170275014628437,
      "grad_norm": 0.4519590735435486,
      "learning_rate": 1.95048992700641e-06,
      "loss": 1.3382,
      "step": 1959
    },
    {
      "epoch": 9.174956114686951,
      "grad_norm": 0.4366214871406555,
      "learning_rate": 1.92789120281312e-06,
      "loss": 1.3113,
      "step": 1960
    },
    {
      "epoch": 9.179637214745465,
      "grad_norm": 0.44564807415008545,
      "learning_rate": 1.9054215852650437e-06,
      "loss": 1.3216,
      "step": 1961
    },
    {
      "epoch": 9.18431831480398,
      "grad_norm": 0.45024698972702026,
      "learning_rate": 1.8830811347085697e-06,
      "loss": 1.3193,
      "step": 1962
    },
    {
      "epoch": 9.188999414862492,
      "grad_norm": 0.4466053247451782,
      "learning_rate": 1.8608699111431372e-06,
      "loss": 1.3144,
      "step": 1963
    },
    {
      "epoch": 9.193680514921006,
      "grad_norm": 0.44546517729759216,
      "learning_rate": 1.838787974221151e-06,
      "loss": 1.3189,
      "step": 1964
    },
    {
      "epoch": 9.19836161497952,
      "grad_norm": 0.4485017657279968,
      "learning_rate": 1.8168353832477947e-06,
      "loss": 1.3376,
      "step": 1965
    },
    {
      "epoch": 9.203042715038034,
      "grad_norm": 0.4487907290458679,
      "learning_rate": 1.7950121971808454e-06,
      "loss": 1.3314,
      "step": 1966
    },
    {
      "epoch": 9.207723815096548,
      "grad_norm": 0.44457539916038513,
      "learning_rate": 1.7733184746305698e-06,
      "loss": 1.3246,
      "step": 1967
    },
    {
      "epoch": 9.212404915155062,
      "grad_norm": 0.45234352350234985,
      "learning_rate": 1.7517542738595071e-06,
      "loss": 1.3324,
      "step": 1968
    },
    {
      "epoch": 9.217086015213575,
      "grad_norm": 0.44961783289909363,
      "learning_rate": 1.7303196527823585e-06,
      "loss": 1.3519,
      "step": 1969
    },
    {
      "epoch": 9.221767115272089,
      "grad_norm": 0.45592254400253296,
      "learning_rate": 1.7090146689658083e-06,
      "loss": 1.3364,
      "step": 1970
    },
    {
      "epoch": 9.226448215330603,
      "grad_norm": 0.460465669631958,
      "learning_rate": 1.687839379628381e-06,
      "loss": 1.3465,
      "step": 1971
    },
    {
      "epoch": 9.231129315389117,
      "grad_norm": 0.4440554678440094,
      "learning_rate": 1.666793841640285e-06,
      "loss": 1.3189,
      "step": 1972
    },
    {
      "epoch": 9.23581041544763,
      "grad_norm": 0.4493771195411682,
      "learning_rate": 1.6458781115232359e-06,
      "loss": 1.3234,
      "step": 1973
    },
    {
      "epoch": 9.240491515506143,
      "grad_norm": 0.4530016779899597,
      "learning_rate": 1.6250922454503548e-06,
      "loss": 1.3211,
      "step": 1974
    },
    {
      "epoch": 9.245172615564657,
      "grad_norm": 0.4415639638900757,
      "learning_rate": 1.604436299245965e-06,
      "loss": 1.3204,
      "step": 1975
    },
    {
      "epoch": 9.249853715623171,
      "grad_norm": 0.4418812394142151,
      "learning_rate": 1.5839103283854906e-06,
      "loss": 1.3197,
      "step": 1976
    },
    {
      "epoch": 9.254534815681685,
      "grad_norm": 0.4474630653858185,
      "learning_rate": 1.5635143879952575e-06,
      "loss": 1.3566,
      "step": 1977
    },
    {
      "epoch": 9.2592159157402,
      "grad_norm": 0.4447353184223175,
      "learning_rate": 1.5432485328523815e-06,
      "loss": 1.3272,
      "step": 1978
    },
    {
      "epoch": 9.263897015798713,
      "grad_norm": 0.44466260075569153,
      "learning_rate": 1.5231128173846087e-06,
      "loss": 1.3395,
      "step": 1979
    },
    {
      "epoch": 9.268578115857226,
      "grad_norm": 0.44978243112564087,
      "learning_rate": 1.5031072956701697e-06,
      "loss": 1.3233,
      "step": 1980
