{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 471,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0021231422505307855,
      "grad_norm": 6.233692311689662,
      "learning_rate": 0.0,
      "loss": 1.3677,
      "step": 1
    },
    {
      "epoch": 0.004246284501061571,
      "grad_norm": 6.0360948097932825,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 1.4092,
      "step": 2
    },
    {
      "epoch": 0.006369426751592357,
      "grad_norm": 5.936790765452034,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.4005,
      "step": 3
    },
    {
      "epoch": 0.008492569002123142,
      "grad_norm": 6.107653491301953,
      "learning_rate": 1.25e-06,
      "loss": 1.4024,
      "step": 4
    },
    {
      "epoch": 0.010615711252653927,
      "grad_norm": 5.6416859423179195,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.3824,
      "step": 5
    },
    {
      "epoch": 0.012738853503184714,
      "grad_norm": 5.386232950715365,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.4027,
      "step": 6
    },
    {
      "epoch": 0.014861995753715499,
      "grad_norm": 5.260462158475368,
      "learning_rate": 2.5e-06,
      "loss": 1.2939,
      "step": 7
    },
    {
      "epoch": 0.016985138004246284,
      "grad_norm": 4.361483321471081,
      "learning_rate": 2.916666666666667e-06,
      "loss": 1.3433,
      "step": 8
    },
    {
      "epoch": 0.01910828025477707,
      "grad_norm": 4.258300193735322,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.3352,
      "step": 9
    },
    {
      "epoch": 0.021231422505307854,
      "grad_norm": 4.06506696391315,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.3413,
      "step": 10
    },
    {
      "epoch": 0.02335456475583864,
      "grad_norm": 2.9631037995819964,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.2789,
      "step": 11
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 2.967044465905521,
      "learning_rate": 4.583333333333333e-06,
      "loss": 1.2588,
      "step": 12
    },
    {
      "epoch": 0.027600849256900213,
      "grad_norm": 2.9160761923797542,
      "learning_rate": 5e-06,
      "loss": 1.3127,
      "step": 13
    },
    {
      "epoch": 0.029723991507430998,
      "grad_norm": 2.3291890524964733,
      "learning_rate": 5.416666666666667e-06,
      "loss": 1.2187,
      "step": 14
    },
    {
      "epoch": 0.03184713375796178,
      "grad_norm": 2.7317141874955646,
      "learning_rate": 5.833333333333334e-06,
      "loss": 1.2349,
      "step": 15
    },
    {
      "epoch": 0.03397027600849257,
      "grad_norm": 2.6034871595090623,
      "learning_rate": 6.25e-06,
      "loss": 1.1956,
      "step": 16
    },
    {
      "epoch": 0.036093418259023353,
      "grad_norm": 5.248341696418897,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.1591,
      "step": 17
    },
    {
      "epoch": 0.03821656050955414,
      "grad_norm": 2.7756655727491086,
      "learning_rate": 7.083333333333335e-06,
      "loss": 1.1315,
      "step": 18
    },
    {
      "epoch": 0.040339702760084924,
      "grad_norm": 1.8984519497623262,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.1172,
      "step": 19
    },
    {
      "epoch": 0.04246284501061571,
      "grad_norm": 2.0903589430903784,
      "learning_rate": 7.916666666666667e-06,
      "loss": 1.0286,
      "step": 20
    },
    {
      "epoch": 0.044585987261146494,
      "grad_norm": 2.3441882371051936,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.1418,
      "step": 21
    },
    {
      "epoch": 0.04670912951167728,
      "grad_norm": 1.8672005663239521,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.1111,
      "step": 22
    },
    {
      "epoch": 0.04883227176220807,
      "grad_norm": 2.4191480386142583,
      "learning_rate": 9.166666666666666e-06,
      "loss": 1.0186,
      "step": 23
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 1.7753789195215919,
      "learning_rate": 9.583333333333335e-06,
      "loss": 1.0289,
      "step": 24
    },
    {
      "epoch": 0.05307855626326964,
      "grad_norm": 1.2273993769357439,
      "learning_rate": 1e-05,
      "loss": 1.0566,
      "step": 25
    },
    {
      "epoch": 0.055201698513800426,
      "grad_norm": 1.0344250013340117,
      "learning_rate": 9.999876512522269e-06,
      "loss": 1.0465,
      "step": 26
    },
    {
      "epoch": 0.05732484076433121,
      "grad_norm": 1.1494263728238339,
      "learning_rate": 9.999506056188736e-06,
      "loss": 1.087,
      "step": 27
    },
    {
      "epoch": 0.059447983014861996,
      "grad_norm": 1.1301602633871957,
      "learning_rate": 9.99888864929809e-06,
      "loss": 1.0851,
      "step": 28
    },
    {
      "epoch": 0.06157112526539278,
      "grad_norm": 1.1626122948756041,
      "learning_rate": 9.99802432234714e-06,
      "loss": 1.0547,
      "step": 29
    },
    {
      "epoch": 0.06369426751592357,
      "grad_norm": 1.0163867469393915,
      "learning_rate": 9.996913118029306e-06,
      "loss": 1.0567,
      "step": 30
    },
    {
      "epoch": 0.06581740976645435,
      "grad_norm": 1.1007621470115974,
      "learning_rate": 9.995555091232516e-06,
      "loss": 1.0389,
      "step": 31
    },
    {
      "epoch": 0.06794055201698514,
      "grad_norm": 1.063026640935385,
      "learning_rate": 9.99395030903649e-06,
      "loss": 1.0391,
      "step": 32
    },
    {
      "epoch": 0.07006369426751592,
      "grad_norm": 0.8808959073050446,
      "learning_rate": 9.992098850709434e-06,
      "loss": 0.9964,
      "step": 33
    },
    {
      "epoch": 0.07218683651804671,
      "grad_norm": 0.892414136911897,
      "learning_rate": 9.990000807704114e-06,
      "loss": 0.9936,
      "step": 34
    },
    {
      "epoch": 0.07430997876857749,
      "grad_norm": 1.0929906845130524,
      "learning_rate": 9.987656283653344e-06,
      "loss": 0.9981,
      "step": 35
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 1.470611000535341,
      "learning_rate": 9.985065394364869e-06,
      "loss": 1.0313,
      "step": 36
    },
    {
      "epoch": 0.07855626326963906,
      "grad_norm": 0.9257456902237308,
      "learning_rate": 9.982228267815644e-06,
      "loss": 0.9702,
      "step": 37
    },
    {
      "epoch": 0.08067940552016985,
      "grad_norm": 0.9953711624405185,
      "learning_rate": 9.979145044145506e-06,
      "loss": 0.9635,
      "step": 38
    },
    {
      "epoch": 0.08280254777070063,
      "grad_norm": 1.0617273624322012,
      "learning_rate": 9.975815875650265e-06,
      "loss": 0.9483,
      "step": 39
    },
    {
      "epoch": 0.08492569002123142,
      "grad_norm": 1.0473143836003516,
      "learning_rate": 9.972240926774167e-06,
      "loss": 0.9986,
      "step": 40
    },
    {
      "epoch": 0.0870488322717622,
      "grad_norm": 0.8457919738818875,
      "learning_rate": 9.968420374101782e-06,
      "loss": 0.9587,
      "step": 41
    },
    {
      "epoch": 0.08917197452229299,
      "grad_norm": 1.3767525767731605,
      "learning_rate": 9.964354406349272e-06,
      "loss": 0.9625,
      "step": 42
    },
    {
      "epoch": 0.09129511677282377,
      "grad_norm": 0.8914617384358068,
      "learning_rate": 9.960043224355081e-06,
      "loss": 0.9615,
      "step": 43
    },
    {
      "epoch": 0.09341825902335456,
      "grad_norm": 0.7787343722669178,
      "learning_rate": 9.955487041070003e-06,
      "loss": 1.0288,
      "step": 44
    },
    {
      "epoch": 0.09554140127388536,
      "grad_norm": 0.8359830251689131,
      "learning_rate": 9.95068608154667e-06,
      "loss": 1.0315,
      "step": 45
    },
    {
      "epoch": 0.09766454352441614,
      "grad_norm": 0.923320510217848,
      "learning_rate": 9.945640582928438e-06,
      "loss": 1.0166,
      "step": 46
    },
    {
      "epoch": 0.09978768577494693,
      "grad_norm": 0.8371546373170314,
      "learning_rate": 9.940350794437663e-06,
      "loss": 0.9798,
      "step": 47
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 0.9837185052909639,
      "learning_rate": 9.934816977363404e-06,
      "loss": 0.9823,
      "step": 48
    },
    {
      "epoch": 0.1040339702760085,
      "grad_norm": 0.7240963389932448,
      "learning_rate": 9.929039405048502e-06,
      "loss": 0.9676,
      "step": 49
    },
    {
      "epoch": 0.10615711252653928,
      "grad_norm": 0.8361769763648854,
      "learning_rate": 9.923018362876093e-06,
      "loss": 1.0051,
      "step": 50
    },
    {
      "epoch": 0.10828025477707007,
      "grad_norm": 0.7854708795752825,
      "learning_rate": 9.916754148255501e-06,
      "loss": 0.9925,
      "step": 51
    },
    {
      "epoch": 0.11040339702760085,
      "grad_norm": 0.7406975656834304,
      "learning_rate": 9.91024707060755e-06,
      "loss": 1.0003,
      "step": 52
    },
    {
      "epoch": 0.11252653927813164,
      "grad_norm": 0.7421357632855965,
      "learning_rate": 9.903497451349286e-06,
      "loss": 0.9285,
      "step": 53
    },
    {
      "epoch": 0.11464968152866242,
      "grad_norm": 0.7102816852549364,
      "learning_rate": 9.896505623878088e-06,
      "loss": 0.9876,
      "step": 54
    },
    {
      "epoch": 0.11677282377919321,
      "grad_norm": 0.7777264555968619,
      "learning_rate": 9.889271933555214e-06,
      "loss": 1.0259,
      "step": 55
    },
    {
      "epoch": 0.11889596602972399,
      "grad_norm": 0.7349378770101916,
      "learning_rate": 9.881796737688732e-06,
      "loss": 1.0533,
      "step": 56
    },
    {
      "epoch": 0.12101910828025478,
      "grad_norm": 0.6987463564433962,
      "learning_rate": 9.874080405515874e-06,
      "loss": 0.9906,
      "step": 57
    },
    {
      "epoch": 0.12314225053078556,
      "grad_norm": 0.8065087328611837,
      "learning_rate": 9.866123318184803e-06,
      "loss": 0.992,
      "step": 58
    },
    {
      "epoch": 0.12526539278131635,
      "grad_norm": 0.8411662078974967,
      "learning_rate": 9.857925868735774e-06,
      "loss": 0.9868,
      "step": 59
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 0.7703519212712204,
      "learning_rate": 9.84948846208173e-06,
      "loss": 0.9797,
      "step": 60
    },
    {
      "epoch": 0.12951167728237792,
      "grad_norm": 0.8160023784995357,
      "learning_rate": 9.840811514988294e-06,
      "loss": 0.9056,
      "step": 61
    },
    {
      "epoch": 0.1316348195329087,
      "grad_norm": 0.7403495030999158,
      "learning_rate": 9.831895456053197e-06,
      "loss": 0.968,
      "step": 62
    },
    {
      "epoch": 0.1337579617834395,
      "grad_norm": 0.8368650004711656,
      "learning_rate": 9.822740725685087e-06,
      "loss": 0.991,
      "step": 63
    },
    {
      "epoch": 0.13588110403397027,
      "grad_norm": 0.7631095210629854,
      "learning_rate": 9.81334777608179e-06,
      "loss": 1.0121,
      "step": 64
    },
    {
      "epoch": 0.13800424628450106,
      "grad_norm": 0.7855099822305834,
      "learning_rate": 9.803717071207965e-06,
      "loss": 1.0218,
      "step": 65
    },
    {
      "epoch": 0.14012738853503184,
      "grad_norm": 0.7211799308460066,
      "learning_rate": 9.793849086772198e-06,
      "loss": 0.9605,
      "step": 66
    },
    {
      "epoch": 0.14225053078556263,
      "grad_norm": 0.7158533314697694,
      "learning_rate": 9.783744310203492e-06,
      "loss": 0.9648,
      "step": 67
    },
    {
      "epoch": 0.14437367303609341,
      "grad_norm": 0.8116378282708514,
      "learning_rate": 9.77340324062719e-06,
      "loss": 0.9627,
      "step": 68
    },
    {
      "epoch": 0.1464968152866242,
      "grad_norm": 0.6729942687368299,
      "learning_rate": 9.76282638884034e-06,
      "loss": 0.9398,
      "step": 69
    },
    {
      "epoch": 0.14861995753715498,
      "grad_norm": 0.7925561163372088,
      "learning_rate": 9.752014277286433e-06,
      "loss": 1.0496,
      "step": 70
    },
    {
      "epoch": 0.15074309978768577,
      "grad_norm": 0.8168812732930426,
      "learning_rate": 9.740967440029628e-06,
      "loss": 0.9914,
      "step": 71
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 0.6727362552724726,
      "learning_rate": 9.729686422728353e-06,
      "loss": 1.0114,
      "step": 72
    },
    {
      "epoch": 0.15498938428874734,
      "grad_norm": 0.7119941049177072,
      "learning_rate": 9.718171782608355e-06,
      "loss": 0.959,
      "step": 73
    },
    {
      "epoch": 0.15711252653927812,
      "grad_norm": 0.8057011161059664,
      "learning_rate": 9.706424088435183e-06,
      "loss": 0.9365,
      "step": 74
    },
    {
      "epoch": 0.1592356687898089,
      "grad_norm": 0.6871742220004184,
      "learning_rate": 9.694443920486083e-06,
      "loss": 0.9802,
      "step": 75
    },
    {
      "epoch": 0.1613588110403397,
      "grad_norm": 0.7191162102363715,
      "learning_rate": 9.682231870521347e-06,
      "loss": 0.8989,
      "step": 76
    },
    {
      "epoch": 0.16348195329087048,
      "grad_norm": 1.8088460965549538,
      "learning_rate": 9.669788541755072e-06,
      "loss": 0.9524,
      "step": 77
    },
    {
      "epoch": 0.16560509554140126,
      "grad_norm": 0.7467517862627305,
      "learning_rate": 9.657114548825372e-06,
      "loss": 0.969,
      "step": 78
    },
    {
      "epoch": 0.16772823779193205,
      "grad_norm": 0.6507183031203463,
      "learning_rate": 9.644210517764014e-06,
      "loss": 0.9671,
      "step": 79
    },
    {
      "epoch": 0.16985138004246284,
      "grad_norm": 0.7623621079641432,
      "learning_rate": 9.631077085965501e-06,
      "loss": 1.0642,
      "step": 80
    },
    {
      "epoch": 0.17197452229299362,
      "grad_norm": 0.7822356119042599,
      "learning_rate": 9.617714902155576e-06,
      "loss": 0.9911,
      "step": 81
    },
    {
      "epoch": 0.1740976645435244,
      "grad_norm": 0.7827100099968081,
      "learning_rate": 9.60412462635919e-06,
      "loss": 1.0467,
      "step": 82
    },
    {
      "epoch": 0.1762208067940552,
      "grad_norm": 0.7379573454971106,
      "learning_rate": 9.590306929867896e-06,
      "loss": 0.9685,
      "step": 83
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 0.659254589941974,
      "learning_rate": 9.576262495206689e-06,
      "loss": 0.9991,
      "step": 84
    },
    {
      "epoch": 0.18046709129511676,
      "grad_norm": 0.6623198264855394,
      "learning_rate": 9.561992016100293e-06,
      "loss": 0.9784,
      "step": 85
    },
    {
      "epoch": 0.18259023354564755,
      "grad_norm": 0.6847143494642806,
      "learning_rate": 9.547496197438896e-06,
      "loss": 1.0074,
      "step": 86
    },
    {
      "epoch": 0.18471337579617833,
      "grad_norm": 0.7114741953342526,
      "learning_rate": 9.532775755243334e-06,
      "loss": 1.0078,
      "step": 87
    },
    {
      "epoch": 0.18683651804670912,
      "grad_norm": 0.7322798670773077,
      "learning_rate": 9.517831416629717e-06,
      "loss": 0.9434,
      "step": 88
    },
    {
      "epoch": 0.18895966029723993,
      "grad_norm": 0.7355300151781767,
      "learning_rate": 9.502663919773516e-06,
      "loss": 0.9226,
      "step": 89
    },
    {
      "epoch": 0.1910828025477707,
      "grad_norm": 0.6728390445271868,
      "learning_rate": 9.487274013873104e-06,
      "loss": 0.9917,
      "step": 90
    },
    {
      "epoch": 0.1932059447983015,
      "grad_norm": 0.7065250007964868,
      "learning_rate": 9.471662459112747e-06,
      "loss": 0.9152,
      "step": 91
    },
    {
      "epoch": 0.19532908704883228,
      "grad_norm": 0.7619468736243152,
      "learning_rate": 9.455830026625053e-06,
      "loss": 0.9537,
      "step": 92
    },
    {
      "epoch": 0.19745222929936307,
      "grad_norm": 0.7903432346339875,
      "learning_rate": 9.439777498452883e-06,
      "loss": 0.9674,
      "step": 93
    },
    {
      "epoch": 0.19957537154989385,
      "grad_norm": 0.7446836330388329,
      "learning_rate": 9.423505667510724e-06,
      "loss": 0.9492,
      "step": 94
    },
    {
      "epoch": 0.20169851380042464,
      "grad_norm": 0.6684138178927302,
      "learning_rate": 9.40701533754552e-06,
      "loss": 1.0012,
      "step": 95
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 0.7232014896126353,
      "learning_rate": 9.390307323096972e-06,
      "loss": 0.9491,
      "step": 96
    },
    {
      "epoch": 0.2059447983014862,
      "grad_norm": 0.7677419037215932,
      "learning_rate": 9.373382449457305e-06,
      "loss": 0.9484,
      "step": 97
    },
    {
      "epoch": 0.208067940552017,
      "grad_norm": 0.7553292492265486,
      "learning_rate": 9.356241552630503e-06,
      "loss": 0.9329,
      "step": 98
    },
    {
      "epoch": 0.21019108280254778,
      "grad_norm": 0.731417101445953,
      "learning_rate": 9.338885479291012e-06,
      "loss": 1.002,
      "step": 99
    },
    {
      "epoch": 0.21231422505307856,
      "grad_norm": 0.7218309778871758,
      "learning_rate": 9.321315086741916e-06,
      "loss": 1.0027,
      "step": 100
    },
    {
      "epoch": 0.21443736730360935,
      "grad_norm": 0.7064792931836008,
      "learning_rate": 9.303531242872606e-06,
      "loss": 1.0383,
      "step": 101
    },
    {
      "epoch": 0.21656050955414013,
      "grad_norm": 0.7789379913410127,
      "learning_rate": 9.285534826115884e-06,
      "loss": 0.9571,
      "step": 102
    },
    {
      "epoch": 0.21868365180467092,
      "grad_norm": 0.7300691645038767,
      "learning_rate": 9.2673267254046e-06,
      "loss": 0.9049,
      "step": 103
    },
    {
      "epoch": 0.2208067940552017,
      "grad_norm": 0.6966548166340003,
      "learning_rate": 9.248907840127726e-06,
      "loss": 0.9682,
      "step": 104
    },
    {
      "epoch": 0.2229299363057325,
      "grad_norm": 0.6855280784617488,
      "learning_rate": 9.230279080085933e-06,
      "loss": 0.9296,
      "step": 105
    },
    {
      "epoch": 0.22505307855626328,
      "grad_norm": 0.7528839990687416,
      "learning_rate": 9.211441365446661e-06,
      "loss": 0.9324,
      "step": 106
    },
    {
      "epoch": 0.22717622080679406,
      "grad_norm": 0.7238165951427732,
      "learning_rate": 9.192395626698656e-06,
      "loss": 0.9746,
      "step": 107
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 0.7323285515771358,
      "learning_rate": 9.173142804606012e-06,
      "loss": 0.9424,
      "step": 108
    },
    {
      "epoch": 0.23142250530785563,
      "grad_norm": 0.7312649696483972,
      "learning_rate": 9.153683850161706e-06,
      "loss": 0.9893,
      "step": 109
    },
    {
      "epoch": 0.23354564755838642,
      "grad_norm": 0.6801476689460046,
      "learning_rate": 9.13401972454062e-06,
      "loss": 1.0064,
      "step": 110
    },
    {
      "epoch": 0.2356687898089172,
      "grad_norm": 0.6569930110967293,
      "learning_rate": 9.114151399052064e-06,
      "loss": 0.9498,
      "step": 111
    },
    {
      "epoch": 0.23779193205944799,
      "grad_norm": 0.6834181446007654,
      "learning_rate": 9.094079855091797e-06,
      "loss": 0.9686,
      "step": 112
    },
    {
      "epoch": 0.23991507430997877,
      "grad_norm": 0.7531028590215507,
      "learning_rate": 9.073806084093556e-06,
      "loss": 0.941,
      "step": 113
    },
    {
      "epoch": 0.24203821656050956,
      "grad_norm": 0.7710859996596717,
      "learning_rate": 9.053331087480075e-06,
      "loss": 0.947,
      "step": 114
    },
    {
      "epoch": 0.24416135881104034,
      "grad_norm": 0.7181827447543798,
      "learning_rate": 9.032655876613636e-06,
      "loss": 0.9993,
      "step": 115
    },
    {
      "epoch": 0.24628450106157113,
      "grad_norm": 0.8046003364408971,
      "learning_rate": 9.01178147274609e-06,
      "loss": 0.9677,
      "step": 116
    },
    {
      "epoch": 0.2484076433121019,
      "grad_norm": 0.7504304096891474,
      "learning_rate": 8.990708906968431e-06,
      "loss": 0.9713,
      "step": 117
    },
    {
      "epoch": 0.2505307855626327,
      "grad_norm": 0.7331303659304669,
      "learning_rate": 8.969439220159861e-06,
      "loss": 0.9716,
      "step": 118
    },
    {
      "epoch": 0.2526539278131635,
      "grad_norm": 0.784369916505515,
      "learning_rate": 8.947973462936366e-06,