    },
    {
      "epoch": 9.27325921591574,
      "grad_norm": 0.449532687664032,
      "learning_rate": 1.4832320214376472e-06,
      "loss": 1.3264,
      "step": 1981
    },
    {
      "epoch": 9.277940315974254,
      "grad_norm": 0.4369581341743469,
      "learning_rate": 1.4634870480658035e-06,
      "loss": 1.3071,
      "step": 1982
    },
    {
      "epoch": 9.282621416032768,
      "grad_norm": 0.4466133713722229,
      "learning_rate": 1.4438724285834594e-06,
      "loss": 1.3364,
      "step": 1983
    },
    {
      "epoch": 9.287302516091282,
      "grad_norm": 0.4498167037963867,
      "learning_rate": 1.4243882156693477e-06,
      "loss": 1.3335,
      "step": 1984
    },
    {
      "epoch": 9.291983616149794,
      "grad_norm": 0.4429709017276764,
      "learning_rate": 1.4050344616519662e-06,
      "loss": 1.33,
      "step": 1985
    },
    {
      "epoch": 9.296664716208308,
      "grad_norm": 0.439984530210495,
      "learning_rate": 1.3858112185094418e-06,
      "loss": 1.3171,
      "step": 1986
    },
    {
      "epoch": 9.301345816266823,
      "grad_norm": 0.43798401951789856,
      "learning_rate": 1.3667185378694048e-06,
      "loss": 1.3388,
      "step": 1987
    },
    {
      "epoch": 9.306026916325337,
      "grad_norm": 0.4463445544242859,
      "learning_rate": 1.3477564710088098e-06,
      "loss": 1.3281,
      "step": 1988
    },
    {
      "epoch": 9.31070801638385,
      "grad_norm": 0.43593892455101013,
      "learning_rate": 1.328925068853837e-06,
      "loss": 1.3263,
      "step": 1989
    },
    {
      "epoch": 9.315389116442365,
      "grad_norm": 0.45093831419944763,
      "learning_rate": 1.3102243819797467e-06,
      "loss": 1.323,
      "step": 1990
    },
    {
      "epoch": 9.320070216500877,
      "grad_norm": 0.4551108181476593,
      "learning_rate": 1.2916544606107305e-06,
      "loss": 1.3109,
      "step": 1991
    },
    {
      "epoch": 9.324751316559391,
      "grad_norm": 0.44643521308898926,
      "learning_rate": 1.273215354619789e-06,
      "loss": 1.3328,
      "step": 1992
    },
    {
      "epoch": 9.329432416617905,
      "grad_norm": 0.4388735592365265,
      "learning_rate": 1.2549071135285917e-06,
      "loss": 1.3295,
      "step": 1993
    },
    {
      "epoch": 9.33411351667642,
      "grad_norm": 0.44772210717201233,
      "learning_rate": 1.2367297865073401e-06,
      "loss": 1.3407,
      "step": 1994
    },
    {
      "epoch": 9.338794616734933,
      "grad_norm": 0.44219115376472473,
      "learning_rate": 1.2186834223746612e-06,
      "loss": 1.332,
      "step": 1995
    },
    {
      "epoch": 9.343475716793446,
      "grad_norm": 0.4451064169406891,
      "learning_rate": 1.20076806959743e-06,
      "loss": 1.3138,
      "step": 1996
    },
    {
      "epoch": 9.34815681685196,
      "grad_norm": 0.4520408511161804,
      "learning_rate": 1.1829837762906915e-06,
      "loss": 1.3427,
      "step": 1997
    },
    {
      "epoch": 9.352837916910474,
      "grad_norm": 0.44814303517341614,
      "learning_rate": 1.1653305902174894e-06,
      "loss": 1.3238,
      "step": 1998
    },
    {
      "epoch": 9.357519016968988,
      "grad_norm": 0.44885024428367615,
      "learning_rate": 1.1478085587887488e-06,
      "loss": 1.3449,
      "step": 1999
    },
    {
      "epoch": 9.362200117027502,
      "grad_norm": 0.4502180814743042,
      "learning_rate": 1.1304177290631824e-06,
      "loss": 1.3253,
      "step": 2000
    },
    {
      "epoch": 9.366881217086016,
      "grad_norm": 0.44551151990890503,
      "learning_rate": 1.113158147747101e-06,
      "loss": 1.329,
      "step": 2001
    },
    {
      "epoch": 9.371562317144528,
      "grad_norm": 0.4621962010860443,
      "learning_rate": 1.0960298611943532e-06,