      "loss": 1.0014,
      "step": 119
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 0.7610760319401311,
      "learning_rate": 8.926312695598837e-06,
      "loss": 1.068,
      "step": 120
    },
    {
      "epoch": 0.25690021231422505,
      "grad_norm": 0.7919075855161128,
      "learning_rate": 8.904457988080682e-06,
      "loss": 0.9523,
      "step": 121
    },
    {
      "epoch": 0.25902335456475584,
      "grad_norm": 0.7293682697898348,
      "learning_rate": 8.882410419894983e-06,
      "loss": 0.9536,
      "step": 122
    },
    {
      "epoch": 0.2611464968152866,
      "grad_norm": 0.6770519934059125,
      "learning_rate": 8.860171080081174e-06,
      "loss": 1.0067,
      "step": 123
    },
    {
      "epoch": 0.2632696390658174,
      "grad_norm": 0.7668379033306872,
      "learning_rate": 8.837741067151251e-06,
      "loss": 0.973,
      "step": 124
    },
    {
      "epoch": 0.2653927813163482,
      "grad_norm": 0.7139714751876862,
      "learning_rate": 8.8151214890355e-06,
      "loss": 0.9314,
      "step": 125
    },
    {
      "epoch": 0.267515923566879,
      "grad_norm": 0.6814693095633286,
      "learning_rate": 8.792313463027777e-06,
      "loss": 1.0245,
      "step": 126
    },
    {
      "epoch": 0.26963906581740976,
      "grad_norm": 0.6924465193860749,
      "learning_rate": 8.76931811573033e-06,
      "loss": 0.9371,
      "step": 127
    },
    {
      "epoch": 0.27176220806794055,
      "grad_norm": 0.642073845716499,
      "learning_rate": 8.74613658299813e-06,
      "loss": 0.8808,
      "step": 128
    },
    {
      "epoch": 0.27388535031847133,
      "grad_norm": 0.7174917070331621,
      "learning_rate": 8.72277000988278e-06,
      "loss": 0.9637,
      "step": 129
    },
    {
      "epoch": 0.2760084925690021,
      "grad_norm": 0.6982926048873913,
      "learning_rate": 8.699219550575954e-06,
      "loss": 0.9757,
      "step": 130
    },
    {
      "epoch": 0.2781316348195329,
      "grad_norm": 0.6567002214778425,
      "learning_rate": 8.675486368352376e-06,
      "loss": 0.9503,
      "step": 131
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 0.6794958936952806,
      "learning_rate": 8.651571635512372e-06,
      "loss": 0.9736,
      "step": 132
    },
    {
      "epoch": 0.2823779193205945,
      "grad_norm": 0.6646146029012326,
      "learning_rate": 8.627476533323957e-06,
      "loss": 0.9239,
      "step": 133
    },
    {
      "epoch": 0.28450106157112526,
      "grad_norm": 0.6632510333242262,
      "learning_rate": 8.603202251964492e-06,
      "loss": 0.9977,
      "step": 134
    },
    {
      "epoch": 0.28662420382165604,
      "grad_norm": 0.7248035608452511,
      "learning_rate": 8.578749990461884e-06,
      "loss": 1.0044,
      "step": 135
    },
    {
      "epoch": 0.28874734607218683,
      "grad_norm": 0.7145181226027357,
      "learning_rate": 8.554120956635375e-06,
      "loss": 0.9909,
      "step": 136
    },
    {
      "epoch": 0.2908704883227176,
      "grad_norm": 0.6810283606365963,
      "learning_rate": 8.52931636703587e-06,
      "loss": 0.9278,
      "step": 137
    },
    {
      "epoch": 0.2929936305732484,
      "grad_norm": 0.6390994316344762,
      "learning_rate": 8.504337446885854e-06,
      "loss": 0.9401,
      "step": 138
    },
    {
      "epoch": 0.2951167728237792,
      "grad_norm": 0.68090912832389,
      "learning_rate": 8.47918543001886e-06,
      "loss": 0.9274,
      "step": 139
    },
    {
      "epoch": 0.29723991507430997,
      "grad_norm": 0.7071789703121685,
      "learning_rate": 8.453861558818542e-06,
      "loss": 0.9887,
      "step": 140
    },
    {
      "epoch": 0.29936305732484075,
      "grad_norm": 0.7352751295567939,
      "learning_rate": 8.428367084157292e-06,
      "loss": 0.9288,
      "step": 141
    },
    {
      "epoch": 0.30148619957537154,
      "grad_norm": 0.6595005202531524,
      "learning_rate": 8.402703265334455e-06,
      "loss": 1.0007,
      "step": 142
    },
    {
      "epoch": 0.3036093418259023,
      "grad_norm": 0.6735904870127594,
      "learning_rate": 8.376871370014139e-06,
      "loss": 1.0216,
      "step": 143
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 0.6975929380086214,
      "learning_rate": 8.350872674162578e-06,
      "loss": 0.8988,
      "step": 144
    },
    {
      "epoch": 0.3078556263269639,
      "grad_norm": 0.7224054080414788,
      "learning_rate": 8.324708461985124e-06,
      "loss": 0.9669,
      "step": 145
    },
    {
      "epoch": 0.3099787685774947,
      "grad_norm": 0.7281952438832896,
      "learning_rate": 8.298380025862805e-06,
      "loss": 0.9126,
      "step": 146
    },
    {
      "epoch": 0.31210191082802546,
      "grad_norm": 0.7548928188615618,
      "learning_rate": 8.271888666288488e-06,
      "loss": 0.9399,
      "step": 147
    },
    {
      "epoch": 0.31422505307855625,
      "grad_norm": 0.6597939741370626,
      "learning_rate": 8.245235691802644e-06,
      "loss": 1.0276,
      "step": 148
    },
    {
      "epoch": 0.31634819532908703,
      "grad_norm": 0.7311599924151362,
      "learning_rate": 8.218422418928709e-06,
      "loss": 1.0213,
      "step": 149
    },
    {
      "epoch": 0.3184713375796178,
      "grad_norm": 0.6841346093353339,
      "learning_rate": 8.191450172108058e-06,
      "loss": 0.9467,
      "step": 150
    },
    {
      "epoch": 0.3205944798301486,
      "grad_norm": 0.7131209569882775,
      "learning_rate": 8.164320283634585e-06,
      "loss": 0.9844,
      "step": 151
    },
    {
      "epoch": 0.3227176220806794,
      "grad_norm": 0.6987651677449753,
      "learning_rate": 8.137034093588885e-06,
      "loss": 0.9849,
      "step": 152
    },
    {
      "epoch": 0.3248407643312102,
      "grad_norm": 0.705165665042376,
      "learning_rate": 8.109592949772076e-06,
      "loss": 1.0129,
      "step": 153
    },
    {
      "epoch": 0.32696390658174096,
      "grad_norm": 0.7169171021772854,
      "learning_rate": 8.081998207639212e-06,
      "loss": 1.0058,
      "step": 154
    },
    {
      "epoch": 0.32908704883227174,
      "grad_norm": 0.6104059708044309,
      "learning_rate": 8.054251230232333e-06,
      "loss": 0.9988,
      "step": 155
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 0.6934560008115752,
      "learning_rate": 8.026353388113142e-06,
      "loss": 0.979,
      "step": 156
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.6725043581752848,
      "learning_rate": 7.998306059295302e-06,
      "loss": 1.0304,
      "step": 157
    },
    {
      "epoch": 0.3354564755838641,
      "grad_norm": 0.6223984704621439,
      "learning_rate": 7.97011062917637e-06,
      "loss": 0.9363,
      "step": 158
    },
    {
      "epoch": 0.3375796178343949,
      "grad_norm": 0.706091085660486,
      "learning_rate": 7.941768490469368e-06,
      "loss": 1.0168,
      "step": 159
    },
    {
      "epoch": 0.33970276008492567,
      "grad_norm": 0.7327606621828687,
      "learning_rate": 7.913281043133978e-06,
      "loss": 0.9492,
      "step": 160
    },
    {
      "epoch": 0.34182590233545646,
      "grad_norm": 0.6772432140445678,
      "learning_rate": 7.884649694307413e-06,
      "loss": 0.9167,
      "step": 161
    },
    {
      "epoch": 0.34394904458598724,
      "grad_norm": 0.6801663140277004,
      "learning_rate": 7.855875858234894e-06,
      "loss": 0.9572,
      "step": 162
    },
    {
      "epoch": 0.346072186836518,
      "grad_norm": 0.6863907703405828,
      "learning_rate": 7.826960956199796e-06,
      "loss": 0.9643,
      "step": 163
    },
    {
      "epoch": 0.3481953290870488,
      "grad_norm": 0.6932142414615767,
      "learning_rate": 7.797906416453445e-06,
      "loss": 1.0056,
      "step": 164
    },
    {
      "epoch": 0.3503184713375796,
      "grad_norm": 0.6590692919655603,
      "learning_rate": 7.768713674144578e-06,
      "loss": 1.0103,
      "step": 165
    },
    {
      "epoch": 0.3524416135881104,
      "grad_norm": 0.7043688709406302,
      "learning_rate": 7.739384171248436e-06,
      "loss": 0.9201,
      "step": 166
    },
    {
      "epoch": 0.35456475583864117,
      "grad_norm": 0.7342438164454753,
      "learning_rate": 7.709919356495555e-06,
      "loss": 0.9547,
      "step": 167
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 0.6051801915416819,
      "learning_rate": 7.6803206853002e-06,
      "loss": 1.0027,
      "step": 168
    },
    {
      "epoch": 0.35881104033970274,
      "grad_norm": 0.7036658747712503,
      "learning_rate": 7.650589619688468e-06,
      "loss": 0.9752,
      "step": 169
    },
    {
      "epoch": 0.3609341825902335,
      "grad_norm": 0.6579507052321748,
      "learning_rate": 7.620727628226081e-06,
      "loss": 0.9079,
      "step": 170
    },
    {
      "epoch": 0.3630573248407643,
      "grad_norm": 0.6196419289472935,
      "learning_rate": 7.590736185945843e-06,
      "loss": 0.9451,
      "step": 171
    },
    {
      "epoch": 0.3651804670912951,
      "grad_norm": 0.7001997909038171,
      "learning_rate": 7.560616774274775e-06,
      "loss": 0.9347,
      "step": 172
    },
    {
      "epoch": 0.3673036093418259,
      "grad_norm": 0.6351437976372771,
      "learning_rate": 7.5303708809609514e-06,
      "loss": 0.9299,
      "step": 173
    },
    {
      "epoch": 0.36942675159235666,
      "grad_norm": 0.664760060599698,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9716,
      "step": 174
    },
    {
      "epoch": 0.37154989384288745,
      "grad_norm": 0.6524778235549052,
      "learning_rate": 7.469505631561318e-06,
      "loss": 0.9444,
      "step": 175
    },
    {
      "epoch": 0.37367303609341823,
      "grad_norm": 0.7096570464027714,
      "learning_rate": 7.4388892819139625e-06,
      "loss": 0.981,
      "step": 176
    },
    {
      "epoch": 0.37579617834394907,
      "grad_norm": 0.701946788192956,
      "learning_rate": 7.408152463352249e-06,
      "loss": 0.9394,
      "step": 177
    },
    {
      "epoch": 0.37791932059447986,
      "grad_norm": 0.7062701533777472,
      "learning_rate": 7.3772966941210585e-06,
      "loss": 0.9795,
      "step": 178
    },
    {
      "epoch": 0.38004246284501064,
      "grad_norm": 0.7371591165235177,
      "learning_rate": 7.346323498340839e-06,
      "loss": 0.9207,
      "step": 179
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 0.732545654535989,
      "learning_rate": 7.3152344059323165e-06,
      "loss": 1.0082,
      "step": 180
    },
    {
      "epoch": 0.3842887473460722,
      "grad_norm": 0.7128716261827475,
      "learning_rate": 7.284030952540937e-06,
      "loss": 0.9764,
      "step": 181
    },
    {
      "epoch": 0.386411889596603,
      "grad_norm": 0.6497530527994643,
      "learning_rate": 7.252714679461001e-06,
      "loss": 0.9936,
      "step": 182
    },
    {
      "epoch": 0.3885350318471338,
      "grad_norm": 0.7225290801130503,
      "learning_rate": 7.221287133559537e-06,
      "loss": 0.9304,
      "step": 183
    },
    {
      "epoch": 0.39065817409766457,
      "grad_norm": 0.6765347357945422,
      "learning_rate": 7.189749867199899e-06,
      "loss": 0.9094,
      "step": 184
    },
    {
      "epoch": 0.39278131634819535,
      "grad_norm": 0.7174562419985759,
      "learning_rate": 7.1581044381650735e-06,
      "loss": 0.9801,
      "step": 185
    },
    {
      "epoch": 0.39490445859872614,
      "grad_norm": 0.7472567362872526,
      "learning_rate": 7.126352409580749e-06,
      "loss": 1.0142,
      "step": 186
    },
    {
      "epoch": 0.3970276008492569,
      "grad_norm": 0.6407928497506329,
      "learning_rate": 7.094495349838093e-06,
      "loss": 1.0047,
      "step": 187
    },
    {
      "epoch": 0.3991507430997877,
      "grad_norm": 0.7053895517849673,
      "learning_rate": 7.062534832516288e-06,
      "loss": 1.0126,
      "step": 188
    },
    {
      "epoch": 0.4012738853503185,
      "grad_norm": 0.7061208624812129,
      "learning_rate": 7.0304724363048025e-06,
      "loss": 0.9254,
      "step": 189
    },
    {
      "epoch": 0.4033970276008493,
      "grad_norm": 0.6892392457356276,
      "learning_rate": 6.998309744925411e-06,
      "loss": 0.8733,
      "step": 190
    },
    {
      "epoch": 0.40552016985138006,
      "grad_norm": 0.6824843754838162,
      "learning_rate": 6.9660483470539704e-06,
      "loss": 0.9369,
      "step": 191
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 0.7207059177507389,
      "learning_rate": 6.933689836241939e-06,
      "loss": 0.9403,
      "step": 192
    },
    {
      "epoch": 0.40976645435244163,
      "grad_norm": 0.651073565486602,
      "learning_rate": 6.901235810837668e-06,
      "loss": 0.9733,
      "step": 193
    },
    {
      "epoch": 0.4118895966029724,
      "grad_norm": 0.6694293166716937,
      "learning_rate": 6.868687873907458e-06,
      "loss": 0.9576,
      "step": 194
    },
    {
      "epoch": 0.4140127388535032,
      "grad_norm": 0.6526111024271177,
      "learning_rate": 6.836047633156361e-06,
      "loss": 0.9173,
      "step": 195
    },
    {
      "epoch": 0.416135881104034,
      "grad_norm": 0.670504287795627,
      "learning_rate": 6.8033167008487784e-06,
      "loss": 0.9082,
      "step": 196
    },
    {
      "epoch": 0.4182590233545648,
      "grad_norm": 0.6649654248466491,
      "learning_rate": 6.77049669372882e-06,
      "loss": 0.9684,
      "step": 197
    },
    {
      "epoch": 0.42038216560509556,
      "grad_norm": 0.6971039334316886,
      "learning_rate": 6.737589232940445e-06,
      "loss": 0.998,
      "step": 198
    },
    {
      "epoch": 0.42250530785562634,
      "grad_norm": 0.6625856942226671,
      "learning_rate": 6.704595943947385e-06,
      "loss": 0.9324,
      "step": 199
    },
    {
      "epoch": 0.42462845010615713,
      "grad_norm": 0.6467812509098547,
      "learning_rate": 6.671518456452859e-06,
      "loss": 0.9503,
      "step": 200
    },
    {
      "epoch": 0.4267515923566879,
      "grad_norm": 0.6751623960134163,
      "learning_rate": 6.638358404319064e-06,
      "loss": 0.969,
      "step": 201
    },
    {
      "epoch": 0.4288747346072187,
      "grad_norm": 0.6538986247088467,
      "learning_rate": 6.605117425486483e-06,
      "loss": 0.9637,
      "step": 202
    },
    {
      "epoch": 0.4309978768577495,
      "grad_norm": 0.6283492223913785,
      "learning_rate": 6.571797161892965e-06,
      "loss": 0.919,
      "step": 203
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 0.6754135565342441,
      "learning_rate": 6.538399259392637e-06,
      "loss": 0.9515,
      "step": 204
    },
    {
      "epoch": 0.43524416135881105,
      "grad_norm": 0.6200332586740341,
      "learning_rate": 6.504925367674595e-06,
      "loss": 1.0323,
      "step": 205
    },
    {
      "epoch": 0.43736730360934184,
      "grad_norm": 0.6424305197480263,
      "learning_rate": 6.471377140181419e-06,
      "loss": 0.9402,
      "step": 206
    },
    {
      "epoch": 0.4394904458598726,
      "grad_norm": 0.709344876257473,
      "learning_rate": 6.437756234027512e-06,
      "loss": 0.9472,
      "step": 207
    },
    {
      "epoch": 0.4416135881104034,
      "grad_norm": 0.6954520680490753,
      "learning_rate": 6.40406430991723e-06,
      "loss": 0.9094,
      "step": 208
    },
    {
      "epoch": 0.4437367303609342,
      "grad_norm": 0.6630556744494306,
      "learning_rate": 6.370303032062869e-06,
      "loss": 0.9094,
      "step": 209
    },
    {
      "epoch": 0.445859872611465,
      "grad_norm": 0.6422331097269288,
      "learning_rate": 6.336474068102444e-06,
      "loss": 0.9728,
      "step": 210
    },
    {
      "epoch": 0.44798301486199577,
      "grad_norm": 0.7033435384287802,
      "learning_rate": 6.302579089017328e-06,
      "loss": 0.9465,
      "step": 211
    },
    {
      "epoch": 0.45010615711252655,
      "grad_norm": 0.7227177031033113,
      "learning_rate": 6.268619769049713e-06,
      "loss": 0.9227,
      "step": 212
    },
    {
      "epoch": 0.45222929936305734,
      "grad_norm": 0.7229370830386173,
      "learning_rate": 6.234597785619906e-06,
      "loss": 0.9222,
      "step": 213
    },
    {
      "epoch": 0.4543524416135881,
      "grad_norm": 0.6798315368100751,
      "learning_rate": 6.200514819243476e-06,
      "loss": 0.9065,
      "step": 214
    },
    {
      "epoch": 0.4564755838641189,
      "grad_norm": 0.679227834446727,
      "learning_rate": 6.166372553448241e-06,
      "loss": 0.9163,
      "step": 215
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 0.7284520215353752,
      "learning_rate": 6.132172674691119e-06,
      "loss": 0.9645,
      "step": 216
    },
    {
      "epoch": 0.4607218683651805,
      "grad_norm": 0.6536710419909815,
      "learning_rate": 6.097916872274815e-06,
      "loss": 1.0051,
      "step": 217
    },
    {
      "epoch": 0.46284501061571126,
      "grad_norm": 0.6349521879178258,
      "learning_rate": 6.063606838264384e-06,
      "loss": 0.9397,
      "step": 218
    },
    {
      "epoch": 0.46496815286624205,
      "grad_norm": 0.64300216518759,
      "learning_rate": 6.029244267403652e-06,
      "loss": 0.921,
      "step": 219
    },
    {
      "epoch": 0.46709129511677283,
      "grad_norm": 0.6577888839915914,
      "learning_rate": 5.9948308570315e-06,
      "loss": 0.9415,
      "step": 220
    },
    {
      "epoch": 0.4692144373673036,
      "grad_norm": 0.6441869969003124,
      "learning_rate": 5.960368306998023e-06,
      "loss": 1.0097,
      "step": 221
    },
    {
      "epoch": 0.4713375796178344,
      "grad_norm": 0.6164144128466511,
      "learning_rate": 5.92585831958058e-06,
      "loss": 0.884,
      "step": 222
    },
    {
      "epoch": 0.4734607218683652,
      "grad_norm": 0.6862129278542842,
      "learning_rate": 5.891302599399686e-06,
      "loss": 0.9178,
      "step": 223
    },
    {
      "epoch": 0.47558386411889597,
      "grad_norm": 0.648139964743113,
      "learning_rate": 5.856702853334833e-06,
      "loss": 0.9932,
      "step": 224
    },
    {
      "epoch": 0.47770700636942676,
      "grad_norm": 0.6854865636704949,
      "learning_rate": 5.8220607904401725e-06,
      "loss": 0.9235,
      "step": 225
    },
    {
      "epoch": 0.47983014861995754,
      "grad_norm": 0.6711248584344208,
      "learning_rate": 5.78737812186009e-06,
      "loss": 0.9212,
      "step": 226
    },
    {
      "epoch": 0.4819532908704883,
      "grad_norm": 0.6302334515166729,
      "learning_rate": 5.752656560744692e-06,
      "loss": 0.9562,
      "step": 227
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": 0.6752405762911042,
      "learning_rate": 5.717897822165179e-06,
      "loss": 0.9556,
      "step": 228
    },
    {
      "epoch": 0.4861995753715499,
      "grad_norm": 0.7007828217751222,
      "learning_rate": 5.6831036230291345e-06,
      "loss": 0.976,
      "step": 229
    },
    {
      "epoch": 0.4883227176220807,
      "grad_norm": 0.7096600330939049,
      "learning_rate": 5.648275681995716e-06,
      "loss": 0.9477,
      "step": 230
    },
    {
      "epoch": 0.49044585987261147,
      "grad_norm": 0.7034615904822891,
      "learning_rate": 5.613415719390759e-06,
      "loss": 0.906,
      "step": 231
    },
    {
      "epoch": 0.49256900212314225,
      "grad_norm": 0.735843820490175,
      "learning_rate": 5.578525457121807e-06,
      "loss": 1.0386,
      "step": 232
    },
    {
      "epoch": 0.49469214437367304,
      "grad_norm": 0.72350491411306,
      "learning_rate": 5.543606618593053e-06,
      "loss": 0.9102,
      "step": 233
    },
    {
      "epoch": 0.4968152866242038,
      "grad_norm": 0.6865757038798469,
      "learning_rate": 5.508660928620216e-06,
      "loss": 0.8942,
      "step": 234
    },
    {
      "epoch": 0.4989384288747346,
      "grad_norm": 0.6961697220230768,
      "learning_rate": 5.473690113345343e-06,
      "loss": 0.9593,
      "step": 235
    },
    {
      "epoch": 0.5010615711252654,
      "grad_norm": 0.683780839923492,
      "learning_rate": 5.438695900151537e-06,
      "loss": 0.8619,
      "step": 236
    },
    {
      "epoch": 0.5031847133757962,
      "grad_norm": 0.8892028839705228,
      "learning_rate": 5.403680017577653e-06,
      "loss": 0.9516,