      "loss": 1.3406,
      "step": 2002
    },
    {
      "epoch": 9.376243417203042,
      "grad_norm": 0.4457390010356903,
      "learning_rate": 1.079032915406153e-06,
      "loss": 1.3477,
      "step": 2003
    },
    {
      "epoch": 9.380924517261557,
      "grad_norm": 0.43988901376724243,
      "learning_rate": 1.0621673560309797e-06,
      "loss": 1.3197,
      "step": 2004
    },
    {
      "epoch": 9.38560561732007,
      "grad_norm": 0.4548582136631012,
      "learning_rate": 1.0454332283644508e-06,
      "loss": 1.3194,
      "step": 2005
    },
    {
      "epoch": 9.390286717378585,
      "grad_norm": 0.4404996931552887,
      "learning_rate": 1.0288305773491936e-06,
      "loss": 1.3099,
      "step": 2006
    },
    {
      "epoch": 9.394967817437097,
      "grad_norm": 0.45231539011001587,
      "learning_rate": 1.01235944757474e-06,
      "loss": 1.3293,
      "step": 2007
    },
    {
      "epoch": 9.399648917495611,
      "grad_norm": 0.45580315589904785,
      "learning_rate": 9.960198832773881e-07,
      "loss": 1.3271,
      "step": 2008
    },
    {
      "epoch": 9.404330017554125,
      "grad_norm": 0.4584179222583771,
      "learning_rate": 9.79811928340102e-07,
      "loss": 1.3292,
      "step": 2009
    },
    {
      "epoch": 9.40901111761264,
      "grad_norm": 0.45283496379852295,
      "learning_rate": 9.637356262923725e-07,
      "loss": 1.3134,
      "step": 2010
    },
    {
      "epoch": 9.413692217671153,
      "grad_norm": 0.44626957178115845,
      "learning_rate": 9.477910203101181e-07,
      "loss": 1.3412,
      "step": 2011
    },
    {
      "epoch": 9.418373317729667,
      "grad_norm": 0.4469052255153656,
      "learning_rate": 9.31978153215557e-07,
      "loss": 1.341,
      "step": 2012
    },
    {
      "epoch": 9.42305441778818,
      "grad_norm": 0.44531938433647156,
      "learning_rate": 9.162970674771176e-07,
      "loss": 1.3298,
      "step": 2013
    },
    {
      "epoch": 9.427735517846694,
      "grad_norm": 0.44805341958999634,
      "learning_rate": 9.007478052092732e-07,
      "loss": 1.3138,
      "step": 2014
    },
    {
      "epoch": 9.432416617905208,
      "grad_norm": 0.4502425789833069,
      "learning_rate": 8.853304081724967e-07,
      "loss": 1.3216,
      "step": 2015
    },
    {
      "epoch": 9.437097717963722,
      "grad_norm": 0.45359623432159424,
      "learning_rate": 8.700449177730774e-07,
      "loss": 1.3235,
      "step": 2016
    },
    {
      "epoch": 9.441778818022236,
      "grad_norm": 0.43736910820007324,
      "learning_rate": 8.548913750630661e-07,
      "loss": 1.307,
      "step": 2017
    },
    {
      "epoch": 9.446459918080748,
      "grad_norm": 0.44915610551834106,
      "learning_rate": 8.398698207401412e-07,
      "loss": 1.3311,
      "step": 2018
    },
    {
      "epoch": 9.451141018139262,
      "grad_norm": 0.45041677355766296,
      "learning_rate": 8.249802951474927e-07,
      "loss": 1.3438,
      "step": 2019
    },
    {
      "epoch": 9.455822118197776,
      "grad_norm": 0.44452545046806335,
      "learning_rate": 8.102228382737331e-07,
      "loss": 1.3257,
      "step": 2020
    },
    {
      "epoch": 9.46050321825629,
      "grad_norm": 0.44057121872901917,
      "learning_rate": 7.955974897527752e-07,
      "loss": 1.332,
      "step": 2021
    },
    {
      "epoch": 9.465184318314805,
      "grad_norm": 0.44600802659988403,
      "learning_rate": 7.81104288863721e-07,
      "loss": 1.3383,
      "step": 2022
    },
    {
      "epoch": 9.469865418373317,
      "grad_norm": 0.45320701599121094,
      "learning_rate": 7.66743274530779e-07,
      "loss": 1.3159,
      "step": 2023
    },
    {
      "epoch": 9.474546518431831,
      "grad_norm": 0.44251635670661926,