      "step": 237
    },
    {
      "epoch": 0.505307855626327,
      "grad_norm": 0.6485125727265503,
      "learning_rate": 5.368644195232896e-06,
      "loss": 0.9792,
      "step": 238
    },
    {
      "epoch": 0.5074309978768577,
      "grad_norm": 0.6745197820597727,
      "learning_rate": 5.3335901637113985e-06,
      "loss": 0.9247,
      "step": 239
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 0.6594590567939217,
      "learning_rate": 5.298519654506736e-06,
      "loss": 0.999,
      "step": 240
    },
    {
      "epoch": 0.5116772823779193,
      "grad_norm": 0.7018679426711577,
      "learning_rate": 5.2634343999263985e-06,
      "loss": 0.9351,
      "step": 241
    },
    {
      "epoch": 0.5138004246284501,
      "grad_norm": 0.6909133067899488,
      "learning_rate": 5.228336133006223e-06,
      "loss": 0.8968,
      "step": 242
    },
    {
      "epoch": 0.5159235668789809,
      "grad_norm": 0.6709376382865763,
      "learning_rate": 5.193226587424793e-06,
      "loss": 0.9038,
      "step": 243
    },
    {
      "epoch": 0.5180467091295117,
      "grad_norm": 0.7111428217652355,
      "learning_rate": 5.158107497417795e-06,
      "loss": 0.983,
      "step": 244
    },
    {
      "epoch": 0.5201698513800425,
      "grad_norm": 0.6351568108537181,
      "learning_rate": 5.122980597692372e-06,
      "loss": 0.9938,
      "step": 245
    },
    {
      "epoch": 0.5222929936305732,
      "grad_norm": 0.6817936890017572,
      "learning_rate": 5.087847623341421e-06,
      "loss": 0.9761,
      "step": 246
    },
    {
      "epoch": 0.524416135881104,
      "grad_norm": 0.6900540530544701,
      "learning_rate": 5.052710309757899e-06,
      "loss": 0.9552,
      "step": 247
    },
    {
      "epoch": 0.5265392781316348,
      "grad_norm": 0.6719177167300503,
      "learning_rate": 5.0175703925490936e-06,
      "loss": 0.9249,
      "step": 248
    },
    {
      "epoch": 0.5286624203821656,
      "grad_norm": 0.6998669576753951,
      "learning_rate": 4.982429607450907e-06,
      "loss": 0.9643,
      "step": 249
    },
    {
      "epoch": 0.5307855626326964,
      "grad_norm": 0.6424511913080388,
      "learning_rate": 4.947289690242103e-06,
      "loss": 0.9772,
      "step": 250
    },
    {
      "epoch": 0.5329087048832272,
      "grad_norm": 0.7159365178289718,
      "learning_rate": 4.91215237665858e-06,
      "loss": 0.9305,
      "step": 251
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": 0.7475463786009823,
      "learning_rate": 4.877019402307629e-06,
      "loss": 0.936,
      "step": 252
    },
    {
      "epoch": 0.5371549893842887,
      "grad_norm": 0.7136691758831674,
      "learning_rate": 4.841892502582206e-06,
      "loss": 0.9603,
      "step": 253
    },
    {
      "epoch": 0.5392781316348195,
      "grad_norm": 0.6203396388829158,
      "learning_rate": 4.806773412575211e-06,
      "loss": 1.0014,
      "step": 254
    },
    {
      "epoch": 0.5414012738853503,
      "grad_norm": 0.658243565597372,
      "learning_rate": 4.7716638669937784e-06,
      "loss": 0.923,
      "step": 255
    },
    {
      "epoch": 0.5435244161358811,
      "grad_norm": 0.6728067153766357,
      "learning_rate": 4.736565600073602e-06,
      "loss": 0.9161,
      "step": 256
    },
    {
      "epoch": 0.5456475583864119,
      "grad_norm": 0.7214951443210661,
      "learning_rate": 4.701480345493266e-06,
      "loss": 0.9259,
      "step": 257
    },
    {
      "epoch": 0.5477707006369427,
      "grad_norm": 0.6619249080654683,
      "learning_rate": 4.666409836288603e-06,
      "loss": 0.9654,
      "step": 258
    },
    {
      "epoch": 0.5498938428874734,
      "grad_norm": 0.6714573866349525,
      "learning_rate": 4.631355804767106e-06,
      "loss": 0.9374,
      "step": 259
    },
    {
      "epoch": 0.5520169851380042,
      "grad_norm": 0.6592147190515532,
      "learning_rate": 4.596319982422348e-06,
      "loss": 0.9083,
      "step": 260
    },
    {
      "epoch": 0.554140127388535,
      "grad_norm": 0.6942501349295557,
      "learning_rate": 4.561304099848464e-06,
      "loss": 0.9506,
      "step": 261
    },
    {
      "epoch": 0.5562632696390658,
      "grad_norm": 0.6988620238717118,
      "learning_rate": 4.526309886654659e-06,
      "loss": 0.9977,
      "step": 262
    },
    {
      "epoch": 0.5583864118895966,
      "grad_norm": 0.6921429983455063,
      "learning_rate": 4.491339071379783e-06,
      "loss": 0.8757,
      "step": 263
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": 0.6771212073875934,
      "learning_rate": 4.4563933814069475e-06,
      "loss": 0.932,
      "step": 264
    },
    {
      "epoch": 0.5626326963906582,
      "grad_norm": 0.6420633068956142,
      "learning_rate": 4.4214745428781946e-06,
      "loss": 0.9849,
      "step": 265
    },
    {
      "epoch": 0.564755838641189,
      "grad_norm": 0.7089424922243647,
      "learning_rate": 4.386584280609242e-06,
      "loss": 0.9816,
      "step": 266
    },
    {
      "epoch": 0.5668789808917197,
      "grad_norm": 0.6662051267433905,
      "learning_rate": 4.351724318004286e-06,
      "loss": 0.9407,
      "step": 267
    },
    {
      "epoch": 0.5690021231422505,
      "grad_norm": 0.6301712452097127,
      "learning_rate": 4.316896376970866e-06,
      "loss": 1.0318,
      "step": 268
    },
    {
      "epoch": 0.5711252653927813,
      "grad_norm": 0.6314890663243107,
      "learning_rate": 4.282102177834822e-06,
      "loss": 0.9311,
      "step": 269
    },
    {
      "epoch": 0.5732484076433121,
      "grad_norm": 0.6419289651169521,
      "learning_rate": 4.2473434392553115e-06,
      "loss": 0.9442,
      "step": 270
    },
    {
      "epoch": 0.5753715498938429,
      "grad_norm": 0.6177620934633588,
      "learning_rate": 4.212621878139912e-06,
      "loss": 0.9336,
      "step": 271
    },
    {
      "epoch": 0.5774946921443737,
      "grad_norm": 0.6596837531543023,
      "learning_rate": 4.177939209559828e-06,
      "loss": 0.9878,
      "step": 272
    },
    {
      "epoch": 0.5796178343949044,
      "grad_norm": 0.6453367239167705,
      "learning_rate": 4.143297146665167e-06,
      "loss": 0.9079,
      "step": 273
    },
    {
      "epoch": 0.5817409766454352,
      "grad_norm": 0.6522513374450648,
      "learning_rate": 4.108697400600316e-06,
      "loss": 0.9854,
      "step": 274
    },
    {
      "epoch": 0.583864118895966,
      "grad_norm": 0.6732938314131642,
      "learning_rate": 4.074141680419422e-06,
      "loss": 0.9926,
      "step": 275
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": 0.7092550415636378,
      "learning_rate": 4.039631693001976e-06,
      "loss": 0.967,
      "step": 276
    },
    {
      "epoch": 0.5881104033970276,
      "grad_norm": 0.6524896445355799,
      "learning_rate": 4.005169142968503e-06,
      "loss": 0.9368,
      "step": 277
    },
    {
      "epoch": 0.5902335456475584,
      "grad_norm": 0.644883665492233,
      "learning_rate": 3.970755732596349e-06,
      "loss": 0.9341,
      "step": 278
    },
    {
      "epoch": 0.5923566878980892,
      "grad_norm": 0.6736291474419323,
      "learning_rate": 3.936393161735616e-06,
      "loss": 0.9744,
      "step": 279
    },
    {
      "epoch": 0.5944798301486199,
      "grad_norm": 0.680611909030509,
      "learning_rate": 3.902083127725186e-06,
      "loss": 0.9889,
      "step": 280
    },
    {
      "epoch": 0.5966029723991507,
      "grad_norm": 0.6811819729687757,
      "learning_rate": 3.867827325308882e-06,
      "loss": 0.9783,
      "step": 281
    },
    {
      "epoch": 0.5987261146496815,
      "grad_norm": 0.6365141720169178,
      "learning_rate": 3.83362744655176e-06,
      "loss": 0.9082,
      "step": 282
    },
    {
      "epoch": 0.6008492569002123,
      "grad_norm": 0.6483027093614826,
      "learning_rate": 3.799485180756526e-06,
      "loss": 0.9702,
      "step": 283
    },
    {
      "epoch": 0.6029723991507431,
      "grad_norm": 0.6558888618128743,
      "learning_rate": 3.765402214380095e-06,
      "loss": 0.9624,
      "step": 284
    },
    {
      "epoch": 0.6050955414012739,
      "grad_norm": 0.7423269688448214,
      "learning_rate": 3.731380230950288e-06,
      "loss": 0.9803,
      "step": 285
    },
    {
      "epoch": 0.6072186836518046,
      "grad_norm": 0.665408242644521,
      "learning_rate": 3.6974209109826724e-06,
      "loss": 0.92,
      "step": 286
    },
    {
      "epoch": 0.6093418259023354,
      "grad_norm": 0.6082795254064975,
      "learning_rate": 3.663525931897559e-06,
      "loss": 0.916,
      "step": 287
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": 0.6554479295173997,
      "learning_rate": 3.6296969679371325e-06,
      "loss": 0.9609,
      "step": 288
    },
    {
      "epoch": 0.613588110403397,
      "grad_norm": 0.6737780075210532,
      "learning_rate": 3.595935690082769e-06,
      "loss": 0.8926,
      "step": 289
    },
    {
      "epoch": 0.6157112526539278,
      "grad_norm": 0.6081469745516901,
      "learning_rate": 3.56224376597249e-06,
      "loss": 0.9514,
      "step": 290
    },
    {
      "epoch": 0.6178343949044586,
      "grad_norm": 0.6581642589602584,
      "learning_rate": 3.528622859818582e-06,
      "loss": 0.9565,
      "step": 291
    },
    {
      "epoch": 0.6199575371549894,
      "grad_norm": 0.5923692426637494,
      "learning_rate": 3.495074632325407e-06,
      "loss": 0.9684,
      "step": 292
    },
    {
      "epoch": 0.6220806794055201,
      "grad_norm": 0.6409823550988945,
      "learning_rate": 3.461600740607366e-06,
      "loss": 0.9835,
      "step": 293
    },
    {
      "epoch": 0.6242038216560509,
      "grad_norm": 0.6477471386159979,
      "learning_rate": 3.4282028381070366e-06,
      "loss": 0.8795,
      "step": 294
    },
    {
      "epoch": 0.6263269639065817,
      "grad_norm": 0.6456450294718449,
      "learning_rate": 3.3948825745135196e-06,
      "loss": 0.939,
      "step": 295
    },
    {
      "epoch": 0.6284501061571125,
      "grad_norm": 0.7315407037525199,
      "learning_rate": 3.361641595680937e-06,
      "loss": 0.9801,
      "step": 296
    },
    {
      "epoch": 0.6305732484076433,
      "grad_norm": 0.6452110148561251,
      "learning_rate": 3.3284815435471423e-06,
      "loss": 0.9615,
      "step": 297
    },
    {
      "epoch": 0.6326963906581741,
      "grad_norm": 0.6253835621122691,
      "learning_rate": 3.295404056052616e-06,
      "loss": 0.927,
      "step": 298
    },
    {
      "epoch": 0.6348195329087049,
      "grad_norm": 0.6363789821603231,
      "learning_rate": 3.2624107670595567e-06,
      "loss": 0.9705,
      "step": 299
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": 0.6677508958368374,
      "learning_rate": 3.2295033062711823e-06,
      "loss": 0.9087,
      "step": 300
    },
    {
      "epoch": 0.6390658174097664,
      "grad_norm": 0.7037995039835674,
      "learning_rate": 3.1966832991512232e-06,
      "loss": 0.9857,
      "step": 301
    },
    {
      "epoch": 0.6411889596602972,
      "grad_norm": 0.629654116611617,
      "learning_rate": 3.16395236684364e-06,
      "loss": 0.9507,
      "step": 302
    },
    {
      "epoch": 0.643312101910828,
      "grad_norm": 0.6785540515594622,
      "learning_rate": 3.131312126092544e-06,
      "loss": 0.9398,
      "step": 303
    },
    {
      "epoch": 0.6454352441613588,
      "grad_norm": 0.6407691844743428,
      "learning_rate": 3.098764189162332e-06,
      "loss": 0.9039,
      "step": 304
    },
    {
      "epoch": 0.6475583864118896,
      "grad_norm": 0.6258454088461756,
      "learning_rate": 3.0663101637580626e-06,
      "loss": 0.9242,
      "step": 305
    },
    {
      "epoch": 0.6496815286624203,
      "grad_norm": 0.745199786640981,
      "learning_rate": 3.03395165294603e-06,
      "loss": 0.9536,
      "step": 306
    },
    {
      "epoch": 0.6518046709129511,
      "grad_norm": 0.6324017551337827,
      "learning_rate": 3.0016902550745896e-06,
      "loss": 0.9527,
      "step": 307
    },
    {
      "epoch": 0.6539278131634819,
      "grad_norm": 0.6241906193592306,
      "learning_rate": 2.9695275636951983e-06,
      "loss": 0.9568,
      "step": 308
    },
    {
      "epoch": 0.6560509554140127,
      "grad_norm": 0.6229924568908893,
      "learning_rate": 2.9374651674837128e-06,
      "loss": 1.0229,
      "step": 309
    },
    {
      "epoch": 0.6581740976645435,
      "grad_norm": 0.6204870651207804,
      "learning_rate": 2.9055046501619088e-06,
      "loss": 0.9789,
      "step": 310
    },
    {
      "epoch": 0.6602972399150743,
      "grad_norm": 0.7361247533813169,
      "learning_rate": 2.8736475904192516e-06,
      "loss": 0.9306,
      "step": 311
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": 0.6469024579110451,
      "learning_rate": 2.841895561834927e-06,
      "loss": 0.8995,
      "step": 312
    },
    {
      "epoch": 0.6645435244161358,
      "grad_norm": 0.6798475871777178,
      "learning_rate": 2.810250132800103e-06,
      "loss": 0.9457,
      "step": 313
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.5974407421450565,
      "learning_rate": 2.778712866440464e-06,
      "loss": 0.964,
      "step": 314
    },
    {
      "epoch": 0.6687898089171974,
      "grad_norm": 0.584100105339634,
      "learning_rate": 2.7472853205389997e-06,
      "loss": 0.9053,
      "step": 315
    },
    {
      "epoch": 0.6709129511677282,
      "grad_norm": 0.6342401081454511,
      "learning_rate": 2.715969047459066e-06,
      "loss": 0.9826,
      "step": 316
    },
    {
      "epoch": 0.673036093418259,
      "grad_norm": 0.5786828455381722,
      "learning_rate": 2.6847655940676843e-06,
      "loss": 0.9468,
      "step": 317
    },
    {
      "epoch": 0.6751592356687898,
      "grad_norm": 0.6091300911407583,
      "learning_rate": 2.6536765016591626e-06,
      "loss": 0.9259,
      "step": 318
    },
    {
      "epoch": 0.6772823779193206,
      "grad_norm": 0.6539407752728442,
      "learning_rate": 2.622703305878941e-06,
      "loss": 0.9335,
      "step": 319
    },
    {
      "epoch": 0.6794055201698513,
      "grad_norm": 0.6733794364508308,
      "learning_rate": 2.5918475366477536e-06,
      "loss": 0.8735,
      "step": 320
    },
    {
      "epoch": 0.6815286624203821,
      "grad_norm": 0.6270883315392012,
      "learning_rate": 2.5611107180860395e-06,
      "loss": 0.979,
      "step": 321
    },
    {
      "epoch": 0.6836518046709129,
      "grad_norm": 0.6498694158555346,
      "learning_rate": 2.530494368438683e-06,
      "loss": 0.8834,
      "step": 322
    },
    {
      "epoch": 0.6857749469214437,
      "grad_norm": 0.6869641678831084,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.9647,
      "step": 323
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": 0.6239790954969553,
      "learning_rate": 2.4696291190390494e-06,
      "loss": 0.9792,
      "step": 324
    },
    {
      "epoch": 0.6900212314225053,
      "grad_norm": 0.6134158027251899,
      "learning_rate": 2.4393832257252253e-06,
      "loss": 0.9696,
      "step": 325
    },
    {
      "epoch": 0.692144373673036,
      "grad_norm": 0.6086160560392221,
      "learning_rate": 2.4092638140541586e-06,
      "loss": 0.9228,
      "step": 326
    },
    {
      "epoch": 0.6942675159235668,
      "grad_norm": 0.6498216593510738,
      "learning_rate": 2.3792723717739197e-06,
      "loss": 0.958,
      "step": 327
    },
    {
      "epoch": 0.6963906581740976,
      "grad_norm": 0.625686536550573,
      "learning_rate": 2.349410380311532e-06,
      "loss": 0.8942,
      "step": 328
    },
    {
      "epoch": 0.6985138004246284,
      "grad_norm": 0.6531306674626144,
      "learning_rate": 2.319679314699801e-06,
      "loss": 0.9158,
      "step": 329
    },
    {
      "epoch": 0.7006369426751592,
      "grad_norm": 0.6484034467819224,
      "learning_rate": 2.290080643504446e-06,
      "loss": 0.976,
      "step": 330
    },
    {
      "epoch": 0.70276008492569,
      "grad_norm": 0.6097583458280551,
      "learning_rate": 2.2606158287515662e-06,
      "loss": 0.927,
      "step": 331
    },
    {
      "epoch": 0.7048832271762208,
      "grad_norm": 0.6910687359695524,
      "learning_rate": 2.2312863258554236e-06,
      "loss": 0.9503,
      "step": 332
    },
    {
      "epoch": 0.7070063694267515,
      "grad_norm": 0.6169296505248961,
      "learning_rate": 2.2020935835465567e-06,
      "loss": 0.9448,
      "step": 333
    },
    {
      "epoch": 0.7091295116772823,
      "grad_norm": 0.6206102236780358,
      "learning_rate": 2.1730390438002056e-06,
      "loss": 0.9367,
      "step": 334
    },
    {
      "epoch": 0.7112526539278131,
      "grad_norm": 0.5882669175278411,
      "learning_rate": 2.1441241417651072e-06,
      "loss": 0.9246,
      "step": 335
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": 0.6593545038938035,
      "learning_rate": 2.1153503056925872e-06,
      "loss": 0.9597,
      "step": 336
    },
    {
      "epoch": 0.7154989384288747,
      "grad_norm": 0.6187836424308805,
      "learning_rate": 2.086718956866024e-06,
      "loss": 0.9823,
      "step": 337
    },
    {
      "epoch": 0.7176220806794055,
      "grad_norm": 0.6622306313226891,
      "learning_rate": 2.0582315095306343e-06,
      "loss": 0.8998,
      "step": 338
    },
    {
      "epoch": 0.7197452229299363,
      "grad_norm": 0.6064995088828125,
      "learning_rate": 2.0298893708236307e-06,
      "loss": 0.9494,
      "step": 339
    },
    {
      "epoch": 0.721868365180467,
      "grad_norm": 0.6315356893553601,
      "learning_rate": 2.0016939407046987e-06,
      "loss": 0.934,
      "step": 340
    },
    {
      "epoch": 0.7239915074309978,
      "grad_norm": 0.6132141895487265,
      "learning_rate": 1.9736466118868573e-06,
      "loss": 0.9662,
      "step": 341
    },
    {
      "epoch": 0.7261146496815286,
      "grad_norm": 0.5913159522920567,
      "learning_rate": 1.945748769767667e-06,
      "loss": 0.9388,
      "step": 342
    },
    {
      "epoch": 0.7282377919320594,
      "grad_norm": 0.5911001882219563,
      "learning_rate": 1.9180017923607884e-06,
      "loss": 0.9959,
      "step": 343
    },
    {
      "epoch": 0.7303609341825902,
      "grad_norm": 0.5964411898925175,
      "learning_rate": 1.8904070502279242e-06,
      "loss": 0.9609,
      "step": 344
    },
    {
      "epoch": 0.732484076433121,
      "grad_norm": 0.5956188683958897,
      "learning_rate": 1.8629659064111138e-06,
      "loss": 0.9042,
      "step": 345
    },
    {
      "epoch": 0.7346072186836518,
      "grad_norm": 0.6429639213975425,
      "learning_rate": 1.8356797163654172e-06,
      "loss": 0.9719,
      "step": 346
    },
    {
      "epoch": 0.7367303609341825,
      "grad_norm": 0.625453067786892,
      "learning_rate": 1.8085498278919421e-06,
      "loss": 1.0156,
      "step": 347
    },
    {
      "epoch": 0.7388535031847133,
      "grad_norm": 0.6474903500210286,
      "learning_rate": 1.7815775810712921e-06,
      "loss": 0.9304,
      "step": 348
    },
    {
      "epoch": 0.7409766454352441,
      "grad_norm": 0.643384296734223,
      "learning_rate": 1.754764308197358e-06,
      "loss": 0.9429,
      "step": 349
    },
    {
      "epoch": 0.7430997876857749,
      "grad_norm": 0.6488250778766241,
      "learning_rate": 1.728111333711514e-06,
      "loss": 0.8978,
      "step": 350
    },
    {
      "epoch": 0.7452229299363057,
      "grad_norm": 0.6609986780324201,
      "learning_rate": 1.7016199741371958e-06,
      "loss": 0.918,
      "step": 351
    },
    {
      "epoch": 0.7473460721868365,
      "grad_norm": 0.6128799955400698,
      "learning_rate": 1.6752915380148772e-06,
      "loss": 0.953,
      "step": 352
    },
    {
      "epoch": 0.7494692144373672,
      "grad_norm": 0.6122929864573905,
      "learning_rate": 1.6491273258374241e-06,
      "loss": 0.9491,
      "step": 353
    },
    {
      "epoch": 0.7515923566878981,
      "grad_norm": 0.581921090261184,
      "learning_rate": 1.6231286299858635e-06,
      "loss": 1.0149,
      "step": 354
    },
    {
      "epoch": 0.7537154989384289,
      "grad_norm": 0.6499572625124347,
      "learning_rate": 1.5972967346655449e-06,
      "loss": 0.9035,