      "learning_rate": 7.525144853231469e-07,
      "loss": 1.3332,
      "step": 2024
    },
    {
      "epoch": 9.479227618490345,
      "grad_norm": 0.4494597017765045,
      "learning_rate": 7.384179594548957e-07,
      "loss": 1.2961,
      "step": 2025
    },
    {
      "epoch": 9.483908718548859,
      "grad_norm": 0.4432707130908966,
      "learning_rate": 7.244537347848856e-07,
      "loss": 1.3427,
      "step": 2026
    },
    {
      "epoch": 9.488589818607373,
      "grad_norm": 0.44649192690849304,
      "learning_rate": 7.106218488166505e-07,
      "loss": 1.3363,
      "step": 2027
    },
    {
      "epoch": 9.493270918665887,
      "grad_norm": 0.4434208571910858,
      "learning_rate": 6.969223386983137e-07,
      "loss": 1.3351,
      "step": 2028
    },
    {
      "epoch": 9.4979520187244,
      "grad_norm": 0.4437309503555298,
      "learning_rate": 6.83355241222472e-07,
      "loss": 1.3356,
      "step": 2029
    },
    {
      "epoch": 9.502633118782914,
      "grad_norm": 0.4394586384296417,
      "learning_rate": 6.699205928261066e-07,
      "loss": 1.3243,
      "step": 2030
    },
    {
      "epoch": 9.507314218841428,
      "grad_norm": 0.4349606931209564,
      "learning_rate": 6.566184295904776e-07,
      "loss": 1.3393,
      "step": 2031
    },
    {
      "epoch": 9.511995318899942,
      "grad_norm": 0.4388129711151123,
      "learning_rate": 6.434487872410355e-07,
      "loss": 1.3338,
      "step": 2032
    },
    {
      "epoch": 9.516676418958456,
      "grad_norm": 0.44483059644699097,
      "learning_rate": 6.30411701147321e-07,
      "loss": 1.3174,
      "step": 2033
    },
    {
      "epoch": 9.52135751901697,
      "grad_norm": 0.44136685132980347,
      "learning_rate": 6.175072063228648e-07,
      "loss": 1.3547,
      "step": 2034
    },
    {
      "epoch": 9.526038619075482,
      "grad_norm": 0.450206995010376,
      "learning_rate": 6.047353374251108e-07,
      "loss": 1.3374,
      "step": 2035
    },
    {
      "epoch": 9.530719719133996,
      "grad_norm": 0.4519071877002716,
      "learning_rate": 5.920961287552929e-07,
      "loss": 1.3303,
      "step": 2036
    },
    {
      "epoch": 9.53540081919251,
      "grad_norm": 0.44036248326301575,
      "learning_rate": 5.795896142583801e-07,
      "loss": 1.3202,
      "step": 2037
    },
    {
      "epoch": 9.540081919251024,
      "grad_norm": 0.4422554671764374,
      "learning_rate": 5.672158275229489e-07,
      "loss": 1.3312,
      "step": 2038
    },
    {
      "epoch": 9.544763019309539,
      "grad_norm": 0.4471757709980011,
      "learning_rate": 5.549748017811274e-07,
      "loss": 1.3342,
      "step": 2039
    },
    {
      "epoch": 9.54944411936805,
      "grad_norm": 0.4468821883201599,
      "learning_rate": 5.428665699084789e-07,
      "loss": 1.3261,
      "step": 2040
    },
    {
      "epoch": 9.554125219426565,
      "grad_norm": 0.4374167025089264,
      "learning_rate": 5.308911644239245e-07,
      "loss": 1.3184,
      "step": 2041
    },
    {
      "epoch": 9.558806319485079,
      "grad_norm": 0.43856698274612427,
      "learning_rate": 5.190486174896648e-07,
      "loss": 1.3377,
      "step": 2042
    },
    {
      "epoch": 9.563487419543593,
      "grad_norm": 0.4368426501750946,
      "learning_rate": 5.073389609110579e-07,
      "loss": 1.3091,
      "step": 2043
    },
    {
      "epoch": 9.568168519602107,
      "grad_norm": 0.4541669189929962,
      "learning_rate": 4.95762226136598e-07,
      "loss": 1.3391,
      "step": 2044
    },
    {
      "epoch": 9.57284961966062,
      "grad_norm": 0.4393460750579834,
      "learning_rate": 4.843184442577641e-07,
      "loss": 1.3178,
      "step": 2045
    },
    {
      "epoch": 9.577530719719134,
      "grad_norm": 0.44958001375198364,