      "step": 355
    },
    {
      "epoch": 0.7558386411889597,
      "grad_norm": 0.6018564846421318,
      "learning_rate": 1.5716329158427097e-06,
      "loss": 0.959,
      "step": 356
    },
    {
      "epoch": 0.7579617834394905,
      "grad_norm": 0.6192065507123931,
      "learning_rate": 1.546138441181459e-06,
      "loss": 0.9242,
      "step": 357
    },
    {
      "epoch": 0.7600849256900213,
      "grad_norm": 0.6004306254969125,
      "learning_rate": 1.5208145699811417e-06,
      "loss": 0.9603,
      "step": 358
    },
    {
      "epoch": 0.7622080679405521,
      "grad_norm": 0.6569862768052468,
      "learning_rate": 1.4956625531141495e-06,
      "loss": 0.9735,
      "step": 359
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 0.6439568810696925,
      "learning_rate": 1.470683632964131e-06,
      "loss": 0.9389,
      "step": 360
    },
    {
      "epoch": 0.7664543524416136,
      "grad_norm": 0.6145728176506001,
      "learning_rate": 1.4458790433646264e-06,
      "loss": 0.8859,
      "step": 361
    },
    {
      "epoch": 0.7685774946921444,
      "grad_norm": 0.6326061250773658,
      "learning_rate": 1.4212500095381176e-06,
      "loss": 0.9676,
      "step": 362
    },
    {
      "epoch": 0.7707006369426752,
      "grad_norm": 0.6123548283214884,
      "learning_rate": 1.3967977480355106e-06,
      "loss": 0.9067,
      "step": 363
    },
    {
      "epoch": 0.772823779193206,
      "grad_norm": 0.6475017566755511,
      "learning_rate": 1.3725234666760428e-06,
      "loss": 0.9761,
      "step": 364
    },
    {
      "epoch": 0.7749469214437368,
      "grad_norm": 0.5905257224026246,
      "learning_rate": 1.3484283644876289e-06,
      "loss": 0.9567,
      "step": 365
    },
    {
      "epoch": 0.7770700636942676,
      "grad_norm": 0.6176210891568339,
      "learning_rate": 1.3245136316476253e-06,
      "loss": 0.8845,
      "step": 366
    },
    {
      "epoch": 0.7791932059447984,
      "grad_norm": 0.6212881415222462,
      "learning_rate": 1.3007804494240478e-06,
      "loss": 0.9558,
      "step": 367
    },
    {
      "epoch": 0.7813163481953291,
      "grad_norm": 0.6268225582403189,
      "learning_rate": 1.2772299901172198e-06,
      "loss": 0.9741,
      "step": 368
    },
    {
      "epoch": 0.7834394904458599,
      "grad_norm": 0.6290633940922796,
      "learning_rate": 1.2538634170018727e-06,
      "loss": 0.905,
      "step": 369
    },
    {
      "epoch": 0.7855626326963907,
      "grad_norm": 0.6106693917571974,
      "learning_rate": 1.2306818842696716e-06,
      "loss": 0.9158,
      "step": 370
    },
    {
      "epoch": 0.7876857749469215,
      "grad_norm": 0.6589407432757999,
      "learning_rate": 1.2076865369722246e-06,
      "loss": 0.9218,
      "step": 371
    },
    {
      "epoch": 0.7898089171974523,
      "grad_norm": 0.6181774095804418,
      "learning_rate": 1.184878510964504e-06,
      "loss": 0.9282,
      "step": 372
    },
    {
      "epoch": 0.7919320594479831,
      "grad_norm": 0.634843246523155,
      "learning_rate": 1.1622589328487505e-06,
      "loss": 0.8704,
      "step": 373
    },
    {
      "epoch": 0.7940552016985138,
      "grad_norm": 0.6588681055604763,
      "learning_rate": 1.1398289199188262e-06,
      "loss": 0.9072,
      "step": 374
    },
    {
      "epoch": 0.7961783439490446,
      "grad_norm": 0.6496256026060344,
      "learning_rate": 1.1175895801050185e-06,
      "loss": 0.9369,
      "step": 375
    },
    {
      "epoch": 0.7983014861995754,
      "grad_norm": 0.6157885136577669,
      "learning_rate": 1.09554201191932e-06,
      "loss": 0.9857,
      "step": 376
    },
    {
      "epoch": 0.8004246284501062,
      "grad_norm": 0.693915421413349,
      "learning_rate": 1.0736873044011632e-06,
      "loss": 0.9195,
      "step": 377
    },
    {
      "epoch": 0.802547770700637,
      "grad_norm": 0.6352811715121436,
      "learning_rate": 1.052026537063634e-06,
      "loss": 0.9494,
      "step": 378
    },
    {
      "epoch": 0.8046709129511678,
      "grad_norm": 0.6314980398127243,
      "learning_rate": 1.03056077984014e-06,
      "loss": 0.9829,
      "step": 379
    },
    {
      "epoch": 0.8067940552016986,
      "grad_norm": 0.6171720030728264,
      "learning_rate": 1.0092910930315698e-06,
      "loss": 0.9196,
      "step": 380
    },
    {
      "epoch": 0.8089171974522293,
      "grad_norm": 0.6111752995592951,
      "learning_rate": 9.882185272539107e-07,
      "loss": 1.0179,
      "step": 381
    },
    {
      "epoch": 0.8110403397027601,
      "grad_norm": 0.6063600459587173,
      "learning_rate": 9.673441233863661e-07,
      "loss": 0.9778,
      "step": 382
    },
    {
      "epoch": 0.8131634819532909,
      "grad_norm": 0.6219984593584619,
      "learning_rate": 9.466689125199247e-07,
      "loss": 0.9517,
      "step": 383
    },
    {
      "epoch": 0.8152866242038217,
      "grad_norm": 0.5884991058391071,
      "learning_rate": 9.261939159064465e-07,
      "loss": 0.9125,
      "step": 384
    },
    {
      "epoch": 0.8174097664543525,
      "grad_norm": 0.6336855501453481,
      "learning_rate": 9.059201449082045e-07,
      "loss": 0.9055,
      "step": 385
    },
    {
      "epoch": 0.8195329087048833,
      "grad_norm": 0.6438487189876951,
      "learning_rate": 8.858486009479384e-07,
      "loss": 0.8816,
      "step": 386
    },
    {
      "epoch": 0.821656050955414,
      "grad_norm": 0.6165977153524274,
      "learning_rate": 8.659802754593805e-07,
      "loss": 0.9708,
      "step": 387
    },
    {
      "epoch": 0.8237791932059448,
      "grad_norm": 0.6192150411773142,
      "learning_rate": 8.463161498382949e-07,
      "loss": 0.9328,
      "step": 388
    },
    {
      "epoch": 0.8259023354564756,
      "grad_norm": 0.6252080211561564,
      "learning_rate": 8.268571953939897e-07,
      "loss": 1.0164,
      "step": 389
    },
    {
      "epoch": 0.8280254777070064,
      "grad_norm": 0.6174239147729979,
      "learning_rate": 8.07604373301345e-07,
      "loss": 0.9278,
      "step": 390
    },
    {
      "epoch": 0.8301486199575372,
      "grad_norm": 0.5934957686287227,
      "learning_rate": 7.885586345533397e-07,
      "loss": 0.998,
      "step": 391
    },
    {
      "epoch": 0.832271762208068,
      "grad_norm": 0.6821446761372294,
      "learning_rate": 7.697209199140676e-07,
      "loss": 0.9244,
      "step": 392
    },
    {
      "epoch": 0.8343949044585988,
      "grad_norm": 0.6293471158429464,
      "learning_rate": 7.510921598722765e-07,
      "loss": 0.93,
      "step": 393
    },
    {
      "epoch": 0.8365180467091295,
      "grad_norm": 0.6756858812445671,
      "learning_rate": 7.326732745954001e-07,
      "loss": 0.9123,
      "step": 394
    },
    {
      "epoch": 0.8386411889596603,
      "grad_norm": 0.6596572053271065,
      "learning_rate": 7.144651738841174e-07,
      "loss": 0.9155,
      "step": 395
    },
    {
      "epoch": 0.8407643312101911,
      "grad_norm": 0.6341753704135638,
      "learning_rate": 6.96468757127396e-07,
      "loss": 0.9352,
      "step": 396
    },
    {
      "epoch": 0.8428874734607219,
      "grad_norm": 0.641019320177676,
      "learning_rate": 6.786849132580841e-07,
      "loss": 0.9704,
      "step": 397
    },
    {
      "epoch": 0.8450106157112527,
      "grad_norm": 0.6502762949418774,
      "learning_rate": 6.611145207089897e-07,
      "loss": 1.0109,
      "step": 398
    },
    {
      "epoch": 0.8471337579617835,
      "grad_norm": 0.5883498984829939,
      "learning_rate": 6.437584473694991e-07,
      "loss": 0.937,
      "step": 399
    },
    {
      "epoch": 0.8492569002123143,
      "grad_norm": 0.6044650989839885,
      "learning_rate": 6.266175505426958e-07,
      "loss": 0.9899,
      "step": 400
    },
    {
      "epoch": 0.851380042462845,
      "grad_norm": 0.7032208468983784,
      "learning_rate": 6.096926769030298e-07,
      "loss": 0.977,
      "step": 401
    },
    {
      "epoch": 0.8535031847133758,
      "grad_norm": 0.6599095174918433,
      "learning_rate": 5.929846624544821e-07,
      "loss": 0.8938,
      "step": 402
    },
    {
      "epoch": 0.8556263269639066,
      "grad_norm": 0.6326742176715536,
      "learning_rate": 5.76494332489278e-07,
      "loss": 0.9807,
      "step": 403
    },
    {
      "epoch": 0.8577494692144374,
      "grad_norm": 0.5864803282744446,
      "learning_rate": 5.602225015471175e-07,
      "loss": 0.9364,
      "step": 404
    },
    {
      "epoch": 0.8598726114649682,
      "grad_norm": 0.6074843091694057,
      "learning_rate": 5.441699733749479e-07,
      "loss": 0.9128,
      "step": 405
    },
    {
      "epoch": 0.861995753715499,
      "grad_norm": 0.6249418223588977,
      "learning_rate": 5.283375408872538e-07,
      "loss": 0.9691,
      "step": 406
    },
    {
      "epoch": 0.8641188959660298,
      "grad_norm": 0.6168009522971204,
      "learning_rate": 5.127259861268974e-07,
      "loss": 0.9434,
      "step": 407
    },
    {
      "epoch": 0.8662420382165605,
      "grad_norm": 0.6190463715943864,
      "learning_rate": 4.973360802264859e-07,
      "loss": 0.9756,
      "step": 408
    },
    {
      "epoch": 0.8683651804670913,
      "grad_norm": 0.622941478104092,
      "learning_rate": 4.82168583370285e-07,
      "loss": 0.9562,
      "step": 409
    },
    {
      "epoch": 0.8704883227176221,
      "grad_norm": 0.6538118577546125,
      "learning_rate": 4.6722424475666715e-07,
      "loss": 0.9785,
      "step": 410
    },
    {
      "epoch": 0.8726114649681529,
      "grad_norm": 0.5955630313016704,
      "learning_rate": 4.5250380256110335e-07,
      "loss": 0.9444,
      "step": 411
    },
    {
      "epoch": 0.8747346072186837,
      "grad_norm": 0.6354711324766207,
      "learning_rate": 4.380079838997087e-07,
      "loss": 0.9588,
      "step": 412
    },
    {
      "epoch": 0.8768577494692145,
      "grad_norm": 0.615805033371163,
      "learning_rate": 4.237375047933118e-07,
      "loss": 0.9514,
      "step": 413
    },
    {
      "epoch": 0.8789808917197452,
      "grad_norm": 0.6114715882888384,