      "learning_rate": 4.730076460089716e-07,
      "loss": 1.3473,
      "step": 2046
    },
    {
      "epoch": 9.582211819777648,
      "grad_norm": 0.4427577257156372,
      "learning_rate": 4.618298617674932e-07,
      "loss": 1.3227,
      "step": 2047
    },
    {
      "epoch": 9.586892919836162,
      "grad_norm": 0.45502230525016785,
      "learning_rate": 4.507851215533543e-07,
      "loss": 1.3424,
      "step": 2048
    },
    {
      "epoch": 9.591574019894676,
      "grad_norm": 0.4533569812774658,
      "learning_rate": 4.3987345502927157e-07,
      "loss": 1.3534,
      "step": 2049
    },
    {
      "epoch": 9.59625511995319,
      "grad_norm": 0.44390031695365906,
      "learning_rate": 4.2909489150056994e-07,
      "loss": 1.3026,
      "step": 2050
    },
    {
      "epoch": 9.600936220011702,
      "grad_norm": 0.4406379461288452,
      "learning_rate": 4.184494599151045e-07,
      "loss": 1.3387,
      "step": 2051
    },
    {
      "epoch": 9.605617320070216,
      "grad_norm": 0.4497063159942627,
      "learning_rate": 4.0793718886316665e-07,
      "loss": 1.3011,
      "step": 2052
    },
    {
      "epoch": 9.61029842012873,
      "grad_norm": 0.4548267126083374,
      "learning_rate": 3.975581065774281e-07,
      "loss": 1.3051,
      "step": 2053
    },
    {
      "epoch": 9.614979520187244,
      "grad_norm": 0.446915864944458,
      "learning_rate": 3.873122409328578e-07,
      "loss": 1.3206,
      "step": 2054
    },
    {
      "epoch": 9.619660620245758,
      "grad_norm": 0.4462285041809082,
      "learning_rate": 3.7719961944664985e-07,
      "loss": 1.3205,
      "step": 2055
    },
    {
      "epoch": 9.62434172030427,
      "grad_norm": 0.4510735869407654,
      "learning_rate": 3.672202692781401e-07,
      "loss": 1.3192,
      "step": 2056
    },
    {
      "epoch": 9.629022820362785,
      "grad_norm": 0.4520784318447113,
      "learning_rate": 3.5737421722873423e-07,
      "loss": 1.3301,
      "step": 2057
    },
    {
      "epoch": 9.633703920421299,
      "grad_norm": 0.4418693780899048,
      "learning_rate": 3.476614897418573e-07,
      "loss": 1.325,
      "step": 2058
    },
    {
      "epoch": 9.638385020479813,
      "grad_norm": 0.43810001015663147,
      "learning_rate": 3.380821129028489e-07,
      "loss": 1.3326,
      "step": 2059
    },
    {
      "epoch": 9.643066120538327,
      "grad_norm": 0.44491061568260193,
      "learning_rate": 3.2863611243892923e-07,
      "loss": 1.3267,
      "step": 2060
    },
    {
      "epoch": 9.647747220596841,
      "grad_norm": 0.4490460753440857,
      "learning_rate": 3.193235137190831e-07,
      "loss": 1.3157,
      "step": 2061
    },
    {
      "epoch": 9.652428320655353,
      "grad_norm": 0.4372991919517517,
      "learning_rate": 3.101443417540595e-07,
      "loss": 1.3611,
      "step": 2062
    },
    {
      "epoch": 9.657109420713867,
      "grad_norm": 0.4481930434703827,
      "learning_rate": 3.010986211962219e-07,
      "loss": 1.3173,
      "step": 2063
    },
    {
      "epoch": 9.661790520772382,
      "grad_norm": 0.4486005902290344,
      "learning_rate": 2.921863763395538e-07,
      "loss": 1.3013,
      "step": 2064
    },
    {
      "epoch": 9.666471620830896,
      "grad_norm": 0.43497464060783386,
      "learning_rate": 2.8340763111956437e-07,
      "loss": 1.3223,
      "step": 2065
    },
    {
      "epoch": 9.67115272088941,
      "grad_norm": 0.4525913596153259,
      "learning_rate": 2.7476240911320506e-07,
      "loss": 1.3495,
      "step": 2066
    },
    {
      "epoch": 9.675833820947922,
      "grad_norm": 0.4425511360168457,
      "learning_rate": 2.6625073353884753e-07,
      "loss": 1.328,
      "step": 2067
    },
    {
      "epoch": 9.680514921006436,
      "grad_norm": 0.4437705874443054,
      "learning_rate": 2.578726272561782e-07,
      "loss": 1.3167,
      "step": 2068
    },
    {
      "epoch": 9.68519602106495,
      "grad_norm": 0.45476165413856506,
      "learning_rate": 2.4962811276618146e-07,
      "loss": 1.323,
      "step": 2069
    },
    {
      "epoch": 9.689877121123464,
      "grad_norm": 0.4444226324558258,
      "learning_rate": 2.415172122110343e-07,
      "loss": 1.3449,
      "step": 2070
    },
    {
      "epoch": 9.694558221181978,
      "grad_norm": 0.4542228877544403,
      "learning_rate": 2.3353994737408403e-07,
      "loss": 1.3285,
      "step": 2071
    },
    {
      "epoch": 9.69923932124049,
      "grad_norm": 0.4445129632949829,
      "learning_rate": 2.2569633967977067e-07,
      "loss": 1.3213,
      "step": 2072
    },
    {
      "epoch": 9.703920421299005,
      "grad_norm": 0.4466426372528076,
      "learning_rate": 2.1798641019356025e-07,
      "loss": 1.3333,
      "step": 2073
    },
    {
      "epoch": 9.708601521357519,
      "grad_norm": 0.43596237897872925,
      "learning_rate": 2.1041017962192267e-07,
      "loss": 1.3328,
      "step": 2074
    },
    {
      "epoch": 9.713282621416033,
      "grad_norm": 0.4403690695762634,
      "learning_rate": 2.0296766831223168e-07,
      "loss": 1.3189,
      "step": 2075
    },
    {
      "epoch": 9.717963721474547,
      "grad_norm": 0.4439951479434967,
      "learning_rate": 1.9565889625275946e-07,
      "loss": 1.322,
      "step": 2076
    },
    {
      "epoch": 9.722644821533061,
      "grad_norm": 0.44233083724975586,
      "learning_rate": 1.8848388307257104e-07,
      "loss": 1.3317,
      "step": 2077
    },
    {
      "epoch": 9.727325921591573,
      "grad_norm": 0.45015835762023926,
      "learning_rate": 1.8144264804150214e-07,
      "loss": 1.3242,
      "step": 2078
    },
    {
      "epoch": 9.732007021650087,
      "grad_norm": 0.44355273246765137,
      "learning_rate": 1.7453521007012031e-07,
      "loss": 1.3198,
      "step": 2079
    },
    {
      "epoch": 9.736688121708601,
      "grad_norm": 0.43759721517562866,
      "learning_rate": 1.6776158770963613e-07,
      "loss": 1.348,
      "step": 2080
    },
    {
      "epoch": 9.741369221767116,
      "grad_norm": 0.43967142701148987,
      "learning_rate": 1.611217991518754e-07,
      "loss": 1.3111,
      "step": 2081
    },
    {
      "epoch": 9.74605032182563,
      "grad_norm": 0.43936997652053833,
      "learning_rate": 1.5461586222924596e-07,
      "loss": 1.3315,
      "step": 2082
    },
    {
      "epoch": 9.750731421884144,
      "grad_norm": 0.444767028093338,
      "learning_rate": 1.4824379441464863e-07,
      "loss": 1.3306,
      "step": 2083
    },
    {
      "epoch": 9.755412521942656,
      "grad_norm": 0.44919633865356445,
      "learning_rate": 1.420056128214664e-07,
      "loss": 1.3198,
      "step": 2084
    },
    {
      "epoch": 9.76009362200117,
      "grad_norm": 0.4446408152580261,
      "learning_rate": 1.3590133420350315e-07,
      "loss": 1.3278,
      "step": 2085
    },
    {
      "epoch": 9.764774722059684,
      "grad_norm": 0.4420437812805176,
      "learning_rate": 1.2993097495493935e-07,
      "loss": 1.3339,
      "step": 2086
    },
    {
      "epoch": 9.769455822118198,
      "grad_norm": 0.4489443302154541,
      "learning_rate": 1.2409455111030422e-07,
      "loss": 1.3413,
      "step": 2087
    },
    {
      "epoch": 9.774136922176712,
      "grad_norm": 0.4377923607826233,
      "learning_rate": 1.1839207834439258e-07,
      "loss": 1.3148,
      "step": 2088
    },
    {
      "epoch": 9.778818022235225,
      "grad_norm": 0.4404031038284302,
      "learning_rate": 1.1282357197226478e-07,
      "loss": 1.3239,
      "step": 2089
    },
    {
      "epoch": 9.783499122293739,