      "learning_rate": 4.0969307013210445e-07,
      "loss": 0.9463,
      "step": 414
    },
    {
      "epoch": 0.881104033970276,
      "grad_norm": 0.6095867459156437,
      "learning_rate": 3.958753736408105e-07,
      "loss": 0.9843,
      "step": 415
    },
    {
      "epoch": 0.8832271762208068,
      "grad_norm": 0.6123640433423114,
      "learning_rate": 3.822850978444254e-07,
      "loss": 0.979,
      "step": 416
    },
    {
      "epoch": 0.8853503184713376,
      "grad_norm": 0.6037382407098268,
      "learning_rate": 3.6892291403449963e-07,
      "loss": 0.9593,
      "step": 417
    },
    {
      "epoch": 0.8874734607218684,
      "grad_norm": 0.6421465549828673,
      "learning_rate": 3.557894822359864e-07,
      "loss": 0.9743,
      "step": 418
    },
    {
      "epoch": 0.8895966029723992,
      "grad_norm": 0.6523016163798167,
      "learning_rate": 3.428854511746293e-07,
      "loss": 0.894,
      "step": 419
    },
    {
      "epoch": 0.89171974522293,
      "grad_norm": 0.6131799852928372,
      "learning_rate": 3.302114582449295e-07,
      "loss": 0.9001,
      "step": 420
    },
    {
      "epoch": 0.8938428874734607,
      "grad_norm": 0.592786282667123,
      "learning_rate": 3.177681294786539e-07,
      "loss": 0.9109,
      "step": 421
    },
    {
      "epoch": 0.8959660297239915,
      "grad_norm": 0.6308395651339358,
      "learning_rate": 3.055560795139173e-07,
      "loss": 0.9416,
      "step": 422
    },
    {
      "epoch": 0.8980891719745223,
      "grad_norm": 0.5733021722650089,
      "learning_rate": 2.9357591156481793e-07,
      "loss": 0.9304,
      "step": 423
    },
    {
      "epoch": 0.9002123142250531,
      "grad_norm": 0.6601790561924576,
      "learning_rate": 2.8182821739164534e-07,
      "loss": 0.9506,
      "step": 424
    },
    {
      "epoch": 0.9023354564755839,
      "grad_norm": 0.6448366971138038,
      "learning_rate": 2.7031357727164865e-07,
      "loss": 0.9662,
      "step": 425
    },
    {
      "epoch": 0.9044585987261147,
      "grad_norm": 0.6353764277416116,
      "learning_rate": 2.5903255997037246e-07,
      "loss": 0.9452,
      "step": 426
    },
    {
      "epoch": 0.9065817409766455,
      "grad_norm": 0.6215966155589953,
      "learning_rate": 2.479857227135685e-07,
      "loss": 0.9373,
      "step": 427
    },
    {
      "epoch": 0.9087048832271762,
      "grad_norm": 0.6071667900895601,
      "learning_rate": 2.3717361115966343e-07,
      "loss": 0.9365,
      "step": 428
    },
    {
      "epoch": 0.910828025477707,
      "grad_norm": 0.5662505726716307,
      "learning_rate": 2.2659675937281078e-07,
      "loss": 0.9449,
      "step": 429
    },
    {
      "epoch": 0.9129511677282378,
      "grad_norm": 0.6103475320116462,
      "learning_rate": 2.1625568979651012e-07,
      "loss": 0.9148,
      "step": 430
    },
    {
      "epoch": 0.9150743099787686,
      "grad_norm": 0.6021851948761284,
      "learning_rate": 2.061509132278028e-07,
      "loss": 0.9571,
      "step": 431
    },
    {
      "epoch": 0.9171974522292994,
      "grad_norm": 0.6160276194809644,
      "learning_rate": 1.9628292879203482e-07,
      "loss": 0.9542,
      "step": 432
    },
    {
      "epoch": 0.9193205944798302,
      "grad_norm": 0.5793322886001591,
      "learning_rate": 1.866522239182117e-07,
      "loss": 0.9498,
      "step": 433
    },
    {
      "epoch": 0.921443736730361,
      "grad_norm": 0.6436803146727256,
      "learning_rate": 1.7725927431491375e-07,
      "loss": 0.9164,
      "step": 434
    },
    {
      "epoch": 0.9235668789808917,
      "grad_norm": 0.5802384555049214,
      "learning_rate": 1.6810454394680431e-07,
      "loss": 0.8862,
      "step": 435
    },
    {
      "epoch": 0.9256900212314225,
      "grad_norm": 0.5991103996578487,
      "learning_rate": 1.5918848501170647e-07,
      "loss": 0.9835,
      "step": 436
    },
    {
      "epoch": 0.9278131634819533,
      "grad_norm": 0.6140911819745668,
      "learning_rate": 1.505115379182731e-07,
      "loss": 0.9125,
      "step": 437
    },
    {
      "epoch": 0.9299363057324841,
      "grad_norm": 0.6569732944090182,
      "learning_rate": 1.420741312642282e-07,
      "loss": 0.8964,
      "step": 438
    },
    {
      "epoch": 0.9320594479830149,
      "grad_norm": 0.6189932685484231,
      "learning_rate": 1.338766818151982e-07,
      "loss": 0.9251,
      "step": 439
    },
    {
      "epoch": 0.9341825902335457,
      "grad_norm": 0.641464843833588,
      "learning_rate": 1.2591959448412628e-07,
      "loss": 0.9644,
      "step": 440
    },
    {
      "epoch": 0.9363057324840764,
      "grad_norm": 0.5982347753497826,
      "learning_rate": 1.1820326231126944e-07,
      "loss": 0.9263,
      "step": 441
    },
    {
      "epoch": 0.9384288747346072,
      "grad_norm": 0.5823643996812994,
      "learning_rate": 1.107280664447874e-07,
      "loss": 0.8708,
      "step": 442
    },
    {
      "epoch": 0.940552016985138,
      "grad_norm": 0.6134197078158699,
      "learning_rate": 1.0349437612191259e-07,
      "loss": 0.928,
      "step": 443
    },
    {
      "epoch": 0.9426751592356688,
      "grad_norm": 0.6237353280448422,
      "learning_rate": 9.650254865071428e-08,
      "loss": 0.9015,
      "step": 444
    },
    {
      "epoch": 0.9447983014861996,
      "grad_norm": 0.5971219836614964,
      "learning_rate": 8.975292939244928e-08,
      "loss": 0.938,
      "step": 445
    },
    {
      "epoch": 0.9469214437367304,
      "grad_norm": 0.6774232363303238,
      "learning_rate": 8.324585174449895e-08,
      "loss": 0.9284,
      "step": 446
    },
    {
      "epoch": 0.9490445859872612,
      "grad_norm": 0.6403835297664868,
      "learning_rate": 7.698163712390683e-08,
      "loss": 0.9299,
      "step": 447
    },
    {
      "epoch": 0.9511677282377919,
      "grad_norm": 0.5942848027014522,
      "learning_rate": 7.096059495149855e-08,
      "loss": 0.9471,
      "step": 448
    },
    {
      "epoch": 0.9532908704883227,
      "grad_norm": 0.5872025609513564,
      "learning_rate": 6.518302263659737e-08,
      "loss": 0.9717,
      "step": 449
    },
    {
      "epoch": 0.9554140127388535,
      "grad_norm": 0.6294814684278404,
      "learning_rate": 5.964920556233767e-08,
      "loss": 0.9498,
      "step": 450
    },
    {
      "epoch": 0.9575371549893843,
      "grad_norm": 0.6148892781592927,
      "learning_rate": 5.435941707156389e-08,
      "loss": 0.955,
      "step": 451
    },
    {
      "epoch": 0.9596602972399151,
      "grad_norm": 0.6509786720651843,
      "learning_rate": 4.931391845333089e-08,
      "loss": 0.9547,
      "step": 452
    },
    {
      "epoch": 0.9617834394904459,
      "grad_norm": 0.6501016558558563,
      "learning_rate": 4.451295892999863e-08,
      "loss": 0.9612,
      "step": 453
    },
    {
      "epoch": 0.9639065817409767,
      "grad_norm": 0.5677280542511052,
      "learning_rate": 3.99567756449204e-08,
      "loss": 0.9536,
      "step": 454
    },
    {
      "epoch": 0.9660297239915074,
      "grad_norm": 0.6270377760292427,
      "learning_rate": 3.5645593650728284e-08,
      "loss": 0.9415,
      "step": 455
    },
    {
      "epoch": 0.9681528662420382,
      "grad_norm": 0.5936095931176947,
      "learning_rate": 3.157962589821872e-08,
      "loss": 0.9123,
      "step": 456
    },
    {
      "epoch": 0.970276008492569,
      "grad_norm": 0.5753275305003565,
      "learning_rate": 2.77590732258326e-08,
      "loss": 0.8992,
      "step": 457
    },
    {
      "epoch": 0.9723991507430998,
      "grad_norm": 0.6161560451020908,
      "learning_rate": 2.4184124349734828e-08,
      "loss": 0.9011,
      "step": 458
    },
    {
      "epoch": 0.9745222929936306,
      "grad_norm": 0.5986199087825571,
      "learning_rate": 2.085495585449404e-08,
      "loss": 0.9366,
      "step": 459
    },
    {
      "epoch": 0.9766454352441614,
      "grad_norm": 0.6144363258091632,
      "learning_rate": 1.7771732184357905e-08,
      "loss": 0.9702,
      "step": 460
    },
    {
      "epoch": 0.9787685774946921,
      "grad_norm": 0.6159493368932808,
      "learning_rate": 1.4934605635132383e-08,
      "loss": 0.9388,
      "step": 461
    },
    {
      "epoch": 0.9808917197452229,
      "grad_norm": 0.6326721912348251,
      "learning_rate": 1.2343716346657209e-08,
      "loss": 0.9672,
      "step": 462
    },
    {
      "epoch": 0.9830148619957537,
      "grad_norm": 0.6286300913967682,
      "learning_rate": 9.999192295886973e-09,
      "loss": 0.8955,
      "step": 463
    },
    {
      "epoch": 0.9851380042462845,
      "grad_norm": 0.6469478397795659,
      "learning_rate": 7.90114929056618e-09,
      "loss": 0.969,
      "step": 464
    },
    {
      "epoch": 0.9872611464968153,
      "grad_norm": 0.6109488636303972,
      "learning_rate": 6.04969096350938e-09,
      "loss": 1.0043,
      "step": 465
    },
    {
      "epoch": 0.9893842887473461,
      "grad_norm": 0.672241938807974,
      "learning_rate": 4.444908767484712e-09,
      "loss": 0.9447,
      "step": 466
    },
    {
      "epoch": 0.9915074309978769,
      "grad_norm": 0.6327761576127415,
      "learning_rate": 3.0868819706947327e-09,
      "loss": 0.9163,
      "step": 467
    },
    {
      "epoch": 0.9936305732484076,
      "grad_norm": 0.5827977793629625,
      "learning_rate": 1.9756776528601085e-09,
      "loss": 0.93,
      "step": 468
    },
    {
      "epoch": 0.9957537154989384,
      "grad_norm": 0.5656174378230285,
      "learning_rate": 1.111350701909486e-09,
      "loss": 0.9092,
      "step": 469
    },
    {
      "epoch": 0.9978768577494692,
      "grad_norm": 0.6248670179846091,
      "learning_rate": 4.939438112638861e-10,
      "loss": 0.9335,
      "step": 470
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.584849627731381,
      "learning_rate": 1.2348747773172075e-10,
      "loss": 0.9356,
      "step": 471
    }
  ],
  "logging_steps": 1,
  "max_steps": 471,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 157,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 70004199849984.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}