      "grad_norm": 0.4418086111545563,
      "learning_rate": 1.0738904694919116e-07,
      "loss": 1.3247,
      "step": 2090
    },
    {
      "epoch": 9.788180222352253,
      "grad_norm": 0.4544632136821747,
      "learning_rate": 1.0208851787060215e-07,
      "loss": 1.3373,
      "step": 2091
    },
    {
      "epoch": 9.792861322410767,
      "grad_norm": 0.4450118839740753,
      "learning_rate": 9.692199897206044e-08,
      "loss": 1.3194,
      "step": 2092
    },
    {
      "epoch": 9.79754242246928,
      "grad_norm": 0.4508454203605652,
      "learning_rate": 9.18895041292278e-08,
      "loss": 1.3307,
      "step": 2093
    },
    {
      "epoch": 9.802223522527793,
      "grad_norm": 0.4486829340457916,
      "learning_rate": 8.699104685779835e-08,
      "loss": 1.3247,
      "step": 2094
    },
    {
      "epoch": 9.806904622586307,
      "grad_norm": 0.44039294123649597,
      "learning_rate": 8.222664031350413e-08,
      "loss": 1.3265,
      "step": 2095
    },
    {
      "epoch": 9.811585722644821,
      "grad_norm": 0.44150322675704956,
      "learning_rate": 7.759629729204854e-08,
      "loss": 1.3311,
      "step": 2096
    },
    {
      "epoch": 9.816266822703335,
      "grad_norm": 0.4456532895565033,
      "learning_rate": 7.310003022908407e-08,
      "loss": 1.336,
      "step": 2097
    },
    {
      "epoch": 9.82094792276185,
      "grad_norm": 0.4442593455314636,
      "learning_rate": 6.873785120017906e-08,
      "loss": 1.3007,
      "step": 2098
    },
    {
      "epoch": 9.825629022820364,
      "grad_norm": 0.4418592154979706,
      "learning_rate": 6.450977192077879e-08,
      "loss": 1.3156,
      "step": 2099
    },
    {
      "epoch": 9.830310122878876,
      "grad_norm": 0.4495104253292084,
      "learning_rate": 6.041580374618328e-08,
      "loss": 1.3298,
      "step": 2100
    },
    {
      "epoch": 9.83499122293739,
      "grad_norm": 0.44459906220436096,
      "learning_rate": 5.6455957671508466e-08,
      "loss": 1.3133,
      "step": 2101
    },
    {
      "epoch": 9.839672322995904,
      "grad_norm": 0.4395291805267334,
      "learning_rate": 5.263024433166397e-08,
      "loss": 1.3444,
      "step": 2102
    },
    {
      "epoch": 9.844353423054418,
      "grad_norm": 0.44066277146339417,
      "learning_rate": 4.89386740013198e-08,
      "loss": 1.3244,
      "step": 2103
    },
    {
      "epoch": 9.849034523112932,
      "grad_norm": 0.43951621651649475,
      "learning_rate": 4.5381256594878574e-08,
      "loss": 1.3599,
      "step": 2104
    },
    {
      "epoch": 9.853715623171446,
      "grad_norm": 0.44329434633255005,
      "learning_rate": 4.195800166644781e-08,
      "loss": 1.3467,
      "step": 2105
    },
    {
      "epoch": 9.858396723229959,
      "grad_norm": 0.44371113181114197,
      "learning_rate": 3.866891840982323e-08,
      "loss": 1.3412,
      "step": 2106
    },
    {
      "epoch": 9.863077823288473,
      "grad_norm": 0.43377214670181274,
      "learning_rate": 3.551401565844992e-08,
      "loss": 1.314,
      "step": 2107
    },
    {
      "epoch": 9.867758923346987,
      "grad_norm": 0.4374937415122986,
      "learning_rate": 3.249330188541122e-08,
      "loss": 1.3343,
      "step": 2108
    },
    {
      "epoch": 9.8724400234055,
      "grad_norm": 0.44057637453079224,
      "learning_rate": 2.960678520340099e-08,
      "loss": 1.3129,
      "step": 2109
    },
    {
      "epoch": 9.877121123464015,
      "grad_norm": 0.4521021246910095,
      "learning_rate": 2.685447336469582e-08,
      "loss": 1.3492,
      "step": 2110
    },
    {
      "epoch": 9.881802223522527,
      "grad_norm": 0.44130635261535645,
      "learning_rate": 2.423637376114396e-08,
      "loss": 1.3382,
      "step": 2111
    },
    {
      "epoch": 9.886483323581041,
      "grad_norm": 0.4505654573440552,
      "learning_rate": 2.1752493424148645e-08,
      "loss": 1.3508,
      "step": 2112
    },
    {
      "epoch": 9.891164423639555,
      "grad_norm": 0.452155202627182,
      "learning_rate": 1.94028390246348e-08,
      "loss": 1.3373,
      "step": 2113
    },
    {
      "epoch": 9.89584552369807,
      "grad_norm": 0.4516679048538208,
      "learning_rate": 1.718741687303238e-08,
      "loss": 1.3376,
      "step": 2114
    },
    {
      "epoch": 9.900526623756583,
      "grad_norm": 0.43587708473205566,
      "learning_rate": 1.5106232919276375e-08,
      "loss": 1.3504,
      "step": 2115
    },
    {
      "epoch": 9.905207723815096,
      "grad_norm": 0.456166535615921,
      "learning_rate": 1.31592927527735e-08,
      "loss": 1.3378,
      "step": 2116
    },
    {
      "epoch": 9.90988882387361,
      "grad_norm": 0.45621541142463684,
      "learning_rate": 1.1346601602396644e-08,
      "loss": 1.3327,
      "step": 2117
    },
    {
      "epoch": 9.914569923932124,
      "grad_norm": 0.4448513090610504,
      "learning_rate": 9.668164336473773e-09,
      "loss": 1.3376,
      "step": 2118
    },
    {
      "epoch": 9.919251023990638,
      "grad_norm": 0.4518662691116333,
      "learning_rate": 8.123985462749062e-09,
      "loss": 1.34,
      "step": 2119
    },
    {
      "epoch": 9.923932124049152,
      "grad_norm": 0.44186970591545105,
      "learning_rate": 6.714069128416212e-09,
      "loss": 1.3283,
      "step": 2120
    },
    {
      "epoch": 9.928613224107666,
      "grad_norm": 0.43974027037620544,
      "learning_rate": 5.438419120062932e-09,
      "loss": 1.3303,
      "step": 2121
    },
    {
      "epoch": 9.933294324166178,
      "grad_norm": 0.4513680636882782,
      "learning_rate": 4.297038863687597e-09,
      "loss": 1.3344,
      "step": 2122
    },
    {
      "epoch": 9.937975424224692,
      "grad_norm": 0.4399128556251526,
      "learning_rate": 3.2899314246825906e-09,
      "loss": 1.3169,
      "step": 2123
    },
    {
      "epoch": 9.942656524283207,
      "grad_norm": 0.4567125737667084,
      "learning_rate": 2.417099507817655e-09,
      "loss": 1.3343,
      "step": 2124
    },
    {
      "epoch": 9.94733762434172,
      "grad_norm": 0.4285227954387665,
      "learning_rate": 1.6785454572509908e-09,
      "loss": 1.3151,
      "step": 2125
    },
    {
      "epoch": 9.952018724400235,
      "grad_norm": 0.4506865441799164,
      "learning_rate": 1.0742712565070533e-09,
      "loss": 1.31,
      "step": 2126
    },
    {
      "epoch": 9.956699824458747,
      "grad_norm": 0.44049522280693054,
      "learning_rate": 6.04278528482105e-10,
      "loss": 1.3176,
      "step": 2127
    },
    {
      "epoch": 9.961380924517261,
      "grad_norm": 0.44814878702163696,
      "learning_rate": 2.68568535416458e-10,
      "loss": 1.3382,
      "step": 2128
    },
    {
      "epoch": 9.966062024575775,
      "grad_norm": 0.4420654773712158,
      "learning_rate": 6.714217893333264e-11,
      "loss": 1.326,
      "step": 2129
    },
    {
      "epoch": 9.97074312463429,
      "grad_norm": 0.4389389753341675,
      "learning_rate": 0.0,
      "loss": 1.3419,
      "step": 2130
    },
    {
      "epoch": 9.97074312463429,
      "step": 2130,
      "total_flos": 8.067940197159444e+17,
      "train_loss": 1.4460852847972385,
      "train_runtime": 7405.4046,
      "train_samples_per_second": 415.158,
      "train_steps_per_second": 0.288
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2130,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 128,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.067940197159444e+17,
  "train_batch_size": 180,
  "trial_name": null,
  "trial_params": null
}